Patient-provider relationships
Patient satisfaction
Patient perception of physician’s patient-centeredness
Patient treatment adherence
Provider decision making
Provider’s perspective of patient’s likelihood to adhere to treatment
Other studies have confirmed associations between provider bias (demonstrated via IAT testing) and disparate treatment of their patients ( 63 ). In a systematic literature review, six studies found that higher implicit bias among health care providers was associated with disparities in treatment recommendations, expectations of therapeutic bonds, pain management, and empathy ( 63 ). Seven studies that examined the impact of implicit provider bias on real-world patient–provider interaction found that health care providers with stronger implicit bias demonstrated poorer patient–provider communication and that health care providers with high implicit biases ( a ) provided lower rates of postoperative narcotic prescriptions for Black children than for White children ( 93 ), ( b ) had poorer bonding with Black patients than with White patients ( 55 ), and ( c ) made disparate recommendations for thrombolytic therapy for Black patients and White patients ( 40 ).
A study of 3,756 students at 49 US medical schools demonstrated that high scores of racism as measured by the three variables were significantly correlated with low scores of student intentions to work in underserved areas and to provide care to minority populations ( 74 ).
Implicit bias affects not only patients but also trainees and faculty within health care systems. A 2014 systematic literature review revealed that rates of harassment and discrimination against trainees (24% reported racial discrimination, 33% reported sexual harassment, and 54% reported gender discrimination) have remained unchanged over time ( 31 ). Minority trainees report facing daily bias and microaggressions and having feelings of isolation and substantial stress ( 74 ). Minority medical students reported five-times-higher odds of racial discrimination and isolation than did nonminority peers ( 26 ). Stereotype threat (defined in Table 1 ) is common, particularly among non-White students, interferes with learning, and adds to the cognitive load of minoritized students ( 9 ). Thus, bias in health professions training can affect the performance of racialized minorities. Early and small differences in assessed clinical performance, which may be affected by implicit biases, lead to larger differences in grades and selection for awards [e.g., Alpha Omega Alpha Honor Medical Society (AOA)], ultimately affecting career trajectories of racial minority candidates ( 102 ). For example, significant differences in negative descriptive words on medical students’ evaluations have been found across different racial and gender groups ( 91 ). Membership in AOA, conferred to only 16% of each graduating medical school class, has effectively barred diversity in many specialties and may represent a longstanding form of structural racism ( 7 ).
Literature outside of health care has introduced techniques to manage implicit bias, including stereotype replacement (replacing stereotypical responses to bias with nonstereotypical ones), counter-stereotypic imaging (imagining known counter-stereotypical people), individuation (learning personal attributes of persons present rather than identifying group attributes), perspective taking (taking the perspective of persons present), and increasing opportunities for contact. Several studies have explored the efficacy of these interventions. Strikingly, the only study demonstrating reduction of measured implicit bias was conducted on undergraduate students enrolled in a course using a prejudice-habit-breaking intervention involving instruction of all the aforementioned techniques with effects lasting 8 weeks ( 24 ). Unfortunately, these results may not be generalizable and have not been reproduced. Lai et al. ( 57 ) tested nine interventions and although all immediately reduced implicit preferences, results were sustained for only several hours to days. FitzGerald et al. ( 30 ) conducted in 2019 a systematic review of bias interventions utilizing the IAT or other measures across multiple disciplines. They found that most studies did not provide robust data to support many interventions, although perspective taking was more successful than counter-stereotypic imaging.
Implicit bias has important interactions with structural elements of the health care system. Evidence suggests that implicit bias can reinforce structural dimensions of the health care system that generate disparities. Other evidence suggests that structural dimensions of the health care system and medical education can reinforce implicit bias. These interactions suggest a complex and mutually reinforcing relationship between implicit bias and structural elements of the health care system.
Implicit biases influence the decisions of policy makers in government and health care that result in structural racism ( 70 , 75 , 81 ). Public health responses to the coronavirus disease 2019 (COVID-19) pandemic offer evidence of this dynamic. Despite data demonstrating that non-Hispanic Black populations and Hispanic populations were dying at a younger average age (71.8 years and 67.3 years) than non-Hispanic White patients were (80.9 years), the phase 1b vaccination strategy targeted individuals age 75 and older ( 25 ). Thus, federal public health recommendations ignored or discounted the evidence that an age-based approach would lead to further disparities in COVID-19 infections and mortality, amounting to structural racism against Black and Hispanic populations.
Studies have consistently shown that decision makers burdened with higher cognitive load are more likely to make biased decisions ( 10 ). A more recent study of physicians in the emergency department has confirmed that cognitive stressors such as patient overcrowding and patient load were associated with increased implicit racial bias as measured by a race IAT preshift compared to postshift ( 53 ).
Unfortunately, to date, medical education and educators have not adequately addressed the implicit biases that place marginalized patients at high risk of receiving disparate care and suffering poorer health outcomes. In fact, Phelan et al. ( 84 ) concluded that structural racism is at play in medical education through many medical schools’ formal and hidden curricula ( 52 , 88 ). In contrast to a formal curriculum, which can be measured by the number of hours students receive training related to racial disparities and bias, structured service-learning, minority health activities, cultural awareness programming, and the completion of an IAT, the hidden curriculum is unofficial and often more powerful, consisting of faculty role modeling ( 52 ), institutional priorities around the interracial climate, and experiences of microaggressions.
Most medical students continue to believe that both race and gender (as opposed to sex) are genetic and biological constructs. Even when students are taught otherwise, the practice of race-based medicine reinforces these characterizations. When students are taught about health disparities without the appropriate contextualization of structural racism, historic segregation, the pathologization of gender and sexual orientation, and the medical professions’ complicity in scientific racism, students may assume there is something inherently wrong with racialized minorities rather than with the systems that have harmed them. Students are often taught that race, instead of racism, is an independent risk factor for disease. They learn to associate race with any number of diseases. They are taught to incorporate the race of their patient into the opening line of clinical presentations even though there is no evidence that race is relevant to the establishment of diagnoses. They learn to use race-based algorithms to calculate glomerular filtration rates, pulmonary function testing, hypertension guidelines, and even urinary tract infection diagnoses in pediatric populations ( 2 ). Such messaging only serves to undo any structured teaching on the social construct of race and gender ( 16 ).
As discussed above, there is substantial evidence that implicit bias results in health care disparities through mechanisms including disparate care and trust. But the relationship between implicit bias and outcomes may be bidirectional. Evidence has shown that implicit attitudes are malleable and that such attitudes are learned and strengthened through repeated observation of particular classes of people in valued or devalued circumstances. For example, individuals exposed to less favorable exemplars from a given identity demonstrate increased implicit bias and stereotypes with respect to that entire group ( 20 ). Furthermore, these investigators showed that changing exposure to more favorable exemplars can diminish established implicit bias. This phenomenon has been demonstrated in experiments looking specifically at race- and age-related attitudes ( 21 ). These findings suggest that a practitioner’s implicit bias toward a marginalized group may be augmented or diminished by the clinical outcomes of that group.
The CHANGES study demonstrated that students’ implicit bias against sexual minorities was reduced at 42 medical schools and increased at only 7 schools. Reduced bias was associated with more frequent interaction with LGBT students, faculty, and patients; the perceived quality of that contact; and increased training involving skills in caring for sexual minorities ( 85 ).
The CHANGES study found that changes in student implicit racial attitudes were independently associated with formal curricula related to disparities in health and health care, cultural competence, and minority health; informal curricula (or hidden curricula, defined in Table 1 ), including racial climate and role model behavior; and the amount and favorability of interracial contact during medical school ( 84 ).
Thus, carefully designed structural elements of the learning environment can favorably affect the implicit biases and wellness of students.
A systematic literature review was performed with the goal of assessing the efficacy of extant interventions designed to reduce the explicit and implicit biases of health care providers and of learners across the continuum of health professions education.
We searched three databases (ERIC, PubMed, and MedEdPORTAL) using key terms ( Figure 1 ). The terms “implicit bias,” prejudice,” and “stigma” were often used inter-changeably and the terms “bias” and “biases” yielded more than 100,000 articles, often with little relevance to implicit bias in the health professions. We found, as did FitzGerald et al. ( 30 ) in their systematic review, that indexing in databases for these terms was inconsistent and that titles and abstracts were often imprecise. We conducted repeated searches with and without these terms, comparing the number of search results. We developed a set of terms most frequently encountered in the titles and abstracts of irrelevant articles and defined important terminology ( Table 1 ) to narrow the search. We reviewed the references of landmark articles and used the advanced search function to increase the likelihood that no key articles were missed.
PRISMA flow diagram of the systematic review.
A study had to include health care professionals, assess an intervention (e.g., training, workshop, didactics, contact, program) designed to address explicit or implicit bias held by health care providers, be written in English, and be published between May 2011 and May 2021. We excluded commentaries, theoretical frameworks, editorials, and institutional or societal pledges that address racism, although these were reviewed for context. We did not exclude qualitative studies, studies without comparison groups, or studies outside North America. However, although we did find studies from other countries detailing explicit and implicit biases, we did not find articles with interventions addressing these biases for inclusion in this review. We extracted subjects, intervention format (e.g., lectures, workshops, discussions, panels, interviews), target (e.g., knowledge, skills, attitudes, IAT), and summary of key findings.
We excluded abstracts that did not include original research or bias reduction as an expected outcome; that did not employ a discrete intervention or, like the CHANGES study, retrospectively identified effective interventions; or that studied populations other than health professions students, trainees, or providers. We excluded articles that focused on self-stigma (e.g., from a diagnosis of obesity, HIV, sexually transmitted infection, mental health) and community-based interventions, as they were not focused specifically on the bias of health professionals. Observational studies without discrete interventions were excluded but were reviewed in Section 1 .
Title, abstract, and full-text review were conducted by three authors (M.B.V., A.I.E., and N.A.S.) and coded to consensus.
Twenty-five studies met inclusion criteria ( Table 3 ). None of the studies mentioned in Sections 1 and 2 met inclusion criteria but were reviewed because of their significant contributions to the understanding of the interactions of implicit bias in learning and clinical settings. Most studies (68%) engaged medical students and utilized classroom or web-based interventions. Most studies did not have a control group (72%) and none used actual clinical settings. Three studies focused on interventions for implicit bias of faculty serving on admissions or search committees.
Provider-level implicit bias interventions
Study population | Intervention | Evaluation/outcomes | Limitations | Reference | |
---|---|---|---|---|---|
Interventions without formal measurement of implicit bias/attitudes | Medical students ( = 25) | Study and control groups Study group participated in 5-h dialogues on race and bias | Pre- and postsurveys Paired -tests demonstrated increased knowledge and awareness of racial bias and increased comfort talking about race. | No formal bias measure Self-selected study group of students | |
Faculty who serve on search committees ( = 22) | 2-h reflection-based workshop on unconscious bias | Post-intervention survey evaluated effectiveness and utility of exercise. Most surveyed found workshop helpful in preparing for faculty searches. | Extremely limited evaluation (no pre-/postcomparison) No formal bias measure | ||
Medical students = 615) | 2-day orientation on power, privilege, and bias | Post-intervention survey Surveys demonstrated raised bias awareness. | No formal bias measure No pre-/postcomparison | ||
Medical students ( = 187) | Five 2-h workshops with lectures on bias | Pre- and postsurveys Paired -tests on surveys demonstrated raised awareness of own biases and intent to address bias. | No formal bias measure | ||
Health professions educators = 70) | Introduced new longitudinal case conference curriculum called HER to discuss and address the impact of structural racism and implicit bias on patient care Utilized case-based discussion, evidence-based exercises, and two conceptual frameworks | Tracked conference attendance and postconference surveys Most survey respondents (88% or more) indicated that HER promoted personal reflection on implicit bias, and 7 5 % or more indicated that HER would affect their clinical practice. | No pre-/postcomparison No formal bias measure No control group | ||
Faculty = 66) | 90-min interactive workshop that included a reflective exercise, role-play, brief didactic session, and case-based discussion on use of language in patient charts | Post-intervention survey with four Likert scale questions Participants felt workshop met its objectives (4.8 out of 5.0) and strongly agreed that they would apply skills learned (4.8). | Self-selected study group No measure of bias No control group | ||
Family medicine residents ( = 31) | Training on institutional racism, colonization, and cultural power followed by humanism and instruction on taking health equity time-outs during clinical time | Focus groups conducted 6 months post-intervention Four themes: | No measure of bias No pre-/postcomparison Qualitative analysis only No control group | ||
Medical students ( = 26) | Service-learning plus reflection | Reflection practice questionnaire analysis Students reported recognizing and mitigating bias. | No formal measure of bias used No control group | ||
Medical students ( = 127) | Readings/reflections on weight stigma Standardized patient before and after | Pre-/post-intervention questionnaires Reduced stereotyping, increased empathy, and improved counseling confidence Weak analysis may be biased itself. | No formal bias measurement No control group | ||
Interventions with formal measurement of implicit bias/attitudes | Medical students/elective ( = 218) | Single session in which students completed an IAT followed by discussion | Post IAT survey Implicit bias deniers were significantly more likely to report IAT results with implicit preferences toward self, to believe the IAT is invalid, and to believe that doctors and the health care system provide equal care to all, and were less likely to report having directly observed inequitable care. | Self-selected study group No control group | |
Medical students ( = 180) | Single IAT administration followed by guided reflective discussion and essay writing | Evaluation of reflective essays Students noted raised awareness of bias but were not able to strategize solutions to mitigate bias. | Prompt did not ask for strategies No control group | ||
Medical students ( = 15) | Nine 1.5-h sessions focused on promoting skills to empower students to recognize implicit bias reduction as part of professionalism Three objectives (grounded in implicit bias recognition and transformative learning theory): | Post-intervention focus groups and analysis of semistructured interviews Major themes: | Self-selected small group of students No control group | ||
Medical students ( = 72) | IAT administration followed by small group debrief and discussion on bias | Qualitative analysis of discussion transcripts Students who reach for normative versus personal standards had higher implicit bias post-intervention. | No post IAT measure of bias No control group | ||
Nursing students ( = 75) | Pre/post IAT with debriefing, writing, and teaching of bias management techniques (e.g., internal feedback, humanism) | Postclass survey, conducted 5 weeks after the intervention Learners were extremely likely or likely to ( ) take additional IATs and reflect on the results and ( ) learn more about unconscious bias. | No formal analysis of pre/post IATs, but focus was on acceptance of bias and management No control group | ||
Medical students ( = 78) | Workshops that involved IAT administration, instruction on implicit bias and impact on decision making, and presentation of six strategies to reduce implicit bias | Reduction of implicit bias against Hispanics as measured by an IAT in majority students only No change for minority students was demonstrated. | No control group Nonclinical setting | ||
Medical students, house staff, faculty ( = 468) | Twenty workshops to emphasize skill building and include lectures, guided reflections, and facilitated discussions focused on the following: | Survey response rate was 80%; a paired -test Pre- and postsurveys to evaluate the intervention’s capacity to improve awareness of bias and address it through allyship Demonstrated greatest improvements in understanding of the process of allyship; ability to describe strategies to address, assess, and recognize unconscious bias; and knowledge of managing situations in which prejudice, power, and privilege are involved | Improved confidence in addressing bias but no measure of bias reduction | ||
Faculty on admissions committee ( = 140) | Black-White IAT administered before 2012–2013 medical school admission cycle Study participants received results before start of admission cycle and were surveyed on the impact at the end of cycle in May 2013 | Most survey respondents (67%) thought the IAT might be helpful in reducing bias, 48% were conscious of their individual results when interviewing candidates in the next cycle, and 21 % reported knowledge of their IAT results impacted their admissions decisions in the subsequent cycle. This class is the most diverse to matriculate in the Ohio State University College of Medicine’s history. | Unclear whether other factors affected matriculation of students | ||
Faculty members ( = 281) | Standardized, 20-min educational intervention to educate faculty about implicit biases and strategies for overcoming them | Pre-/postassessments that included the following: The intervention had a small but significant effect on the implicit biases surrounding women and leadership of all participants regardless of age and gender. Faculty experienced significant increases in their perceptions of personal bias (Cohen’s = 0.50 and 0.17; < 0.01 for both questions), perceptions of societal bias (Cohen’s = 0.14, 0.12, and 0.25; < 0.05 for all three questions), and perceptions of bias in academic medicine (Cohen’s = 0.38, 0.57, and 0.58; < 0.001 for all three questions). | Immediate impact only No control group | ||
Medical students ( = 64) | Study participants watched video linking obesity to genetics and environment | Beliefs about Obese Persons, Attitudes toward Obese Persons, and Fat Phobia Scales administered pre- and post-intervention Paired -tests revealed decreased negative stereotypes and beliefs. | No longitudinal results No control group | ||
House staff ( = 69) | Narrative photography to prompt reflection and photovoice of Latino adolescents | Control and intervention groups Measured ethnocultural empathy, health care empathy, patient centeredness, and implicit attitudes using the affect misattribution procedure All measures improved with some note of dose response with more exposure. | Nonclinical setting | ||
Medical students ( = 129) | Workshop to address obesity-related bias using theater reading (intervention group) of play versus lecture (control group) on obesity Students randomly assigned to groups | Obesity-specific IAT, anti-fat attitudes questionnaire pre-/postworkshop Reduced explicit fat bias in theater group with no change in implicit bias or empathy post-intervention or 4 months later | Nonclinical setting | ||
Primary care providers ( = 185) | Study participants randomized to intervention (lecture and contact)/control (lecture and discussion) | Beliefs and Attitudes towards Mental Health Service Users’ Rights Scale Reduced stigmatizing beliefs and attitudes at 1 month in the intervention group but rebound effect at 3 months | No formal measure of bias Nonclinical setting | ||
Medical students ( = 111) | One-time contact-based educational intervention on the stigma of mental illness among medical students and compared this with a multimodal undergraduate psychiatry course | Opening Minds Scale for Health Care Providers to assess changes in stigma Stigma scores for both groups were significandy reduced upon course completion ( < 0.0001) but were not significandy changed following the one-time contact-based educational intervention in the primary analysis. | Nonclinical setting | ||
Medical students ( = 160) | Intergroup contact theory (facilitated contact to reduce bias) plus 50 h of competency-based curriculum on inclusive care of LGBTQ and gender-nonconforming individuals through lectures, standardized patients, discussion, panels, and reflective writing | Had study and control groups Pre and post IATs with debriefings demonstrated reduced implicit preference for straight people. IAT with debriefings were important when used to facilitate curriculum. | Nonclinical setting | ||
Medical students ( = 50) | Three cultural competency training sessions led by LGBTQ2S+ experts and elders from the community Study participants randomized to intervention and control groups Focus group discussions conducted | Pre-/postassessment Lesbian, Gay, and Bisexual Knowledge and Attitudes Scale for Heterosexuals and The Riddle Scale: Attitudes towards Gay, Lesbian, Bisexual, and Trans people survey Measurable and relevant changes in health care students’ perceived knowledge, attitudes, and clinical behavior regarding LGBTQ2S+ populations as a result | Nonclinical setting |
Abbreviations: HER, Health Equity Round; IAT, implicit association test.
The three studies of faculty serving on admissions or search committees reported increased awareness of biases, but none reported bias reduction or long-lasting impact.
Three studies followed subjects 3, 4, and 6 months post-intervention, but only one noted a lasting positive impact ( 96 ).
All studies addressing implicit bias among health care providers raised awareness of implicit bias through didactic instruction, discussions, workshops or other reflection-based techniques (e.g., service-learning, photovoice, contact-based interventions, theater reading; see Table 4 ), or an IAT or similar measure.
Definitions of intervention types used in selected studies
Intervention type | Definition |
---|---|
Allyship training | “An active, consistent, and arduous practice of unlearning and re-evaluating, in which a person of privilege seeks to operate in solidarity with a marginalized group” ( ) “Allyship begins with an awareness of unconscious biases and then moves to actions that address inequities in everyday interactions to create an inclusive culture for example to amplify the voices of those in underrepresented groups and to advocate for equitable practices” ( , p. 6). |
Bias literacy | Promotes a basic understanding of key terms, skills and concepts related to bias as a first step to organizational change ( , p. 64; , p. 22) |
Brave space | “A space where difficult, diverse, and often controversial issues are presented and can be discussed with a common goal of understanding the barriers to equity in health care” ( , p. 87) |
Emotional regulation | “The processes by which we influence which emotions we have, when we have them, and how we experience and express them” ( , p. 282) |
Intergroup contact | The promotion of contact between two groups with the goal of reducing prejudice ( , p. 66) |
Photovoice | “A method that allows participants to use photography to document their experiences and dialogue to eventually influence change” ( , p. 318) |
Service-learning | A “pedagogy of engagement wherein students address a genuine community need by engaging in volunteer service that is connected explicitly to the academic curriculum through structured ongoing reflections” ( , p. 115) |
Theater reading | Play reading with students as active participants ( , p. 232) |
Despite the limitations noted in Section 2 , the IAT continues to be widely utilized. The IAT and other measures ( 32 ) of implicit bias, stigma, and attitudes toward groups of persons were used among subjects to ( a ) demonstrate the existence of participant implicit biases, ( b ) act as a springboard to create cognitive dissonance for oral and/or written reflection and to practice bias management skills, and ( c ) evaluate interventions. Gonzalez et al. ( 37 ) found that using the IAT without priming on its results and without a follow-up debriefing led some subjects (22%) to question the validity of the measure and the existence of implicit biases, and therefore advised judicious use of the IAT and trained facilitators. Subjects who accepted the results of the IAT were not able to develop management strategies for those biases without dedicated instruction.
Despite having low explicit bias based on a self-reported survey, admissions committee members at The Ohio State University College of Medicine ( 14 ) had high levels of implicit preference for White versus Black students as measured by the Black-White IAT. Results were presented to committee members with strategies to reduce implicit bias. The following admissions cycle resulted in an increase in underrepresented minority matriculation from 17% to 20%, a change that was not statistically significant.
Seventy-six percent of studies ( 8 , 13 , 14 , 23 , 28 , 35 – 38 , 48 , 51 , 58 , 59 , 77 , 82 , 94 , 96 , 99 , 109 ) instructed on structural determinants such as structural racism and/or historic oppression of groups so that subjects could explore explicit and implicit biases. All these studies demonstrated an increased awareness of bias, and subjects often voiced a willingness to address their biases. Four studies explored the use of contact with groups with identities such as LGBTQI ( 58 , 59 ) and persons with mental illness ( 27 , 77 ) with positive and negative results, respectively.
In recognition that biases may be immutable in the current health care context but can be managed, educators have used transformative learning theory (TLT) in concert with implicit bias management techniques. TLT transforms the individual’s existing paradigm by disrupting assumptions and then engaging in critical reflection and dialogue to interpret the disruptions ( 68 ). TLT may move learners to an “inclusive, self-reflective and integrative frame of reference” ( 100 , p. 718). This paired approach has had early success. Sherman et al. ( 96 ) engaged both residents and faculty in transformative learning to address issues of race, racism, and Whiteness and created an environment for critical dialogue incorporating practical recommendations for addressing implicit bias in clinical practice. Focus groups 4 months later revealed that subjects noted increased awareness of their biases and sustained commitment to addressing racial bias, to challenging their own clinical decision making, and to engaging leadership in dialogue regarding bias.
