Sex differences have been described in diabetes cardiovascular outcome trials (CVOTs).
We systematically reviewed for baseline sex differences in cardiovascular (CV) risk factors and CV protection therapy in diabetes CVOTs.
Randomized placebo-controlled trials examining the effect of diabetes medications on major adverse cardiovascular events in people ≥18 years of age with type 2 diabetes.
Included trials reported baseline sex-specific CV risks and use of CV protection therapy.
Two reviewers independently abstracted study data.
We included five CVOTs with 46,606 participants. We summarized sex-specific data using mean differences (MDs) and relative risks (RRs) and pooled estimates using random effects meta-analysis. There were fewer women than men in included trials (28.5–35.8% women). Women more often had stroke (RR 1.28; 95% CI 1.09, 1.50), heart failure (RR 1.30; 95% CI 1.21,1.40), and chronic kidney disease (RR 1.33; 95% CI 1.17; 1.51). They less often used statins (RR 0.90; 95% CI 0.86, 0.93), aspirin (RR 0.82; 95% CI 0.71, 0.95), and β-blockers (RR 0.93; 95% CI 0.88, 0.97) and had a higher systolic blood pressure (MD 1.66 mmHg; 95% CI 0.90, 2.41), LDL cholesterol (MD 0.34 mmol/L; 95% CI 0.29, 0.39), and hemoglobin A1c (MD 0.11%; 95% CI 0.09, 0.14 [1.2 mmol/mol; 1.0, 1.5]) than men.
We could not carry out subgroup analyses due to the small number of studies. Our study is not generalizable to low CV risk groups nor to patients in routine care.
There were baseline sex disparities in diabetes CVOTs. We suggest efforts to recruit women into trials and promote CV management across the sexes.
Introduction
There are important sex differences in the risk, pathophysiology, and complications of diabetes including its vascular consequences (1–4). In women, type 1 and type 2 diabetes promote a greater excess relative risk of cardiovascular disease (CVD), stroke, heart failure, and mortality than in men (5–8).
While sex differences in diabetes might be due to biological factors, body composition, fat distribution, or sex hormones (2), variances might also be related to disparities in cardiovascular (CV) risk factor management (e.g., control of blood pressure and cholesterol) (1,4,9). Disparities in CV risk factor management have, in fact, been described across populations with diabetes in routine care (10–15).
Sex differences in CV risk factors and outcomes have also been observed in the clinical trial setting, including recent cardiovascular outcome trials (CVOTs) of populations with diabetes (16–18). In TECOS (Trial Evaluating Cardiovascular Outcomes With Sitagliptin) and EMPA-REG OUTCOME (BI 10773 [Empagliflozin] Cardiovascular Outcome Event Trial in Type 2 Diabetes Mellitus Patients), for example, women had suboptimal risk factor profiles compared with men (16,17). In TECOS, women were at lower risk of major adverse cardiovascular events (MACE) (19). Because clinical trial participants are typically more adherent, educated, and affluent than patients in routine care (20–22), a systematic understanding of disparities in CV risk factor management in trial participants is of interest.
In this systematic review and meta-analysis, we evaluated for baseline differences in CV risk factors and use of CV protection therapy in women and men who participated in major diabetes CVOTs. We hypothesized that even in clinical trial participants, women would have suboptimal risk factor management and use CV protection therapy less often than men.
Research Design and Methods
We followed a prespecified protocol registered at the International Prospective Register of Systematic Reviews (PROSPERO) (CRD 42019137121), as well as the guidelines for reporting systematic reviews and meta-analyses (Supplementary Table 1) (23).
Data Sources and Searches
We included all randomized placebo-controlled trials that evaluated the effect of diabetes medications (oral or injectable) on 3- or 4-point MACE (i.e., CV death, nonfatal myocardial infarction, nonfatal stroke, hospitalization for heart failure, hospitalization for unstable angina for 4-point MACE) in those ≥18 years of age with type 2 diabetes. To be included, trials had to report sex-specific baseline characteristics and CV risk factors (described below) and had to be reported in the English language. We excluded observational studies, case series, case reports, cross-over studies, trials with fewer than 100 participants, and review articles.
