Network Meta‐Analysis Comparing the Outcomes of Treatments for Intermittent Claudication Tested in Randomized Controlled Trials

Abstract

Background

No network meta‐analysis has considered the relative efficacy of cilostazol, home exercise therapy, supervised exercise therapy (SET), endovascular revascularization (ER), and ER plus SET (ER+SET) in improving maximum walking distance (MWD) over short‐ (<1 year), moderate‐ (1 to <2 years), and long‐term (≥2 years) follow‐up in people with intermittent claudication.

Methods and Results

A systematic literature search was performed to identify randomized controlled trials testing 1 or more of these 5 treatments according to Preferred Reporting Items for Systematic Review and Meta‐Analysis guidelines. The primary outcome was improvement in MWD assessed by a standardized treadmill test. Secondary outcomes were adverse events and health‐related quality of life. Network meta‐analysis was performed using the gemtc R statistical package. The Cochrane collaborative tool was used to assess risk of bias. Forty‐six trials involving 4256 patients were included. At short‐term follow‐up, home exercise therapy (mean difference [MD], 89.4 m; 95% credible interval [CrI], 20.9–157.7), SET (MD, 186.8 m; 95% CrI, 136.4–237.6), and ER+SET (MD, 326.3 m; 95% CrI, 222.6–430.6), but not ER (MD, 82.5 m; 95% CrI, −2.4 to 168.2) and cilostazol (MD, 71.1 m; 95% CrI, −24.6 to 167.9), significantly improved MWD (in meters) compared with controls. At moderate‐term follow‐up, SET (MD, 201.1; 95% CrI, 89.8–318.3) and ER+SET (MD, 368.5; 95% CrI, 195.3–546.9), but not home exercise therapy (MD, 99.4; 95% CrI, −174.0 to 374.9) or ER (MD, 84.2; 95% CrI, −35.3 to 206.4), significantly improved MWD (in meters) compared to controls. At long‐term follow‐up, none of the tested treatments significantly improved MWD compared to controls. Adverse events and quality of life were reported inconsistently and could not be meta‐analyzed. Risk of bias was low, moderate, and high in 4, 24, and 18 trials respectively.

Conclusions

This network meta‐analysis suggested that SET and ER+SET are effective at improving MWD over the moderate term (<2 year) but not beyond this. Durable treatments for intermittent claudication are needed.

Nonstandard Abbreviations and Acronyms

CrI

credible interval

ER

endovascular revascularization

HET

home exercise therapy

MCMC

Markov Chain Monte Carlo

MD

mean difference

MWD

maximum walking distance

NMA

network meta‐analysis

SET

supervised exercise therapy

SF‐36

36‐Item Short Form Health Survey

WIQ

Walking Impairment Questionnaire

Clinical Perspective

What Is New?

• This network meta‐analysis of randomized controlled trials suggested that home exercise therapy, supervised exercise therapy, and a combination of supervised exercise therapy with peripheral revascularization significantly improved walking distance in patients with intermittent claudication over the short term compared with controls.

• None of the tested treatments improved walking distance significantly compared with controls in the longer term.

What Are the Clinical Implications?

• Larger and better designed clinical trials testing treatments for intermittent claudication are needed.

