Improving the reporting on health equity in observational research (STROBE-Equity): extension checklist and elaboration

“Marginalized populations have approximately 5 times greater risk of adverse COVID-19 consequences related to social determinants of health that may exacerbate patient comorbidities. Equitable distribution would eliminate vaccination disparities while mitigating the disproportionate effect of the COVID-19 pandemic in underserved populations, which are disadvantaged because of limited access to health care, low socioeconomic status, or race.” 74 “Substantial racial and ethnic disparities persist in cancer survival, reflecting systemic barriers to cancer screening, treatment, and survivorship care and differences in the prevalence of comorbidities.” 75 “The importance of non-traditional risk factors for atherosclerotic diseases, such as socioeconomic deprivation, has also been increasingly recognised. Whether socioeconomic deprivation is as strongly associated with other CVDs [cardiovascular diseases] as with atherosclerosis is uncertain, but it is important to understand as many countries have reported an increase in socioeconomic inequalities.” 76 Describing how the aim of health equity relates to the author’s motivation for conducting the study clarifies its purpose and demonstrates its contribution to addressing health inequities. 77 Understanding these motivations helps to contextualise the research design and explains why specific research questions were prioritised. 78 79 We advise authors to emphasise the relevance of the health condition to populations experiencing inequities, using context specific language and acknowledging the role of intersectionality 80 81 82 and systemic causes of health inequities (eg, racism, sexism, heterosexism, ableism, and colonialism). Authors should provide a rationale that elucidates the underlying theory for why the issue is particularly relevant to these populations, indicating where differences reflect inequities. 83 84 Using a logic model to illustrate this theory can further clarify the study’s equity focused objectives. 85 86

This guideline extends the STROBE reporting guidelines for specific attention to reporting health equity relevant methods, data, and considerations in observational studies. We describe the STROBE-Equity reporting guideline checklist and elaborate on each extension item.

