Introduction

The international incidence and prevalence of peripheral artery disease (PAD) is rising,1 and people with PAD are typically older, current or past cigarette smokers and have multiple comorbidities, including diabetes, coronary artery disease (CAD) and cerebrovascular disease (CVD).2 The care of people with PAD is costly as they have a high annual incidence of visits to primary healthcare providers, emergency departments and vascular specialists; hospital admissions; open and endovascular lower limb revascularisation procedures and minor (below-ankle) and major (above-ankle) lower limb amputations.3 Those with chronic limb-threatening ischaemia (CLTI), an advanced form of PAD manifested by ischaemic rest pain, tissue loss or toe or foot gangrene, suffer a substantial burden of disability and pain and >60% visit the emergency department annually.4–7

International clinical practice guidelines strongly and consistently recommend that people with PAD be prescribed antiplatelet and statin [ie, 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase inhibitor] medications because class-1 evidence supports that the benefit of these medications greatly outweighs their risks.5 8–11 They also strongly recommend that all those with PAD and hypertension are prescribed antihypertensive medications (and many guidelines suggest that these should preferably be angiotensin-targeted agents).5 8–11 These recommendations mirror those for people with CAD and CVD because antiplatelets, statins and antihypertensives reduce risk of myocardial infarction, stroke and death in large, well-designed and conducted randomised controlled trials (RCTs) that enrolled participants with PAD, CAD and/or CVD.5 8–11 RCTs that enrolled PAD patients have also reported that these medications reduce risk of lower limb revascularisation, acute lower limb ischaemia and major lower limb amputation, an outcome rated by many people with PAD as worse than death.12–15

However, several cohort studies have reported that antiplatelet, statin and antihypertensive medications may be underprescribed to adults with PAD, especially when compared with those who have CAD or CVD.16–25 In support of this, a 2007 study conducted in a Canadian tertiary care hospital reported that 69% of people with PAD were not prescribed a statin and 48% with PAD and hypertension were not prescribed an angiotensin-converting enzyme (ACE)inhibitor.26 Further, a recent cross-sectional survey found that less than half of vascular surgeons (the specialists who most commonly medically and surgically manage patients with PAD) routinely initiated or modified statin therapy and fewer than 10% prescribed angiotensin-targeted or other antihypertensive therapy.27

Objectives

No evidence synthesis has examined the frequency of clinician underprescription of and patient non-adherence to guideline-recommended cardiovascular medications among adults with PAD, patient and clinician characteristics that predict underprescription of and non-adherence to these medications, and associations between underprescription of and non-adherence to these medications and adverse health and healthsystem outcomes. The primary objective of this systematic review is, therefore, to meta-analyse reported direct estimates of the incidence and prevalence of healthcare provider underprescription of and patient non-adherence to guideline-recommended medications in adults with PAD. Secondary objectives are to identify and summarise characteristics of the patient and treating clinician that predict clinician underprescription of and patient non-adherence to guideline-recommended medications in multivariable, adjusted analyses and determine whether underprescription and non-adherence is associated with an increased adjusted risk of mortality, major adverse cardiac and limb events (including revascularisation procedures and major amputations), other morbidities, healthcare resource use and costs. We will include adjusted instead of unadjusted predictor estimates because these are recommended by rigorous systematic review methodological guidance documents to examine the independent prognostic value of these predictors over and above (ie, adjusted for) other prognostic factors.28 Results of the work will be used to identify international evidence-care gaps for adults with PAD and inform where implementation interventions may be required to improve healthcare provider prescribing of guideline-recommend cardiovascular medications to people with PAD and patient adherence to these prescribed medications.

MethodsProtocol, reporting and registration

We prespecified our methods following recommendations for conducting systematic reviews and meta-analyses of prognostic factor studies.28–30 This protocol is reported according to the Preferred Reporting Items in Systematic Reviews and Meta-Analyses Protocols statement31 32 (see online supplemental appendix A) and Sex and Gender Equity in Research (SAGER) guidelines33 (see online supplemental appendix B). It is registered on PROSPERO, the international prospective register of systematic reviews (PROSPERO registration number: CRD42022362801). The start date of the study was 26 June 2023 while the planned end date (submission of the manuscript for peer-review) is 1 November 2024.

Clinical questions

We formulated study clinical questions according to suggested frameworks for posing clinical questions for systematic reviews of prognostic factor studies.29 30 34

Primary clinical question

In adults (age ≥18 years) with PAD, what are the pooled cumulative incidence, incidence rate and point or period prevalence of clinician underprescription of and patient non-adherence to guideline-recommended cardiovascular medications?