Gonzalez et al. ( 38 ) describe implicit bias recognition and management (IBRM), a process that promotes conscious awareness of biases and fosters behavioral changes. IBRM supposes that biases are difficult to reduce and should therefore be managed. IBRM has helped medical students interrupt biases in learning and clinical settings. Wu et al. ( 109 ) paired IAT administration with training to improve skills in bias literacy, emotional regulation, and allyship ( Table 4 ). Trainees practiced these skills in clinical vignettes and improved their confidence in addressing bias in real-world settings. All three studies created a brave space to explore biases and emphasized continued practice and development of skills.
These studies have multiple limitations. They often lacked control groups or used pre- and postcomparison designs. They had limited longitudinal follow-up and often were not performed in real-world clinical or learning environments. Many studies did not focus on targeted outcomes, and most did not access the continuum of learners in medical education such as practicing health care providers and leadership. Most interventions had a limited one-time delivery with no opportunity to measure a dose- or time-dependent effect.
Many of the interventions demonstrated successful promotion of awareness of implicit bias held among subjects as well as an interest in mitigating implicit biases among subjects. No intervention in this review, however, achieved sustained reduction of implicit bias among health care professionals or trainees. In addition, no study demonstrated that an intervention improved clinical outcomes, the learning environment, interprofessional team dynamics, patient care, health disparities, patient satisfaction, or satisfaction of health professionals. Studies were hampered by lack of statistical analysis, lack of control group, limited numbers of participants, findings that are not necessarily generalizable from the classroom or web-based setting to the clinical or real-world setting, and heavy reliance on qualitative assessments or nonvalidated instruments. Future studies should also assess whether regularly timed booster interventions manifest in sustained changes over time and should have longer-term follow-up to assess sustainability of initial gains. Future studies should include educational models that use direct clinical observation or standardized patients. Studies should assess health care trainees’ ability to incorporate skills into patient communication and shared decision making, their improvement of clinical delivery practices, their interactions with colleagues, and their teaching practices.
Based on Jones’s ( 54 ) allegory A Gardener’s Tale , we present a conceptual model of implicit biases of health care providers and the key structural factors affecting these biases ( Figure 2 ). In the vicious cycle of health disparity, students, trainees, and providers receive a constant barrage of messaging that reinforces biases. The soil of their work (practice and learning environments) is laden with structural bias from racialized medicine, a biased learning environment, and poor compositional diversity. Furthermore, these trainees and health care providers are under substantial time pressure and cognitive load. These characteristics of the practice and learning environments may be considered structural determinants of implicit bias.
Interactions between structural determinants and provider implicit bias. The vicious cycle: Structural determinants of implicit bias in the practice environment support biased decision making. Structural determinants of health in the community further impair outcomes in marginalized populations, leading to confirmation of the practitioner’s implicit bias. Health disparities are exacerbated. The virtuous cycle: A favorable practice environment regarding structural determinants of implicit bias supports unbiased clinical decision making. Favorable structural determinants of health in the community further enhance patient outcomes, positively reinforcing unbiased practice. Health disparities are reduced.
Biases are now primed as the clinician moves to provide care to patients (see the left side of Figure 2 ). When caring for marginalized patients, the provider’s bias influences communication with the patient, potentially resulting in suboptimal decision making. The patient may sense the bias, may distrust the provider and system, and may decide to not follow through on treatment plans or may modify them. The patient lives in underresourced and unhealthy spaces that contribute to poor outcomes. The provider notes the poor outcomes and their implicit bias is confirmed. Health care disparities are exacerbated. Further exacerbation of the vicious cycle occurs when this dynamic is accompanied with biases toward students, trainees, and providers from marginalized groups. Individuals from these marginalized groups are less likely to succeed, confirming biases about them and perpetuating poor diversity in the health care workforce. The benefits of diversity to education and patient care are lost.
The right side of Figure 2 depicts the virtuous cycle of health equity. A well-resourced provider learning and working within an environment devoid of racialized medicine and bias and characterized by compositional diversity is less likely to display biases against the patient. Compositional diversity also increases the likelihood that the provider shares lived experiences with the patient. The patient notes the absence of provider bias, develops a trusting relationship, adheres to the treatment plan in a well-resourced environment, and returns with improved health outcomes. The patient’s outcome confirms the provider’s more favorable bias. Health care disparities are reduced.
This conceptual model highlights two important dynamics in the perpetuation of implicit bias and its impact on care. First, structural determinants in the health care system and surrounding community contribute to the development of implicit bias toward marginalized patient populations and then reinforce that implicit bias through generation of poorer patient outcomes. Second, interruption of this cycle is possible only through an overall shift toward favorable structural influences on implicit bias. Discrete, time-limited training as the sole intervention to reduce implicit bias is unlikely to result in sustained change; health care providers return to a practice or learning environment that is often replete with structural determinants and patient outcomes that reinforce implicit bias. To avoid the ongoing creation and perpetuation of racist structures in society, systems, and organizations, it is crucial to recognize that these dynamics may enhance the implicit bias of medical leaders and policy makers as well.
To enable provider-level bias interventions to succeed in improving health outcomes, multiple other concurrent approaches should address structural factors inside and outside the health care system that influence these biases ( 80 ).
Structural inequities outside the health care system include poor access to high-quality health care, racialized violence, the carceral state, crowded housing, healthy food scarcity, lack of access to green spaces, environmental toxins, and poorly protected workspaces, among other issues related to geography and place ( 19 , 103 ).
Structural inequities inside the health care system that prime bias include the work and learning environments of students, trainees, and providers ( 104 ). It will be important to address these structural drivers of bias, including time pressures, cognitive load, and the practice of racialized medicine. Racism, sex and gender discrimination, and other forms of discrimination must be rooted out, as they prevent marginalized trainees and faculty from thriving, create stereotype threat for the marginalized, and confirm bias for the nonmarginalized. Bioethical principles of fairness, distributive justice, and reciprocity should be core for public health officials and health care providers, and practitioner and provider trainings in these areas can raise awareness. For example, to address health inequities laid bare by COVID-19, Peek et al. ( 79 ) recommend a multifactorial approach that acknowledges the systemic racism of the health care system and other societal structures as well as the biases of providers ( 67 ).
Addressing compositional diversity in health care is another avenue for treating the structures that influence implicit and explicit biases and eliminate health care disparities. Minority health professionals are underrepresented in the workforce and health professions faculty ( 60 ). Only 6.2% of medical students identify as Hispanic or Latinx, and only 8.4% as Black or African American ( 1 ). Gender parity among medical school students has been achieved. However, women are underrepresented at the faculty instructor level, with substantially less representation at the professor level, and are also underrepresented in hospital leadership, with even starker inequities for female racial and ethnic minorities ( 33 , 88 ). Gender inequalities in salaries have been well documented ( 12 , 62 , 71 ). In academic medicine, Black male faculty are offered lower rates of compensation than their White counterparts and are less likely to be awarded research funding from the National Institutes of Health ( 34 ). Similarly, in 2016, graduate student enrollment in the Association of Schools and Programs of Public Health demonstrated a ≤5% increase over a 20-year period among Asian, Black, Hispanic, and Native American students; only 11.1% of students were Black and 12% were Hispanic. Black, Hispanic, and Native American representation among tenured public health faculty increased <3% during this same 20-year period ( 39 ).
TLT, IBRM, and a skills-based approach offer promise for future interventions in implicit bias management. It is also encouraging that discussions around disparities and inequities have moved from race to racism and have focused on the professional responsibility of providers to root out inequities and manage biases. The extant literature regarding the use of provider-level implicit bias interventions suggests that these interventions can play an important role in concert with other interventions that more broadly address bias and discrimination inside and outside the health care system. Evidence supports the use of provider-level interventions in immediate-impact activities such as decision making on search committees or admissions committees and raising critical awareness of the bioethical principles of fairness, distributive justice, and reciprocity. However, provider-level implicit bias interventions alone have not improved health outcomes. Thus, provider-level implicit bias interventions should be accompanied by interventions that systemically change structures inside and outside the health care system that influence biases and perpetuate health inequities.
The authors extend their heartfelt thanks to Debra A. Werner, the University of Chicago’s Librarian for Science Instruction & Outreach and Biomedical Reference Librarian, for her patient guidance and assistance with the systematic literature review, and Morgan Ealey, Administrative Manager, Section of General Internal Medicine, who helped format the manuscript.
M.E.P. and M.H.C. were supported in part by Bridging the Gap: Reducing Disparities in Diabetes Care National Program Office, funded by the Merck Foundation, and the Chicago Center for Diabetes Translation Research, funded by the National Institute of Diabetes and Digestive and Kidney Diseases (P30 DK092949). M.H.C. was also supported in part by Advancing Health Equity: Leading Care, Payment, and Systems Transformation, a program funded by the Robert Wood Johnson Foundation. M.H.C. is a member of the Blue Cross Blue Shield Health Equity Strategy advisory panel, Bristol Myers Squibb Company Health Equity Initiative advisory board, and The Joint Commission and Kaiser Permanente Bernard J. Tyson National Award for Excellence in Pursuit of Healthcare Equity review panel. The other authors are not aware of any affiliations, memberships, funding, or financial holdings that might be perceived as affecting the objectivity of this review.
BMC Medicine volume 22 , Article number: 336 ( 2024 ) Cite this article
2065 Accesses
87 Altmetric
Metrics details
Dietary guidelines recommend a shift to plant-based diets. Fortified soymilk, a prototypical plant protein food used in the transition to plant-based diets, usually contains added sugars to match the sweetness of cow’s milk and is classified as an ultra-processed food. Whether soymilk can replace minimally processed cow’s milk without the adverse cardiometabolic effects attributed to added sugars and ultra-processed foods remains unclear. We conducted a systematic review and meta-analysis of randomized controlled trials, to assess the effect of substituting soymilk for cow’s milk and its modification by added sugars (sweetened versus unsweetened) on intermediate cardiometabolic outcomes.
MEDLINE, Embase, and the Cochrane Central Register of Controlled Trials were searched (through June 2024) for randomized controlled trials of ≥ 3 weeks in adults. Outcomes included established markers of blood lipids, glycemic control, blood pressure, inflammation, adiposity, renal disease, uric acid, and non-alcoholic fatty liver disease. Two independent reviewers extracted data and assessed risk of bias. The certainty of evidence was assessed using GRADE (Grading of Recommendations, Assessment, Development, and Evaluation). A sub-study of lactose versus sucrose outside of a dairy-like matrix was conducted to explore the role of sweetened soymilk which followed the same methodology.
Eligibility criteria were met by 17 trials ( n = 504 adults with a range of health statuses), assessing the effect of a median daily dose of 500 mL of soymilk (22 g soy protein and 17.2 g or 6.9 g/250 mL added sugars) in substitution for 500 mL of cow’s milk (24 g milk protein and 24 g or 12 g/250 mL total sugars as lactose) on 19 intermediate outcomes. The substitution of soymilk for cow’s milk resulted in moderate reductions in non-HDL-C (mean difference, − 0.26 mmol/L [95% confidence interval, − 0.43 to − 0.10]), systolic blood pressure (− 8.00 mmHg [− 14.89 to − 1.11]), and diastolic blood pressure (− 4.74 mmHg [− 9.17 to − 0.31]); small important reductions in LDL-C (− 0.19 mmol/L [− 0.29 to − 0.09]) and c-reactive protein (CRP) (− 0.82 mg/L [− 1.26 to − 0.37]); and trivial increases in HDL-C (0.05 mmol/L [0.00 to 0.09]). No other outcomes showed differences. There was no meaningful effect modification by added sugars across outcomes. The certainty of evidence was high for LDL-C and non-HDL-C; moderate for systolic blood pressure, diastolic blood pressure, CRP, and HDL-C; and generally moderate-to-low for all other outcomes. We could not conduct the sub-study of the effect of lactose versus added sugars, as no eligible trials could be identified.
Current evidence provides a good indication that replacing cow’s milk with soymilk (including sweetened soymilk) does not adversely affect established cardiometabolic risk factors and may result in advantages for blood lipids, blood pressure, and inflammation in adults with a mix of health statuses. The classification of plant-based dairy alternatives such as soymilk as ultra-processed may be misleading as it relates to their cardiometabolic effects and may need to be reconsidered in the transition to plant-based diets.
ClinicalTrials.gov identifier, NCT05637866.
Peer Review reports
Major dietary guidelines recommend a shift to plant-based diets for public and planetary health [ 1 , 2 , 3 , 4 , 5 , 6 , 7 , 8 ] , while recommending simultaneous reductions in ultra-processed foods [ 2 , 3 , 4 , 5 , 6 , 7 , 8 ]. The shift to plant-based diets has resulted in an explosion of dairy, meat, and egg alternatives with plant protein foods projected to reach almost 10% of the global protein market by 2030 [ 9 ]. Although these foods can aid in the transition to plant-based diets, food classification systems such as the World Health Organization (WHO)-endorsed NOVA classification system classify them as ultra-processed foods to be avoided [ 10 ].
Dairy alternatives are an important example of a food category at the crossroads of these competing recommendations. School milk programs provide > 150 million servings of cow’s milk to children worldwide [ 11 ]. These programs are in addition to the food service and procurement policies of public institutions such as schools, universities, hospitals, long-term care homes, and prisons. Many of these programs and policies do not allow for the free replacement of cow’s milk with nutrient-dense plant milks [ 12 , 13 ]. Although the Dietary Guidelines for Americans [ 1 ], Canada’s Food Guide [ 3 ], and several European food-based dietary guidelines [ 14 ] recognize fortified soymilk [ 1 ] as nutritionally equivalent to cow’s milk, school nutrition programs in the United States (US) [ 12 ] and Europe [ 13 ] only provide funding for cow’s milk. There is a bipartisan bill before the US congress to change this policy and provide funding for fortified soymilk [ 15 ]. A major barrier to the use of fortified soymilk is that it contains added sugars to match the sweetness of cow’s milk at a level which would disqualify it from meeting the Food and Drug Administration’s proposed definition of “healthy” [ 16 ] (although its total sugar content is usually ~ 60% less than that of cow’s milk given the higher sweetness intensity of sucrose vs lactose) [ 17 ] and is classified (irrespective of its sugar content) as an ultra-processed food to be avoided [ 10 , 18 ]. Cow’s milk, on the other hand, enjoys classification as a “healthy,” minimally processed food to be encouraged [ 10 , 18 ].
As industry innovates in response to the growing demand and policy makers develop public health nutrition policies and programs in response to the evolving dietary guidance for more plant-based diets, it is important to understand whether nutrient-dense ultra-processed plant protein foods can replace minimally processed dairy foods without the adverse cardiometabolic effects attributed to added sugars and ultra-processed foods. We conducted a systematic review and meta-analysis of randomized controlled trials of the effect of substituting soymilk for minimally processed cow’s milk and its modification by added sugars (sweetened versus unsweetened) on intermediate cardiometabolic outcomes as a basis for understanding the role of nutrient-dense ultra-processed plant protein foods in the transition to plant-based diets.
We followed the Cochrane Handbook for Systematic Reviews of Interventions to conduct this systematic review and meta-analysis and reported our results by the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines [ 19 , 20 ] (Additional file 1 : Table 1). To explore whether added sugars mediate any effects observed in sweetened soymilk studies, we conducted an additional systematic review and meta-analysis sub-study. This separate investigation followed the same protocol and methodology as our main study. It focused on controlled trials examining the impact of lactose in isocaloric comparisons with fructose-containing sugars (such as sucrose, high-fructose corn syrup [HFCS], or fructose) when not included in a dairy-like matrix, on all outcomes in the main study. The protocol is registered at ClinicalTrials.gov (NCT05637866).
We searched MEDLINE, Embase, and the Cochrane Central Register of Controlled Trials databases through June 2024. The detailed search strategies for the main study and sub-study were based on validated search terms [ 21 ] (Additional file 1 : Tables 2 and 4). Manual searches of the reference lists of included studies supplemented the systematic search.
The main study included randomized controlled trials in human adults with any health status. Included trials had a study duration of ≥ 3 weeks and investigated the effects of soymilk compared with cow’s milk in energy matched conditions on intermediate cardiometabolic outcomes (Additional file 1 : Table 3). Trials that included other comparators that were not cow’s milk or had no viable outcome data were excluded. No restrictions were placed on language. For the sub-study, we included controlled trials involving adults of all health statuses that had a study duration of ≥ 3 weeks and investigated the effects of added sugars compared with lactose on the same intermediate cardiometabolic outcomes (Additional file 1 : Table 5).
A minimum of two investigators (ME, DG, SBM, AA) independently extracted relevant data from eligible studies. Extracted data included study design, sample size, sample characteristics (age, body mass index [BMI], sex, health status), intervention characteristics (soymilk volume, total sugars content, soy protein dose), control characteristics (cow’s milk volume, total sugars content, milk protein dose, milk fat content), baseline outcome levels, background diet, follow-up duration, setting, funding sources, and outcome data. The authors were contacted for missing outcome data when it was indicated that a relevant outcome was measured but not reported. Graphically presented data were extracted from figures using Plot Digitizer [ 22 ].
Outcomes for the main study and sub-study included blood lipids (low-density lipoprotein cholesterol [LDL-C], high-density lipoprotein cholesterol [HDL-C], non-high-density lipoprotein cholesterol [non-HDL-C], triglycerides, and apolipoprotein B [ApoB]), glycemic control (hemoglobin A1c [HbA1c], fasting plasma glucose, 2-h postprandial glucose, fasting insulin, and plasma glucose area under the curve [PG-AUC]), blood pressure (systolic blood pressure and diastolic blood pressure), inflammation (c-reactive protein [CRP]), adiposity (body weight, BMI, body fat, and waist circumference), kidney function and structure (creatinine, creatinine clearance, glomerular filtration rate [GFR], estimated glomerular filtration rate [eGFR], albuminuria, and albumin-creatinine ratio [ACR]), uric acid, and non-alcoholic fatty liver disease (NAFLD) (intrahepatocellular lipid [IHCL], alanine transaminase [ALT], aspartate aminotransferase [AST], and fatty liver index).
Mean differences (MDs) between the intervention and control arm and respective standard errors were extracted for each trial. If these were not provided, they were derived from available data using published formulas [ 19 ]. Mean pairwise difference in change-from-baseline values were preferred over end values. When median data was provided, they were converted to mean data with corresponding variances using methods developed by McGrath et al. [ 23 ]. When no variance data was available, the standard deviation of the MDs was borrowed from a trial similar in size, participants, and nature of intervention. All disagreements were reconciled by consensus or with a senior reviewer (JLS).
Included studies were assessed for the risk of bias independently and in duplicate by at least two investigators (ME, DG, SBM, AA) using the Cochrane Risk of Bias (ROB) 2 Tool [ 24 ]. The assessment was performed across six domains of bias (randomization process, deviations from intended interventions, missing outcome data, measurement of the outcome, selection of the reported result, and overall bias). Crossover studies were assessed for an additional domain of bias (risk of bias arising from period or carryover effects). The ROB for each domain was assessed as “low” (plausible bias unlikely to seriously alter the results), “high” (plausible bias that seriously weakens confidence in results), or “some concern” (plausible bias that raises some doubt about the results). Reviewer discrepancies were resolved by consensus or arbitration by a senior investigator (JLS).
STATA (version 17; StataCorp LP, College Station, TX) was used for all analyses for the main study and sub-study. The principal effect measures were the mean pair-wise differences in change from baseline (or alternatively, end differences) between the intervention arm providing the soymilk and the cow’s milk comparator/control arm in each trial (significance at P MD < 0.05). Results are reported as MDs with 95% confidence intervals (95% CI). As one of our primary research questions relates to the role of added sugars as a mediator in any observed differences between soymilk and cow’s milk, we stratified results by the presence of added sugars in the soymilk (sweetened versus unsweetened) and assessed effect modification by this variable on pooled estimates. Data were pooled using the generic inverse variance method with DerSimonian and Laird random effect models [ 25 ]. Fixed effects were used when less than five trials were available for an outcome [ 26 ]. A paired analysis was applied for crossover designs and for within-individual correlation coefficient between treatment of 0.5 as described by Elbourne et al. [ 27 , 28 ].
Heterogeneity was assessed using the Cochran’s Q statistic and quantified using the I 2 statistic, where I 2 ≥ 50% and P Q < 0.10 were used as evidence of substantial heterogeneity [ 19 ]. Potential sources of heterogeneity were explored using sensitivity analyses. Sensitivity analyses were done via two methods. We conducted an influence analysis by systematically removing one trial at a time and recalculating the overall effect estimate and heterogeneity. A trial was considered influential if its removal explained the substantial heterogeneity or altered the direction, magnitude, or significance of the summary estimate. To determine whether the overall summary estimates were robust to the use of an assumed correlation coefficient for crossover trials, we conducted a second sensitivity analysis by using correlation coefficients of 0.25 and 0.75. If ≥ 10 trials were available, meta-regression analyses were used to assess the significance of each subgroup categorically and when possible, continuously (significance at P < 0.05). A priori subgroup analyses included soy protein dose, follow-up duration, baseline outcome levels, comparator, design, age, health status, funding, and risk of bias.
If ≥ 6 trials are available [ 29 ], dose–response analyses were performed using meta-regression to assess linear (by generalized least squares trend (GLST) estimation models) and non-linear spline curve modeling (by MKSPLINE procedure) dose–response gradients (significance at P < 0.05).
If ≥ 10 studies were available, publication bias was assessed by inspection of contour-enhanced funnel plots and formal testing with Egger’s and Begg’s tests (significance at P < 0.10) [ 30 , 31 , 32 ]. If evidence of publication bias was suspected, the Duval and Tweedie trim-and-fill method was performed to adjust for funnel plot asymmetry by imputing missing study data and assess for small-study effects [ 33 ].
The Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach was used to assess the certainty of evidence. The GRADE Handbook and GRADEpro V.3.2 software were used [ 34 , 35 ]. A minimum of two investigators (ME, DG, SBM) independently performed GRADE assessments for each outcome [ 36 ]. Discrepancies were resolved by consensus or arbitration by the senior author (JLS). The overall certainty of evidence was graded as either high, moderate, low, or very low. Randomized trials are initially graded as high by default and then downgraded or upgraded based on prespecified criteria. Reasons for downgrading the evidence included study limitations (risk of bias assessed by the Cochrane ROB Tool), inconsistency of results (substantial unexplained interstudy heterogeneity, I 2 > 50% and P Q < 0.10), indirectness of evidence (presence of factors that limit the generalizability of the results), imprecision (the 95% CI for effect estimates overlap with the MID for benefit or harm), and publication bias (evidence of small-study effects). The evidence was upgraded if a significant dose–response gradient was detected. We defined the importance of the magnitude of the pooled effect estimates using prespecified MIDs (Additional file 1 : Table 6) with GRADE guidance [ 36 , 37 , 38 , 39 , 40 , 41 , 42 , 43 , 44 , 45 , 46 , 47 , 48 , 49 , 50 ] according to five levels: very large (≥ 10 MID); large (≥ 5 MID); moderate (≥ 2 MID); small important (≥ 1 MID); and trivial/unimportant (< 1 MID) effects.
