Background

Peripheral artery disease (PAD) is associated with serious adverse medical events and substantial healthcare spending [1, 2]. Significant disparities exist in the prevalence, diagnosis, and outcomes of PAD based on race and sex. While limited data comparing racial and ethnic differences in PAD prevalence is available [3], prevalence rates vary by geographic regions globally [4]. PAD prevalence in the United States (US) is higher among Black patients [3, 5] who also experience worse outcomes [6]. Additionally, Black, Hispanic, and Native American patients in the US are more likely to undergo amputations as a result of PAD [7,8,9,10], while individuals of Asian or Pacific Islander race experience a higher mortality burden when hospitalized for PAD [9].

Disparities by sex are evident as well. Global PAD prevalence is higher in women than in men [4]. In the US, women with PAD present at an older age and with more severe disease, and female sex is associated with more advanced PAD-related disability. However, women are also less likely to receive optimal medical therapy (i.e., statins) or surgical intervention than their male counterparts [11,12,13,14]. Notably, short-term complications after interventions [11] and above-the-knee amputations are more prevalent among women than men [14, 15]. Among US women with PAD, Black and Native American women experience higher mortality than White and Hispanic women [14].

In addition to substantial morbidity, PAD imposes a significant financial burden on patients and society. In the US, the direct medical costs of PAD amount to $6.3 billion [16]. Disparities in PAD diagnosis and treatment extend to differences in costs and utilization: among hospitalized patients with PAD, costs and length of stay differ significantly based on a patient’s race/ethnicity [9].

Endovascular interventions for PAD have shown promise in clinical trials [17, 18], but these trials often lack diverse patient groups that accurately represent the affected population [19,20,21]. Disparities in PAD care and the need to enhance diversity in clinical trials have been noted in previous studies [11, 22, 23], and multiple calls to address this lack of diversity exist [12, 24]. Therefore, this study seeks to identify and summarize the demographic representation and enrollment strategies employed in clinical trials of lower-extremity endovascular interventions for PAD. This review includes trials of patients with PAD undergoing lower-extremity endovascular interventions, specifically targeting the superficial femoral artery (SFA), femoropopliteal artery (FPA), popliteal artery, and tibial artery.

Methods

This review followed the 2020 Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. The systematic review methodology was published in the International Prospective Register of Systematic Reviews (PROSPERO: CRD42022378304) and Long et al. (2023) [25].

Data sources and searches

Several sources were searched, including ClinicalTrials.gov, MEDLINE via OVID, EMBASE via OVID, Cochrane Controlled Register of Trials (CENTRAL), National Institutes of Health grants, and the World Health Organization International Clinical Trials Registry Platform (WHO ICTRP), which was accessed through Dr.Evidence™ (Santa Monica, CA) [26,27,28]. Additionally, Google Scholar was searched for protocols or publications that may not have been indexed in the trial registry. Manual searches of references of eligible publications were also performed. A comprehensive overview of the search strategy used in this study is available as a supplementary file (see Supplementary File, Table 1).

Eligibility criteria

This review included any randomized controlled trials (RCTs) with a parallel group design that compared clinical outcomes of lower-extremity endovascular interventions, including patency rate, target lesion revascularization (TLR), all-cause mortality, amputation rates, amputation-free survival, minor or major amputations, serious adverse events/major adverse limb events (MALEs), change in ankle-brachial index, or improvement in Rutherford category. The inclusion criteria for RCTs in this review were: a sample size greater than 50 patients; published in English between January 2012 and December 2022; and inclusion of 12-month outcome data. Studies were excluded if they did not report the clinical outcomes of interest, if they reported the clinical outcomes of interest outside the 12-month period, or lacked a clinical trial registration number. Non-controlled studies, including those with a single-group assignment, single-arm design, or pragmatic study design, were excluded. The full list of eligibility criteria has also been published in Long et al., 2023 [25]. The search terms were applied following the population, interventions, comparators, outcomes, and setting (PICOS) framework, as detailed in Table 1.