To reduce publication bias, we sought unpublished sex-specific baseline data by contacting the corresponding authors of known diabetes CVOTs completed by April 2019. Corresponding authors were informed of our study objectives. If authors did not respond to our initial correspondence within 1 month, we sent follow-up correspondence 1–2 months later. For authors who expressed interest in providing data, we sent one additional reminder to facilitate data acquisition.
Study Selection
Our literature search strategy was created by a health sciences librarian with expertise in systematic reviews (B. Dishan, St. Joseph’s Health Care London). The search strategy used Medical Subject Headings (MeSH) terms and keywords informed by our study aim, population of interest, and the CVOTs known to be completed up to April 2019 (Supplementary Table 2). We validated our final search strategy through the retrieval of a key set of previously identified relevant articles.
We searched electronic databases including MEDLINE, Embase, CINAHL, Cochrane Library, Scopus, BIOSIS, Web of Science, and ClinicalTrials.gov for relevant CVOTs published since 2008 (when the U.S. Food and Drug Administration provided new guidance to pharmaceutical agencies about the development of diabetes medications) (24).
Data Extraction and Quality Assessment
Using standard forms, one reviewer (K.K.C.) screened the abstracts and/or full-text articles of all identified citations against selection criteria. Citations were marked as include, exclude, or uncertain. If citations were marked uncertain, full-text articles were reviewed in detail. We used the Cochrane Collaboration tool to assess risk of bias in included studies (25).
Data Synthesis and Analysis
Using a structured data abstraction form, two reviewers (K.K.C., S. Elsie [St. Joseph’s Health Care London]) independently abstracted authors, year of publication, sample size, diabetes medications, and sex-specific baseline characteristics (age; percent using statin, renin-angiotensin-aldosterone system [RAAS] blocking medication, β-blocker, and aspirin; percent with a history of myocardial infarction, stroke, heart failure, chronic kidney disease [CKD] [estimated glomerular filtration rate <60 mL/min/1.73 m2 by MDRD formula], and peripheral arterial disease [PAD]; mean [SD] LDL cholesterol [LDL-C], systolic blood pressure, and hemoglobin A1c). Abstractors resolved any disagreements by discussion. If there were unresolved discrepancies, we engaged a third reviewer (B.Z.).
Our primary interest was to examine for sex differences in baseline CV risk factors and use of guideline-recommended CV protection therapy. We pooled sex-specific baseline data across included trials using random effects meta-analyses with inverse variance weighting. Continuous variables were summarized using mean differences (MDs), and dichotomous variables were summarized using relative risks (RRs) and 95% CIs. We used I2 and Cochran Q test to explore heterogeneity. All analyses were performed using Stata version 13.1 (StataCorp, College Station, TX).
Results
We identified 1,192 relevant citations between 1 January 2008 and 10 April 2019 (Supplementary Fig. 1). After eliminating duplicates as well as citations based upon their title or abstract, we reviewed 25 full-text articles in detail.
Two trials published baseline characteristics by sex (EMPA-REG OUTCOME, TECOS) (16,17). The TECOS trial published sex-specific data in two related manuscripts (16,19).
We contacted the corresponding authors of 17 additional diabetes CVOTs to request unpublished sex-specific baseline data. We received data from three additional trials: Examination of Cardiovascular Outcomes with Alogliptin Versus Standard of Care (EXAMINE), Canagliflozin Cardiovascular Assessment Study (CANVAS) Program (CANVAS and CANVAS-R), and Liraglutide Effect and Action in Diabetes: Evaluation of Cardiovascular Outcome Results (LEADER) (18,26,27).
Characteristics of Included Trials
Included trials had between 5,380 and 14,724 participants, all of whom had established CVD or risk factors for CVD (Table 1). The inclusion criteria of the trials in our study did not vary by sex. The diabetes medications under investigation in reported trials were empagliflozin, canagliflozin, liraglutide, alogliptin, and sitagliptin. There were consistently fewer female participants (28.5–35.8% were women). Risk of bias was low.