Secondary clinical question

In adults (age ≥18 years) with PAD, does the pooled clinician underprescription of and patient non-adherence to guideline-recommended medications vary by country, characteristics of the treating clinician or patient, population-based design or study risks of bias?

In adults (age ≥18 years) with PAD, which characteristics of the treating clinician and patient increase the pooled adjusted odds of underprescription of or non-adherence to guideline-recommended cardiovascular medications?

In adults (age ≥18 years) with PAD, is clinician underprescription of or patient non-adherence to guideline-recommended medications associated with an increased pooled adjusted odds of mortality, major adverse cardiac and limb events (including revascularisation procedures and major amputations), other morbidities, healthcare resource use and cost?

Definitions

We will define underprescription as not prescribing one or more guideline-recommended cardiovascular medications to adults with PAD. We will define patient medication non-adherence as not initially filling a prescription, failing to follow its medication instructions for use and/or failure to refill the prescription and therefore continue taking it despite being recommended by their healthcare provider.35 We will define PAD as per the 2016 American College of Cardiology/American Heart Association (ACC/AHA) guideline as atherosclerotic disease of the lower limb arteries, including the aortoiliac, femoropopliteal and infrapopliteal arterial segments, and excluding nonatherosclerotic disease of the lower extremity (eg, fibromuscular dysplasia).5 However, alternate definitions of PAD used by study authors will also be accepted.

Clinical practice guideline-recommended cardiovascular medications for PAD will be defined as antiplatelets (eg, aspirin, clopidogrel), statins and antihypertensives [eg, ACE-inhibitors, angiotensin receptor blockers (ARBs), beta-blockers, calcium-channel blockers (CCBs), thiazide diuretics] (for people with PAD and concurrent hypertension). These are medications that are consistently recommended across multiple international evidence-based PAD clinical practice guidelines.5 8–11 35 36 Since there is some variation in specific recommendations, we will accept individual study authors’ definition of underprescription where underprescription was defined as per a certain published guideline and setting (see online supplemental appendix C for a comparison of medical therapy recommendations across PAD guidelines).

Antiplatelet therapy, antihypertensive drugs (for those with hypertension and PAD) and statins have been recommended in various ACC/AHA guidelines, including the 2005 PAD guideline.36 Some discrepancies exist between the European guidelines, American guidelines and the recently published Canadian guideline.11 37 All three recommend antiplatelets for symptomatic PAD; however, they differ with regard to asymptomatic PAD. The European Society of Cardiology-European Society for Vascular Surgery and Canadian Cardiovascular Society guidelines do not recommend antiplatelets in asymptomatic patients, while the ACC/AHA guideline does.11 The recommendation to treat hypertension with an antihypertensive in people with PAD has been consistent across guidelines for years.36 The most recent American, Canadian and European guidelines recommend prescribing statins to all PAD patients. Medications that are consistently recommended across guidelines include antiplatelet therapy (eg, aspirin, clopidogrel) for symptomatic PAD, antihypertensive therapy (eg, ACE-inhibitors, ARBs, beta-blockers, CCBs, thiazide diuretics) for PAD and concurrent hypertension, and statins in patients with an LDL cholesterol ≥2.5 mmol/L/≥100 mg/dL.5 8–11

Information sources

We will search MEDLINE; EMBASE and Evidence-Based Medicine Reviews (which includes ACP Journal Club; the Cochrane Central Register of Controlled Trials, Database of Systematic Reviews, and Methodology Register Database; Database of Abstracts of Reviews of Effects; Health Technology Assessment Database; and National Health Service Economic Evaluation Database) from 1 January 2006, without restrictions. We will start our search in 2006 as this is the year after publication of the first PAD treatment clinical practice guideline by ACC/AHA.38 To identify additional citations, we will use the PubMed ‘related articles’ feature and manually search bibliographies of included studies and relevant review articles identified during the search.