Table 1 Study PICOS framework
Full size table

Data extraction, risk of bias, and statistical analysis

Title and abstract screening, as well as full-text screening, were performed independently by two reviewers (CMJ and AMM). Disagreements regarding the eligibility of the studies were resolved by a third reviewer (AOW). Data extraction was conducted using a data extraction form specifically developed for this review. Two reviewers (CMJ and AMM) performed data extraction, and a third reviewer verified the data for quality assurance and resolved any discrepancies or inaccuracies (AOW). Two independent reviewers (AOW and CMJ) evaluated the methodological quality of eligible studies for potential bias using the Cochrane risk-of-bias tool for randomized trials (RoB 1) [29]. This tool evaluates the quality of RCTs across several domains: random sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting, and other sources of bias. Each domain was rated as “low risk of bias,” “high risk of bias,” or “unclear risk of bias.” The overall risk of bias was determined by considering all domains. The RoB 1 tool was customized in Covidence (Melbourne, Australia) [30]. Any disagreement was resolved independently by a third reviewer (AOW) or through consensus (see Supplementary file, Table 2).

The extracted trial characteristics included: clinical trial registry source (Clinicaltrials.gov, WHO ICTRP, etc.), reporting of study results, indexing of peer-reviewed or study protocol to trial registry, intervention and comparator, allocation concealment, start and end dates of the trial, follow-up time, sample size, study sites (number of sites, geographic location, hospital setting versus other, urban versus rural), recruitment status (active not recruiting, completed, recruiting, suspended, not yet recruiting, or unknown), type of randomization (1:1, 2:1, 3:1, or not reported), blinding (single, double, or not reported), trial phase, and principal investigator (PI) characteristics (sex, affiliation, country). PI sex was determined through information on trial registry source (i.e., Clinicaltrials.gov) and internet searches of PI names.

For RCT protocol characteristics, the following data were extracted: site of patient recruitment (hospitals or clinics, academic institutions, community settings), withdrawal processes (participant withdrawal by choice, administrative withdrawal, study discontinuation), strategies for follow-up of patients (telephone, letter, office, or clinic visits), availability of participant facing materials in other languages, information on barriers to transportation, patient reimbursement or compensation, types of reimbursement or compensation, patient navigation or coaching strategies adopted, information on cultural competency training for clinical research associates or PIs, information on methods for handling missed or late visits, and reasons for excluding patients (missed visits, investigator removal, defaulted clinical follow-up, surgery, death, withdrawal, early termination).

Data were extracted to assess the demographic representativeness of the study, including baseline demographic characteristics (age, race/ethnicity, sex, geographic region) of patients enrolled and those excluded (due to withdrawal, loss to follow-up). Information on the baseline clinical characteristics (intermittent claudication, critical limb ischemia, Rutherford classification, diabetes, hyperlipidemia, hypertension, smoking status, obesity, coronary artery disease, history of congestive heart failure, chronic obstructive pulmonary disease, and other relevant characteristics) of patients enrolled and excluded were also extracted. Furthermore, data on the reporting of demographic characteristics by clinical outcome were extracted, including patency rate/vessel patency, TLR, all-cause mortality/death, amputation (amputation rates, amputation-free survival, minor or major amputations), and serious adverse events/major adverse events. The review assessed the reporting of clinical outcomes by demographic characteristics (age, sex, and race).

Descriptive analysis was used to summarize the features of the trial, publication (e.g., outcomes reported, how analyses were performed), study protocol characteristics, and the reporting of demographic characteristics in the included trials. Meta-regression used the proportion of women enrolled in each study and the mean age of participants in each study as continuous outcomes. Covariates assessed in separate models for each outcome were study year, study location (non-European vs. European), population type (PAD and CLI vs. PAD only), trial length (years), duration of enrollment (months), and the number of study locations. The coefficients represent the difference in outcome (proportion of women or mean age) for a one-unit increment in continuous covariates (study year, trial length, duration of enrollment, or number of study locations), or between locations (non-European vs. European) and population type (PAD and CLI vs. PAD only). The threshold for statistical significance was set at 0.05, meaning that there is a 5% chance of rejecting the null hypothesis when it is true (a type I error). All meta-regression analysis was performed using STATA version 17 (StataCorp LLC, College Station, TX, USA).