Characteristics of included diabetes CVOTs
Source . | n . | Trial entry criterion . | Diabetes medication . | % women . |
---|---|---|---|---|
Neal et al. (27) | 10,142 | ≥30 years with established CVD or ≥50 years with risk factors for CVD | Canagliflozin | 35.8% |
Zinman et al. (17) | 7,020 | ≥18 years with established CVD | Empagliflozin | 28.5% |
Bethel et al. (16) | 14,724 | ≥50 years with established CVD | Sitagliptin | 29.3% |
White et al. (26) | 5,380 | ACS 15–90 days prior to randomization | Alogliptin | 32.1% |
Marso et al. (18) | 9,340 | ≥50 years with established CVD or ≥60 years with risk factors for CVD | Liraglutide | 35.7% |
Source . | n . | Trial entry criterion . | Diabetes medication . | % women . |
---|---|---|---|---|
Neal et al. (27) | 10,142 | ≥30 years with established CVD or ≥50 years with risk factors for CVD | Canagliflozin | 35.8% |
Zinman et al. (17) | 7,020 | ≥18 years with established CVD | Empagliflozin | 28.5% |
Bethel et al. (16) | 14,724 | ≥50 years with established CVD | Sitagliptin | 29.3% |
White et al. (26) | 5,380 | ACS 15–90 days prior to randomization | Alogliptin | 32.1% |
Marso et al. (18) | 9,340 | ≥50 years with established CVD or ≥60 years with risk factors for CVD | Liraglutide | 35.7% |
ACS, acute coronary syndrome.
Sex Differences in Baseline Characteristics
At baseline, women were older (MD 1.03 years; 95% CI 0.18, 1.89). They less often had a history of myocardial infarction than men (RR 0.71; 95% CI 0.59, 0.86) but more often had stroke (1.28; 1.09, 1.50), heart failure (1.30; 1.21, 1.40), and CKD (1.33; 1.17, 1.51). A similar percentage of women and men had PAD (1.12; 0.97, 1.30) (Fig. 1).
Forest plot of baseline sex differences in heart failure, stroke, myocardial infarction, PAD, and CKD. The solid line represents the line of no difference, and the dotted line highlights the pooled relative risk.
Forest plot of baseline sex differences in heart failure, stroke, myocardial infarction, PAD, and CKD. The solid line represents the line of no difference, and the dotted line highlights the pooled relative risk.
Despite an imbalance in their baseline comorbidities, we found women in included trials less often used β-blockers (RR 0.93; 95% CI 0.88, 0.97), aspirin (0.82; 0.71, 0.95), and statins (0.90; 0.86, 0.93) than men, with similar qualitative differences in all trials (I2 90%, 99%, and 93%, respectively) (Fig. 2). There was no sex difference in the baseline use of RAAS blockers (RR 1.00; 95% CI 0.99, 1.05). Women had higher systolic blood pressure (MD 1.66 mmHg; 95% CI 0.90, 2.41), higher LDL-C (0.34 mmol/L; 0.29, 0.39), and higher hemoglobin A1c than men (MD 0.11%; 95% CI 0.09, 0.14 [1.2 mmol/mol; 1.0, 1.5) (Fig. 3). Although there was evidence of heterogeneity between CVOTs (I2 51–85%), all within-trial differences were in the same direction.
Forest plot of baseline sex differences in statins, RAAS blockade, β-blockers, and aspirin. The solid line represents the line of no difference, and the dotted line highlights the pooled relative risk. Note that LEADER and TECOS reported use of aspirins/antiplatelets.
Forest plot of baseline sex differences in statins, RAAS blockade, β-blockers, and aspirin. The solid line represents the line of no difference, and the dotted line highlights the pooled relative risk. Note that LEADER and TECOS reported use of aspirins/antiplatelets.
Forest plot of baseline sex differences in systolic blood pressure, LDL-C, and hemoglobin A1c. The solid line represents the line of no difference, and the dotted line illustrates the pooled mean difference. Note that data on systolic blood pressure was not available from EXAMINE.