Search strategy

We created the MEDLINE and EMBASE search strategies with the assistance of an information scientist/medical librarian (RS). Using a combination of Medical Subject Heading (MeSH) terms and keywords, search filters were constructed covering the themes PAD and underprescription/non-adherence. For PAD, we extracted disease-related keywords and MeSH subject headings used in a recent meta-analysis examining an exercise intervention for PAD.39 For underprescription/non-adherence, we extracted keywords and MeSH subject headings used in a systematic review examining medication underuse in older adults.40 We then used those terms to search for additional relevant studies in PubMed and extracted the MeSH terms that those studies were indexed under. After the MEDLINE search strategy was created, we submitted it to another information scientist/medical librarian to peer-review it using the Peer-Review of Electronic Search Strategies (PRESS) guideline41 (see box 1 for our PRESS’d MEDLINE search strategy). Subsequently, we searched for Emtree terms that were similar to the above MeSH terms in EMBASE and created a list of non-MeSH/non-Emtree keywords for PAD guideline-recommended medications and underprescription/non-adherence (box 1).

Box 1 PRESS’d search strategiesOvid MEDLINE

Arterial Occlusive Diseases/

Arteriolosclerosis/

Arteriosclerosis/

Arteriosclerosis Obliterans/

Intermittent Claudication/

Intermittent Claudic*.tw,kf.

arteriosclero*.tw,kf.

exp Peripheral Vascular Diseases/

(limb adj2 isch?em*).tw,kf.

(periph* adj2 arter* adj2 disease*).tw,kf.

or/1–10

(under utili* or underutili*).tw,kf.

“under use*".tw,kf.

underusage.tw,kf.

underuse*.tw,kf.

under usage.tw,kf.

underprescri*.tw,kf.

under prescri*.tw,kf.

(under treat* or undertreat*).tw,kf.

((inadequate or deficien* or insufficien* or substandard or suboptimal) adj3 (treatment or management or control or therap*)).tw,kf.

Health Services Accessibility/ or “Delivery of Health Care”/ or Practice Patterns, Physicians'/

Guideline Adherence/ or Prescriptions/ or Drug Prescriptions/ or Drug Utilization/

Medication Adherence/ or “Treatment Adherence and Compliance”/

((prescription or prescribing) adj2 (rate* or practice*)).tw,kf.

adheren*.tw,kf.

((treatment or practice) adj2 pattern*).tw,kf.

(noncomplian* or nonadheren*).tw,kf.

((treatment or prescribing or therapy) adj3 complian*).tw,kf. or complian*.ti.

or/12–28

11 and 29

limit 30 to yr=“2006 -Current”

exp animals/ not humans/

31 not 32

33 use medall

Ovid EMBASE

exp peripheral occlusive artery disease/

intermittent claudication/ or Intermittent Claudic*.tw.

(limb adj2 isch?em*).tw.

(periph* adj2 arter* adj2 disease*).tw.

arteriolosclerosis/ or arteriosclerosis/ or arteriosclero*.tw.

or/35–39

(under utili* or underutili*).tw.

“under use*".tw.

underusage.tw.

underuse*.tw.

under usage.tw.

underprescri*.tw.

under prescri*.tw.

(under treat* or undertreat*).tw.

((inadequate or deficien* or insufficien* or substandard or suboptimal) adj3 (treatment or management or control or therap*)).tw.

*health care access/ or unmet medical need/

*health care delivery/

*clinical practice/

((treatment or practice) adj2 pattern*).tw.

((prescription or prescribing) adj2 (rate* or practice*)).tw.

protocol compliance/

drug utilization/

*“drug use”/ or *prescription/

((treatment or prescribing or therapy) adj3 adheren*).tw. or adheren*.ti.

((treatment or prescribing or therapy) adj3 complian*).tw. or complian*.ti.

(noncomplian* or nonadheren*).tw.

or/41–60

40 and 61

(exp animal/ or nonhuman/) not exp human/

62 not 63

limit 64 to yr=“2006 -Current”

65 use emczd

34 or 66

Data management and selection process

The titles and abstracts of citations identified during the search will be imported into Rayyan Systematic Review Software (https://www.rayyan.ai/).42 Two investigators (DdL and MP) will use Rayyan to remove duplicates, independently review titles and abstracts of articles identified by the search and select any article deemed potentially relevant by either investigator for full-text review. These two investigators will subsequently review the full text of all potentially relevant citations and select studies for inclusion in the systematic review. Disagreements regarding study inclusion will be resolved via consensus or arbitration by the senior investigator (DJR). Chance-corrected agreement between investigators regarding full-text inclusion will be calculated using a kappa statistic.43

Eligibility criteria and outcomes

We will use the following inclusion criteria30 34:

The study included adults (age ≥18 years) with PAD.

The study reported one or more of the following outcomes (or these outcomes could be calculated from the data provided):

Cumulative incidence, incidence rate or point or period prevalence of clinician underprescription of or patient non-adherence to guideline-recommended medications in adults with PAD.