Figure 1 in Appendix shows the flow of the literature for the main analysis. We identified 522 reports through database and manual searches. A total of 17 reports [ 51 , 52 , 53 , 54 , 55 , 56 , 57 , 58 , 59 , 60 , 61 , 62 , 63 , 64 , 65 , 66 , 67 ] met the inclusion criteria and contained data for LDL (10 trials, n = 312), HDL-C (8 trials, n = 271), non-HDL-C (7 trials, n = 243), triglycerides (9 trials, n = 278), HbA1c (1 trial, n = 25), fasting plasma glucose (5 trials, n = 147), 2-h plasma glucose (1 trial, n = 28), fasting insulin (4 trials, n = 119), systolic blood pressure (5 trials, n = 158), diastolic blood pressure (5 trials, n = 158), CRP (5 trials, n = 147), body weight (6 trials, n = 163), BMI (6 trials, n = 173), body fat (1 trial, n = 43), waist circumference (3 trials, n = 90), creatinine (1 trial, n = 25), eGFR (1 trial, n = 25), ALT (1 trial, n = 24), and AST (1 trial, n = 24) involving 504 participants. No trials were available for ApoB, PG-AUC, creatinine clearance, eGFR, albuminuria, ACR, uric acid, IHCL, or fatty liver index.
Additional file 1 : Fig. 1 shows the flow of literature for the sub-study. We identified 1010 reports through database and manual searches. After excluding 305 duplicates, a total of 705 reports were reviewed by title and abstract. No reports met the inclusion criteria and therefore no data was available for analysis.
Table 1 shows the characteristics of the included trials. The trials were conducted in a variety of locations, with most conducted in Iran (7/17 trials, 41%), followed by the US (3/17 trials, 18%), Italy (2/17 trials, 12%), Brazil (1/17 trials, 6%), Scotland (1/17 trials, 6%), Sweden (1/17 trials, 6%), Spain (1/17 trials, 6%), and Australia (1/17 trials, 6%). All trials took place in outpatient settings (17/17, 100%). The median trial size was 25 participants (range, 7–60 participants). The median age of the participants was 48.5 years (range, 20–70 years) and the median BMI was 27.9 kg/m 2 (range, 20–31.1 kg/m 2 ). The trials included participants with hypercholesterolemia (4/17 trials, 25%), overweight or obesity (4/17 trials, 25%), type 2 diabetes (2/17 trials, 12%), hypertension (1/17 trials, 6%), rheumatoid arthritis (1/17 trials, 6%), or were healthy (3/17 trials, 18%) or post-menopausal (2/17 trials, 12%). Both trials with crossover design (10/17 trials, 59%) and parallel design (7/17 trials, 41%) were included. The intervention included sweetened (11/17 trials, 65%) and unsweetened (6/17 trials, 35%) soymilk.
The median soymilk dose was 500 mL/day (range, 240–1000 mL/day) with a median soy protein of 22 g/day (range, 2.5–70 g/day) or 6.6 g/250 mL (range, 2.6–35 g/250 mL) and median total (added) sugars of 17.2 g/day (range, 4.0–32 g/day) or 6.9 g/250 mL (range, 1–16 g/250 mL) in the sweetened soymilk. The comparators included skim (0% milk fat) (2/17 trials, 12%), low-fat (1% milk fat) (4/17 trials, 24%), reduced fat (1.5–2.5% milk fat) (7/17 trials, 41%), and whole (3% milk fat) (1/17 trials, 6%) cow’s milk. Three trials did not report the milk fat content of cow’s milk used. The median cow’s milk dose was 500 mL/day (range, 236–1000 mL/day) with a median milk protein of 24 g/day (range, 3.3–70 g/day) or 8.3 g/250 mL (range, 3.4–35 g/250 mL) and median total (lactose) sugars of 24 g/day (range, 11.5–49.2 g/day) or 12 g/250 mL (range, 10.8–12.8 g/250 mL). The median study duration was 4 weeks (range, 4–16 weeks). The trials received funding from industry (1/17 trials, 6%), agency (8/17 trials, 47%), both industry and agency (4/16 trials, 25%), or they did not report the funding source (4/17 trials, 24%).
Additional file 1 : Fig. 2 shows the ROB assessments of the included trials. Two trials were assessed as having some concerns from period or carryover effects: Bricarello et al. [ 53 ] and Steele [ 67 ]. All other trials were judged as having an overall low risk of bias. There was no evidence of serious risk of bias across the included trials.
Figure 2 and Additional file 1 : Figs. 3–6 show the effect of substituting soymilk for cow’s milk on markers of blood lipids. The substitution resulted in a small important reduction in LDL-C (10 trials; MD: − 0.19 mmol/L; 95% CI: − 0.29 to − 0.09 mmol/L; P MD < 0.001; no heterogeneity: I 2 = 0.0%; P Q = 0.823), a trivial increase in HDL-C (8 trials; MD: 0.05 mmol/L; 95% CI: 0.00 to 0.09 mmol/L; P MD = 0.036; no heterogeneity: I 2 = 0.0%; P Q = 0.053), a moderate reduction in non-HDL-C (7 trials; MD: − 0.26 mmol/L; 95% CI: − 0.43 to − 0.10 mmol/L; P MD = 0.002; no heterogeneity: I 2 = 0.0%; P Q = 0.977), and no effect on triglycerides. There were no interactions by added sugars in soymilk for any blood lipid markers ( P = 0.49–0.821).
Figure 2 and Additional file 1 : Figs. 7–10 show the effect of substituting soymilk for cow’s milk on markers of glycemic control. The substitution had no effect on HbA1c, fasting plasma glucose, 2-h plasma glucose, or fasting insulin. There was no interaction by added sugars in soymilk for fasting plasma glucose ( P = 0.747) but there was an interaction for fasting insulin ( P = 0.026), where a lack of effect remained in both groups with neither the sweetened soymilk (non-significant increasing effect) nor the unsweetened soymilk (non-significant decreasing effect) showing an effect on fasting insulin. We could not assess this interaction for HbA1c or 2-h plasma glucose, as there was only one trial available for each outcome.
Figure 2 and Additional file 1 : Figs. 11 and 12 show the effect of substituting soymilk for cow’s milk on blood pressure. The substitution resulted in a moderate reduction in both systolic blood pressure (5 trials; MD: − 8.00 mmHg; 95% CI: − 14.89 to − 1.11 mmHg; P MD = 0.023; substantial heterogeneity: I 2 = 86.89%; P Q ≤ 0.001) and diastolic blood pressure (5 trials; MD: − 4.74 mmHg; 95% CI: − 9.17 to − 0.31 mmHg; P MD = 0.036; substantial heterogeneity: I 2 = 77.3%; P Q = 0.001). There were no interactions by added sugars in soymilk for blood pressure ( P = 0.747 and 0.964).
Figure 2 and Additional file 1 : Fig. 13 show the effect of substituting soymilk for cow’s milk on markers of inflammation. The substitution resulted in a small important reduction in CRP (5 trials; MD: − 0.81 mg/dL; 95% CI: − 1.26 to − 0.37 mg/dL; P MD = < 0.001; no heterogeneity: I 2 = 0.0%; P Q = 0.814). There was no interaction by added sugars in soymilk for CRP ( P = 0.275).
Figure 2 and Additional file 1 : Figs. 14–17 show the effect of substituting soymilk for cow’s milk on markers of adiposity. The substitution had no effect on body weight, BMI, body fat, or waist circumference. There were no interactions by added sugars in soymilk for any adiposity outcome ( P = 0.664–0.733).
Figure 2 and Additional file 1 : Figs. 18 and 19 show the effect of substituting soymilk for cow’s milk on markers of kidney function. The substitution had no effect on creatinine or eGFR. We could not assess the interaction by added sugars in soymilk for creatinine or eGFR, as there was only one trial available for each outcome which included soymilk without added sugars.
Figure 2 and Additional file 1 : Figs. 20 and 21 show the effect of substituting soymilk for cow’s milk on markers of NAFLD. The substitution had no effect on ALT or AST. We could not assess heterogeneity or the interaction by added sugars in soymilk for ALT or AST, as there was only one trial available for each outcome which included soymilk without added sugars.
Additional file 1 : Figs. 22–33 present the influence analyses across all outcomes. The removal of Bricarello et al. [ 53 ] or Steele [ 67 ] each resulted in loss of significant effect for HDL-C. The removal of Onning et al. [ 62 ] or Steele [ 67 ] each resulted in a partial explanation of heterogeneity for triglycerides. The removal of Hasanpour et al. [ 56 ] explained the heterogeneity for fasting insulin. The removal of Keshavarz et al. [ 57 ] or Miraghajani et al. [ 59 ] each resulted in a loss of significant effect for systolic blood pressure and the removal of Rivas et al. [ 63 ] resulted in a partial explanation of the heterogeneity for systolic blood pressure. The removal of Hasanpour et al. [ 56 ], Keshavarz et al. [ 57 ], Miraghajani et al. [ 59 ], or Rivas et al. [ 63 ] each resulted in a loss of significant effect for diastolic blood pressure and the removal of Rivas et al. [ 63 ] resulted in a partial explanation of heterogeneity for diastolic blood pressure. The removal of Mohammad-Shahi et al. [ 58 ] resulted in loss of significant effect for CRP.
Additional file 1 : Table 8 shows the sensitivity analyses for the different correlation coefficients (0.25 and 0.75) used in paired analyses of crossover trials for all outcomes. The different correlation coefficients did not alter the direction, magnitude, or significance of the effect or evidence for heterogeneity, with the following exceptions: loss of significance for the effect of the substitution on HDL-C (8 trials; MD: 0.04 mmol/L; 95% CI: − 0.10 to 0.01 mmol/L; P MD = 0.107; I 2 = 0.0%; P Q = 0.670) with the use of 0.25 and (8 trials; MD: 0.05 mmol/L; 95% CI: − 0.10 to 0.01 mmol/L; P MD = 0.089; I 2 = 0.0%; P Q = 0.640) with the use of 0.75.
Additional file 1 : Figs. 34–36 present the subgroup analyses and continuous meta-regression analyses for LDL-C. Subgroup analysis was not conducted for any other outcome as there were < 10 trials included. There was no significant effect modification by health status, BMI, age, comparator, baseline LDL-C, study design, follow-up duration, funding source, dose of soy protein, or risk of bias for LDL-C. However, there were tendencies towards a greater reduction in LDL-C by point estimates in groups with certain health statuses (hypercholesterolemic and overweight/obesity), a higher baseline LDL-C, and a higher soy protein dose (> 25 g/day).
Additional file 1 : Figs. 37–42 present linear and non-linear dose–response analyses for LDL-C, HDL-C, non-HDL-C, triglycerides, body weight, and BMI. There was no dose–response seen for the effect of substituting soymilk for cow’s milk, with the exception of a positive linear dose–response for triglycerides ( P linear = 0.038). We did not downgrade the certainty of evidence as the greater reduction in triglycerides seen at lower doses of soy protein was lost at higher doses. There were no dose–response analyses performed for the remaining outcomes because there were < 6 trials available for each.
Additional file 1 : Fig. 43 presents the contour-enhanced funnel plot for assessment of publication bias for LDL-C. There was no asymmetry at the visual inspection and no evidence (Begg’s test = 0.721, Egger’s test = 0.856) of funnel plot asymmetry for LDL-C. No other publication bias analyses could be performed as there were < 10 trials available for each.
Additional file 1 : Table 9 shows the reported adverse events and acceptability of study beverages. Adverse events were reported in nine trials. In one trial by Gardner et al. [ 55 ], one participant experienced a recurrence of a cancer; however, it was considered to be unrelated to the short-term consumption of the study milks. Three trials (Miraghajani et al., Hasanpour et al., and Mohammad-Shahi, et al.) [ 56 , 58 , 59 ] reported one to two withdrawals due to digestive difficulties related to soymilk consumption. Two trials (Sirtori et al. 1999 and 2002) [ 65 , 66 ] reported one or more participants with digestive difficulties related to cow’s milk consumption. Two trials (Nourieh et al. and Keshavarz et al.) [ 57 , 61 ] each reported two participant withdrawals related to digestive problems that were not specific to either study beverage. Of these, four trials indicated that most participants found the soymilk and cow’s milk acceptable and tolerable. One trial, by Onning et al. [ 62 ], incorporated a sensory evaluation of appearance, consistency, flavor, and overall impression, which showed declining scores for both types of milk over the 3-week test period.
Additional file 1 : Table 10 presents the GRADE assessment. The certainty of evidence for the effect of substituting soymilk for cow’s milk was high for LDL-C, non-HDL-C, fasting plasma glucose, and waist circumference. The certainty of evidence was moderate for HDL-C, triglycerides, fasting insulin, systolic blood pressure, diastolic blood pressure, CRP, body weight, and BMI owing to a downgrade for imprecision of the pooled effect estimates and was moderate for body fat owing to a downgrade for indirectness. The certainty of evidence was low for HbA1c, 2-h plasma glucose, creatinine, eGFR, ALT, and AST owing to downgrades for indirectness and imprecision.
We conducted a systematic review and meta-analysis of 17 trials that examined the effect of substituting soymilk (median dose of 22 g/day or 6.6 g/250 mL serving of soy protein per day and 17.2 g/day or 6.9 g/250 mL of total [added] sugars in the sweetened soymilk) for cow’s milk (median dose of 24 g/day or 8.3 g/250 mL of milk protein and 24 g/day or 12 g/250 mL of total sugars [lactose]) and its modification by added sugars (sweetened versus unsweetened soymilk) on 19 intermediate cardiometabolic outcomes over a median follow-up period of 4 weeks in adults of varying health status. The substitution of soymilk for cows’ milk led to moderate reductions in non-HDL-C (− 0.26 mmol/L or ~ − 7%) and systolic blood pressure (− 8.00 mmHg) and diastolic blood pressure (− 4.74 mmHg); small important reductions in LDL-C (− 0.19 mmol/L or ~ − 6%) and CRP (− 0.81 mg/L or ~ 22%); and a trivial increase in HDL-C (0.05 mmol/L or ~ 4%), with no adverse effects on other intermediate cardiometabolic outcomes. There was no meaningful interaction by added sugars in soymilk, with sweetened and unsweetened soymilk showing similar effects across outcomes. There was no dose–response relationship seen across the outcomes for which dose–response analyses were performed.
Our findings agree with previous evidence syntheses of soy. Regulatory authorities such as the United States Food and Drug Administration (FDA) and Health Canada have conducted comprehensive evaluations of the randomized controlled trials of the effect of soy protein from different sources on total-C and LDL-C, resulting in approved health claims for soy protein (based on an intake of 25 g/day of soy protein irrespective of source) for cholesterol reduction [ 68 ] and coronary heart disease risk reduction [ 69 ]. Updated systematic reviews and meta-analyses of the 46 randomized controlled trials included in the re-evaluation of the FDA health claim [ 70 ] showed reductions in LDL-C of − 3.2% [ 71 ]. This reduction has been stable since the health claim was first approved in 1999 [ 72 ] and is smaller but consistent with our findings specifically for soymilk. No increase in HDL-C, however, was detected. Previous systematic reviews and meta-analyses of randomized controlled trials of soy protein and soy isoflavones have also shown significant but smaller reductions in systolic blood pressure (1.70 mmHg) and diastolic blood pressure (− 1.27 mmHg) [ 73 ] than was found in the current analysis. These reductions in LDL-C and blood pressure are further supported by reductions in clinical events with updated pooled analyses of prospective cohort studies showing that legumes including soy are associated with reduced incidence of total cardiovascular disease and coronary heart disease [ 74 ].
Systematic reviews and meta-analyses that specifically isolated the effect of soymilk (as a single food matrix) in its intended substitution for cow’s milk are lacking. Sohouli and coworkers [ 75 ] conducted a systematic review and meta-analysis of 18 randomized controlled trials in 665 individuals of varying health status that assessed the effect of soymilk in comparison with a mix of comparators on intermediate cardiometabolic outcomes but did not isolate its substitution with cow’s milk. This synthesis showed similar improvements in LDL-C (− 0.24 mmol/L), systolic blood pressure (− 7.38 mmHg), diastolic blood pressure (− 4.36 mmHg), and CRP (− 1.07, mg/L), while also showing reductions in waist circumference and TNF-α [ 75 ]. The substitution of legumes that includes soy for various animal protein sources and more specifically legumes/nuts (the only exposure available) for dairy in syntheses of prospective cohort studies has also shown reductions in incident total cardiovascular disease and all-cause mortality [ 76 ].
Indirect evidence from dietary patterns that contain soy foods including soymilk in substitution for different animal sources of protein including cow’s milk further supports our findings. Systematic reviews and meta-analyses of randomized trials of the Portfolio diet and vegetarian and vegan dietary patterns have shown additive reductions in LDL-C, non-HDL-C, blood pressure, and CRP when soy foods including soymilk are combined with other foods that target these same intermediate risk factors with displacement of different animal sources of protein including cow’s milk [ 77 , 78 ]. These reductions have also been shown to translate to reductions in clinical events with systematic reviews and meta-analyses of prospective cohort studies showing that adherence to these dietary patterns is associated with reductions in incident coronary heart disease, total cardiovascular disease, and all-cause mortality [ 79 , 80 , 81 ].
The potential mechanism mediating the effects of soy remains unclear. Specific components within the soy food matrix, including soy protein and phytochemicals like isoflavones [ 82 ], have been implicated. The well-established lipid-lowering effect of soy [ 72 ] may be attributed to the 7S globulin fraction of soy protein, which exerts its primary action by upregulating LDL-C receptors predominantly within the liver, thereby augmenting the clearance of LDL-C from circulation [ 82 ]. The isoflavone, fiber, fatty acids, and anti-nutrient components may also exert some mediation [ 83 ]. The reduction in blood pressure has been most linked to the soy isoflavones [ 83 ]. There is evidence that soy isoflavones may modulate the renin–angiotensin–aldosterone system (RAAS), with the capacity to inhibit the production of angiotensin II and aldosterone, thereby contributing to the regulation of blood pressure [ 73 ]. Another blood pressure lowering mechanism may involve the ability of soy isoflavones to enhance endothelial function by mitigating oxidative stress and inflammation, consequently promoting the release of the relaxing factor nitric oxide (NO) [ 73 ]. This potential mechanism of isoflavones may also explain the reductions seen in inflammation.
Our evidence synthesis had several strengths. First, we completed a comprehensive and reproducible systematic search and selection process of the available literature examining the effect of substituting soymilk for cow’s milk on intermediate cardiometabolic outcomes. Second, we synthesized the totality of available evidence from a large body of randomized controlled trials, which gives the greatest protection against systematic error. Third, we included an extensive and comprehensive list of outcomes to fully capture the impact of soymilk on cardiometabolic health. Fourth, we only included randomized controlled trials that compared soymilk to cow’s milk directly, to increase the specificity of our conclusion. Finally, we included a GRADE assessment to explore the certainty of available evidence.
There were also several limitations. First, we could not conduct the sub-study of the effect of lactose versus added sugars outside of a dairy-like matrix, as no eligible trials could be identified. Although this analysis is important for isolating the effect of added sugars as a mediator of any adverse effects, we did not observe any meaningful interaction by added sugars in soymilk. Second, there was serious imprecision in the pooled estimates across many of the outcomes with the 95% confidence intervals overlapping the MID in each case, with the exception of LDL-C, non-HDL-C, fasting plasma glucose, and waist circumference. The certainty of evidence for HDL-C, triglycerides, HbA1c, fasting plasma glucose, 2-h plasma glucose, fasting insulin, systolic blood pressure, diastolic blood pressure, CRP, body weight, BMI, body fat, creatinine, eGFR, ALT, and AST was downgraded for this reason. Third, there was evidence of indirectness related to insufficient trials for HbA1c, 2-h plasma glucose, creatinine, eGFR, ALT, and AST, which limits generalizability. Each outcome with data from only 1 trial was downgraded for this reason. Another source of indirectness could be the median follow-up duration of 4 weeks (range, 4–16 weeks). This time frame may be sufficient for observing certain effects, but other outcomes may require a longer period to manifest changes. Despite acknowledging this variation in response time among different outcomes, we did not further downgrade for this aspect of indirectness. Instead, we tailored our conclusions to reflect short-to-moderate term effects. Finally, although publication bias was not suspected, we were only able to make this assessment for LDL-C, as there were < 10 trials for all other outcomes.
Considering these strengths and limitations, we assessed the certainty of evidence as high for LDL-C and non-HDL-C; moderate for systolic blood pressure, diastolic blood pressure, CRP, and HDL-C; and moderate-to-low for all outcomes where significant effects were not observed.
This work has important implications for plant protein foods in the recommended shift to more plant-based diets. Major international dietary guidelines in the US [ 1 ], Canada [ 3 ], and Europe [ 4 , 5 , 6 ] recommend fortified soymilk as the only suitable replacement for cow’s milk. Our findings support this recommendation showing soymilk including sweetened soymilk (up to 7 g added sugars per 250 mL) does not have any adverse effects compared with cow’s milk across 19 intermediate cardiometabolic outcomes with benefits for lipids, blood pressure, and inflammation. This evidence suggests that it may be misleading as it relates to their cardiometabolic effects to classify fortified soymilk as an ultra-processed food to be avoided while classifying cow’s milk as a minimally processed food to be encouraged (based on the WHO-endorsed NOVA classification system [ 10 ]). It also suggests that it may be misleading not to allow fortified soymilk that is sweetened with small amounts of sugars to be classified as “healthy” (based on the FDA’s new proposed definition that only permits this claim on products with added sugars ≤ 2.5 g or 5% daily value (DV) per 250 mL serving [ 16 ]). The proposed FDA criteria would prevent this claim on soymilk products designed to be iso-sweet analogs of cow’s milk (in which 5 g or 10% daily value [DV] of added sugars from sucrose in soymilk is equivalent to the 12 g of lactose in cow’s milk per 250 mL serving, as sucrose is 1.4 sweeter than lactose [ 17 ]). To prevent confusion, policy makers may want to exempt fortified soymilk from classification as an ultra-processed food and allow added sugars up to 10% DV for the definition of “healthy,” as has been proposed by the FDA for sodium and saturated fat in dairy products (including soy-based dairy alternatives) to account for accepted processing and preservation methods [ 16 ]. These policy considerations would balance the need to limit nutrient-poor energy-dense foods with the need to promote nutrient-dense foods like fortified soymilk in the shift to healthy plant-based diets.
