Representation of women in heart failure trials: does it matter?

Abstract

Heart failure (HF) is a leading cause of hospitalisation, morbidity, and mortality in men and women, accounting for 46 076 annual HF deaths in women and 2.6 million women living with HF between 2015 and 2018 in the USA. Sex differences across the HF spectrum are well defined and pertain to risk factors, aetiology, provision of evidencebased therapies, referral to services, treatment response and clinical outcomes in both the acute and chronic HF syndrome setting. Much of our evidence base for the management of HF is derived from randomised clinical trials (RCTs) that inform best practice for the treatment of HF and shape guideline recommendations. The value of such trials in informing the management of HF in both men and women depends on representativeness of trial populations. Underenrolment of women in HF trials is well documented, including in landmark trials that have informed care. Since 2000, multiple studies have examined the recruitment of women in HF RCTs and reported that enrolment of women has varied between 21% and 29%, which is significantly below the prevalence of HF at the population level. In an attempt to quantify the representativeness in trials, recent studies have used the ratio of trial participation to disease prevalence ratio (PPR). A PPR <0.8 is considered low and indicates underrepresentation. A recent analysis of 740 cardiovascular trials (102 trials in HF) registered between 2010 and 2017 has shown the lowest PPR of 0.48 in HF trials. This is despite the fact that legislature such as the National Institutes of Health Revitalization Act stipulates the inclusion of women and men in clinical trials proportionate to the sexrelated prevalence of the disease under investigation, to provide data on the treatment effect of interventions/treatments studied in both women and men. Recent studies have tried to understand the factors responsible for low enrolment of women in HF trials. The low enrolment rates of women in cardiovascular clinical trials have historically been attributed to agerecruitment bias since cardiovascular disease is seen predominantly in older women. However, recent multivariable analyses have revealed trial characteristics such as ambulatory recruitment, sexspecific exclusion criteria, drug, device and surgical interventions, exclusively male trial leadership and trial coordination in North America, Europe and Asia to be independently associated with underenrolment of women in HFrEF RCTs. Moreover, poor awareness of HF trials, concerns around greater perceived risks from trial participation and childcare responsibilities have been reported as additional barriers to more equitable participation of women in clinical trials. It is important to highlight that randomised control trials led by women have greater odds of enrolling a representative sample of women and women steering committee members. 8 9 In the present study, Morgan et al have undertaken a systemic review of HF trials published in seven highimpact clinical journals (impact factor >20) to explore the reasons behind sex disparities in enrolment in HF RCTs. Furthermore, in an attempt to determine whether recruitment to RCTs was significantly below the population level prevalence of HF in women, large HF registries and population statistics were identified using the same criteria. The systemic review included data from 146 HF RCTs and 2 48 620 patients (2000–2020). The median proportion of women recruited into RCTs was 25.8% (IQR 21.3–36.0), which was significantly lower than the median proportion of women recruited into registries (40.2% IQR 32.3–52.8) or from populationlevel data (49.0% IQR 38.2–53.4) suggesting significant underenrolment of women. Average age of patients enrolled was lower in trials than in registries and population datasets, and for each condition, the enrolment of women was lower in trials than in registries and population datasets. This is concordant with the pragmatic eligibility criteria and processes of registries and population datasets that result in more representative patient populations. The largest gap between trial, registry and population enrolment of women was seen in ischaemic cardiomyopathy 17.9% (IQR 11.7–21.2) vs 37.7% (IQR 33.2–41.3) in registries and severe systolic dysfunction (left ventricular ejection fraction (LVEF) <35%) 21.4% (IQR 17.7–25.7) compared with registries 25.4% (IQR 23.7–27.2) and population statistics 28.2% (IQR 26.0–30.4). Trials involving an invasive procedure or surgical treatment recruited the lowest proportion of women (21.2%). Enrolment of women in HFpEF trials was greater than in HFrEF trials, reflecting the women preponderance of HFpEF. Women represented more than half (51.6% (IQR 48.6–52.0)) of HFpEF trial participants, with smaller gaps relative to registry (54.8% (IQR 50.8–61.0)) or population datasets (56.5% (IQR 52.2–64.8)). The estimated PPR of 0.9 in HFpEF trials suggested representative enrolment of women in these trials. While prior studies have reported many of the findings in the current study, including low representation of women in HF trials and more women in HFpEF trials, 4 the strengths of this study include a systematic assessment of the prevalence of HF at a registry and population level to place the findings of the study into a broader context, and the analysis of variations in women enrolment across different degrees of LV function as well as HF an aetiology and intervention type. There are several limitations of the current study, particularly in restricting analysis to journals with impact factor of >20 that may bias results towards prominent RCTs led in North America and Europe and towards large RCT focused towards pharmacological treatments and interventions, rather than implementation of evidence based therapies/healthcare delivery that may not necessarily be published in such high impact journals. All analyses were descriptive, and independent associations between trial of disease characteristics and enrolment of women were not explored. Furthermore, although the data presented provides compelling evidence of sex disparities in recruitment to HF trials, the contribution of different biases—such as selection, gender and age biases—to such disparities remains poorly defined. Given the retrospective nature of the article, patient level characteristics such as consent, age, disease severity, study factors such as screening and recruitment strategies for the trials were not accounted for. Division of Cardiology, Department of Medicine, Loma Linda University Health, Loma Linda, California, USA Department of Medicine, Population Health Research Institute, Research Institute of St. Joseph’s, McMaster University, Hamilton, Ontario, Canada Keele Cardiovascular Research Institute, Keele University, StokeonTrent, UK