ORs, risk ratios (RRs) or HRs [and surrounding standard errors (SEs) or 95% confidence intervals (CIs)] adjusted for the presence of other clinician (eg, specialty, years of training) and patient (eg, age, rural vs urban residence) risk and confounding factors and relating one or more potential risk factor of interest to the clinician underprescription of or patient non-adherence to guideline-recommended medications for PAD.

ORs, RRs, HRs or other measures (and surrounding SEs or 95% CIs) describing differences in mortality, major adverse cardiac and limb events (including revascularisation procedures and major amputations), other morbidities, healthcare resource use and costs associated with clinician underprescription of or patient non-adherence to guideline-recommended medication for PAD and adjusted for the presence of other risk factors or confounding factors.

The study design was observational (ie, cohort, case–control or cross-sectional, including studies nested within RCTs44 45).

We will exclude studies that were (1) grey literature; (2) published only as an abstract; (3) only enrolled patients before the year 2006; (4) only reported unadjusted risk factors for underprescription or non-adherence or unadjusted associations between underprescription or non-adherence and outcomes or (5) did not distinguish between clinician underprescription and patient non-adherence (eg, reported underuse without a description).

Data items and collection process

Two investigators will independently extract data in duplicate using a data extraction tool piloted on a random sample of five included studies (see table 1 for data items to be extracted). Where reported comparisons between the frequency of prescription of guideline-recommended medications to patients with PAD instead of CAD or CVD, these will also be extracted as well. Three investigators will independently extract data when they are only presented visually (eg, a bar graph) and then their results will be averaged.

Table 1

Data items to be extracted from included studies

Risk of bias assessment

Two investigators will independently evaluate the risk of bias of studies reporting incidence and prevalence estimates using the Joanna Briggs Institute’s critical appraisal checklist of studies reporting prevalence data.29 The Joanna Briggs checklist includes questions about whether the sample frame was appropriate to address the target population, participants were sampled in an appropriate way, sample size was adequate, study participants (ie, both patients and treating clinicians) and setting was described in detail, the data analysis was conducted with sufficient coverage of the identified sample, valid methods were used for the identification of the condition, the condition was measured in a standard and reliable way and the statistical analyses were appropriate.29 Those studies that reported risk factors for clinician underprescription of or patient non-adherence to guideline-recommended medications for PAD or associations between underprescription and outcomes will also be independently evaluated by two investigators using the Quality in Prognosis Studies tool.46 47 This tool includes questions regarding study participation and attrition; potential risk factor and outcome description and measurement; confounding measurement and account and methods and reporting of statistical analyses.46 47 For those studies that used administrative data, we will also examine whether the study authors considered the accuracy (sensitivity and specificity) of the codes used to define variables. Disagreements regarding risk of bias assessments will be resolved by consensus or arbitration by the senior investigator.

Qualitative data synthesis

We will perform a narrative synthesis of the included studies and their reported data before considering meta-analyses.48 We will first tabulate characteristics of the included studies, including their design, data source, setting, recruitment period, included treating clinicians and patients and reported outcomes. This tabulation will help us identify potentially duplicate data and where meta-analyses may be appropriate.

Quantitative data synthesis and statistical analyses

Where it was not reported, we will calculate the cumulative incidence, incidence rate and point or period prevalence of clinician underprescription of and patient non-adherence to guideline-recommended medications for PAD. Cumulative incidence will be calculated using the following formula:

where the total population at risk will be defined as the number of adults with PAD. Incidence rate will be determined using the formula:

Point or period prevalence will be determined using the formula:

The SE and 95% CI of these proportions will be determined using the Clopper-Pearson exact binomial method. As evidence-based guidelines for PAD vary somewhat by time and across countries, we will report estimates of clinician underprescription according to the clinical practice guideline setting and time during which it was published.

Where we identify multiple studies that provide non-overlapping or non-duplicated data estimates of clinician underprescription of or patient non-adherence to guideline-recommended medications for PAD, incidence or prevalence estimates will be pooled using DerSimonian and Laird random-effects models.49 These pooled analyses will be done according to setting and clinical practice guideline source. As suggested by Barendregt et al, we will first transform these proportional estimates using a double arcsine transformation prior to meta-analyses.29 50 The data will then be back-transformed to incidence and prevalence estimates after meta-analyses.29

We will use the OR (for dichotomous outcomes) or (standardised) mean difference (for continuous outcomes) as the summary measures of choice for pooled risk factor and outcome analyses. Similar adjusted risk factor estimates and outcome associations will be pooled using DerSimonian and Laird random-effects models.49 Where the OR was not reported, we will pool RRs or HRs instead. When adjusted estimates were calculated from the same data source across several studies, we will include the estimate derived from the largest study. As a sensitivity analysis, we will also recalculate the estimate using that derived from the potentially overlapping study that reported the most adjusted estimates as studies may have variably adjusted their estimates for potentially confounding factors.