In conclusion, the evidence provides a good indication that substituting either sweetened or unsweetened soymilk for cow’s milk in adults with varying health statuses does not have the adverse effects on intermediate cardiometabolic outcomes attributed to added sugars and ultra-processed foods in the short-to-moderate term. There appear even to be advantages with small to moderate reductions in established markers of blood lipids (LDL-C, non-HDL-C) that are in line with approved health claims for cholesterol and coronary heart disease risk reduction, as well as small to moderate reductions in blood pressure and inflammation (CRP). Sources of uncertainty include imprecision and indirectness in several of the estimates. There remains a need for more well-powered randomized controlled trials of the effect of substituting soymilk for cow’s milk on less studied intermediate cardiometabolic outcomes, especially established markers of glycemic control, kidney structure and function, and NAFLD. There is also a need for trials comparing lactose versus added sugars outside of a dairy-like matrix to understand better the role of added sugars at different levels in substitution for lactose across outcomes. In the meantime, our findings support the use of fortified soymilk with up to 7 g added sugars per 250 mL as a suitable replacement for cow’s milk and suggest that its classification as ultra-processed and/or not healthy based on small amounts of added sugars may be misleading and need to be reconsidered to facilitate the recommended transition to plant-based diets.
All data generated or analyzed during this study are included in this published article and its Additional file 1 : information files.
Grading of Recommendations, Assessment, Development, and Evaluation
Non-high-density lipoprotein cholesterol
Low-density lipoprotein cholesterol
C-reactive protein
High-density lipoprotein cholesterol
World Health Organization
United States
Preferred Reporting Items for Systematic Reviews and Meta-Analysis
High-fructose corn syrup
Body mass index
Apolipoprotein B
Hemoglobin A1c
Plasma glucose area under the curve
Glomerular filtration rate
Estimated glomerular filtration rate
Albumin-creatinine ratio
Non-alcoholic fatty liver disease
Intrahepatocellular lipid
Alanine transaminase
Aspartate aminotransferase
Mean difference
Risk of bias
95% Confidence interval
Generalized least squares trend
Food and Drug Administration
Tumor necrosis factor alpha
Renin-angiotensin-aldosterone system
Nitric oxide
Daily value
Dietary guidelines for Americans, 2020–2025. 2020 [9:[Available from: www.dietaryguidelines.gov .
Canada, Health. Canada’s Food Guide. Ottawa; 2019. https://food-guide.canada.ca/en/ .
Canada’s food guide Ottawa 2019 [Available from: https://food-guide.canada.ca/en/ .
Blomhoff R, Andersen R, Arnesen EK, Christensen JJ, Eneroth H, Erkkola M, Gudanaviciene I, Halldórsson ÞI, Höyer-Lund A, Lemming EW. Nordic nutrition recommendations 2023: integrating environmental aspects. Nordisk Ministerråd; 2023.
García EL, Lesmes IB, Perales AD, Arribas VM, del Puy Portillo Baquedano M, Velasco AMR, Salvo UF, Romero LT, Porcel FBO, Laín SA. Report of the Scientific Committee of the Spanish Agency for Food Safety and Nutrition (AESAN) on sustainable dietary and physical activity recommendations for the Spanish population. Wiley Online Library; 2023. Report No.: 2940–1399.
Brink E, van Rossum C, Postma-Smeets A, Stafleu A, Wolvers D, van Dooren C, et al. Development of healthy and sustainable food-based dietary guidelines for the Netherlands. Public Health Nutr. 2019;22(13):2419–35.
Article PubMed PubMed Central Google Scholar
Lichtenstein AH, Appel LJ, Vadiveloo M, Hu FB, Kris-Etherton PM, Rebholz CM, et al. 2021 dietary guidance to improve cardiovascular health: a scientific statement from the American Heart Association. Circulation. 2021;144(23):e472–87.
Article PubMed Google Scholar
Willett W, Rockström J, Loken B, Springmann M, Lang T, Vermeulen S, et al. Food in the Anthropocene: the EAT–Lancet Commission on healthy diets from sustainable food systems. The lancet. 2019;393(10170):447–92.
Article Google Scholar
Bartashus J, Srinivasan G. Plant-based foods poised for explosive growth. Bloomberg Intelligence. 2021.
Monteiro CA, Cannon G, Lawrence M, Costa Louzada Md, Pereira Machado P. Ultra-processed foods, diet quality, and health using the NOVA classification system. Rome: FAO; 2019. p. 48.
International Dairy Federation. The contribution of school milk programmes to the nutrition of children worldwide. Brussels: Belgium; 2020.
Google Scholar
USDA Food and Nutrition Service. Special Milk Program [Available from: https://www.fns.usda.gov/smp/special-milk-program .
The European Parliament. European Parliament resolution of 9 May 2023 on the implementation of the school scheme [Available from: https://www.europarl.europa.eu/doceo/document/TA-9-2023-0135_EN.html .
European Commission. Summary of FBDG recommendations for milk and dairy products for the EU, Iceland, Norway, Switzerland and the United Kingdom. [Available from: https://knowledge4policy.ec.europa.eu/health-promotion-knowledge-gateway/food-based-dietary-guidelines-europe-table-7_en .
Addressing Digestive Distress in Stomachs of Our Youth (ADD SOY) Act, House of Representatives, 1st Sess.; 2023. https://troycarter.house.gov/sites/evo-subsites/troycarter.house.gov/files/evo-media-document/add-soy-act.pdf .
Food and Drug Administration. Food labeling: nutrient content claims; definition of term “healthy”. In: Department of Health and Human Services (HHS); 2022. https://www.federalregister.gov/documents/2022/09/29/2022-20975/food-labeling-nutrient-content-claims-definition-of-term-healthy .
Helstad S. Chapter 20 - corn sweeteners. In: Serna-Saldivar SO, editor. Corn. 3rd ed. Oxford: AACC International Press; 2019. p. 551–91.
Chapter Google Scholar
Messina M, Sievenpiper JL, Williamson P, Kiel J, Erdman JW. Perspective: soy-based meat and dairy alternatives, despite classification as ultra-processed foods, deliver high-quality nutrition on par with unprocessed or minimally processed animal-based counterparts. Adv Nutr. 2022;13(3):726–38.
Article PubMed PubMed Central CAS Google Scholar
Higgins J, Thomas J, Chandler J. Cochrane handbook for systematic reviews of interventions version 6.2. 2021.
Moher D, Liberati A, Tetzlaff J, Altman DG, Group* P. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Ann Intern Med. 2009;151(4):264–9.
BMJ Best Practice. Search strategies [Available from: https://bestpractice.bmj.com/info/toolkit/learn-ebm/study-design-search-filters/ .
Rohatgi A. WebPlotDigitizer 4.6; 2022. https://automeris.io/WebPlotDigitizer/ .
McGrath S, Zhao X, Steele R, Thombs BD, Benedetti A, Collaboration DESD. Estimating the sample mean and standard deviation from commonly reported quantiles in meta-analysis. Stat Methods Med Res. 2020;29(9):2520–37.
Sterne JAC, Savovic J, Page MJ, Elbers RG, Blencowe NS, Boutron I, et al. RoB 2: a revised tool for assessing risk of bias in randomised trials. BMJ. 2019;366: l4898.
DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials. 1986;7(3):177–88.
Article PubMed CAS Google Scholar
Tufanaru C, Munn Z, Stephenson M, Aromataris E. Fixed or random effects meta-analysis? Common methodological issues in systematic reviews of effectiveness. Int J Evid Based Healthc. 2015;13(3):196–207.
Elbourne DR, Altman DG, Higgins JP, Curtin F, Worthington HV, Vail A. Meta-analyses involving cross-over trials: methodological issues. Int J Epidemiol. 2002;31(1):140–9.
Balk EM, Earley A, Patel K, Trikalinos TA, Dahabreh IJ. Empirical assessment of within-arm correlation imputation in trials of continuous outcomes. 2013.
Fu R, Gartlehner G, Grant M, Shamliyan T, Sedrakyan A, Wilt TJ, et al. Conducting quantitative synthesis when comparing medical interventions: AHRQ and the Effective Health Care Program. J Clin Epidemiol. 2011;64(11):1187–97.
Peters JL, Sutton AJ, Jones DR, Abrams KR, Rushton L. Contour-enhanced meta-analysis funnel plots help distinguish publication bias from other causes of asymmetry. J Clin Epidemiol. 2008;61(10):991–6.
Egger M, Smith GD, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ. 1997;315(7109):629–34.
Begg CB, Mazumdar M. Operating characteristics of a rank correlation test for publication bias. Biometrics. 1994;50(4):1088–101.
Duval S, Tweedie R. Trim and fill: a simple funnel-plot-based method of testing and adjusting for publication bias in meta-analysis. Biometrics. 2000;56(2):455–63.
Schünemann H, Brożek J, Guyatt G, Oxman A. GRADE handbook. Grading of Recommendations Assessment, Development and Evaluation, Grade Working Group. 2013.
McMaster University and Evidence Prime. GRADEpro GDT: GRADEpro Guideline Development Tool [Software]. gradepro.org .
Brunetti M, Shemilt I, Pregno S, Vale L, Oxman AD, Lord J, et al. GRADE guidelines: 10. Considering resource use and rating the quality of economic evidence. J Clin Epidemiol. 2013;66(2):140–50.
Guyatt GH, Oxman AD, Kunz R, Brozek J, Alonso-Coello P, Rind D, et al. GRADE guidelines 6. Rating the quality of evidence—imprecision. J Clin Epidemiol. 2011;64(12):1283–93.
Guyatt GH, Oxman AD, Kunz R, Woodcock J, Brozek J, Helfand M, et al. GRADE guidelines: 8. Rating the quality of evidence—indirectness. J Clin Epidemiol. 2011;64(12):1303–10.
Guyatt GH, Oxman AD, Kunz R, Woodcock J, Brozek J, Helfand M, et al. GRADE guidelines: 7. Rating the quality of evidence—inconsistency. J Clin Epidemiol. 2011;64(12):1294–302.
Guyatt GH, Oxman AD, Montori V, Vist G, Kunz R, Brozek J, et al. GRADE guidelines: 5. Rating the quality of evidence—publication bias. J Clin Epidemiol. 2011;64(12):1277–82.
Guyatt GH, Oxman AD, Santesso N, Helfand M, Vist G, Kunz R, et al. GRADE guidelines: 12. Preparing summary of findings tables-binary outcomes. J Clin Epidemiol. 2013;66(2):158–72.
Guyatt GH, Oxman AD, Sultan S, Glasziou P, Akl EA, Alonso-Coello P, et al. GRADE guidelines: 9. Rating up the quality of evidence. J Clin Epidemiol. 2011;64(12):1311–6.
Guyatt GH, Oxman AD, Vist G, Kunz R, Brozek J, Alonso-Coello P, et al. GRADE guidelines: 4. Rating the quality of evidence—study limitations (risk of bias). J Clin Epidemiol. 2011;64(4):407–15.
Guyatt GH, Thorlund K, Oxman AD, Walter SD, Patrick D, Furukawa TA, et al. GRADE guidelines: 13. Preparing summary of findings tables and evidence profiles-continuous outcomes. J Clin Epidemiol. 2013;66(2):173–83.
Kaminski-Hartenthaler A, Gartlehner G, Kien C, Meerpohl JJ, Langer G, Perleth M, et al. GRADE-Leitlinien: 11. Gesamtbeurteilung des Vertrauens in Effektschätzer für einen einzelnen Studienendpunkt und für alle Endpunkte. Zeitschrift für Evidenz, Fortbildung und Qualität im Gesundheitswesen. 2013;107(9):638–45.
Langendam M, Carrasco-Labra A, Santesso N, Mustafa RA, Brignardello-Petersen R, Ventresca M, et al. Improving GRADE evidence tables part 2: a systematic survey of explanatory notes shows more guidance is needed. J Clin Epidemiol. 2016;74:19–27.
Santesso N, Carrasco-Labra A, Langendam M, Brignardello-Petersen R, Mustafa RA, Heus P, et al. Improving GRADE evidence tables part 3: detailed guidance for explanatory footnotes supports creating and understanding GRADE certainty in the evidence judgments. J Clin Epidemiol. 2016;74:28–39.
Santesso N, Glenton C, Dahm P, Garner P, Akl EA, Alper B, et al. GRADE guidelines 26: informative statements to communicate the findings of systematic reviews of interventions. J Clin Epidemiol. 2020;119:126–35.
Balshem H, Helfand M, Schünemann HJ, Oxman AD, Kunz R, Brozek J, et al. GRADE guidelines: 3. Rating the quality of evidence. J Clin Epidemiol. 2011;64(4):401–6.
Schünemann HJ, Higgins JPT, Vist GE, Glasziou P, Akl EA, Skoetz N, Guyatt GH, Group, Cochrane GRADEing Methods and Group, the Cochrane Statistical Methods. Chapter 14: completing ‘summary of findings’ tables and grading the certainty of the evidence. Cochrane handbook for systematic reviews of interventions. 2019. p. 375–402.
Azadbakht L, Nurbakhsh S. Effect of soy drink replacement in a weight reducing diet on anthropometric values and blood pressure among overweight and obese female youths. Asia Pac J Clin Nutr. 2011;20(3):383–9.
PubMed CAS Google Scholar
Beavers KM, Serra MC, Beavers DP, Cooke MB, Willoughby DS. Soymilk supplementation does not alter plasma markers of inflammation and oxidative stress in postmenopausal women. Nutr Res. 2009;29(9):616–22.
Bricarello LP, Kasinski N, Bertolami MC, Faludi A, Pinto LA, Relvas WG, et al. Comparison between the effects of soy milk and non-fat cow milk on lipid profile and lipid peroxidation in patients with primary hypercholesterolemia. Nutrition. 2004;20(2):200–4.
Faghih S, Hedayati M, Abadi A, Kimiagar M. Comparison of the effects of cow’s milk, fortified soy milk, and calcium supplement on plasma adipocytokines in overweight and obese women. Iranian Journal of Endocrinology and Metabolism. 2009;11(6):692–8.
Gardner CD, Messina M, Kiazand A, Morris JL, Franke AA. Effect of two types of soy milk and dairy milk on plasma lipids in hypercholesterolemic adults: a randomized trial. J Am Coll Nutr. 2007;26(6):669–77.
Hasanpour A, Babajafari S, Mazloomi SM, Shams M. The effects of soymilk plus probiotics supplementation on cardiovascular risk factors in patients with type 2 diabetes mellitus: a randomized clinical trial. BMC Endocr Disord. 2023;23(1):36.
Keshavarz SA, Nourieh Z, Attar MJ, Azadbakht L. Effect of soymilk consumption on waist circumference and cardiovascular risks among overweight and obese female adults. Int J Prev Med. 2012;3(11):798–805.
PubMed PubMed Central Google Scholar
Mohammad-Shahi M, Mowla K, Haidari F, Zarei M, Choghakhori R. Soy milk consumption, markers of inflammation and oxidative stress in women with rheumatoid arthritis: a randomised cross-over clinical trial. Nutr Diet. 2016;73(2):139–45.
Miraghajani MS, Esmaillzadeh A, Najafabadi MM, Mirlohi M, Azadbakht L. Soy milk consumption, inflammation, coagulation, and oxidative stress among type 2 diabetic patients with nephropathy. Diabetes Care. 2012;35(10):1981–5.
Mitchell JH, Collins AR. Effects of a soy milk supplement on plasma cholesterol levels and oxidative DNA damage in men—a pilot study. Eur J Nutr. 1999;38(3):143–8.
Nourieh Z, Keshavarz SA, Attar MJH, Azadbakht L. Effects of soy milk consumption on inflammatory markers and lipid profiles among non-menopausal overweight and obese female adults. Int J Prev Med. 2012;3:798.
Onning G, Akesson B, Oste R, Lundquist I. Effects of consumption of oat milk, soya milk, or cow’s milk on plasma lipids and antioxidative capacity in healthy subjects. Ann Nutr Metab. 1998;42(4):211–20.
Rivas M, Garay RP, Escanero JF, Cia P Jr, Cia P, Alda JO. Soy milk lowers blood pressure in men and women with mild to moderate essential hypertension. J Nutr. 2002;132(7):1900–2.
Ryan-Borchers TA, Park JS, Chew BP, McGuire MK, Fournier LR, Beerman KA. Soy isoflavones modulate immune function in healthy postmenopausal women. Am J Clin Nutr. 2006;83(5):1118–25.
Sirtori CR, Pazzucconi F, Colombo L, Battistin P, Bondioli A, Descheemaeker K. Double-blind study of the addition of high-protein soya milk v. cows’ milk to the diet of patients with severe hypercholesterolaemia and resistance to or intolerance of statins. Br J Nutr. 1999;82(2):91–6.
Sirtori CR, Bosisio R, Pazzucconi F, Bondioli A, Gatti E, Lovati MR, et al. Soy milk with a high glycitein content does not reduce low-density lipoprotein cholesterolemia in type II hypercholesterolemic patients. Ann Nutr Metab. 2002;46(2):88–92.
Steele M. Effect on serum cholesterol levels of substituting milk with a soya beverage. Aust J Nutr Diet. 1992;49(1):24–8.
Summary of Health Canada’s assessment of a health claim about soy protein and cholesterol lowering Ottawa: Health Canada; 2015 [Available from: https://www.canada.ca/en/health-canada/services/food-nutrition/food-labelling/health-claims/assessments/summary-assessment-health-claim-about-protein-cholesterol-lowering.html .
Food and Drug Administration. Food labeling health claims; soy protein and coronary heart disease. Fed Regist. 1999;64:57699–733.
Food and Drug Administration. Food labeling health claims; soy protein and coronary heart disease. Fed Regist. 2017;82:50324–46.
Blanco Mejia S, Messina M, Li SS, Viguiliouk E, Chiavaroli L, Khan TA, et al. A meta-analysis of 46 studies identified by the FDA demonstrates that soy protein decreases circulating LDL and total cholesterol concentrations in adults. J Nutr. 2019;149(6):968–81.
Jenkins DJA, Blanco Mejia S, Chiavaroli L, Viguiliouk E, Li SS, Kendall CWC, et al. Cumulative meta-analysis of the soy effect over time. J Am Heart Assoc. 2019;8(13):e012458.
Mosallanezhad Z, Mahmoodi M, Ranjbar S, Hosseini R, Clark CCT, Carson-Chahhoud K, et al. Soy intake is associated with lowering blood pressure in adults: a systematic review and meta-analysis of randomized double-blind placebo-controlled trials. Complement Ther Med. 2021;59:102692.
Viguiliouk E, Glenn AJ, Nishi SK, Chiavaroli L, Seider M, Khan T, et al. Associations between dietary pulses alone or with other legumes and cardiometabolic disease outcomes: an umbrella review and updated systematic review and meta-analysis of prospective cohort studies. Adv Nutr. 2019;10(Suppl_4):S308–19.
Sohouli MH, Lari A, Fatahi S, Shidfar F, Găman M-A, Guimaraes NS, et al. Impact of soy milk consumption on cardiometabolic risk factors: a systematic review and meta-analysis of randomized controlled trials. Journal of Functional Foods. 2021;83:104499.
Neuenschwander M, Stadelmaier J, Eble J, Grummich K, Szczerba E, Kiesswetter E, et al. Substitution of animal-based with plant-based foods on cardiometabolic health and all-cause mortality: a systematic review and meta-analysis of prospective studies. BMC Med. 2023;21(1):404.
Chiavaroli L, Nishi SK, Khan TA, Braunstein CR, Glenn AJ, Mejia SB, et al. Portfolio dietary pattern and cardiovascular disease: a systematic review and meta-analysis of controlled trials. Prog Cardiovasc Dis. 2018;61(1):43–53.
Viguiliouk E, Kendall CW, Kahleova H, Rahelic D, Salas-Salvado J, Choo VL, et al. Effect of vegetarian dietary patterns on cardiometabolic risk factors in diabetes: a systematic review and meta-analysis of randomized controlled trials. Clin Nutr. 2019;38(3):1133–45.
Glenn AJ, Guasch-Ferre M, Malik VS, Kendall CWC, Manson JE, Rimm EB, et al. Portfolio diet score and risk of cardiovascular disease: findings from 3 prospective cohort studies. Circulation. 2023;148(22):1750–63.
Glenn AJ, Lo K, Jenkins DJA, Boucher BA, Hanley AJ, Kendall CWC, et al. Relationship between a plant-based dietary portfolio and risk of cardiovascular disease: findings from the Women’s Health Initiative prospective cohort study. J Am Heart Assoc. 2021;10(16): e021515.
Lo K, Glenn AJ, Yeung S, Kendall CWC, Sievenpiper JL, Jenkins DJA, Woo J. Prospective association of the portfolio diet with all-cause and cause-specific mortality risk in the Mr. OS and Ms. OS study. Nutrients. 2021;13(12):4360. https://doi.org/10.3390/nu13124360 .
Jenkins DJ, Mirrahimi A, Srichaikul K, Berryman CE, Wang L, Carleton A, et al. Soy protein reduces serum cholesterol by both intrinsic and food displacement mechanisms. J Nutr. 2010;140(12):2302S-S2311.
Ramdath DD, Padhi EM, Sarfaraz S, Renwick S, Duncan AM. Beyond the cholesterol-lowering effect of soy protein: a review of the effects of dietary soy and its constituents on risk factors for cardiovascular disease. Nutrients. 2017;9(4):324. https://doi.org/10.3390/nu9040324 .
Download references
Aspects of this work were presented at the following conferences: Canadian Nutrition Society (CNS), Quebec City, Canada, May 4–6, 2023; 40th International Symposium on Diabetes and Nutrition, Pula, Croatia, June 15–18, 2023; and Nutrition 2023—American Society for Nutrition (ASN), Boston, USA, July 22–25, 2023.
@Toronto_3D_Unit.
This work was supported by the United Soybean Board (the United States Department of Agriculture Soybean Checkoff Program [funding reference number, 2411–108-0101]) and the Canadian Institutes of Health Research (funding reference number, 129920) through the Canada-wide Human Nutrition Trialists’ Network (NTN). The Diet, Digestive tract, and Disease (3D) Centre, funded through the Canada Foundation for Innovation and the Ministry of Research and Innovation’s Ontario Research Fund, provided the infrastructure for the conduct of this work. ME was funded by a CIHR Canada Graduate Scholarship and Toronto 3D PhD Scholarship award. DG was funded by an Ontario Graduate Scholarship. TAK and AZ were funded by a Toronto 3D Postdoctoral Fellowship Award. LC was funded by a Toronto 3D New Investigator Award. SA-C was funded by a CIHR Canadian Graduate Scholarship. DJAJ was funded by the Government of Canada through the Canada Research Chair Endowment. None of the sponsors had any role in study design; in the collection, analysis, and interpretation of data; in the writing of the report; and in the decision to submit the article for publication. But one of the co-authors, Mark Messina, who was involved in all aspects of the study except data collection or analysis, is the Director of Nutrition Science and Research at the Soy Nutrition Institute Global, an organization that receives partial funding from the principal funder, the United Soybean Board (USB).
Authors and affiliations.