The STROBE-Equity extension is designed to guide the reporting of health equity data and considerations in observational studies of any topic that aims to investigate differences in health that are avoidable and considered unjust (see box 1 for definitions of health equity relevant observational studies). The STROBE-Equity extension applies to the three observational study designs addressed in the STROBE statement: cohort, case-control, and cross sectional. It is intended to complement the STROBE statement rather than replace it. Similarly, it can be used alongside the RECORD (REporting of studies Conducted using Observational Routinely collected Data) guidelines. 52 This extension focuses on transparent reporting, not to inform study design or implementation. Given the sensitive nature of health equity research, authors are encouraged to carefully consider how data are reported to avoid reinforcing stigma or discrimination. The guideline does not apply to qualitative studies as they require a different reporting structure. 53 54 A study that does not involve any intervention (experimental or otherwise) on the part of the investigator. 37 Studies that investigate outcomes related to populations experiencing, or that may experience, inequities and/or examine variations in outcomes among groups experiencing different levels of inequities or no inequities at all. In some cases, these observational studies may meet both criteria by focusing on populations experiencing inequities and assessing subgroup differences. Importantly, this definition does not require the studies to explicitly state that the objective of the observational study is linked to health equity. 38 It is intended for studies that address unfair or unjust differences and may not apply to differences that are unavoidable (eg, biological). For example, a study reporting differences in breast cancer incidence between age groups may reflect biological variation, as age is a non-modifiable risk factor. In contrast, a study reporting inequities in breast cancer mortality between ethnic groups due to delays in diagnosis or unequal access to treatment would qualify as health equity relevant. PROGRESS-Plus items can be used to identify factors that are known to socially stratify health opportunities and outcomes. 6 39 PROGRESS stands for: Place of residence, Race or ethnicity or culture or language, Occupation or out of work, Gender or sex, Religion, Education, Socioeconomic status, and Social capital. “Plus” recognises additional factors, including personal characteristics, time dependent situations, and relationships. These characteristics may be associated with each other (eg, educational attainment and socioeconomic status) and may interact with each other (eg, socioeconomic status and ethnicity) or with systems of oppression, power, and privilege, to drive the experience of health inequities (intersectionality). 40 The importance of these characteristics may depend on the context and setting. EQUALSS GUIDE MULTIPLE 41 is another tool that can be used, which includes: ethnicity (and race), Qualifications and education, Underserved area, Age, Language and religion, Sex, Sexual orientation, Gender identification, Underrepresented groups (inclusion groups), Income and wealth, Disability (physical, mental, and learning), Employment and occupation, and Multiple disadvantage. Various terms have been used to describe populations experiencing health inequities—that is, “inclusion health populations,” 42 43 44 “equity-seeking,” “equity-deserving,” “underserved,” “marginalised,” or “disadvantaged” populations. 45 46 In this reporting guideline, the term “populations experiencing inequities” is utilised to acknowledge individuals with firsthand experience of inequity, whether current or historical. This term emphasises inclusivity and recognises that inequities are not inherent to a given population, can evolve over time, and are context dependent. Those with past and/or present experiences offer a comprehensive perspective on accessing care and services. The importance of using terms that best fit the study context and meet the needs of the populations experiencing inequities involved or affected by the study are recognised. Statistics and information related to health inequities as well as other characteristics associated with inequities. 47 The Patient-Centered Outcomes Research Institute’s (PCORI) definition of individuals with lived experience of inequities was adopted, which includes patients, caregivers, clinicians, and other healthcare partners involved in patient care. 48 These are groups with legitimate interests in the health issue under consideration: patients, health research funders, healthcare service purchasers, peer reviewed journal editors, policy makers, research leaders, review creators and commissioners, product developers, programme overseers, care providers, and the general public. 49 The term interest holders provides an alternative to stakeholders, given its colonial roots and extensive consultation with the MuSE Consortium to find a favoured replacement. 50 Individuals who are likely to use the research results to make informed decisions about health policies, programmes, or practices. 51 Knowledge users may originate from different interest holder groups to ensure usefulness and relevance of the research. According to the MuSE Consortium (previously known as Multi-Stakeholder Engagement), 49 engagement can take two forms: shared decision making, where individuals with lived experience of inequities contribute equally to decisions, or providing advice or feedback, where the perspectives of interest holders are sought and considered by the larger research team.