Intermittent claudication is a common cause of walking impairment, reduced health‐related quality of life (QOL) and low physical activity. ¹ The main treatment options for intermittent claudication are the medication cilostazol, supervised exercise therapy (SET), home exercise therapy (HET), and endovascular revascularization (ER). ² Despite a number of randomized controlled trials, systematic reviews, and traditional and network meta‐analyses (NMA), there remains controversy regarding the most appropriate therapy. ³ , ⁴ , ⁵ , ⁶ Assessment of the available evidence regarding the most effective treatment for intermittent claudication is difficult because of the absence of head‐to‐head comparison of all available therapies in 1 trial or in previous meta‐analyses. ³ , ⁴ , ⁵ Most randomized clinical trials have included a small number of participants or involved a limited number of the available therapies, meaning standard meta‐analyses are not ideally suited to testing the most effective treatment. NMA have been designed to combine all available evidence for randomized trials, including treatments not compared head to head. A number of prior NMA have compared treatments for intermittent claudication but these have been subject to a number of limitations. ⁴ , ⁷ , ⁸ , ⁹ These include not incorporating all available treatments options, not examining outcomes at different follow‐up times, and including therapies, such as older endovascular approaches, which are not contemporary. ⁴ , ⁷ , ⁸ , ⁹ Assessing long‐term, as well as short‐term outcomes is important because recent evidence suggests that both exercise therapy and endovascular revascularization are not durable. ⁶ , ¹⁰ Furthermore, long‐term results from 1 clinical trial have been reported since the publication of the most recent NMA and need to be incorporated into the available evidence. ¹¹ In order to address these limitations this NMA examined data from randomized controlled trials published in the past 2 decades that tested cilostazol, exercise therapy, and ER for treating intermittent claudication.

METHODS

All data used in this study are available in the article and supplementary files.

Search Strategy

The systematic review was performed according to the Preferred Reporting Items for Systematic Review and Meta‐Analysis with an extension for NMA statement (PRISMA‐NMA) and the study protocol was registered in the PROSPERO (International Prospective Register of Systematic Reviews) database (Registration Number: CRD42020197141). The literature search was conducted by 1 author (S.T.). The databases PubMed (Medline), Cochrane Central Register for Controlled Trials, and Web of Science were searched on April 21, 2020. The full search strategy included terms related to intermittent claudication, treatment, and walking capacity (Data S1).

Study Selection

Randomized controlled trials testing cilostazol, SET, HET, ER and combination of any of these strategies for treating intermittent claudication were eligible for inclusion. Control patients were defined as those under attention control, best or optimal medical treatment alone, or receiving placebo. HET was defined as either written exercise advice or a structured program consisting of exercise sessions that were mainly performed without supervision at home. SET was defined as an exercise program performed under direct supervision of a trained health professional. ER was defined to include angioplasty, stenting, or related endovascular procedures. All included trials had to report maximum walking distance (MWD) measured in meters during a standardized treadmill test. Trials that were published as full texts or abstracts were eligible for inclusion if the minimum data requirement (MWD at entry and at least 1 follow‐up time) was published or available from the corresponding author. When multiple publications arising from the same clinical trial were identified, data for all applicable time points were included. Trials published in languages other than English, nonrandomized or crossover trials, and observational studies and trials where baseline and at least 1 follow‐up time point MWD were not available were excluded. In order to compare contemporary treatments, because of the substantial evolution in the techniques used for endovascular revascularization over time, trials published before 2000 were excluded. Unpublished studies and abstracts where minimum data were not available were also excluded. Eligibility was determined by 2 authors (S.T. and J.P.), with discrepancies resolved by discussion with the senior author (J.G.).