Department of Nutritional Sciences, Temerty Faculty of Medicine, University of Toronto, Toronto, ON, Canada
M. N. Erlich, D. Ghidanac, S. Blanco Mejia, T. A. Khan, L. Chiavaroli, A. Zurbau, S. Ayoub-Charette, L. A. Leiter, R. P. Bazinet, D. J. A. Jenkins, C. W. C. Kendall & J. L. Sievenpiper
Toronto 3D Knowledge Synthesis and Clinical Trials Unit, Clinical Nutrition and Risk Factor Modification Centre, St. Michael’s Hospital, Toronto, ON, Canada
M. N. Erlich, D. Ghidanac, S. Blanco Mejia, T. A. Khan, L. Chiavaroli, A. Zurbau, S. Ayoub-Charette, L. A. Leiter, D. J. A. Jenkins, C. W. C. Kendall & J. L. Sievenpiper
Li Ka Shing Knowledge Institute, St. Michael’s Hospital, Toronto, ON, Canada
L. Chiavaroli, L. A. Leiter, D. J. A. Jenkins & J. L. Sievenpiper
Royal College of Surgeons in Ireland, Dublin, Ireland
Soy Nutrition Institute Global, Washington, DC, USA
Department of Medicine, Temerty Faculty of Medicine, University of Toronto, Toronto, ON, Canada
L. A. Leiter, D. J. A. Jenkins & J. L. Sievenpiper
Division of Endocrinology and Metabolism, Department of Medicine, St. Michael’s Hospital, Toronto, ON, Canada
College of Pharmacy and Nutrition, University of Saskatchewan, Saskatoon, SK, Canada
C. W. C. Kendall
You can also search for this author in PubMed Google Scholar
The authors’ responsibilities were as follows: JLS designed the research (conception, development of overall research plan, and study oversight); ME and DG acquired the data; ME, SBM, TAK, and SAC performed the data analysis; JLS, ME, DG, SBM, AA, TAK, and LC interpreted the data; JLS and ME drafted the manuscript, have primary responsibility for the final content, and take responsibility for the integrity of the data and accuracy of the data analysis; JLS, MNE, DG, SBM, TAK, LC, AZ, SAC, AA, MM, LAL, RPB, CWCK, and DJD contributed to the project conception and critical revision of the manuscript for important intellectual content and read and approved the final version of the manuscript. The corresponding author attests that all listed authors meet the authorship criteria and that no others meeting the criteria have been omitted. All authors read and approved the final manuscript.
Correspondence to J. L. Sievenpiper .
Ethics approval and consent to participate.
Not applicable.
Competing interests.
TAK reports receiving grants from Institute for the Advancement of Food and Nutrition Sciences (IAFNS, formerly ILSI North America) and National Honey Board (USDA Checkoff program). He has received honorariums from Advancement of Food and Nutrition Sciences (IAFNS), the International Food Information Council (IFIC), the Calorie Control Council (CCC), the International Sweeteners Association (ISA), and AmCham Dubai. He has received funding from the Toronto 3D Knowledge Synthesis and Clinical Trials foundation. LC has received research support from the Canadian Institutes of health Research (CIHR), Protein Industries Canada (a Government of Canada Global Innovation Clusters), The United Soybean Board (USDA soy “Checkoff” program), and the Alberta Pulse Growers Association. AZ is a part-time research associate at INQUIS Clinical Research, Ltd., a contract research organization. She has received consulting fees from Glycemic Index Foundation Inc. SA-C has received an honorarium from the International Food Information Council (IFIC) for a talk on artificial sweeteners, the gut microbiome, and the risk for diabetes. MM was employed by the Soy Nutrition Institute Global, an organization that receives funding from the United Soybean Board (USB) and from members involved in the soy industry. RPB has received industrial grants, including those matched by the Canadian government, and/or travel support or consulting fees largely related to work on brain fatty acid metabolism or nutrition from Arctic Nutrition, Bunge Ltd., Dairy Farmers of Canada, DSM, Fonterra Inc, Mead Johnson, Natures Crops International, Nestec Inc. Pharmavite, Sancero Inc., and Spore Wellness Inc. Moreover, Dr. Bazinet is on the executive of the International Society for the Study of Fatty Acids and Lipids and held a meeting on behalf of Fatty Acids and Cell Signaling, both of which rely on corporate sponsorship. Dr. Bazinet has given expert testimony in relation to supplements and the brain. DJAJ has received research grants from Saskatchewan & Alberta Pulse Growers Associations, the Agricultural Bioproducts Innovation Program through the Pulse Research Network, the Advanced Foods and Material Network, Loblaw Companies Ltd., Unilever Canada and Netherlands, Barilla, the Almond Board of California, Agriculture and Agri-food Canada, Pulse Canada, Kellogg’s Company, Canada, Quaker Oats, Canada, Procter & Gamble Technical Centre Ltd., Bayer Consumer Care, Springfield, NJ, Pepsi/Quaker, International Nut & Dried Fruit Council (INC), Soy Foods Association of North America, the Coca-Cola Company (investigator initiated, unrestricted grant), Solae, Haine Celestial, the Sanitarium Company, Orafti, the International Tree Nut Council Nutrition Research and Education Foundation, the Peanut Institute, Soy Nutrition Institute (SNI), the Canola and Flax Councils of Canada, the Calorie Control Council, the Canadian Institutes of Health Research (CIHR), the Canada Foundation for Innovation (CFI), and the Ontario Research Fund (ORF). He has received in-kind supplies for trials as a research support from the Almond board of California, Walnut Council of California, the Peanut Institute, Barilla, Unilever, Unico, Primo, Loblaw Companies, Quaker (Pepsico), Pristine Gourmet, Bunge Limited, Kellogg Canada, and WhiteWave Foods. He has been on the speaker’s panel, served on the scientific advisory board and/or received travel support and/or honoraria from Lawson Centre Nutrition Digital Series, Nutritional Fundamentals for Health (NFH)-Nutramedica, Saint Barnabas Medical Center, The University of Chicago, 2020 China Glycemic Index (GI) International Conference, Atlantic Pain Conference, Academy of Life Long Learning, the Almond Board of California, Canadian Agriculture Policy Institute, Loblaw Companies Ltd, the Griffin Hospital (for the development of the NuVal scoring system), the Coca-Cola Company, Epicure, Danone, Diet Quality Photo Navigation (DQPN), Better Therapeutics (FareWell), Verywell, True Health Initiative (THI), Heali AI Corp, Institute of Food Technologists (IFT), Soy Nutrition Institute (SNI), Herbalife Nutrition Institute (HNI), Saskatchewan & Alberta Pulse Growers Associations, Sanitarium Company, Orafti, the International Tree Nut Council Nutrition Research and Education Foundation, the Peanut Institute, Herbalife International, Pacific Health Laboratories, Barilla, Metagenics, Bayer Consumer Care, Unilever Canada and Netherlands, Solae, Kellogg, Quaker Oats, Procter & Gamble, Abbott Laboratories, Dean Foods, the California Strawberry Commission, Haine Celestial, PepsiCo, the Alpro Foundation, Pioneer Hi-Bred International, DuPont Nutrition and Health, Spherix Consulting and WhiteWave Foods, the Advanced Foods and Material Network, the Canola and Flax Councils of Canada, Agri-Culture and Agri-Food Canada, the Canadian Agri-Food Policy Institute, Pulse Canada, the Soy Foods Association of North America, the Nutrition Foundation of Italy (NFI), Nutra-Source Diagnostics, the McDougall Program, the Toronto Knowledge Translation Group (St. Michael’s Hospital), the Canadian College of Naturopathic Medicine, The Hospital for Sick Children, the Canadian Nutrition Society (CNS), the American Society of Nutrition (ASN), Arizona State University, Paolo Sorbini Foundation, and the Institute of Nutrition, Metabolism and Diabetes. He received an honorarium from the United States Department of Agriculture to present the 2013 W.O. Atwater Memorial Lecture. He received the 2013 Award for Excellence in Research from the International Nut and Dried Fruit Council. He received funding and travel support from the Canadian Society of Endocrinology and Metabolism to produce mini cases for the Canadian Diabetes Association (CDA). He is a member of the International Carbohydrate Quality Consortium (ICQC). His wife, Alexandra L Jenkins, is a director and partner of INQUIS Clinical Research for the Food Industry, his 2 daughters, Wendy Jenkins and Amy Jenkins, have published a vegetarian book that promotes the use of the foods described here, The Portfolio Diet for Cardiovascular Risk Reduction (Academic Press/Elsevier 2020 ISBN:978–0-12–810510-8), and his sister, Caroline Brydson, received funding through a grant from the St. Michael’s Hospital Foundation to develop a cookbook for one of his studies. He is also a vegan. CWCK has received grants or research support from the Advanced Food Materials Network, Agriculture and Agri-Foods Canada (AAFC), Almond Board of California, Barilla, Canadian Institutes of Health Research (CIHR), Canola Council of Canada, International Nut and Dried Fruit Council, International Tree Nut Council Research and Education Foundation, Loblaw Brands Ltd, the Peanut Institute, Pulse Canada, and Unilever. He has received in-kind research support from the Almond Board of California, Barilla, California Walnut Commission, Kellogg Canada, Loblaw Companies, Nutrartis, Quaker (PepsiCo), the Peanut Institute, Primo, Unico, Unilever, and WhiteWave Foods/Danone. He has received travel support and/or honoraria from the Barilla, California Walnut Commission, Canola Council of Canada, General Mills, International Nut and Dried Fruit Council, International Pasta Organization, Lantmannen, Loblaw Brands Ltd, Nutrition Foundation of Italy, Oldways Preservation Trust, Paramount Farms, the Peanut Institute, Pulse Canada, Sun-Maid, Tate & Lyle, Unilever, and White Wave Foods/Danone. He has served on the scientific advisory board for the International Tree Nut Council, International Pasta Organization, McCormick Science Institute, and Oldways Preservation Trust. He is a founding member of the International Carbohydrate Quality Consortium (ICQC), Executive Board Member of the Diabetes and Nutrition Study Group (DNSG) of the European Association for the Study of Diabetes (EASD), is on the Clinical Practice Guidelines Expert Committee for Nutrition Therapy of the EASD, and is a Director of the Toronto 3D Knowledge Synthesis and Clinical Trials foundation. JLS has received research support from the Canadian Foundation for Innovation, Ontario Research Fund, Province of Ontario Ministry of Research and Innovation and Science, Canadian Institutes of health Research (CIHR), Diabetes Canada, American Society for Nutrition (ASN), National Honey Board (U.S. Department of Agriculture [USDA] honey “Checkoff” program), Institute for the Advancement of Food and Nutrition Sciences (IAFNS), Pulse Canada, Quaker Oats Center of Excellence, INC International Nut and Dried Fruit Council Foundation, The United Soybean Board (USDA soy “Checkoff” program), Protein Industries Canada (a Government of Canada Global Innovation Cluster), Almond Board of California, European Fruit Juice Association, The Tate and Lyle Nutritional Research Fund at the University of Toronto, The Glycemic Control and Cardiovascular Disease in Type 2 Diabetes Fund at the University of Toronto (a fund established by the Alberta Pulse Growers), The Plant Protein Fund at the University of Toronto (a fund which has received contributions from IFF among other donors), The Plant Milk Fund at the University of Toronto (a fund established by the Karuna Foundation through Vegan Grants), and The Nutrition Trialists Network Fund at the University of Toronto (a fund established by donations from the Calorie Control Council and Physicians Committee for Responsible Medicine). He has received food donations to support randomized controlled trials from the Almond Board of California, California Walnut Commission, Danone, Nutrartis, Soylent, and Dairy Farmers of Canada. He has received travel support, speaker fees and/or honoraria from Danone, FoodMinds LLC, Nestlé, Abbott, General Mills, Nutrition Communications, International Food Information Council (IFIC), Arab Beverages, International Sweeteners Association, Association Calorie Control Council, and Phynova. He has or has had ad hoc consulting arrangements with Perkins Coie LLP, Tate & Lyle, Ingredion, and Brightseed. He is on the Clinical Practice Guidelines Expert Committees of Diabetes Canada, European Association for the study of Diabetes (EASD), Canadian Cardiovascular Society (CCS), and Obesity Canada/Canadian Association of Bariatric Physicians and Surgeons. He serves as an unpaid member of the Board of Trustees of IAFNS. He is a Director at Large of the Canadian Nutrition Society (CNS), founding member of the International Carbohydrate Quality Consortium (ICQC), Executive Board Member of the Diabetes and Nutrition Study Group (DNSG) of the EASD, and Director of the Toronto 3D Knowledge Synthesis and Clinical Trials foundation. His spouse is an employee of AB InBev. All other authors declare that they have no competing interests.
Publisher’s note.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
12916_2024_3524_moesm1_esm.docx.
Additional file 1: This file contains Additional file 1 material, including the PRISMA checklist, further details on the search process, and additional results.
Flow of literature on the effect of substituting soymilk for cow’s milk on intermediate cardiometabolic outcomes. Exclusion criteria: duplicate, abstract only (conference abstract), non-human (animal study), in vitro, review/position paper/commentary/letter, observational (observational study), no soymilk (intervention was not soymilk), children (participants < 18 years of age), no suitable comparator (comparator was not cow’s milk), isolated soy protein (an ISP powder was given to participants), acute (follow-up of < 3 weeks), combined intervention (effects of intervention and comparator could not be isolated), wrong endpoint (no data for outcomes of interest), alternative publication (repeated data from original publication)
A summary plot for the effect of substituting soymilk for cow’s milk on intermediate cardiometabolic outcomes. Analyses were conducted using generic, inverse variance random-effects models (at least 5 trials available), or fixed-effects models (fewer than 5 trials available). Between-study heterogeneity was assessed by the Cochrane Q statistic, where P Q < 0.100 was considered statistically significant, and quantified by the I 2 statistic, where I 2 ≥ 50% was considered evidence of substantial heterogeneity. The GRADE of randomized controlled trials are rated as “high” certainty of evidence and can be downgraded by 5 domains and upgraded by 1 domain. The white squares represent no downgrades, the filled black squares indicate a single downgrade or upgrades for each outcome, and the black square with a white “2” indicates a double downgrade for each outcome. Because all included trials were randomized or nonrandomized controlled trials, the certainty of the evidence was graded as high for all outcomes by default and then downgraded or upgraded based on prespecified criteria. Criteria for downgrades included risk of bias (downgraded if most trials were considered to be at high ROB); inconsistency (downgraded if there was substantial unexplained heterogeneity: I 2 ≥ 50%; P Q < 0.10); indirectness (downgraded if there were factors absent or present relating to the participants, interventions, or outcomes that limited the generalizability of the results); imprecision (downgraded if the 95% CI crossed the minimally important difference (MID) for harm or benefit); and publication bias (downgraded if there was evidence of publication bias based on the funnel plot asymmetry and/or significant Egger or Begg test ( P < 0.10)), with confirmation by adjustment using the trim-and-fill analysis of Duval and Tweedie. The criteria for upgrades included a significant dose–response gradient. For the interpretation of the magnitude, we used the MIDs to assess the importance of magnitude of our point estimate using the effect size categories according to the new GRADE guidance. Then, we used the MIDs to assess the importance of the magnitude of our point estimates using the effect size categories according to the GRADE guidance as follows: a large effect (≥ 5 × MID); moderate effect (≥ 2 × MID); small important effect (≥ 1 × MID); and trivial/unimportant effect (< 1 MID). *HDL-C values reversed to show benefit. **LDL-C was not downgraded for imprecision, as the degree to which the upper 95% CI crosses the MID is not clinically meaningful. Additionally, the moderate change in non-HDL-C, with high certainty of evidence, substantiates the high certainty of the LDL-C results.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Reprints and permissions
Cite this article.
Erlich, M.N., Ghidanac, D., Blanco Mejia, S. et al. A systematic review and meta-analysis of randomized trials of substituting soymilk for cow’s milk and intermediate cardiometabolic outcomes: understanding the impact of dairy alternatives in the transition to plant-based diets on cardiometabolic health. BMC Med 22 , 336 (2024). https://doi.org/10.1186/s12916-024-03524-7
Download citation
Received : 20 December 2023
Accepted : 09 July 2024
Published : 22 August 2024
DOI : https://doi.org/10.1186/s12916-024-03524-7
Anyone you share the following link with will be able to read this content:
Sorry, a shareable link is not currently available for this article.
Provided by the Springer Nature SharedIt content-sharing initiative
ISSN: 1741-7015
An interview with dario corradini..
Interdisciplinary science has quickly become a new norm in physics, and cross-pollination between subjects is leading to better questions and more innovative ideas. It’s fitting, then, that the new chief editor of one of physics’ broadest journals seeks to expand Physical Review E into the interstices between physics, biology, chemistry, and computer science.
We spoke with Dario Corradini about the nature of the scientific review process, the increasing internationality of the journal’s leadership and readership, and more.
You’ve been Chief Editor of Physical Review E since April. What’s in store for the journal under your leadership?
We want to expand further into a few topics that we already cover, including soft matter, machine learning and artificial intelligence, and biophysics. To attract more papers in these areas, we are putting in place different strategies tailored to each topic.
We’ve also started thinking about revamping the image of the journal. We’ve served our audiences well over the years, but we want to be more nimble and able to follow the most recent trends in research, so as to be valuable for the next generations of researchers.
Lastly, we want to get even broader in the geographical representation of not just our authors, but our editorial board — especially across Asia. I’ve already contacted several people at different career levels from different parts of the globe who have agreed to serve on our editorial board next year.
You’ve worked and studied in several different countries. Does that inform your role as chief editor?
Yes, definitely. I started in Rome for my Ph.D., then spent two years in Boston, three in Paris, then came back to the U.S. for a position as associate editor of Physical Review X , which I held for nine years. All this gave me lots of colleagues in both the U.S. and Europe, and within the statistical physics community, which is one of the biggest communities served by Physical Review E . Having lived and worked in different countries has also made me appreciate that science is a truly universal endeavor transcending language and cultural barriers.
What was it like transitioning from research into publishing?
When you do research, you’re used to working on a very specific problem or set of problems. You know everything about your project — it’s like your baby, and you’re very focused. But when you become an editor, your horizons expand a lot. You’re exposed to so much more physics. It's definitely daunting at first.
But over time, I developed a sense that my role now is not to understand every single technical detail. For that, I rely on my referees, the technical experts. As an editor, I have to understand if the paper is a good fit for the journal. Not just to support the journal, but to help researchers reach the best audience for their work. There’s a psychology element to this role, too. I have to know what the referees are saying, but I also have to figure out what they aren’t saying. It’s really an investigative job.
Speaking of expanding your horizons, Physical Review E has one of the broadest scopes in the Physical Review portfolio. Why is that valuable to physics research?
It’s true. Among the Physical Review journals, Physical Review E is arguably the broadest in scope. We could have different journals for each niche topic, but physics is ultimately “one.” Many of the concepts used in a certain area can be translated to other contexts, but we all rely on the same physical principles.
I would actually go so far as to say that science is one. Biologists, chemists, mathematicians, and engineers may have different approaches, but they are not alien to each other. Lots of the most interesting research these days is at the interfaces of these different fields. At Physical Review E , we encourage submissions on physics-adjacent research for this reason. If a variety of tangentially related topics are published in the same journal, our readers can more clearly see the connections between fields, and ultimately ask better questions.
If you could chat with each author who submits to Physical Review E , what would you say?
First of all, thank you. Thank you for trusting us with your research. I would also ask for your help. I can sense that there is a crisis with referees. There are not enough researchers offering their time as referees, and peer review only works when people are willing to review papers. I think it’s really important for younger people to act as referees, so it’s equally important for established researchers to train and mentor younger scientists to be good referees. Part of being in the scientific community means offering your time as an expert referee, because all of your papers are published thanks to others who volunteered their time.
We ask authors to suggest referees when they submit, and we really mean it. But we also ask that they avoid suggesting the usual suspects. The big names in the field can only give so much of the time that’s asked of them. Outside of the top experts, there are many researchers who make great referees, including younger scientists.
Cypress Hansen is a science writer in the San Diego area.
If you embrace scientific discovery, truth and integrity, partnership, inclusion, and lifelong curiosity, this is your professional home.
Helpful Links
Navigate OHIO
Connect With Us
Ohio University Libraries has purchased Sage Research Methods, a platform that includes textbooks, foundation research guidelines, data sets, code books, peer-reviewed case studies and more with updates through 2029.
If members of the OHIO community are looking to explore a new research methodology, hoping to reduce textbook costs or needing a case study for a course, Sage Research Methods can help.
The platform boasts more than 500 downloadable datasets, with code books and instructional materials that provide step by step walk-throughs of the data analysis. The platform also includes quantitative data sets that come with software guides that can assist in understanding tools like SPSS, R, Stata, and Python.
Ohio University now has access to more than 1,000 book titles (including the quantitative social sciences "little green books") that support a variety of research methodologies and approaches from beginner to expert. The OHIO package also includes peer-reviewed case studies with accompanying discussion questions and multiple-choice quiz questions and can be embedded into Canvas courses. Further, the collection includes a Diversifying and Decolonizing Research subcollection that highlights the importance of inclusive research, perspectives from marginalized populations and cultures, and minimizing bias in data analysis.
Highlighted features within Sage Research Methods include:
Learn more about Sage Research Methods.
University Libraries strives to support the OHIO community in and out of the classroom by supporting varying pedagogic approaches and finding ways to make learning more affordable for our students. Further, the Libraries aims to provide access and discoverability to research materials to support Ohio University’s innovative research enterprise. Purchasing Sage Research Methods supports both initiatives as this resource can be used by all students, faculty and staff at Ohio University for research support and instructors for course materials.
Students, faculty and staff Interested in learning more about any of the resources mentioned above are encouraged to reach out to Head of Learning Services and Education Librarian Dr. Chris Guder , Head of Research Services and Health Sciences Librarian Hanna Schmillen or a subject librarian.
Be sure to explore Sage Research Methods on your own; the platform can be accessed through Ohio University Libraries . In addition, there are training sessions and videos from Sage on its training website.
Johannes Schneider-Thoma Roles: Conceptualization, Methodology, Writing – Review & Editing Shimeng Dong Roles: Methodology, Project Administration, Writing – Original Draft Preparation Orestis Efthimiou Roles: Formal Analysis Spyridon Siafis Roles: Conceptualization, Methodology Wulf Peter Hansen Roles: Resources Elfriede Scheuring Roles: Resources Karl Heinz Möhrmann Roles: Resources Stefan Leucht Roles: Conceptualization, Supervision
Sexual dysfunctions are common yet underreported side effects of antipsychotics for schizophrenia, affecting 30-80% of treated individuals. These side effects can severely impact social interactions and treatment adherence for individuals with schizophrenia, but comprehensive comparative evidence assessing the risk profiles of different antipsychotics is lacking. This study aims to address this gap using network meta-analysis that integrates data from both randomized-controlled trials (RCTs) and non-randomized studies (NRS).