To develop the STROBE-Equity extension, we adhered to a published protocol, 55 informed by the Enhancing the Quality and Transparency of Health Research (EQUATOR) Network’s methodological framework for developing reporting guidelines. 56 We followed two parallel streams: one for the global guidance and another for creating Indigenous related guidance, led by Indigenous scholars. This paper presents the development of the global reporting guideline, covering phases 1-3 as specified in the protocol. Phases 4 and 5, which involve evaluating the outcomes of collaboration between the global and Indigenous streams and outline the knowledge translation plan, will be reported separately. We reported patient and public engagement and the Delphi process according to the Guidance for Reporting Involvement of Patients and the Public 2.0 (GRIPP2) 57 and ACcurate COnsensus Reporting Document (ACCORD) 58 reporting guidelines (see supplementary appendix 1). A diverse, global multidisciplinary team (n=57) developed the STROBE-Equity extension, including members of the public, authors of observational studies, statisticians, social scientists, epidemiologists, methodologists, funders, healthcare practitioners, ethicists, and knowledge users. We sought diversity in our research team and steering committees across geography, genders, ethnicities, and career stages (see supplementary appendix 2). The group included knowledge users from the Pan American Health Organization/World Health Organization, the Royal College of Physicians and Surgeons of Canada, scientific journal editors, the Canadian Drug Agency, and the Public Health Agency of Canada. All the team members were part of the technical oversight committee, which oversaw the whole project. In parallel, we established a knowledge user committee (n=12) and a patient and public steering committee (n=3). Both committees were consulted at all stages of the guideline development process. An executive committee (n=10) comprising the co-principal investigators, trainees, and researchers was responsible for leading the empirical studies. Recognising the Indigenous right to self-determination as outlined in the United Declaration on the Rights of Indigenous peoples (Article 3), 59 we established an Indigenous steering committee (n=6) comprising an Indigenous Elder, researchers, clinicians, and trainees. This Indigenous stream provided a culturally safe way to co-produce ethical and relevant guidance for reporting health equity in Indigenous related health research. We describe our values and positionality statements in supplementary appendix 2. In phase 1 the STROBE-Equity project team proposed 36 extension items to the STROBE reporting guideline that would be required to ensure transparent and complete reporting of health equity data and considerations. These items formed the interim guidance for the STROBE-Equity extension. 2 A scoping review of equity guidance found that all 36 items proposed in the interim guidance 2 were supported by existing guidance, 60 61 and 13 additional items were identified ( fig 1 ). Then a methodological review of health equity data and reporting and considerations in 320 observational studies indicated that the reporting of study design features ranged from 0% (0/320) to 95% (307/320) of the studies, with more than half of the items being reported by less than one quarter of the studies. 15 62 We conducted eight key informant interviews with eight members of the STROBE-Equity project group from diverse backgrounds, genders, and ethnicities. These interviews identified six key themes that guided reflection, consideration, and confirmation of final checklist items. These themes highlighted the importance of clarity, feasibility, contextual relevance, and the inclusion of diverse perspectives to improve the reporting of health equity in observational research. A detailed table in supplementary appendix 3 outlines the themes and their impact on the development of the extension items. Flow of proposed STROBE-Equity extension items. For items where no extension to the STROBE reporting guideline was proposed, survey participants were able to provide feedback on their level of agreement. “Items combined” refers to collapsing at least two items into a single item to enhance clarity and reduce redundancy. *Methods review aimed to inform current reporting practices in observational studies, whereas key informant interviews identified principles to guide the consensus meeting and finalise the extension checklist, but neither were intended to identify new items. STROBE=STrengthening the Reporting of OBservational studies in Epidemiology In phase 2, we gathered feedback from diverse interest holders (eg, researchers, members of the public, individuals with lived experiences of inequities) with different backgrounds by deploying a global online survey from June 2023 to January 2024 (see supplementary appendix 4 for recruitment details and survey questions). The survey consisted of 24 recommendations for new items to STROBE and 17 items for which no extension was proposed. We received 162 responses to the global online survey (see supplementary appendix 5), with representation from different self-identifying knowledge users, including research methodologists (36%, 58/162), epidemiologists (20%, 33/162), clinicians (14%, 22/162), scientific journal editors (10%, 17/162), policy makers (7%, 12/162), and patients (6%, 9/162). A third of the survey participants self-identified as individuals with lived experiences of health inequities. Overall, 76% (123/162) had more than 10 years of experience in healthcare research. Content expertise varied from public health and health promotion (32%, 52/162) to clinical research (15%, 24/162), equity oriented research (8%, 13/162), and health systems research (8%, 13/162). We obtained 25% (41/162) of respondents from people based in low and middle income countries (see supplementary appendix 5). The survey showed >80% agreement for 22 out of the 24 proposed extension items (see supplementary appendix 5). However, 63% (85/135) to 75% (91/121) of respondents indicated that 17 items without proposed extension items should have extension items, suggesting a strong interest in further refinement. Open ended comments indicated endorsement for developing guidance describing minimal standards for health equity reporting. Additionally, participants felt a need for better definitions for populations experiencing health inequities. In phase 3, we convened a consensus panel of the team members and all steering committee members and held two online consensus meetings at different times to be inclusive of multiple time zones. A pre-consensus survey evaluated inclusion of items with <70% agreement and helped focus the consensus discussion on items where opinions differed. Survey participants contributed detailed comments to support their judgments, which were synthesised into main themes and presented at the consensus meeting. We then held two 1.5-hour online consensus meetings to discuss and finalise the checklist items. The consensus threshold was defined a priori as ≥80% agreement among 80% of the respondents. The consensus meetings endorsed 21 items with ≥80% agreement. Two items were rejected based on consensus that they were adequately covered by the original STROBE item, and one item was contested (see details reported in supplementary appendix 6). We established a synthesis working group to consolidate the feedback from the consensus meetings and discuss how to increase the feasibility of health equity reporting while safeguarding items the working group considered essential for good practice. This team consisted of 10 project members who convened for two one-hour meetings and discussed drafts by email. The groups’ focus was to improve clarity of wording, maximise parsimony, and maintain essential items. This team was guided by values and principles expressed during the consensus panel discussions and by themes identified in the key informant study. Supplementary appendix 7 provides additional details. In accordance with the GRIPP2 reporting checklists (see supplementary appendix 1), 63 we involved both patients and members of the public at various stages of the guideline’s development. Our patient and public steering committee was led by three of our team members (JT, RGS, and HE) who were also decision makers in the development of the STROBE-Equity extension. JT and RGS have lived experience of health inequities and participation in health equity research, and they were also involved from the funding acquisition stage. They convened every two months to guide the project’s design and received regular project updates through quarterly technical advisory meetings and steering committee sessions, with study progress shared through project newsletters. The patient and public partners played a pivotal role in establishing the need for these guidelines and refining the STROBE-Equity guideline by identifying areas for integrating equity considerations. They also contributed to designing the three empirical studies and discussions regarding interpreting the key findings. The patient and public partners received annual financial compensation for their contributions during the project. Additional patient and public perspectives were gathered through the key informant interviews and the online global survey.