Data Extraction

Primary outcome data were extracted on a customized spreadsheet by 2 authors (S.T. and S.W.) and findings were validated by 2 authors (J.P. and M.I.). Secondary outcomes were extracted by 2 authors (S.T. and P.H.) and the data were validated by 1 author (J.P.). Study characteristics were extracted by 3 authors (S.T., J.P., and C.S.). Any inconsistencies were resolved through discussion and confirmed with the senior researcher (J.G.). The primary outcome was MWD (in meters) measured in a standardized treadmill test. In trials that did not report MWD, other data including peak walking time, absolute claudication distance, or maximum walking time were used to derive MWD considering the type of treadmill test used. Secondary outcomes included QOL and serious adverse events (SAE). Because multiple different QOL questionnaires were used in the included trials, it was not possible to perform a pooled analysis. QOL data were tabulated and findings reported qualitatively. SAEs were reported as the numbers of patients who had a myocardial infarction, stroke, events requiring hospital admission, lower limb revascularization, any other vascular procedure, any amputation, or death. In patients who were randomized to ER, only secondary lower limb revascularization procedures were considered as SAEs. The following additional data were extracted from the included trials: age, sex, body mass index, smoking, hypertension, diabetes mellitus, ankle brachial pressure index, medications, sample size, type of treadmill test, design of HET or SET program, type of ER, and duration of follow‐up. Primary outcome data were categorized depending on the follow‐up time as <1 year (short‐term follow‐up), ≥1 to <2 year (moderate‐term follow‐up), and ≥2 year (long‐term follow‐up). Mean difference (MD) was calculated as the difference in mean values between the specified follow‐up and baseline. SD of change scores was imputed using the formula ¹² : ${\text{sd}}_{\text{change}}=\surd \left[{\text{sd}}_{\text{baseline}}^2+{\text{sd}}_{\text{Final}}^2‐\left(2\times \text{Corr}\times {\text{sd}}_{\text{baseline}}\times {\text{sd}}_{\text{Final}}\right)\right]$. Pearson correlation coefficient (Corr in the formula) was calculated using individual participant data from 1 trial ¹³ as used previously. ¹⁴ The relative effect was expressed as MD±95% credible interval (CrI). CrI denotes the predictive distribution or interval within which a potential unobserved fixed effects parameter value could fall with a particular probability. CrI is the Bayesian statistics equivalent of CI. The statistical analyses were overseen by a senior researcher (J.M.) and the network model was validated by another senior researcher with extensive statistical expertise (R.J.).

Statistical Analysis

The Bayesian random effects NMA was performed using the R statistical package gemtc, which uses the arm‐based model and provides effect size estimates for multiple comparisons. ¹⁵ The package was available through R studio version 3.4.4 from the Comprehensive R Archive Network (CRAN) at https://cran.r‐project.org/web/packages/gemtc/index.html. The package contains functions that use Just Another Gibbs Sampler, a program for analyzing Bayesian hierarchical models using Markov Chain Monte Carlo (MCMC) simulation. ¹⁶ The gemtc package uses Just Another Gibbs Sampler software to develop a random effects model assuming consistency with variance scaling factor of 2.5 and noninformative normal prior distribution. Between‐trial SDs were assumed to follow a noninformative uniform distribution and a weakly informative prior distribution. MCMC simulations were run using 3 chains with different initial values for 100 000 iterations. Convergence of the resulting model was assessed using trace plots. Data were presented according to short‐term, moderate‐term, and long‐term follow‐up periods. Network diagrams of all available treatments from the included trials were plotted to compare and illustrate multiple treatment arms at different follow‐up times. MCMC simulations were performed to estimate the posterior distributions of MWD data between different arms by running the simulations long enough to reach accurate estimates for the model. Convergence of the network models derived from MCMC simulations were assessed using trace and density plots. The stability of the MCMC simulation output was tested with Gelman‐Rubin‐Brooks plots and diagnostics were performed to determine Potential Scale Reduction Factor score. A Potential Scale Reduction Factor score of <1.05 is suggestive of good convergence in the network model. Inconsistencies within the network model were assessed using a more sophisticated node split method; which estimates the random effects of different comparison arms when using direct alone, indirect alone, and all available network evidence separately. ¹⁷ Any significant differences in estimates between direct and indirect evidence were considered as potential presence of large inconsistencies. In addition, ranking probability was calculated for all available treatment strategies for each follow‐up time. ¹⁸ A P value of ≤0.05 was considered as statistically significant.

Risk of Bias Assessment

Two authors (S.T. and J.P.) independently assessed the risk of bias of all included trials using the Cochrane Collaboration's tool, which assessed key aspects of the reports including random sequence generation, statistical sample size estimate, the number of patients who completed the study, the primary outcome, blinding of outcome assessors, intent‐to‐treat principle, and other biases. ¹⁹ Quality parameters were marked as (+), (–), and (?) for positive, negative, and unclear risk of bias respectively. Any study having 3 or more negative quality assessments were considered to have high risk of bias. Trials with 1 to 2 negative quality assessments were considered to have moderate risk of bias and those with no negative quality assessments were considered to have low risk of bias. Any inconsistencies were resolved through discussion between the authors until a consensus was reached.