This systematic review will include both RCTs and NRS focusing on participants with schizophrenia or schizophrenia-like psychoses, without restrictions on symptoms, gender, ethnicity, age, or setting. For interventions, all second-generation antipsychotics will be included. The primary outcome will be the occurrence of at least one sexual adverse event of any kind. Secondary outcomes will be the occurrence of any sexual adverse event evaluated in men and women separately, and any adverse event related to the three phases of sexual response cycle separately: desire (e.g. libido, sexual thoughts), arousal (e.g. erection, lubrication) and orgasm (e.g. ejaculation, anorgasmia), and any adverse effect related to breast dysfunction and menstruation irregularities. Study selection and data extraction will be performed independently by two reviewers. The Cochrane Risk of Bias tool 1 and ROBINS-I will be employed to evaluate the risk of bias for RCTs and NRS, respectively. Single-arm meta-analysis of proportions will synthesize the average frequency of sexual adverse events in treated participants. Pairwise and network meta-analysis of RCTs and NRS will be used to evaluate comparative tolerability. Subgroup and sensitivity analyses will explore possible heterogeneity in results and validate the findings’ robustness. The quality of the evidence will be evaluated using GRADE.
This study will provide vital insights into the sexual side effects of antipsychotics by combining evidence from clinical trials and real-world practice, facilitating better decision-making in choosing the optimal antipsychotic for individuals.
Sexual side effects, antipsychotics, schizophrenia, meta-analysis
Schizophrenia is a prevalent severe mental illness with worldwide distribution, affecting approximately 1% of the population during their lifetime due to its start during early adulthood ( McGrath et al. 2008 ). Antipsychotics, which are critical for both acute management and prevention of relapse in schizophrenia ( DGPPN e.V. for the Guideline Group 2019 ), are often prescribed over long periods, potentially lifelong. These medications, however, are associated with various side effects, including sexual dysfunctions.
Sexual dysfunctions induced by antipsychotics can manifest as disturbances in sexual desire, erection and ejaculation, vaginal lubrication, and orgasmic dysfunctions as well as partly related disorders of the menstruation cycle and the breast (such as gynecomastia and galactorrhea) ( Kelly and Conley 2004 ; La Torre et al. 2013 ; Montejo et al. 2018 ). These dysfunctions are not only common —mostly reported in 30-80% of treated individuals with prevalence rates varying from 0 to over 90% ( La Torre et al. 2013 ) —but also highly distressing and a frequent cause of non-adherence to treatment ( Perkins 2002 ; Lambert et al. 2004 ). Non-adherence significantly elevates the risk of relapse of psychotic symptoms. Moreover, sexual side effects critically interfere with normal participation in social life in terms of having close and satisfying personal relations in a romantic partnership, which is one of the most important unmet needs of individuals with schizophrenia ( Jager and McCann 2017 ). Therefore, sexual side effects severely diminish the quality of life for those affected ( Bebbington et al. 2009 ; Olfson et al. 2005 ).
Despite the significant clinical impact of sexual side effects induced by antipsychotics, there is a lack of comprehensive meta-analyses addressing this critical issue, particularly no network meta-analyses presenting differences between antipsychotics in this regard. Existing reviews include several narrative reviews and some pairwise meta-analyses (mainly Cochrane reviews) that only focused on specific antipsychotics and invested sexual side effects as secondary outcomes (risperidone ( Hunter et al. 2003 ; Jayaram and Hosalli 2005 ; Komossa et al. 2011 ), sertindole ( Komossa et al. 2009 ; Lewis et al. 2005 ), paliperidone ( Harrington and English 2010 ), or amisulpride ( Men et al. 2018 )). Moreover, some meta-analyses only included observational studies ( Zhao et al. 2020 ; Korchia et al. 2023 ) or had a small number of studies ( Trinchieri et al. 2021 ). One single-arm meta-analysis combined both randomized and observational data and calculated overall percentages of sexual dysfunctions with each antipsychotic across 34 studies ( Serretti and Chiesa 2011 ). However, as the authors report themselves, this approach is not suitable to make statements for differences between antipsychotics in propensity to cause sexual side effects. In summary, the existing evidence leaves us with an incomplete and only impressionistic picture which is limited in terms of available trials, number of events and use of inappropriate methods.
This study aims to fill this knowledge gap by providing evidence-based insights on sexual adverse events associated with antipsychotic to guide the selection of the optimal drug for individual needs. Therefore, to summarize according to the PICO(S) scheme, we will conduct a comprehensive network meta-analysis combining data from randomized-controlled trials and real-world observational studies ( S tudy design) to compare all second-generation antipsychotics ( I ntervention) with each other ( C omparator) on their propensity to cause sexual side effects ( O utcome) in patients with schizophrenia ( P opulation).
We report this systematic review and network meta-analysis protocol according to the Preferred Reporting Items for Systematic review and Meta-analysis Protocols (PRISMA-P) checklist, and the PRISMA extension for network meta-analysis ( Hutton et al. 2015 ). The PRISMA-P Checklist can be found in the extended data. This protocol has been registered with PROSPERO (registration number: CRD42024510190) and will be updated with any necessary amendments.
Study designs
We will include randomized controlled trials (RCTs) and non-randomized studies (NRS). RCTs identified with high risk of bias in sequence generation will be considered as quasi-randomized studies and grouped with NRS. The inclusion of NRS is not limited to specific study designs because as stressed by the Cochrane handbook ( Reeves et al. 2022 ), design labels are used very inconsistently and the risk of bias of a certain NRS can be only assessed when the specific study features are known. Accordingly, studies will first be classified by design, followed by a careful assessment of bias risk for each study and studies with critical risk of bias will be excluded from the analysis. We will also exclude studies from mainland China that are not conducted by international pharmaceutical companies or published in international scientific journals due to significant concerns regarding methodological and reporting quality ( Leucht et al. 2022 ). Both open-label and blinded studies will be included; however, open-label and single-blind studies will be excluded in a sensitivity analysis to address potential bias in expectations of sexual side effects. The minimum study duration will be 3 weeks because shorter studies usually do not focus on clinical efficacy and tolerability of antipsychotics but on more experimental research questions. For cross-over studies, only data from the first phase will be used to avoid carry-over effects, which are common in schizophrenia.
Participants
We will include trials in which at least 80% of the participants are diagnosed with schizophrenia or related disorders (such as schizophreniform or schizoaffective disorders) without restrictions in terms of symptoms (acute episode or maintenance phase), gender, ethnicity, age, or setting. These inclusion criteria are adopted because occurrence of side effects can be considered largely independent of psychopathology and they will increase the data availability for these typically underreported outcomes ( Zorzela et al. 2016 ). Of note, we will record potentially important population characteristics for each trial and consider them in the assessments of heterogeneity and transitivity ae well as in subgroup and sensitivity analyses.
Interventions
All second-generation antipsychotics (SGAs), which are predominantly prescribed for schizophrenia in Europe, Japan and the USA, will be included in this study, namely amisulpride, aripiprazole, asenapine, blonanserine, brexpiprazole, cariprazine, clozapine, iloperidone, lumateperone, lurasidone, olanzapine, olanzapine-samidorphan, paliperidone, perospirone, quetiapine, risperidone, sertindole, ziprasidone, zotepine. Only SGAs are included because those were investigated in recent clinical research adhering to standardized procedures. These standards include systematic documentation of adverse events according to protocols like the Good-Clinical-Practice guideline and use of standardized nomenclatures of adverse events such as MedDRA ( International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use 2022 ). Furthermore, reporting of studies involving SGAs typically comply with guidelines like CONSORT for RCTs ( Schulz et al. 2010 ) and STROBE for NRS ( Vandenbroucke et al. 2007 ), ensuring detailed information on study design and outcomes. Moreover, study authors and pharmaceutical companies of these trials are very likely to keep electronic records and are contactable to provide necessary additional information, which is very important for this review. However, we will include first-generation antipsychotics (FGAs), placebo and no treatment when they were used as comparators in RCTs and NRS of SGAs.
We will include all these compounds, when used in monotherapy, in any form of administration (e.g. oral or intramuscular depot). Primarily, different applications of the same drug will be combined because side effects predominantly follow the pharmacodynamic profile of the specific compounds and not its pharmacokinetics, as observed in previous reviews ( Huhn et al. 2019 ; Schneider-Thoma et al. 2022 ), but considered separate interventions in sensitivity analysis. For RCTs, we will only include fixed-dose studies within the target to maximum range according to a recent consensus reached after a two-step Delphi survey among international experts in the treatment of schizophrenia ( McAdam et al. 2023 ); all flexible-dose treatment regimens (as long as they overlap with the target to maximum range) will be included as these allow investigators to titrate doses to optimal levels for individual participants. Similarly, NRS that rely on observed clinical data will be treated as having flexible dose. In sensitivity analyses, we will exclude flexible dose RCTs and NRS in which the applied doses were outside the target to maximum range for some participants, to control for potential effects of extremely low or high doses.
Comparators
In network meta-analysis there is no formal comparator as all interventions will be compared with each other.
Outcome measures
Primary outcome
The primary outcome will be “Any sexual side effect”. We will use the occurrence of at least one sexual adverse event of any kind provided by the original authors, for example from specific questionnaires for sexual side effects. In case the occurrence of any sexual side effect is not explicitly reported, we will use the highest number of participants reported for any specific sexual adverse event, in line with methodologies used in previous reviews ( Serretti and Chiesa 2011 ; Huhn et al. 2019 ; Schneider-Thoma et al. 2022 ).
1. Any sexual adverse event in men and women separately.
2. Any adverse events related to the “desire” phase of sexual response cycle, such as libido decrease, loss of sexual thoughts.
3. Any adverse events related to the “arousal” phase of sexual response cycle, such as erectile dysfunction, vaginal lubrication decrease.
4. Any adverse events related to the “orgasm” phase of sexual response cycle, such as ejaculation dysfunction, anorgasmia.
5. Any adverse related to breast dysfunction, such as gynecomastia, galactorrhea.
6. Any adverse related to menstruation irregularities, such as amenorrhea.
Of note, there is discussion whether breast dysfunction and menstruation irregularities should be considered as sexual side effects because they are not part of the sexual function per se. However, they are frequently mentioned in parallel to dysfunctions of the sexual response cycle, included in some scales for sexual side effects ( Serretti and Chiesa 2011 ; Boer et al. 2014 ) and very bothersome for participants, and therefore we decided to address them as secondary outcomes.
Timing of outcome measurement will be at study endpoint.
Electronic searches
As recommended by the PRISMA harms checklist ( Zorzela et al. 2016 ) and the Cochrane handbook ( Reeves et al. 2022 ), we search for any study that might have reported adverse events in general and not only for studies mentioning specific sexual adverse events in title/abstract because it is impossible to report all adverse events in searchable/indexable parts of publications. For RCTs, we search the Cochrane Schizophrenia Group’s Study-Based Register of trials ( Shokraneh and Adams 2020 ) for published and unpublished reports. Following the methods from Cochrane ( Higgins JPT, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, Welch VA 2022 ) the Information Specialist compiles this register from systematic searches in MEDLINE, Embase, Allied and Complementary Medicine (AMED), Cumulative Index to Nursing and Allied Health Literature (CINAHL), PsycINFO, PubMed, US National Institute of Health Ongoing Trials Register ClinicalTrials.gov , World Health Organization International Clinical Trials Registry Platform ( www.who.int/ictrp ), ProQuest Dissertations and Theses A&I. The register also includes hand searches and conference proceedings and does not place any limitations on language, date, document type or publication status. For NRS, we search multiple electronic databases including ClinicalTrials.gov , Embase, MEDLINE, PsycINFO, Science Citation Index-Expanded, and WHO International Clinical Trials Registry Platform (ICTRP) with no date/time, language, document type, and publication status limitations. The search string contains terms for schizophrenia and the included antipsychotics. The detailed search strategies can be found in the extended data.
Reference lists and other sources
As additional hand searches, we will check the included studies in previously published relevant systematic reviews. Moreover, because adverse events are often underreported, we will contact the corresponding authors of each included study for unpublished information about adverse events.
Using Rayyan ( Ouzzani et al. 2016 ), title and abstracts of identified references are screened in duplicate by two reviewers with regard to the eligibility criteria above. Any disagreements between the two reviewers are solved by discussion. Then, again in duplicate, two reviewers will inspect the full articles of references selected in title/abstract screening for eligibility and for availability of sexual side effects. Any disagreements will be solved by discussion among the two reviewers or with a third, experienced reviewer (JST, SL). If a decision cannot be made, the study authors will be contacted for clarification.
− General information, such as author name, year of publication, treatment arms and sample size.
− Methodology, such as study design, blinding, duration of study, diagnostic criteria used, study population (Intention-to-treat, observed cases) for which adverse events are reported.
− Participant characteristics, such as age, weight, number of men/women, diagnosis details, plasma prolactin level.
− Intervention characteristics, such as doses, form of application, percent co-medication with antidepressants.
− Outcome measures.
Risk of bias will be assessed for each included study by two reviewers in duplicate referring to the Cochrane Collaboration’s risk of bias tools for randomized controlled studies (RoB tool 1) and non-randomized studies (Risk Of Bias In Non-randomized Studies – of Interventions, ROBINS-I). Disagreements in the assessment will be discussed among the two reviewers and, if needed with a third, experienced reviewer (JST, SL). We will exclude NRS judged as carrying an overall critical risk of bias from the primary analysis. RCTs judged at high risk of bias RCTs and NRS judged at serious risk of bias, we will exclude in a sensitivity analysis.
Overview of the step-wise process for data synthesis of randomized and non-randomized data
First, we will conduct frequentist random effects single-arm and pairwise meta-analyses with RCTs and NRS as subgroups to synthesize estimates of overall prevalence and comparative tolerability and to assess heterogeneity. In the next step, we will conduct network meta-analysis of RCTs (including assessment of transitivity and evaluation of consistency). If the network meta-analysis of RCTs is internally consistent, we proceed with comparing the different estimates from RCTs (direct, indirect, mixed evidence) to the estimates of NRS. If there are no indications for systematic differences between RCT and NRS estimates, we proceed with combined network meta-analysis (again including assessment of transitivity and inconsistency).
Of note, if the requirements for network meta-analysis of RCTs or joint network meta-analysis of RCTs and NRS are not met, we will not proceed to the next step and use pairwise meta-analysis or network meta-analysis of RCTs for data synthesis.
Details of synthesis
Estimation procedures
For estimating the proportion of patients experiencing side effects in antipsychotics, we will use the number of participants experiencing sexual adverse events and non-events among pa exposed to antipsychotics or placebo/no treatment. We will meta-analyze the data using generalized linear mixed models ( Schwarzer et al. 2019 ).
For comparative pairwise and network meta-analysis, the number of participants experiencing sexual adverse events will be synthesized using odds ratios (OR) because ORs have better mathematical properties for meta-analysis, particularly in the case of studies with varying prevalence rates ( Doi et al. 2022 ) and because it is the only measure available in case-control studies. If available in the original publication, we will use reported ORs that are adjusted for possible confounders, such as differences in age and sex between the compared groups. If not available, we will calculate ORs based on the number of participants with events and the number of participants assessed (considering that some sexual adverse events only occur in men or women).
For the pairwise meta-analysis we aim to use a random effects meta-analysis model. However, if the data are sparse, i.e. if there are many studies with few or zero events in one or more of their arms, the usual inverse variance model for meta-analysis has limitations. In that case we will use models that can better handle rare events, such as the Mantel-Haenszel model and Bayesian approaches, as per methodological recommendations ( Efthimiou 2018 ).
Network meta-analysis of RCTs will be performed in a frequentist framework using a random effects model. We will assume a common heterogeneity parameter across the various treatment comparisons. For combined network meta-analysis of RCTs and NRS, different several statistical models are available. The selection of the most suitable model will be decided after careful consideration of the actual data, the distribution of studies by designs and the risk of bias assessment ( Efthimiou et al. 2017 ). In case of sparse data, we will explore the use of a Bayesian model or a frequentist model based on the Mantel-Haenszel approach ( Efthimiou et al. 2019 ).
Assessment of heterogeneity
Heterogeneity (variability in relative treatment effects within the same treatment comparison) will be assessed within and across study designs by visual inspection of forest plots and estimating the statistical heterogeneity τ, i.e. the standard deviation of random effects, and I 2 . We will employ empirical distributions to characterize the amount of heterogeneity as low, moderate or high ( Turner et al. 2012 ). Substantial heterogeneity indicates important differences in clinical and methodological characteristics of the studies which warrant further investigation, such as checking for mistakes in data entry and for potential effect modifiers and bias factors. Moreover, to assess how much heterogeneity affects the clinical interpretation of the relative treatment effects with respect to the extra uncertainty anticipated in a future study, we will produce prediction intervals.
Assessment of the transitivity assumption in network meta-analysis
Joint analysis of treatments can be misleading if the network is substantially intransitive. Intransitivity can arise when design, population or treatment characteristics that may modify the relative effects between interventions are distributed differently between comparisons. For the case of relative treatment effects in terms of sexual side effects, there is no clear a-priori-evidence, but several characteristics, may play a role (e.g. study design, blinding, gender, age, dose, antidepressant co-medication). Therefore, we will investigate if relevant characteristics are similarly distributed across studies grouped by comparison.
Assessment of inconsistency
Consistency, i.e. the agreement between direct evidence and indirect evidence of a network meta-analysis, will be statistically evaluated globally, by using the design-by-treatment test ( Higgins et al. 2012 ) and locally, via the back-calculation method ( König et al. 2013 ). In case of evidence of inconsistency, we will investigate possible sources of it (mistakes in data entry, clear differences in study characteristics).
Investigation of heterogeneity and inconsistency
Substantial heterogeneity, intransitivity or inconsistency will prevent network meta-analysis. Small or moderate amounts will be further explored by subgroup, network meta-regression, and sensitivity analyses.
We a priori plan to investigate the impact of following potential effect modifiers via Bayesian network meta-regression analyses of the primary outcome: percentage women, mean age, prolactin level, percent co-medication with antidepressants (which also cause sexual side effects), and duration of study. Additionally, we will perform separate network meta-analyses for sexual adverse events occurring in men and women (see secondary outcome).
Moreover, we will explore the robustness of results (with regard to the inclusion of studies with differences in study design, population and intervention characteristics in the primary analysis) by sensitivity analyses. The following sensitivity analyses of the primary outcomes are predefined: exclusion of (1) non-double-blind studies, (2) studies that report only observed-case analyses, (3) RCTs with high risk/NRS with serious risk of bias, (4) flexible-dose studies in which the range of applied doses exceeded the recommended (target to maximum) dose range ( McAdam et al. 2023 ), (5) studies that did not use specific questionnaires to assess sexual side effects, (6) studies in acutely ill patients (because acute psychosis might interfere with sexual functioning). Moreover, we will perform network meta-analysis with oral and depot applications of the same compound as separate interventions.
Small study effects and publication bias
For assessment of small study effects and publication bias, we will employ a comparison-adjusted funnel plot method to explore the association between study size and effect size ( Chaimani and Salanti 2012 ). Moreover, comparisons with 10 or more studies will be plotted in a contour-enhanced funnel plot ( Peters et al. 2008 ). Similarly, we will plot a contour enhanced funnel plot of all SGAs combined versus placebo.
Assessment of the confidence in estimates
The quality of the evidence of the primary outcome will be evaluated using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) framework extended to NMA ( Puhan et al. 2014 ).
Statistical software
Analyses will be performed in R using the packages “meta” for single-arm and pairwise meta-analysis ( Balduzzi et al. 2019 ), “netmeta” for network meta-analyses ( Rücker et al. 2020 ), “crossnma” for combined network meta-analyses of RCTs and NRS ( Hamza and Salanti 2022 ). Bayesian analyses will be performed using self-programmed routines in “rjags” ( Plummer et al. 2023 ). R software and the mentioned packages are freely available https://cran.r-project.org/bin/windows/base/
Study status: search and the selection process are ongoing currently.
Despite its significant clinical relevance, there is a lack of scientific comparison between different antipsychotics regarding their sexual side effects. This project will address this gap by providing a comprehensive synthesis of evidence from clinical trials and real-world clinical practice. This information is relevant for clinicians and guideline developers in selecting the most appropriate medication for individuals. Additionally, this review is of high importance for future clinical research, regarding both RCTs and NRS, as it will report the current state of evidence concerning sexual side effects of antipsychotics and identify existing limitations. Finally, this review will be among the first to integrate randomized and non-randomized evidence in a network meta-analysis, thereby advancing methodological approaches in evidence-based medicine.
We collaborate with members of the patient organization “BASTA - Bündnis für psychisch erkrankte Menschen” and the relatives’ organization “Landesverband Bayern der Angehörigen psychisch erkrankter Menschen e.V.” in this project. They contributed in identifying the research idea and developing this review protocol from their perspective as people with lived experience with the disease schizophrenia and the treatment with antipsychotics. They will be updated regularly about the state of the project and help with any upcoming questions. Moreover, they will be involved in interpreting the results and in preparing a lay summary of the results so that other patients and relatives of patients can be directly informed about the scientific results with a text that can be understood, e.g. using the BASTA-newsblog ( http://www.bastagegenstigma.de/ ).
This review does not require ethical approval.
SL is the principal investigator, obtained funding, and supervises the study. JST, SD, SS and SL designed the study and provided clinical and methodological advice. JST and SD drafted the manuscript and registered the protocol with PROSPERO before. OE provided substantial methodological and statistical advice. WPH, ES and KHM provided the patient perspective when designing the study. All authors critically reviewed the manuscript for important intellectual content and approved its final version.
Underlying data.
No data associated with this article.
Figshare: Sexual side effects of antipsychotic drugs in schizophrenia: Protocol for a systematic review with single-arm, pairwise and network meta-analysis of randomized controlled trials and non-randomized studies, https://doi.org/10.6084/m9.figshare.26396275.v2 ( Dong, 2024 ).
• PRISMA-p checklist
• Search strategy
Data are available under the terms of the Creative Commons Attribution 4.0 International License (CC-BY 4.0).
We would like to thank Dr Farhad Shokraneh, Systematic Review Consultants LTD, for designing the searches and AR, who wants to stay anonymous, for providing the patient perspective in the design of this review.
Views | Downloads | |
---|---|---|
F1000Research | - | - |
PubMed Central | - | - |
Reviewer status, comments on this article.
All Comments (0)
Provide sufficient details of any financial or non-financial competing interests to enable users to assess whether your comments might lead a reasonable person to question your impartiality. Consider the following examples, but note that this is not an exhaustive list:
Sign up for content alerts and receive a weekly or monthly email with all newly published articles
Register with F1000Research
Already registered? Sign in
Not now, thanks
The email address should be the one you originally registered with F1000.
You registered with F1000 via Google, so we cannot reset your password.
To sign in, please click here .
If you still need help with your Google account password, please click here .
You registered with F1000 via Facebook, so we cannot reset your password.
If you still need help with your Facebook account password, please click here .
If your email address is registered with us, we will email you instructions to reset your password.
If you think you should have received this email but it has not arrived, please check your spam filters and/or contact for further assistance.
BMC Pulmonary Medicine volume 24 , Article number: 412 ( 2024 ) Cite this article
26 Accesses
Metrics details
Ibuprofen is one of the most commonly used analgesic and antipyretic drugs in children. However, its potential causal role in childhood asthma pathogenesis remains uncertain. In this systematic review, we assessed the association between ibuprofen administration in children and the risk of developing or exacerbating asthma.