Table 1 shows the original STROBE checklist (22 items) alongside the STROBE-Equity extension items. STROBE-Equity introduces 10 items that are specific to health equity relevant observational studies. The items apply to the three study designs covered by the original STROBE statement. Identical to the STROBE statement, the items in the STROBE-Equity guidelines are organised into sections: Title and abstract (1.1), Introduction (2.1), Methods (4.1-12.1), Results (13.1-14.1), and Discussion (18.1). Authors of health equity relevant observational studies should address all items of the STROBE-Equity checklist (although some information may be best placed in supplementary materials, if journals have space constraints). Equity extension to the STROBE (STrengthening the Reporting of OBservational studies in Epidemiology) statement

Study title : “Trends in Racial and Ethnic Disparities in Barriers to Timely Medical Care Among Adults in the US, 1999 to 2018” 64 Abstract objective : “To identify variation in several measures of health equity from 1993 to 2017 in the United States and to test whether there are significant time trends.” 65 Abstract methods : “We use population at risk and mortality data covering 1991-2019 to calculate the mortality rate at age 65-69 at the Local Authority level. We use the 2019 Index of Multiple Deprivation to examine geographical inequalities.” 66 Abstract results : “Those with a foreign first language had lower testing rates, higher incidence, and higher positivity.” 67 Abstract discussion : “Management of high fasting plasma glucose remains a major challenge for older people with T1DM [type 1 diabetes mellitus], and targeted clinical guidelines are needed.” 68 Reporting health equity objectives in the title or the abstract helps patients, policy makers, and practitioners to more easily identify health equity relevant studies and understand how they relate to health equity. 60 This is especially important because health equity studies often focus on specific populations, 15 69 such as those defined by PROGRESS-Plus (see box 1 ), and the findings should clearly describe the populations studied and any limitations in applicability to other settings. Clear reporting ensures that the study’s reference and potential limitations in the applicability of the findings to other settings are easily understood. Moreover, making equity relevant studies easily searchable in databases such as Medline or Embase as indexing terms (eg, MeSH or Emtree) are influenced by the terms used in the title and abstract, which also guide search strategies. We recommend that authors incorporate specific terms such as “health equity” or other more specific terms in the title, when appropriate. 70 71 Additionally, the relevance of health equity should be highlighted across different sections of the abstract (ie, as an objective, emphasising equity relevant results, or interpreting findings while considering the impact on health equity). To further improve searchability, authors (and journals) are encouraged to describe their work using appropriate standardised subject terms (MeSH or Emtree). For example, MeSH terms for the concept of health equity include “health equity”, “health inequities”, and “health status disparities” for the purposes of indexing. 72 Each concept of a robust search strategy should consist of text words and subject terms. 73