Results

Search results

Of the 2,374 materials identified, 59 materials (comprising 35 RCTs, 14 publications of RCTs, and 10 protocols) met the inclusion criteria (Fig. 1). All records were unique and reflected different studies.

Fig. 1
figure 1

Study Identification Cohort. The number of studies identified via databases and registries, screened, excluded, and included for the final review

Full size image

Characteristics of RCTs

The 35 RCTs comprised a total of 4,338 trial participants across nine countries (Table 2). The lead PIs were mostly male (31, 89%) and most often affiliated with hospitals (24, 69%), followed by academic institutions (8, 23%). The most common country affiliations of the PIs were Germany (12, 34%), the US (11, 31%), and China (7, 20%). Among the 35 RCTs, 11 (31%) were completed and 6 (17%) reported study results; the remainder were either active, recruiting, not yet recruiting, suspended, or of unknown status. The most common interventions used were drug-coated balloon/drug-eluting balloon (DCB/DEB) (23, 66% of RCTs), followed by drug-eluting stent/drug-coated stent (DES/DCS) (6, 17%), percutaneous transluminal angioplasty (PTA) (3, 8.6%), and bare metal stent (BMS) (3, 8.6%).

Table 2 Summary of clinical trials for lower extremity endovascular interventions for the treatment of PAD
Full size table

Characteristics of RCT protocols

Among the 10 study protocols identified, the majority lacked information relevant to population disparities (Table 3). Four protocols (40%) included information on barriers to transportation, and three (30%) outlined strategies to address these barriers. None of the protocols mentioned patient navigation/coaching strategies, cultural competency training for clinical research associates, or relationship-building/social marketing activities. Seven protocols (70%) discussed follow-up strategies, which included telephone and office/clinic visits. Overall, 7 (70%) of the published protocols planned to recruit patients from hospitals, and 2 (20%) indicated the availability of trial materials in other languages.

Table 3 Characteristics of the included clinical trial study protocols for lower extremity endovascular interventions for the treatment of PAD
Full size table

Approximately, 23 (66%) and 7 (20%) of the trials assessed for methodological quality were rated high and low for blinding of participants and personnel. More than half (54%) and 16 (46%) were rated low and unsure regarding allocation concealment (see Supplementary Table 2).

Characteristics of trial publications

The 14 trial publications comprised a total sample size of 3,964 patients (Table 4). All studies reported age and sex; the overall mean (standard deviation [SD]) age of patients was 68.5 (9.4) years, and two-thirds of patients (67%) were male. Race was provided in 4 of 14 (29%) studies. Among the publications that reported on race/ethnicity (48%), 75% of patients were White, followed by Asian (16%), Black (4.3%), Hispanic (3.0%), other (2.0%), and American Indian/Alaska Native or Native Hawaiian/Pacific Islander (< 1%). None of the publications reported on other demographic characteristics, such as socio-economic status, marital status, or immigration status. Regarding the reporting of treatment effects or outcomes by demographic characteristics, only 4 (29%) publications reported clinical outcomes by sex, age, or race (and 2 did so by sex only); 2 (14%) publications reported primary patency by sex, while one publication reported clinically-driven target lesion revascularization (CD-TLR) by sex.

Table 4 Characteristics of included publications, by reporting of demographic characteristics
Full size table

Meta-regression by demographic characteristics

Across all 14 trial publications, women were underrepresented, accounting for 33% of participants. The meta-regression analysis revealed that 5.9% more women were enrolled in non-European trials (36%) than in European trials (30%). However, meta-regression analysis shows the proportion of women enrolled in the trials increased over time, a finding that was not statistically significant (Table 5). While the proportion of women enrolled varied by study population type, trial length, enrollment duration, or the number of study locations, a significantly higher proportion of women were enrolled in studies in non-European countries (US, China, Singapore, New Zealand) compared to European countries (Table 5). Figure 2 shows the proportion of women increased between 2012 and 2019 (reflected by the trial start year); however, this finding is non-significant.