We searched MEDLINE, Embase, Cochrane Library, CINAHL, Web of Science, and Scopus from inception to May 2022, with no language limits; searched relevant reviews; and performed citation searching. We included studies of any design that were primary empirical peer-reviewed publications, where ibuprofen use in children 0–18 years was reported. Screening was performed in duplicate by blinded review. In total, 24 studies met our criteria. Data were extracted according to PRISMA guidelines, and the risk of bias was assessed using RoB2 and NOS tools. Quantitative data were pooled using fixed effect models, and qualitative data were pooled using narrative synthesis. Primary outcomes were asthma or asthma-like symptoms. The results were grouped according to population (general, asthmatic, and ibuprofen-hypersensitive), comparator type (active and non-active) and follow-up duration (short- and long-term).
Comparing ibuprofen with active comparators, there was no evidence of a higher risk associated with ibuprofen over both the short and long term in either the general or asthmatic population. Comparing ibuprofen use with no active alternative over a short-term follow-up, ibuprofen may provide protection against asthma-like symptoms in the general population when used to ease symptoms of fever or bronchiolitis. In contrast, it may cause asthma exacerbation for those with pre-existing asthma. However, in both populations, there were no clear long-term follow-up effects.
Ibuprofen use in children had no elevated risk relative to active comparators. However, use in children with asthma may lead to asthma exacerbation. The results are driven by a very small number of influential studies, and research in several key clinical contexts is limited to single studies. Both clinical trials and observational studies are needed to understand the potential role of ibuprofen in childhood asthma pathogenesis.
Peer Review reports
Asthma is a noncommunicable disease affecting approximately 235 million people worldwide and is characterised by inflammation and narrowing of the small airways in the lungs, leading to any combination of cough, wheeze, shortness of breath, and chest tightness [ 1 ]. The prevalence of asthma has increased in many countries in recent decades, especially among children, making asthma a serious global public health problem [ 2 , 3 ]. The reason for increasing asthma prevalence in children is uncertain, but there is likely a complex interaction of multiple risk factors, including environmental (e.g., increased air pollution, changes to housing conditions) and lifestyle factors (e.g., decreased physical activity, changes in diet, increased childhood obesity) [ 4 ].
Increased early-life use of pharmacological agents, such as analgesics and antipyretics, could be causal factors in childhood asthma pathogenesis. Due to fears of a causal relationship between aspirin use and Reye’s syndrome [ 5 ] and the risk of aspirin-induced asthma [ 6 ], aspirin use in children has dramatically decreased in recent decades. Consequently, drugs such as ibuprofen and paracetamol have become increasingly popular for treating fever and pain in children. In the United Kingdom, the National Health Service describes both paracetamol and ibuprofen as safe for treating pain and high temperature in babies and children [ 7 ]. However, caution is advised for ibuprofen use in children with asthma [ 8 ], while no such warning is supplied for paracetamol [ 9 ], suggesting that ibuprofen may be linked to asthma development or exacerbation in those with pre-existing asthma.
Ibuprofen is a non-steroidal anti-inflammatory drug (NSAID) that is frequently prescribed or administered over-the-counter (OTC) to treat fever and pain. Links between childhood ibuprofen use and asthma development or exacerbation are being investigated [ 10 , 11 , 12 , 13 , 14 , 15 , 16 ]. Ibuprofen’s inhibition of the cyclooxygenase system can lead to activation of the lipoxygenase system, resulting in bronchospasm [ 6 , 17 ], which could precipitate asthma. Additionally, empirical evidence exists demonstrating ibuprofen-induced asthma exacerbation in children with asthma and self-reported aspirin allergy [ 18 ].
Despite these points, two recent systematic reviews did not identify a risk difference between ibuprofen and paracetamol in asthma development or exacerbation in children [ 14 , 16 ]. However, one of these reviews limited the scope to randomised controlled trials (RCTs) [ 14 ], and the other to a relatively narrow age range of less than 2 years [ 16 ], restricting the generalisability of the findings.
We conducted a systematic review to assess the association between ibuprofen administration in children and the risk of developing or exacerbating asthma. The aim was to expand on previous reviews by looking across the entire age range of childhood from 0 to 18 years, including both interventional and observational studies, and assessing the association separately for clinically distinct paediatric subpopulations: general, asthmatic, and ibuprofen-hypersensitive.
We registered our review on PROSPERO on 8 July 2022 (CRD42022344838). The protocol was written according to PRISMA-P guidelines [ 19 , 20 ] and made publicly available on OSF prior to registration with PROSPERO. Further methodological details can be found in our online protocol ( https://doi.org/10.17605/OSF.IO/Z37KW ).
A full list of eligibility criteria is provided in Supplementary Methods S1.1 (Supplementary Tables 1–2). The numeric results from studies included in our review were grouped by population for synthesis: (i) general population of children (i.e., studies not limiting eligibility to specific clinical subpopulations; however, some study-specific exclusion will always occur, for example, children with severe asthma, ibuprofen hypersensitivity, or other contraindications for safety reasons; children with conditions that could interfere with ibuprofen administration or absorption, such as inability to swallow or frequent vomiting; children receiving treatments that could interfere with the outcome assessment, such as leukotriene receptor antagonist and other anti-asthmatic treatments); (ii) children with asthma; and (iii) children with ibuprofen hypersensitivity.
We searched six bibliographic databases (MEDLINE, Embase, Cochrane Library, CINAHL, Web Of Science, Scopus) to identify records on 21-May-2022, and our searches were independently peer-reviewed using the PRESS Checklist [ 21 , 22 ] by an outreach librarian at the Bodleian Health Care Libraries, University of Oxford ( https://doi.org/10.17605/OSF.IO/R3AV6 ). All search strategies are provided in full in Supplementary Methods S1.2. Additional information sources included relevant reviews that were identified during screening [ 10 , 11 , 12 , 13 , 14 , 15 , 16 ] and backwards citation searching using the citationchaser tool [ 23 ]. EPPI-Reviewer [ 24 ] was used for de-duplication, and screening was performed independently in duplicate, with disagreements settled by discussion between both reviewers.
Data extraction and bias assessment were performed by one reviewer and then verified by a second reviewer, with disagreements settled by discussion. Our primary outcomes of interest were asthma, asthma-like symptoms, or asthma exacerbation [ 2 ]. For risk of bias assessment, the Cochrane risk of bias tool (RoB2) was used for RCTs [ 25 ], and the Newcastle-Ottawa Scale (NOS) [ 26 ] was used for observational studies. The results from these assessments were used to decide which studies to include in primary syntheses (Supplementary Figs. 1–2). Our approach to assessing meta-biases (outcome reporting and publication biases) is detailed in Supplementary Methods S1.3.
A narrative synthesis was performed when outcomes were too heterogeneous to synthesise quantitatively. Otherwise, meta-analysis was performed using the R package meta [ 27 ]. Given the sparsity of the data for quantitative synthesis, we report the common effect model results as primary results. For completion, we report additional analysis outputs, e.g., both odds and risk ratios; both common and random effects model effect sizes; I 2 , tau 2 , and chi 2 for heterogeneity. Due to the sparsity of the results, subgroup analyses were not performed.
For meta-analysis of dichotomous data, ORs were pooled using Peto’s method [ 28 ] due to zero events in some arms. Where multiple outcomes from a study were available, the primary analysis was performed by selecting the outcomes with the expected lowest risk of bias. To test the robustness of the primary analysis, sensitivity analyses were performed using alternative combinations of studies’ numeric results.
Of the 820 records screened, 18 relevant studies were identified, with a further 6 from relevant reviews (Supplementary Fig. 3). The study characteristics for all 24 studies are summarised in Table 1 . Relevant numeric results were grouped by population: (i) general population of children (Table 2 ), (ii) children with asthma (Table 3 ), and (iii) children with ibuprofen hypersensitivity (Table 4 ). For the general population and children with asthma, data synthesis was performed for (i) ibuprofen versus an active comparator (Fig. 1 ) and (ii) ibuprofen versus baseline (i.e., children not taking an alternative antipyretic or analgesic). To increase homogeneity, the results were also grouped based on the duration of follow-up, in line with a recent similar systematic review [ 16 ]: short duration of ≤ 28 days or long duration of > 28 days.
Synthesis of results of ibuprofen versus active comparators. The active comparator for Kokki 2010 was ketoprofen; for all other studies, the active comparator was paracetamol. ( a ) General population of children over a short duration. ( b ) Children with asthma over a long duration. Abbreviations: OR = odds ratio; 95% CI = 95% confidence interval
In total, 13 numeric results from 9 studies relevant to assessing ibuprofen use in a general population of children were identified [ 29 , 30 , 31 , 32 , 33 , 34 , 35 , 36 , 37 ] (Table 2 ).
There were six results from six interventional studies (all RCTs) and two results from one observational cohort study that compared ibuprofen use with an active comparator in the general population. The main active comparator was paracetamol, with one study [ 29 ] using ketoprofen (Table 2 ). The durations of study for the interventional RCT were all short (≤ 28 days). Two of these results were from publications based on the same dataset, the Boston University Fever Study [ 30 , 31 ], of which the original publication was selected for primary analysis.
The synthesis of five results comparing ibuprofen with active comparators (four paracetamol, one ketoprofen) resulted in a common effect OR = 0.87; 95% CI=[0.55, 1.37], demonstrating a lack of significant difference between ibuprofen and active comparators (Fig. 1 a). Our sensitivity analyses were in agreement with this primary result (Supplementary Fig. 4).
A single observational study [ 36 ] assessed ibuprofen relative to paracetamol over both short and long durations (Table 2 ) in a general population of children. Over a short duration (14 days), no significant difference in wheezing was identified, but over a long duration (1 year), they observed a significant advantage to ibuprofen over paracetamol, with a reduction in health care practitioner visits for wheezing illness consistent with bronchiolitis or asthma.
Taken together, these interventional and observational results suggest that there is no difference between ibuprofen and active comparators in the general population over a short duration (≤ 28 days). This finding is driven largely by a single study, the Boston University Fever Study [ 31 ], conducted almost 30 years ago on a large sample ( n = 83,915) of children aged 6 months to 12 years. Over longer follow-up durations of one year, there is evidence from only a single cohort study [ 36 ] to suggest that there may be a reduction in wheezing when ibuprofen is prescribed, rather than paracetamol, for a first episode of bronchiolitis in children aged 0–12 months.
Five numeric results from three studies relevant to assessing ibuprofen relative to baseline (children not taking an alternative antipyretic or analgesic) in the general population were identified (Table 2 ). All outcomes were from observational studies. Due to the sparsity and substantive heterogeneity of the results, quantitative synthesis was not possible.
Two studies looked at general populations over short durations (≤ 28 days) [ 33 , 36 ]. Both studies suggest that ibuprofen might decrease wheezing when taken for either acute febrile illness or bronchiolitis (Table 2 ).
Two studies looked at general populations of children over long durations [ 35 , 36 ] and produced conflicting results. One study [ 36 ] compared those prescribed ibuprofen for a first episode of bronchiolitis to those not prescribed ibuprofen (or another drug) and followed up participants over a 1-year duration, observing a positive impact of ibuprofen prescription. The second study [ 35 ] compared children administered ibuprofen to those not administered ibuprofen during the first postnatal year and followed-up participants at a 3–5 year duration, observing a negative impact of ibuprofen on asthma development, and at a 7–10 year duration, observing no difference between cohorts (Table 2 ).
Taken together, ibuprofen use in the general population of children during acute febrile illness or bronchiolitis might decrease wheezing when assessed in the short-term (≤ 28 days), with both observational studies reporting strong significant effects (Table 2 ). Over longer durations, the two observational studies identified in this review have substantive heterogeneity in design, analysis, and outcome, preventing meaningful synthesis. Additionally, their numeric findings are inconsistent (Table 2 ).
Five numeric results from four studies relevant to assessing ibuprofen in asthmatic paediatric populations were identified [ 38 , 39 , 40 , 41 ] (Table 3 ).
Three results across three studies compared ibuprofen with an active comparator (paracetamol in all cases) in asthmatic populations (Table 3 ). One interventional study assessed outcomes over a short duration [ 39 ] and found no difference between treatments. While further analyses in this paper did suggest a favourable outcome for ibuprofen relative to paracetamol, the results from this second post-hoc Boston Fever Study report are at very high risk of bias in the selection of the reported result (Supplementary Fig. 2).
Two studies looked at the comparison between ibuprofen and paracetamol in asthmatic populations over long durations [ 38 , 41 ]. The RCT study [ 41 ] identified no difference between drugs (OR = 0.90 [ 0.57, 1.41]). In contrast, the observational cohort study [ 38 ] identified a significant disadvantage for ibuprofen relative to paracetamol in asthmatic populations (aOR = 2.10 [1.17, 3.76]). These conflicting results for ibuprofen relative to paracetamol in asthmatic populations over long durations are challenging to resolve due to the different experimental designs. However, there are also several similarities in their designs: use of the same active comparator, inclusion of asthmatic populations of children with similar age ranges (Sheehan: 1–4.9 years; Fu: 1–5 years) over similar follow-up durations (Sheehan: 46 weeks; Fu: 52 weeks), and use of asthma exacerbation as the outcome. As an exploratory analysis, we synthesised these results, which resulted in a common effect OR = 1.24; 95% CI=[0.87, 1.77], suggesting an overall non-significant effect, which is consistent with the RCT study result alone (Fig. 1 b).
Taken together, these interventional and observational results suggest that there is no difference in asthma exacerbation between ibuprofen and paracetamol in asthmatic populations over short or long durations.
Only a single study looked at an asthmatic population over both short and long durations [ 40 ]. Over a short duration, this study found that ibuprofen increased asthma exacerbation. Over a long duration, they found no effect of ibuprofen on asthma exacerbation in the asthmatic population.
Four drug provocation studies were identified that studied ibuprofen-hypersensitive children where ibuprofen was ingested and adverse events reported as part of hypersensitivity diagnosis [ 42 , 43 , 44 , 45 ]. A range of respiratory adverse effects were reported that included asthma, coughing, wheezing, dyspnoea, and respiratory distress (Table 4 ). Across the four studies, there was a total of 10 children with respiratory adverse events reported in a total of 80 children. Thus, in children with ibuprofen hypersensitivity, the average rate of respiratory adverse events following ibuprofen ingestion was 12.5%.
Seven studies were identified that reported the relationship between ibuprofen and asthma in children, which were not synthesised in this review [ 18 , 46 , 47 , 48 , 49 , 50 , 51 ]: five studies reported on single cases, and two group analysis studies had substantive differences in methodology and outcomes relative to other studies included in this review.
One crossover RCT [ 46 ] assessed the prevalence of ibuprofen-sensitive asthma in children with mild or moderate persistent asthma using bronchoprovocation challenge and found a prevalence of 2%. Another non-randomised controlled study [ 18 ] assessed the impact of short-term ibuprofen treatment on pulmonary function in children with mild to moderate stable asthma and self-reported aspirin allergy. Relative to a healthy control group, the asthmatic group exhibited a drop in FEV1 (forced expiratory volume in the first second) of 18.85% and an increase in FeNO (fractional exhaled nitric oxide) of 20.76 ppb. A summary of the results from these two studies is provided in Supplementary Table 3.
Four case reports of severe adverse events to ibuprofen were identified [ 47 , 48 , 50 , 51 ], and in all cases, the children had pre-existing asthma. Last, in a case series of fatal asthma in Finland, a single death due to ibuprofen ingestion was reported in a child with severe asthma and a known allergy to ibuprofen [ 49 ].
Here, we assessed the association between ibuprofen use and asthma in children aged 0–18 years. Both observational and interventional studies were reviewed in the general population as well as the asthmatic population. Studies that benchmarked ibuprofen against an active comparator almost exclusively used paracetamol, and in both populations of children, the combined evidence suggested no difference in asthma-related adverse events between ibuprofen and paracetamol (or ketoprofen) use. A single observational study suggested a potential benefit of ibuprofen over paracetamol prescription in response to bronchiolitis in the general paediatric population after a one-year follow-up. When ibuprofen use was assessed relative to no alternative drug administration, differences emerged between the general and asthmatic populations. In the short-term follow-up (1–14 days) to ibuprofen use, two observational studies reported favourable effects in the general population, while one observational and one interventional study observed unfavourable effects in the asthmatic population. Over a longer follow-up period (12 weeks to 10 years), no clear effect emerged for either population.
The majority of research on the association between ibuprofen use and asthma-related adverse events in children has been conducted in the general population, benchmarked relative to paracetamol, and participants followed-up over a short duration [ 29 , 30 , 31 , 32 , 34 , 36 , 37 ]. The aggregate result from five RCTs conducted in this context is driven primarily by the Boston University Fever Study [ 31 ], conducted almost 30 years ago on children aged 6 months to 12 years. While a single observational study [ 36 ] conducted five years ago corroborates this finding, research is sparse. Furthermore, only a single study comparing ibuprofen with paracetamol use with a short-term follow-up was conducted in children with asthma [ 39 ], and this study was a second post-hoc analysis publication of the same Boston University Fever Study dataset. Given the increased vulnerability of the asthmatic population to respiratory adverse events from ibuprofen use that was observed in our review, there is a clear lack of research comparing the short-term effects of ibuprofen relative to alternative analgesics and antipyretics such as paracetamol in children with asthma.
Two studies [ 38 , 41 ] assessing differences between ibuprofen and paracetamol use over longer follow-up periods in asthmatic populations report conflicting results. Due to several study similarities, we tentatively synthesised the two results, and no aggregate difference between ibuprofen and paracetamol was observed. However, in the RCT [ 41 ], the median dose of trial medication (ibuprofen or paracetamol) was 5.5 doses (IQR = 1–15) and matched between trial arms. In the retrospective cohort study [ 38 ], it could not be determined by the original investigators whether patients took the medication prescribed. Additionally, the observational study did not control for upper respiratory tract infections, a well-documented source of confounding by indication [ 35 , 52 ], which were not well-matched between the ibuprofen and paracetamol cohorts. For these reasons, the RCT finding alone or the synthesised outcome of no difference between drugs seems most justifiable.
Comparing the asthmatic and general populations for short-term asthma-relevant outcomes after ibuprofen use, no conflicts in results were observed. The two observational studies in the general population [ 33 , 36 ] both observed reductions in asthma-related outcomes, while one observational [ 40 ] and one interventional [ 18 ] study in the asthmatic population both observed increases in asthma-related outcomes. These findings highlight the importance of avoiding naïve pooling of results from studies in these different paediatric populations.
It is noteworthy that all RCTs reviewed compared ibuprofen with an active comparator. Of the studies comparing ibuprofen with a baseline of no alternative drug, three were cohort studies [ 35 , 36 , 40 ], and one was cross-sectional [ 33 ]. One non-randomised interventional study [ 18 ] compared an asthmatic sample with a healthy control sample. This highlights one of the limitations of the RCT design approach in assessing adverse events in the youngest children [ 53 , 54 ]. As a recent RCT feasibility study found [ 55 ], almost three quarters of parents surveyed described the use of a placebo comparator treatment as unacceptable for treating their child’s fever or pain. This ethical unacceptability of using a placebo arm in clinical trials for treating pain and fever in young children [ 55 , 56 ] introduces an ambiguity into these active comparator RCT studies, as a lack of difference among active comparators does not exclude the possibility that both ibuprofen and active comparator use may be associated with parallel increases in asthma exacerbations [ 41 , 56 ]. It has been argued that, given that ibuprofen and paracetamol have different mechanisms of action, it is unlikely that their use could be associated with similar increases in the rate of asthma-related complications that are known to be determined by disparate mechanisms of disease [ 41 , 56 ]. However, this speculation requires careful examination and empirical support. Observational studies with comparator groups in which an active treatment was not prescribed or taken can be used as a baseline control to assess the impact of ibuprofen alone, acknowledging the challenges of inferring causality in observational studies. It is these advantages and disadvantages of both RCTs and observational designs that require a review of the association between ibuprofen use and asthma-related outcomes in children to consider and attempt to synthesise all study design types. This feature of our review adds substantially to two recent systematic reviews in this area [ 14 , 56 ] that either limited the study designs to RCTs [ 14 ] or limited the population to those under 2 years [ 56 ].
We identified four drug provocation trials in which ibuprofen hypersensitivity was confirmed in children by controlled administration of ibuprofen [ 42 , 43 , 44 , 45 ] and respiratory adverse events were recorded. The average percentage of children with confirmed ibuprofen hypersensitivity who displayed respiratory adverse events was 12.5%. Relative to other adverse events, such as angio-oedema and urticaria (which were by far the most common adverse events), asthma and asthma-like respiratory events were less commonly reported. While adverse respiratory reactions to ibuprofen ingestion in those with ibuprofen hypersensitivity can be quite severe, as reported in a handful of case reports [ 47 , 48 , 50 , 51 ], fatalities appear to be very rare. In this review, only a single case of ibuprofen-induced asthma fatality was identified [ 49 ].
The number of studies in this review that were relevant to important clinical populations and contexts was unfortunately sparse. Only a single publication was identified for each of the following three contexts: the general population where ibuprofen is compared with an active comparator with a follow-up duration longer than 1 month [ 36 ]; the asthmatic population where ibuprofen is compared with an active comparator with a short-term follow-up [ 39 ]; and the asthmatic population where ibuprofen is compared with a baseline of no active comparator with a follow-up duration longer than 1 month [ 40 ]. These limitations hinder the generalisability of findings to several important clinical contexts and are an ongoing issue to be addressed.
Here, we found that research is most lacking for populations of children with pre-existing asthma, who are the population at most risk for potential respiratory adverse events following ibuprofen use. Our review highlights the importance of assessing both interventional and observational studies and analysing the general population and asthmatic population separately. Continued investigation into the role of early-life ibuprofen use and its short-term and long-term impact on childhood asthma is needed.
All data (data collection form, risk of bias assessment forms, and data used for all analyses) are publicly available on the project’s OSF site: https://doi.org/10.17605/OSF.IO/ZBDS7 . All code used for the meta-analysis is publicly available on Zenodo: https://doi.org/10.5281/zenodo.11258287 .
WHO. Asthma [Internet]. 2023 [cited 2023 Apr 4]. https://www.who.int/news-room/fact-sheets/detail/asthma
Reddel HK, Bacharier LB, Bateman ED, Brightling CE, Brusselle GG, Buhl R et al. Global Initiative for Asthma Strategy 2021: executive summary and rationale for key changes. European Respiratory Journal [Internet]. 2022 Jan 1 [cited 2022 Apr 27];59(1). https://erj.ersjournals.com/content/59/1/2102730
Vos T, Murray CJL, GBD 2016 Disease and Injury Incidence and Prevalence Collaborators. Global, regional, and national incidence, prevalence, and years lived with disability for 328 diseases and injuries for 195 countries, 1990–2016: a systematic analysis for the global burden of Disease Study 2016. Lancet. 2017;390(10100):1211–59.
Article Google Scholar
Beasley R, Semprini A, Mitchell EA. Risk factors for asthma: is prevention possible? Lancet. 2015;386(9998):1075–85.