“We calculated hazard ratios (HR) and 95%CI [confidence interval] to estimate the difference in SARS-CoV-2 incidence between ethnic groups, using the Dutch origin group as a reference, adjusting for current age in years and sex.” 142 “Thirdly, as boys are more likely to receive a diagnosis of neurodevelopmental disorders than girls, we evaluated the association by sex of the infants.” 143 Additional analyses are often conducted to assess whether effects or associations differ across groups experiencing inequities (eg, based on race, gender, socioeconomic status, or other relevant factors), 144 145 or to assess how potential biases could obscure the true impact of inequity related characteristics on outcomes. 146 Reporting these analyses allows readers to understand how differences between groups were evaluated and whether any findings could indicate a reduction or increase in health inequities. Authors should clearly state whether subgroup analyses were preplanned or exploratory, as post hoc analyses can introduce bias. 147 148 They should describe the methods used to examine differences between groups and the assumptions underlying them. For example, authors should report whether interaction tests were conducted and specify how they were reported, 149 and report if any sensitivity analyses were performed to assess the robustness of the findings. Consideration should be given to how the choice of presenting analyses—whether in absolute or relative terms—could substantially influence judgments about whether health inequities are widening or narrowing. 150 151

“The 8.5% of BAME [Black, Asian, and minority ethnic] is the sum of approximately 2% Bangladeshi, Indian and Pakistani (BIP), and 6.5% of other minority ethnic groups (non-BIP: White Other, Mixed, Black, Asian, and Arab).” 158 “We included 3,574,903 individuals who did not have atrial fibrillation at index age 45 years or older, of whom 1,727,703 were men (48.3%) and 1,847,200 were women (51.7%).” 159 Disaggregating baseline characteristics of study participants and examining how these characteristics relate to the exposure allows readers to better assess the applicability 160 161 and representation 162 163 164 165 166 of different populations in the study. Depending on the study design, these data can highlight potential biases in healthcare practices 120 167 and provide insight into the effectiveness of recruitment strategies focused on specific populations, if used. Additionally, stratifying follow-up time by relevant characteristics offers a clearer understanding of how follow-up duration and missing data may differ across population groups, 168 169 improving the overall interpretation of the findings. Authors should present baseline characteristics disaggregated by characteristics of interest associated with health inequities, such as socioeconomic status or ethnicity, and explain how these factors relate to the exposure of interest. Authors should report the number of participants with missing data and follow-up time stratified by these characteristics.