Table 5 Meta-regression results
Full size table
Fig. 2
figure 2

Meta-Analysis Bubble Plot of the Proportion of Women Enrolled by Study Start Year. The bubbles are drawn with sizes proportional to the contribution of individual studies towards the linear prediction

Full size image

The mean age of participants did not significantly differ by study year, location, study population type, trial length, duration of enrollment, or the number of study locations (Table 5).

Discussion

Previous studies have emphasized the poor representation of women and racially/ethnically diverse or underrepresented minorities (URMs) in cardiovascular trials [22, 45, 46]. Efforts have been made to address this disparity by implementing innovative trial designs that prioritize diverse enrollment recruitment processes and minimize sex-specific exclusion criteria [46]. For instance, the ELEGANCE registry, a global clinical peripheral vascular disease (PVD) registry, was specifically designed to enroll diverse patient populations that have been historically underrepresented in PVD trials [47]. As of December 2022, the registry achieved an enrollment of 44% women and 47% URMs in the US [47]. This registry’s focus on diverse enrollment is crucial for enhancing the generalizability of study findings and providing optimal individualized care for all patients with PAD.

This analysis revealed a limited representation of female physicians participating as PIs in clinical trials. Previous studies have shown that race concordance between patients and providers can lead to better patient-clinician relationships, better disease management, and improved outcomes [48,49,50]. This suggests that increasing the diversity of PIs and study teams could impact the level of comfort and trust of the diverse patients these studies aim to recruit. To increase diversity in clinical research teams, it is imperative to invest in equity initiatives that prioritize promoting demographic representativeness among physicians and fostering diverse participation in clinical research globally (and more specifically, RCTs) [47, 51]. If successful, such initiatives would improve patient-physician concordance and help to enhance the diversity of clinical trial participants, improving the validity and relevance of research findings.

This review supports previous findings that demonstrate a lack of reporting and representation of participant sub-groups beyond age, sex, and race. Information pertaining to income, education, language proficiency, immigrant status, or other relevant characteristics were absent in the published RCTs [24, 46, 52,53,54,55]. The absence of such information hinders our ability to generalize treatment outcomes to specific sub-groups and understand the potential moderating effects of these factors [56, 57]. Representation of diverse sub-groups is crucial as it promotes inclusivity and ensures comprehensive reporting in clinical trials, enabling the application of trial findings to diverse populations and informing equitable healthcare practices. To encourage consistency in how such results are reported, some journals, such as those published by the American Heart Association, provide guidance for authors submitting manuscripts that report health differences by race/ethnicity [58].

Insufficient attention has been given to addressing the geographic and regional variability in PAD RCTs. This variability is likely influenced by local policies that can significantly impact the conducting and reporting of clinical trials. Regional policies, including regulatory requirements, reimbursement practices, and research infrastructure requirements, can create barriers and affect the feasibility of conducting and reporting trial data. Such policies may introduce increased costs or burdens that hinder participation or data collection, ultimately impacting the generalizability of treatment outcomes. It is crucial to acknowledge and account for these regional policy differences to ensure the validity and applicability of trial findings across diverse geographical settings.

In the identified protocols, there was a reliance on traditional recruitment strategies that primarily targeted participants from clinics and academic settings. Careful site selection can help increase the diversity of both patient populations and the research team. Additionally, it is important to recognize the need for broader inclusion and adoption of non-traditional recruitment strategies to enhance the representation of diverse and URMs in clinical trials. Thus, movement toward the inclusion and adoption of non-traditional recruitment strategies are necessary to boost the inclusion of diverse and under-represented groups. Expanding the eligibility criteria beyond traditional parameters, providing training on implicit bias and cultural competence, and increasing the diversity of funding committees and reviewers may help increase diversity in trials [46].