Article PubMed Google Scholar
Schrör K. Aspirin and Reye syndrome: a review of the evidence. Paediatr Drugs. 2007;9(3):195–204.
Szczeklik A, Stevenson DD. Aspirin-induced asthma: advances in pathogenesis and management. J Allergy Clin Immunol. 1999;104(1):5–13.
Article PubMed CAS Google Scholar
NHS. nhs.uk. 2020 [cited 2023 Apr 4]. Medicines for babies and children. https://www.nhs.uk/conditions/baby/health/medicines-for-babies-and-children/
NHS. nhs.uk. 2022 [cited 2023 Apr 4]. Who can and cannot take ibuprofen for children. https://www.nhs.uk/medicines/ibuprofen-for-children/who-can-and-cannot-take-ibuprofen-for-children/
NHS. nhs.uk. 2022 [cited 2023 Apr 4]. Who can and cannot take paracetamol for children. https://www.nhs.uk/medicines/paracetamol-for-children/who-can-and-cannot-take-paracetamol-for-children/
Kanabar D, Dale S, Rawat M. A review of ibuprofen and acetaminophen use in febrile children and the occurrence of asthma-related symptoms. Clin Ther. 2007;29(12):2716–23.
Kanabar DJ. A clinical and safety review of Paracetamol and Ibuprofen in children. Inflammopharmacology. 2017;25(1):1–9.
Article PubMed PubMed Central CAS Google Scholar
Kauffman RE, Lieh-Lai M. Ibuprofen and increased morbidity in children with asthma: fact or fiction? Paediatr Drugs. 2004;6(5):267–72.
Pierce CA, Voss B. Efficacy and safety of ibuprofen and acetaminophen in children and adults: a meta-analysis and qualitative review. Ann Pharmacother. 2010;44(3):489–506.
Sherbash M, Furuya-Kanamori L, Nader JD, Thalib L. Risk of wheezing and asthma exacerbation in children treated with Paracetamol versus Ibuprofen: a systematic review and meta-analysis of randomised controlled trials. BMC Pulm Med. 2020;20(1):72.
Southey ER, Soares-Weiser K, Kleijnen J. Systematic review and meta-analysis of the clinical safety and tolerability of ibuprofen compared with paracetamol in paediatric pain and fever. Current medical research and opinion [Internet]. 2009 Sep [cited 2022 May 19];25(9). https://pubmed.ncbi.nlm.nih.gov/19606950/
Tan E, Braithwaite I, McKinlay CJD, Dalziel SR. Comparison of Acetaminophen (paracetamol) with Ibuprofen for Treatment of Fever or Pain in children younger than 2 years: a systematic review and Meta-analysis. JAMA Netw Open. 2020;3(10):e2022398.
Article PubMed PubMed Central Google Scholar
Szczeklik A. The cyclooxygenase theory of aspirin-induced asthma. Eur Respir J. 1990;3(5):588–93.
Su YM, Huang CS, Wan KS. Short-term ibuprofen treatment and pulmonary function in children with asthma. Indian Pediatr. 2015;52(8):691–3.
Moher D, Shamseer L, Clarke M, Ghersi D, Liberati A, Petticrew M, et al. Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015 statement. Syst Reviews. 2015;4(1):1.
Shamseer L, Moher D, Clarke M, Ghersi D, Liberati A, Petticrew M, et al. Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015: elaboration and explanation. BMJ. 2015;349:g7647.
McGowan J, Sampson M, Salzwedel DM, Cogo E, Foerster V, Lefebvre C. PRESS peer review of electronic search strategies: 2015 Guideline Statement. J Clin Epidemiol. 2016;75:40–6.
McGowan J, Sampson M, Salzwedel DM, Cogo E, Lefebvre C, PRESS Peer Review of Electronic Search Strategies. : 2015 Guideline Explanation and Elaboration (PRESS E&E). CADTH Methods and Guidelines [Internet]. 2016 [cited 2022 Mar 3]; https://www.cadth.ca/press-peer-review-electronic-search-strategies-2015-guideline-explanation-and-elaboration
Haddaway NR, Grainger MJ, Gray CT. citationchaser: An R package and Shiny app for forward and backward citations chasing in academic searching [Internet]. Zenodo; 2021 [cited 2023 Jan 17]. https://zenodo.org/record/4543513
Thomas J, Graziosi S, Brunton J, Ghouze Z, O’Driscoll P, Bond M et al. EPPI-Reviewer: advanced software for systematic reviews, maps and evidence synthesis [Internet]. EPPI-Centre, UCL Social Research Institute, University College London; 2022. http://eppi.ioe.ac.uk/cms/Default.aspx?tabid=2914
Sterne JAC, Savović J, Page MJ, Elbers RG, Blencowe NS, Boutron I, et al. RoB 2: a revised tool for assessing risk of bias in randomised trials. BMJ. 2019;366:l4898.
Wells G, Shea B, O’Connell D, Peterson J, Welch V, Losos M et al. The Newcastle–Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses [Internet]. 2008 [cited 2022 May 23]. http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp
Balduzzi S, Rücker G, Schwarzer G. How to perform a meta-analysis with R: a practical tutorial. Evid Based Ment Health. 2019;22(4):153–60.
Yusuf S, Peto R, Lewis J, Collins R, Sleight P. Beta blockade during and after myocardial infarction: an overview of the randomized trials. Prog Cardiovasc Dis. 1985;27(5):335–71.
Kokki H, Kokki M. Ketoprofen versus Paracetamol (acetaminophen) or ibuprofen in the management of fever: results of two randomized, double-blind, double-dummy, parallel-group, repeated-dose, multicentre, phase III studies in children. Clin Drug Investig. 2010;30(6):375–86.
Lesko SM, Mitchell AA. The safety of acetaminophen and ibuprofen among children younger than two years old. Pediatrics. 1999;104(4):e39.
Lesko SM, Mitchell AA. An assessment of the safety of pediatric ibuprofen. A practitioner-based randomized clinical trial. JAMA. 1995;273(12):929–33.
Luo S, Ran M, Luo Q, Shu M, Guo Q, Zhu Y, et al. Alternating Acetaminophen and Ibuprofen versus monotherapies in improvements of distress and reducing refractory fever in Febrile children: a Randomized Controlled Trial. Paediatr Drugs. 2017;19(5):479–86.
Matok I, Elizur A, Perlman A, Ganor S, Levine H, Kozer E. Association of Acetaminophen and Ibuprofen Use with Wheezing in Children with Acute Febrile illness. Ann Pharmacother. 2017;51(3):239–44.
McIntyre J, Hull D. Comparing efficacy and tolerability of Ibuprofen and Paracetamol in fever. Arch Dis Child. 1996;74(2):164–7.
Sordillo JE, Scirica CV, Rifas-Shiman SL, Gillman MW, Bunyavanich S, Camargo CA, et al. Prenatal and infant exposure to acetaminophen and ibuprofen and the risk for wheeze and asthma in children. J Allergy Clin Immunol. 2015;135(2):441–8.
Walsh P, Rothenberg SJ. Wheezing after the use of acetaminophen and or ibuprofen for first episode of bronchiolitis or respiratory tract infection. PLoS ONE. 2018;13(9):e0203770.
Wong A, Sibbald A, Ferrero F, Plager M, Santolaya ME, Escobar AM, et al. Antipyretic effects of dipyrone versus ibuprofen versus acetaminophen in children: results of a multinational, randomized, modified double-blind study. Clin Pediatr (Phila). 2001;40(6):313–24.
Fu LS, Lin CC, Wei CY, Lin CH, Huang YC. Risk of acute exacerbation between acetaminophen and ibuprofen in children with asthma. PeerJ. 2019;7:e6760.
Lesko SM, Louik C, Vezina RM, Mitchell AA. Asthma morbidity after the short-term use of ibuprofen in children. Pediatrics. 2002;109(2):E20.
Lo PC, Tsai YT, Lin SK, Lai JN. Risk of asthma exacerbation associated with nonsteroidal anti-inflammatory drugs in childhood asthma: a nationwide population-based cohort study in Taiwan. Med (Baltim). 2016;95(41):e5109.
Article CAS Google Scholar
Sheehan WJ, Mauger DT, Paul IM, Moy JN, Boehmer SJ, Szefler SJ, et al. Acetaminophen versus Ibuprofen in Young children with mild persistent asthma. N Engl J Med. 2016;375(7):619–30.
Corzo JL, Zambonino MA, Muñoz C, Mayorga C, Requena G, Urda A, et al. Tolerance to COX-2 inhibitors in children with hypersensitivity to nonsteroidal anti-inflammatory drugs. Br J Dermatol. 2014;170(3):725–9.
Ertoy Karagol HI, Yilmaz O, Topal E, Ceylan A, Bakirtas A. Nonsteroidal anti-inflammatory drugs-exacerbated respiratory disease in adolescents. Int Forum Allergy Rhinol. 2015;5(5):392–8.
Guvenir H, Dibek Misirlioglu E, Vezir E, Toyran M, Ginis T, Civelek E et al. Nonsteroidal anti-inflammatory drug hypersensitivity among children. Allergy Asthma Proc. 2015;36(5):386–93.
Yilmaz Topal O, Kulhas Celik I, Turgay Yagmur I, Toyran M, Civelek E, Karaatmaca B, et al. Results of NSAID provocation tests and difficulties in the classification of children with nonsteroidal anti-inflammatory drug hypersensitivity. Ann Allergy Asthma Immunol. 2020;125(2):202–7.
Debley JS, Carter ER, Gibson RL, Rosenfeld M, Redding GJ. The prevalence of ibuprofen-sensitive asthma in children: a randomized controlled bronchoprovocation challenge study. J Pediatr. 2005;147(2):233–8.
Goraya JS, Virdi VS. To the editor: exacerbation of asthma by ibuprofen in a very young child. Pediatr Pulmonol. 2001;32(3):262.
King G, Byrne A, Fleming P. A case of severe NSAID exacerbated respiratory disease (NERD) following a dental procedure in a child. Eur Arch Paediatr Dent. 2016;17(4):277–81.
Malmström K, Kaila M, Kajosaari M, Syvänen P, Juntunen-Backman K. Fatal asthma in Finnish children and adolescents 1976–1998: validity of death certificates and a clinical description. Pediatr Pulmonol. 2007;42(3):210–5.
Menendez R, Venzor J, Ortiz G. Failure of zafirlukast to prevent ibuprofen-induced anaphylaxis. Ann Allergy Asthma Immunol. 1998;80(3):225–6.
Palmer GM. A teenager with severe asthma exacerbation following ibuprofen. Anaesth Intensive Care. 2005;33(2):261–5.
Schnabel E, Heinrich J. Respiratory tract infections and not Paracetamol medication during infancy are associated with asthma development in childhood. J Allergy Clin Immunol. 2010;126(5):1071–3.
CRD. Chapter 4: systematic reviews of adverse effects. Systematic reviews: CRD’s guidance for undertaking reviews in health care. York Publishing Services; 2009.
Reeves B, Deeks J, Higgins J, Shea B, Tugwell P, Wells G. Chapter 24: Including non-randomized studies on intervention effects. In: Higgins J, Thomas J, Chandler J, Cumpston M, Li T, Page M, editors. Cochrane Handbook for Systematic Reviews of Interventions [Internet]. 6.3. Cochrane; 2022 [cited 2023 Apr 7]. https://training.cochrane.org/handbook/current/chapter-24
Riley J, Braithwaite I, Shirtcliffe P, Caswell-Smith R, Hunt A, Bowden V, et al. Randomized controlled trial of asthma risk with Paracetamol use in infancy–a feasibility study. Clin Exp Allergy. 2015;45(2):448–56.
Tan E, Braithwaite I, McKinlay C, Riley J, Hoare K, Okesene-Gafa K, et al. Randomised controlled trial of Paracetamol or Ibuprofen, as required for fever and pain in the first year of life, for prevention of asthma at age 6 years: Paracetamol or Ibuprofen in the primary prevention of asthma in Tamariki (PIPPA Tamariki) protocol. BMJ Open. 2020;10(12):e038296.
Download references
We thank Imran Lodhi, Fiona Murray-Zmijewski, Frederic Esclassan, and Bill Laughey for reviewing and advising on improvements for this systematic review. We thank Carolyn Smith, the Outreach Librarian at University of Oxford’s Bodleian Libraries, who performed the search strategy PRESS Peer Review.
This work was funded by Reckitt. Employees of Reckitt were involved in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of a manuscript; and decision regarding where to submit the manuscript for publication.
Authors and affiliations.
Department of Paediatrics, University of Oxford, Oxford, UK
Luke Baxter, Maria M. Cobo, Aomesh Bhatt & Rebeccah Slater
Colegio de Ciencias Biologicas y Ambientales, Universidad San Francisco de Quito USFQ, Quito, Ecuador
Maria M. Cobo
Reckitt, Dansom Lane, Hull, HU8 7DS, UK
Olutoba Sanni
Reckitt (Global Headquarters), Turner House, 103-105 Bath Road, Slough, Berkshire, SL1 3UH, UK
Nutan Shinde
You can also search for this author in PubMed Google Scholar
LB: conceptualization, methodology, data curation, formal analysis, investigation, visualization, writing – original draft, writing – review & editing. MC: methodology, data curation, investigation, validation, writing – review & editing. AB: conceptualization, methodology, data curation, investigation, writing – review & editing. RS: conceptualization, methodology, data curation, investigation, writing – review & editing. OS: conceptualization, methodology, project administration, writing – review & editing. NS: conceptualization, methodology, project administration, writing – review & editing.
Correspondence to Luke Baxter .
Ethics approval and consent to participate.
Not Applicable.
Competing interests.
OS and NS are current employees of Reckitt and may hold equity interest in Reckitt. LB and RS were compensated by Reckitt for activities related to execution of the study. MC and AB declare no competing interests. No other disclosures were reported.
Publisher’s note.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Below is the link to the electronic supplementary material.
Rights and permissions.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .
Reprints and permissions
Cite this article.
Baxter, L., Cobo, M.M., Bhatt, A. et al. The association between ibuprofen administration in children and the risk of developing or exacerbating asthma: a systematic review and meta-analysis. BMC Pulm Med 24 , 412 (2024). https://doi.org/10.1186/s12890-024-03179-3
Download citation
Received : 24 October 2023
Accepted : 22 July 2024
Published : 26 August 2024
DOI : https://doi.org/10.1186/s12890-024-03179-3
Anyone you share the following link with will be able to read this content:
Sorry, a shareable link is not currently available for this article.
Provided by the Springer Nature SharedIt content-sharing initiative
ISSN: 1471-2466
APA Style provides a foundation for effective scholarly communication because it helps writers present their ideas in a clear, concise, and inclusive manner. When style works best, ideas flow logically, sources are credited appropriately, and papers are organized predictably. People are described using language that affirms their worth and dignity. Authors plan for ethical compliance and report critical details of their research protocol to allow readers to evaluate findings and other researchers to potentially replicate the studies. Tables and figures present information in an engaging, readable manner.
The style and grammar guidelines pages present information about APA Style as described in the Publication Manual of the American Psychological Association, Seventh Edition and the Concise Guide to APA Style, Seventh Edition . Any updates to APA Style are noted on the applicable topic pages. If you are still using the sixth edition, helpful resources are available in the sixth edition archive .
COMMENTS
Hence, even if one were to confer it a hybrid status wherein it can both prevent and detect bias, the extent of bias that has long been documented in peer-reviewed journals reveals major weaknesses in peer review. Recent high-profile COVID-19 -related retractions 31 and commentary 32 further confirms these weaknesses. Consequently, we need to ...
The subject of this column is the nature of bias in both quantitative and qualitative research. To that end, bias will be defined and then both the processes by which it enters into research will be entertained along with discussions on how to ameliorate this problem.
Underpowered research or questionable methodological or statistical decisions can be identified and addressed through peer-review prior to study conduct. Since journals commit to publication upon review of study plans, rather than finished papers, registered reports may reduce the incentive for authors to 'spin' and for reviewers to request ...
Conceptually, the peer review process can lead to distortion of the results from the viewpoint of the evidence user, akin to bias. Peer review bias can be defined as a violation of impartiality in the evaluation of a submission. We propose that this transgression of neutrality standards can affect the dissemination of research (ie, denying ...
16,646. As of Dec 2022, 16,646 reviewers have completed the training. 91%. 91% of reviewers thought that the training substantially improved their ability to identify bias in peer review. 93%. 93% of reviewers stated the training made them substantially more comfortable intervening against bias. Data for the Jan 2022 Advisory Council cycle.
Abstract. This narrative review provides an overview on the topic of bias as part of Plastic and Reconstructive Surgery 's series of articles on evidence-based medicine. Bias can occur in the planning, data collection, analysis, and publication phases of research. Understanding research bias allows readers to critically and independently review ...
Stories of research funding bodies and journal peer reviewers rejecting proposed qualitative methods or study findings due to "bias" are not uncommon. Usually, I find this relates to a perception by peer reviewers that the way data have/will be collected or analyzed is too closely aligned with the personal agenda of the researcher(s).
Future research is needed to refine our methodology, but our empirically grounded form of bias-adjusted meta-analysis could be implemented as follows: 1.) collate studies for the same true effect ...
With regard to gender bias in publications, data suggest that women are under-represented in peer review and same-gender preference exists strongly for males and females. 9 For racial bias, we found data in National Institutes of Health research awards, which showed that African Americans/Black and Asians are less likely to receive National ...
Reducing and challenging bias in peer review is critically important to ensure the integrity of the process and to help advance equity, diversity, and inclusion in our scientific ... "The big consequences of small biases: A simulation of peer review." Research Policy, 44(6): 1266-1270. [online] Available at: <https://www.sciencedirect.com ...
Understanding Bias in Peer Review. In the 1600's, a series of practices came into being known collectively as the "scientific method.". These practices encoded verifiable experimentation as a path to establishing scientific fact. Scientific literature arose as a mechanism to validate and disseminate findings, and standards of scientific ...
The peer review process can also introduce bias. A compelling ethical and moral rationale necessitates improving the peer review process. A double-blind peer review system is supported on equipoise and fair-play principles. Triple-and quadruple-blind systems have also been described but are not commonly used.
Peer Review as a Source of Bias. Scientific experiments in general, and biomedical research specifically, are subject to bias and confounding. Domains for the different types of bias and confounding have been established as to their role before, during, and after the intervention is delivered. 6 Dedicated tools have been developed to evaluate ...
Plain Language Summary. Quantitative bias analysis methods can be used to evaluate the impact of biases on observational study results. However, little is known about the full range and characteristics of available methods in the peer-reviewed literature that can be used to conduct quantitative bias analysis using information reported in manuscripts and other publicly available sources without ...
Research on bias in peer review examines scholarly communication and funding processes to assess the epistemic and social legitimacy of the mechanisms by which knowledge communities vet and self-regulate their work. Despite vocal concerns, a closer look at the empirical and methodological limitations of research on bias raises questions about ...
Reducing bias in peer review is critically important to ensure the integrity of our editorial process and embrace diversity, equity, and inclusion in our scientific communities. As a practical matter, we will encourage associate editors to take gender, racial, and geographical diversity into consideration when selecting reviewers for submitted ...
Abstract. Scholarly journals are often blamed for a gender gap in publication rates, but it is unclear whether peer review and editorial processes contribute to it. This article examines gender bias in peer review with data for 145 journals in various fields of research, including about 1.7 million authors and 740,000 referees.
This review provides a brief description of the function, history, and scope of peer review; articulates and critiques the con-ception of bias unifying research on bias in peer review; characterizes and examines the empirical, methodologi-cal, and normative claims of bias in peer review research; and assesses possible alternatives to the status ...
The concept of implicit bias, also termed unconscious bias, and the related Implicit Association Test (IAT) rests on the belief that people act on the basis of internalised schemas of which they are unaware and thus can, and often do, engage in discriminatory behaviours without conscious intent.1 This idea increasingly features in public discourse and scholarly inquiry with regard to ...
Blinding can help reduce bias in peer review. In double-blind peer review, the identities of authors and reviewers are concealed from each other. Some journals have even introduced triple-blind peer review, where the authors' identity is also hidden from the journal editors. These types of blinding efforts help reviewers focus on the content ...
Peer review has an important role in improving the quality of research papers. It is the "lifeblood of research in academia […] the social structure that subjects research to the critical assessment of other researchers" (Bourdieu, 1975).This structure relies on self-regulated interactions within the scientific community, in which a journal editor appoints peer reviewers with expertise ...
Background Loss to follow-up in long-term epidemiological studies is well-known and often substantial. Consequently, there is a risk of bias to the results. The motivation to take part in an epidemiological study can change over time, but the ways to minimize loss to follow-up are not well studied. The Citizen Science approach offers researchers to engage in direct discussions with study ...
Objectives To estimate age standardised suicide rate ratios in male and female physicians compared with the general population, and to examine heterogeneity across study results. Design Systematic review and meta-analysis. Data sources Studies published between 1960 and 31 March 2024 were retrieved from Embase, Medline, and PsycINFO. There were no language restrictions. Forward and backwards ...
1. INTRODUCTION. Although expressions of explicit bias have declined in the United States over time, implicit bias has remained unrelenting. Health care providers hold negative explicit and implicit biases against many marginalized groups of people, including racial and ethnic minoritized populations, disabled populations, and gender and sexual minorities, among others (29, 63).
We conducted a systematic review and meta-analysis of 17 trials that examined the effect of substituting soymilk (median dose of 22 g/day or 6.6 g/250 mL serving of soy protein per day and 17.2 g/day or 6.9 g/250 mL of total [added] sugars in the sweetened soymilk) for cow's milk (median dose of 24 g/day or 8.3 g/250 mL of milk protein and 24 g/day or 12 g/250 mL of total sugars [lactose ...
Biologists, chemists, mathematicians, and engineers may have different approaches, but they are not alien to each other. Lots of the most interesting research these days is at the interfaces of these different fields. At Physical Review E, we encourage submissions on physics-adjacent research for this reason. If a variety of tangentially ...
The OHIO package also includes peer-reviewed case studies with accompanying discussion questions and multiple-choice quiz questions and can be embedded into Canvas courses. Further, the collection includes a Diversifying and Decolonizing Research subcollection that highlights the importance of inclusive research, perspectives from marginalized ...
The Cochrane Risk of Bias tool 1 and ROBINS-I will be employed to evaluate the risk of bias for RCTs and NRS, respectively. ... [version 1; peer review: awaiting peer review]. F1000Research 2024, 13:973 (https://doi.org ... They contributed in identifying the research idea and developing this review protocol from their perspective as people ...
Ibuprofen is one of the most commonly used analgesic and antipyretic drugs in children. However, its potential causal role in childhood asthma pathogenesis remains uncertain. In this systematic review, we assessed the association between ibuprofen administration in children and the risk of developing or exacerbating asthma. We searched MEDLINE, Embase, Cochrane Library, CINAHL, Web of Science ...
People are described using language that affirms their worth and dignity. Authors plan for ethical compliance and report critical details of their research protocol to allow readers to evaluate findings and other researchers to potentially replicate the studies. Tables and figures present information in an engaging, readable manner.