Forest plot of baseline sex differences in systolic blood pressure, LDL-C, and hemoglobin A1c. The solid line represents the line of no difference, and the dotted line illustrates the pooled mean difference. Note that data on systolic blood pressure was not available from EXAMINE.
Conclusions
In this systematic review and meta-analysis of diabetes CVOTs, we identified several interesting findings. Only two CVOTs published baseline characteristics by sex, a finding that has also been described in systematic reviews of CV prevention and stroke (28,29). In the reported diabetes CVOTs, the participation of women was low, as we have described previously (30). We also found that women in these CVOTs more often had a baseline history of stroke, CKD, and heart failure than men. Despite their comorbidities and guideline recommendations that statins be considered in high-risk patients with type 2 diabetes (31–33), that aspirin be prescribed in established CVD (34), and that β-blockers be prescribed in those with coronary artery disease and heart failure (34,35), fewer women than men were using statins, aspirin, and β-blockers across included trials. Accordingly, women had higher LDL-C and systolic blood pressure. Although we did not capture their baseline use of antihyperglycemic medications, women also had a slightly higher hemoglobin A1c than men at baseline.
There may be a number of reasons for the sex disparities observed in the reported CVOTs. Differences might be related to patient, provider, or health care system factors (4,9,12–15,36). With respect to the lower number of women in reported CVOTs, it is possible that some trials favored recruitment of men because of their higher absolute risk of CVD. There may have been bias in the referral of women for trials by health care providers (28). Women might have also faced barriers to trial participation due to family, work, or social reasons (28,37). A lower number of female participants have also been described in previous systematic reviews of CV prevention (28), stroke (29), and heart failure (38). Encouragingly, the recently published REWIND (Researching cardiovascular Events with a Weekly INcretin in Diabetes) trial of dulaglutide recruited the largest proportion of women in a diabetes CVOTs to date (46% women) (39). Whether related to REWIND’s recruitment strategies or to the inclusion of those with newly diagnosed diabetes or subclinical cardiovascular disease (e.g., carotid artery disease, left ventricular dysfunction), trialists might continue to make efforts to screen and enroll more women in clinical trials.
The baseline sex differences we observed in heart failure, CKD, and stroke have been described previously. Heart failure with preserved ejection fraction, in particular, appears more common in women, possibly related to differences in body composition, hormone or reproductive factors, endothelial inflammation, and microvascular function (40) or to the presence of diabetes and hypertension (potent risk factors for heart failure in women) (41). A higher prevalence of CKD in women has also been described, which might in part be due to the longer life expectancy of women or overdiagnosis with estimating equations (42). Stroke may be more common in women due to suboptimal CV risk factor management, the presence of diabetes (excess risk of stroke is higher in women than men), or growing rates of obesity and the metabolic syndrome in women (5,43).
Despite their comorbidities and the presence of sex-based clinical practice guidelines on CV protection in women (44), we observed disparities in the baseline use of statins, aspirin, and β-blockers, as well as higher LDL-C, blood pressure, and hemoglobin A1c in women in reported trials. It is possible that providers might not recognize that women with diabetes can be at high risk of CVD, and thus women received suboptimal CV risk factor management. Women and men might have different health beliefs that could affect adherence to medications. They may additionally experience different drug side effects that could result in differential discontinuation of therapies (36,45). Further, women might face barriers to attending clinic appointments to receive CV protection for social reasons (i.e., children and family). Although we do recognize that fewer women had baseline myocardial infarction than men (indication for both statins and β-blockers) (34), we do not feel that this difference is enough to explain variance in the use of β-blockers, aspirin, and statins between the sexes. All trial participants had established type 2 diabetes and CVD or risk factors for CVD (indication for statin) (31). Women also had heart failure more often than men (indication for β-blockers) (35), along with stroke (indication for aspirin and statin) and CKD (indication for statin) (32). Sex differences in CV protection therapy, LDL-C, blood pressure, and hemoglobin A1c has also been described in patients with diabetes previously, including in contemporary routine care settings in regions across the world (10,11,13,46–47). Of note, in these trials we did not observe baseline differences in the use of RAAS blockade, a therapy indicated for control of blood pressure and organ protection in diabetes (48). As sex disparities in the use of RAAS blockade have been described in patients with diabetes in routine care previously (11), this finding might be suggestive of recent change.