The STROBE-Equity extension was developed to address a recognised need for improved reporting of health equity data and considerations in observational studies. It is designed to complement the original STROBE statement and checklist. The STROBE-Equity extension aligns with the existing guidelines for reporting on health equity in randomised controlled trials (CONSORT (Consolidated Standards of Reporting Trials)-Equity) and systematic reviews (PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses)-Equity)) developed by members of our team, and it clarifies the process for conducting equity relevant research. We will include the STROBE-Equity extension in the EQUATOR Network’s library of reporting guidelines. To promote its adoption, we will engage the editors in chief of scholarly journals in epidemiology and global health, encouraging them to endorse and require the extension for equity relevant studies. We have established a knowledge translation working group to develop strategies to ensure widespread dissemination and use, such as road testing 180 the reporting guideline, translations, and producing various patient oriented tools such as digital stories. We are also surveying our team members about the perceived benefits, challenges, and impact of our integrated knowledge translation approach on the development process. We followed a prespecified process to gather evidence on the necessity of the reporting guideline extension 55 and adhered to widely accepted methods for developing reporting guidelines. 56 Also, we engaged a diverse group of 57 people, employed empirical studies, and obtained input from 162 individuals through an online survey, capturing a wide range of perspectives. The robustness and inclusivity of our approach, combined with high agreement with the extension items, enhances the guideline’s applicability across diverse populations and settings. Nonetheless, we recognise that no guideline can comprehensively include the perspectives of all possible populations and situations where minor adaptations to the guideline may arise. Additionally, our collective understanding of health equity in research is rapidly evolving with frequent new methodological developments, reflecting advances in understanding how health equity can be conceptualised 181 and the factors at different levels that create inequities 182 and how to overcome them. 183 We recognise that these guidelines may need to be updated to encompass evolving concepts, such as intersectionality, 40 as well as considerations for certain populations. We acknowledge that the ability of researchers to report on health equity data and considerations is limited by the availability, quality, and type of data they can collect (ie, culturally sensitive data, lack of standardisation). 39 184 Reporting these challenges could help bring these obstacles to light and foster opportunities for developing innovative solutions. The STROBE-Equity extension provides essential guidelines for reporting of health equity data and considerations in observational studies. We recommend using the STROBE-Equity extension alongside the main STROBE statement to enhance the clarity and completeness of health equity reporting in research. We encourage all interest holders to promote using this extension to improve the evidence base for health equity.

“To account for differences in baseline characteristics between the 2 groups [males and females], a propensity score-matching (PSM) model was developed using logistic regression to derive two well-matched groups for comparative outcome analysis [from registry data]. Variables included in the PSM model included age, race, and key comorbidities (hypertension, diabetes, chronic obstructive lung disease, heart failure, obesity, nicotine dependence, and history of stroke).” 136 Reporting whether characteristics associated with health inequities were considered in the selection process provides insights about the representation of the study sample to populations most affected, and it informs the applicability of the study’s results to these populations. 137 138 Doing so allows readers to assess whether imbalances in these factors could have influenced the study findings or biased their interpretation. In certain contexts, characteristics such as race or ethnicity in the US are often proxies for social determinants, such as systemic racism. 139 140 Thus, reporting them helps with contextualising the results with a more nuanced and deeper understanding of potential biases and inequities 141 and informs the relevance of the findings to other settings. Authors should describe the rationale for including or excluding specific populations with characteristics associated with inequities in the selection process. This may include the eligibility criteria, propensity score when matching in case-control and controlled cohort studies, or adjusting eligibility criteria to ensure balanced groups. In line with the original STROBE statement, authors should describe the source of the study population (eg, the general population of a region or country), method of recruitment (eg, referral or self-selection through advertisements), and follow-up procedures. Additionally, they should report whether efforts to minimise non-response and loss to follow-up were made, as well comment on the consistency of study procedures for all the included participants.

“We excluded individuals with missing information on race/ethnicity (n=112) and those who identified as Alaskan Native or American Indian or ‘Other’ Race (n=6,673) due to small numbers.” 64 People experiencing health inequities may be less inclined to participate in studies and more inclined to withdraw than other populations. 152 153 154 155 Explaining why participants were lost to follow-up, excluded, or failed to respond and how these factors relate to characteristics of people experiencing inequities, allows readers to judge the representativeness of the final sample. Additionally, this detailed reporting helps identify whether biases related to inequities were potentially introduced to the study 156 and highlight what needs to be done in future studies for populations experiencing inequities. Authors should explain the reasons behind participant loss, exclusions, or non-responses, especially about characteristics linked to inequities. Authors could present these data through a flow diagram or through narrative description in the text. 157 Such a flow diagram could become challenging to create if the equity factor involves more than two levels or if multiple characteristics of interest, such as poverty and rurality, are considered. Using tables may more effectively convey this information.