The effective management of PAD requires a multifaceted approach with strategies anchored by several factors, such as patients, healthcare systems and providers, and scientific advancements. To address the complexities associated with PAD, it is important for trial protocols to integrate approaches that address each of these components [59]. Collaborative initiatives among various stakeholders (academia, regulatory bodies, industry stakeholders, and healthcare payors) are crucial in facilitating the conduct of clinical trials focused on cardiovascular conditions, including PAD [60]. Such inter-agency collaborations foster the timely introduction of innovative therapies and enhance the overall management of cardiovascular diseases.

A major strength of this study is the inclusion of research from around the globe, rather than just a single country or geographic location, which increases the external validity of the findings. This was accomplished by using a variety of databases from different sources, thereby maximizing the inclusion of published trials and increasing the volume of included studies that evaluated the diversity of clinical trials. Additionally, this study used three types of data sources that centered on RCTs (trial registrations, protocols, and peer-reviewed publications), which offer stakeholders comprehensive information about the diversity of clinical trials from trial design, trial reporting, and trial outcomes on studies that are in progress or have been completed. The findings offer insights to inform policy and clinical decision-making in RCTs.

This study has several limitations. The study was limited by the small number of studies identified, which potentially threatens the generalizability of the study findings. It may be that expanding the search criteria would include more studies; the requirement of a sample size of at least 50 participants may have excluded studies with more diverse patient populations. This study observed missing data or inconsistencies between the reporting of information in clinical trial registries versus publications. Other studies have reported on the quality of clinical trial data submission and indicated a need to improve the reporting of results posted in trial registries [61]. For instance, in Clinicaltrials.gov, some studies reported extensive details regarding locations, patient population, included protocols, and results, while other studies reported limited information on trial features. Unless additional details are provided in the publications, the variability in the quality of reporting is a limitation. The use of non-study level variables (proportion of women, mean age) in a meta-regression should be interpreted with caution since they are subject to ecological fallacy [62]. Lastly, the variability in the methodological quality ratings (blinding, allocation concealment, etc.), could potentially introduce a source of bias in the study results, impacting the reliability of the conclusions drawn from this analysis.

Despite these limitations, this review holds implications for clinical practice, policy, and future research. First, these findings highlight potential issues that can undermine the reliability and validity of study findings in lower-extremity PAD RCTs. Addressing these issues is crucial to enhancing the evidence-base for clinical decision-making and improving clinical outcomes for the management of PAD. Additionally, the observed inequities in clinical trial study populations emphasize the importance of health equity for URMs. Regulatory and decision-making bodies globally have promoted guidelines aimed at improving representation in clinical trials [63,64,65]. Countries and regions without universally-accepted guidelines promoting clinical diversity should pursue the development of such guidelines, using existing resources as guides. In the US, the Food and Drug Administration recently released the final guidance on Clinical Trial Diversity Plans [66] driven by legislative mandates. Approaches for inclusive trials have been reported in the literature [47, 67,68,69,70,71]. Standardization efforts are needed to ensure transparency, accountability, and progress in achieving health equity while considering the cultural and social context of trial locations.

Future research must encourage investigators and life sciences industry representatives to increase investments and diversify resources to improve the design of clinical research. This includes expanding the inclusion of regions and populations underrepresented in clinical trials. Integrating a health equity lens into trial design is crucial, with a focus on ensuring fair and equitable representation of diverse populations. It is equally important to emphasize the reporting and the interpretation of trial results by key clinical outcomes through an equity perspective [72]. In addition to addressing representation, it is essential to consider the potential burden and costs that participants may incur when participating in clinical trials. Direct costs (e.g., travel expenses to the trial site) and indirect costs (e.g., productivity loss) can have an impact on participant motivation and retention. Thus, PIs should explore existing incentives (e.g., travel reimbursement) and develop strategies to boost retention in clinical trials [73]. Future research should consider exploring the role of demographic characteristics beyond age, sex, and race in treatment outcomes.