There are several strengths to our study. We examined a large number of individuals who participated in five diabetes CVOTs in recent times. We followed a prespecified protocol as well as recommended guidelines for the conduct and reporting of systematic reviews and meta-analyses. Weaknesses included our inability to carry out meaningful subgroup analyses (i.e., examination of disparities by age) due to the relative small sample size. Further, we could not determine if the increased prevalence of CV risks including heart failure in women was due to heart failure with preserved ejection fraction or heart failure with reduced ejection fraction, as we used aggregate data from clinical trials and did not have access to echocardiogram results. Over the course of CVOTs, it is possible that investigators might have optimized cardioprotective medications or initiated these therapies because of incident MACE. We could not capture these prescribing changes. Although we attempted to reduce publication bias by requesting unpublished data from eligible diabetes CVOTs, it is possible that by sharing our study objectives we may have introduced bias (investigators may have been dissuaded from participating if they had a study of sex differences planned or if the topic was not of interest). Our study is also not generalizable to low-risk CV groups nor to high-risk CV patients in real-life practices.
Our results, however, might be helpful to diabetes investigators, patients, and providers. With few published diabetes CVOTs reporting sex-specific baseline data, we suggest that trialists both examine and publish sex-specific characteristics and outcomes. Where diabetes CVOTs have found that drugs including SGLT2 inhibitors meaningfully reduce the risk of heart and kidney failure (27,49–51), it is important to include more women with these comorbidities in future trials and perform sex-based analyses to help translate trial results into effective clinical care strategies across the sexes (40). We also suggest that providers be increasingly attentive to CV risks in women. Further, it is important to continue to promote efforts to adopt guideline-based CV care for both women and men, educate women on modifiable risk factors for the development of CVD, and promote equitable access to health care across the sexes. Our study highlights that even in clinical trial participants, there remains a female disadvantage in diabetes. Study investigators should make special efforts to enroll females in clinical trials and should report sex-specific characteristics and outcomes in their studies (9,28). We also suggest ongoing efforts to promote equitable care across the sexes.
Article Information
Acknowledgments. The authors thank Dr. John Buse (University of North Carolina); David Ørsted, Benjamin Wolthers, and Søren Rasmussen (Novo Nordisk); Dr. William White (University of Connecticut School of Medicine); Jie Liu (The George Institute of Global Health); and Yuyin Liu (Baim Institute for Clinical Research) for providing and summarizing unpublished baseline data. They also thank Brad Dishan (St. Joseph’s Health Care) for creating the detailed search strategy and Sharlene Elsie (St. Joseph’s Health Care) and Alexandra Ouedraogo (ICES Western) for providing administrative support.
Duality of Interest. Unrelated to this work, K.K.C. received a Diabetes Canada Junior Investigator Award in 2017, sponsored by AstraZeneca. She has also attended conferences sponsored by Merck. M.W. is a consultant to Kirin and Amgen. B.N. has received research support from Janssen and has participated in advisory boards and/or continuing medical education programs with Janssen; consultancy, honoraria, and travel support have been paid to his institution. B.Z. has received fees for participation in advisory boards for AstraZeneca, Boehringer Ingelheim, Eli Lilly, Novo Nordisk, Merck, Janssen, and Sanofi. No other potential conflicts of interest relevant to this article were reported.
Author Contributions. K.K.C. contributed to the conception and design of the study, the acquisition of data, and the interpretation of the results and drafted the manuscript. M.W. contributed to the design of the study, the analysis, and result interpretation and revised the manuscript critically for its content. B.N. contributed to the design of the study and the interpretation of results and reviewed the manuscript for its content. B.Z. conceptualized the study, helped to acquire data, interpreted the results, and revised the manuscript. All authors gave final approval of the version to be published.