We will inspect forest plots, calculate I2 inconsistency statistics and conduct tests of homogeneity to assess for interstudy heterogeneity in the above estimates.51–53 We will consider I2 statistics >25%, >50% and >75% to represent low, moderate and high degrees of heterogeneity, respectively.52 In the presence of at least low interstudy heterogeneity in our pooled estimates of incidence and prevalence, we will conduct subgroup meta-analyses and meta-regression. We will use the following predictor variables to explore heterogeneity in these stratified meta-analyses and meta-regressions: country; percentages of patient sex, race and socioeconomic status and patients with CAD, CVD, PAD, pulmonary disease, diabetes, chronic kidney disease, cancer and a past or present smoking history; percentages of clinicians’ sex, practice type (eg, primary community care vs tertiary care centre), clinician training (medicine, nursing) and clinician subspecialty (general practice, nurse practitioner, vascular surgery, general internal medicine, cardiology other) and population-based design versus not.

We will evaluate for evidence of small study effects potentially due to publication bias by visually inspecting funnel plots of incidence and prevalence of underprescription and using Egger’s tests.54 We will use the study sample size instead of the inverse of the SE on the y-axis as this may perform more favourably in these analyses.29 55 Statistical analyses will be performed by a trained meta-analyst using Stata V.13.0 (StataCorp).

Certainty in the cumulative evidence

We will use Grading of Recommendations, Assessment, Development and Evaluation to assess certainty in the estimates of associations between the reported risk factors and clinician underprescription and patient non-adherence and between underprescription/non-adherence and outcomes.56 We will first assess the risk of bias, imprecision, inconsistency, indirectness and publication bias associated with the evidence for the reported risk factors.57–61 Estimate certainty will then be adjudicated as high (further research is very unlikely to change the estimate), moderate (further research could have an important impact, which may change the estimate) or low (further research is very likely to have an important impact, which is likely to change the estimate).

Patient and public involvement

There is no patient involvement in the development of this systematic review.

Ethics and dissemination

No ethics approval is required for this study as it includes previously published data. International clinical practice guidelines have strongly and consistently recommended that antiplatelets, statins and antihypertensives be prescribed to adults with PAD to prevent morbidity, mortality, lower limb revascularisation and major amputation. This study seeks to determine how often these medications are underprescribed by clinicians to these patients and how often patients do not adhere to them after prescription. We also seek to compare the frequency with which these medications are prescribed to those with PAD instead of CAD or CVD, identify patient and treating clinician characteristics that predict underprescription of and non-adherence to these guideline-recommended medications in adults with PAD, and estimate outcomes associated with underprescription of and non-adherence to these medications in people with PAD. Finally, as sex-based differences in PAD mortality have been observed,62 we will also examine whether the above varies by patient sex.

This proposed systematic review has both strengths and limitations. The strengths of our study include the creation of a detailed protocol in accordance with rigorous systematic review conduct and reporting and SAGER guidelines; the piloted and peer-reviewed search strategy; and our extensive preplanned meta-analyses, stratified meta-analyses and meta-regressions. A limitation is likely a reliance on studies using administrative health data, which may put our meta-analyses at variable risk for misclassification bias. An additional concern with administrative data studies is that their measurement of complications has been suggested to have high specificity, but low sensitivity.63 A final important limitation is the slight inconsistencies that exist between evidence-based guidelines for PAD across time and countries. To account for this, we will report data for underprescription according to the clinical practice guideline setting and time during which it was published.

The aim of this systematic review will be to identify evidence-care gaps for PAD, compare these gaps across different countries and settings and identify those patients at highest risk for clinician underprescription and patient non-adherence and physician characteristics related to underprescribing and non-adherence. We will also seek to quantify the importance of these gaps, notably how underprescription of and non-adherence to these medications influences PAD patient outcomes and the burden on the healthcare system. If our study identifies that an important gap exists between clinical practice guideline recommendations and healthcare provider and patient behaviours, it may justify design and testing of implementation strategies to improve prescription of guideline-recommended cardiovascular medications to adults with PAD and possibly patient adherence to these medications after prescription.