RESULTS

Included Trials and Participants

After screening 46 clinical trials involving a total of 4256 patients were included (Figure 1). Forty‐two trials involving 3515 patients reported MWD (in meters) at short‐term follow‐up. ⁶ , ¹³ , ²⁰ , ²¹ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ , ³¹ , ³² , ³³ , ³⁴ , ³⁵ , ³⁶ , ³⁷ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ , ⁴² , ⁴³ , ⁴⁴ , ⁴⁵ , ⁴⁶ , ⁴⁷ , ⁴⁸ , ⁴⁹ , ⁵⁰ , ⁵¹ , ⁵² , ⁵³ , ⁵⁴ , ⁵⁵ , ⁵⁶ , ⁵⁷ , ⁵⁸ , ⁵⁹ Fourteen trials involving 1327 patients reported MWD (in meters) at moderate‐term follow‐up ⁶ , ⁶² and 6 trials involving 601 patients reported MWD (in meters) at long‐term follow‐up. ⁶ , ⁵⁴ , ⁵⁵ , ⁵⁷ , ⁶² , ⁶³ Details such as the specific treatment strategy and outcome assessments are shown in Table S1. The baseline characteristics of the participants are shown in Table 1.

Figure 1. Preferred Reporting Items of Systematic Review and Meta‐Analyses (PRISMA) flow diagram.

A total of 4021 studies were screened and 46 randomized controlled trials (RCT) were ultimately included. MWD indicates maximum walking distance.

Table 1.

Baseline Characteristics of All Patients Included in All Treatment Arms of the Included Trials

Reference	Intervention	No. of Patients	Age (y) in Mean±SD/Median (IQR)	Male Sex (%)	Body Mass Index Mean±SD/Median (IQR)	Ankle‐Brachial Pressure Index in Mean±SD/Median (IQR)	Currently Smoking (%)	Diabetes Mellitus (%)	Coronary Heart Disease (%)	Hypertension (%)	Dylipidemia (%)	Cilostazol (%)	Statins (%)	Angiotensin‐Converting Enzyme Inhibitors (%)	Angiotensin Receptor Blockers (%)	Aspirin (%)	Anti‐Platelet Drugs (%)
57	Control	28	69 (61, 75)	69.0	25 (36, 50)	0.6 (0.5, 0.7)	67.9	21.4	17.8	15.0	92.8	NR	NR	NR	NR	NR	46.4
	ER	28	68 (56, 72)	68.0	26 (23, 28)	0.6 (0.5, 0.7)	71.4	14.3	7.1	8.0	85.7	NR	NR	NR	NR	NR	17.9
55	ER	60	74.5 (67.8, 79.0)	NR	NR	0.7 (0.6, 0.7)	6.7	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
	SET	60	75.0 (67.0, 80.0)	NR	NR	0.7 (0.6, 0.8)	15.0	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
	SET+ER	58	75.0 (70, 81)	NR	NR	0.6 (0.6, 0.7)	12.1	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
62	SET	75	NR	NR	NR	0.6 (0.6, 0.7)	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
	ER	76	NR	NR	NR	0.6 (0.6, 0.7)	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
6	ER	79	68±7.0	51.9	26±5.0	0.7±0.2	30.4	14.0	NR	NR	NR	NR	NR	NR	NR	NR	NR
	Control	79	68±6.0	53.2	26±4.0	0.7±0.1	27.8	16.0	NR	NR	NR	NR	NR	NR	NR	NR	NR
63	ER	48	71.3±5.3	45.1	NR	0.6±0.1	7.0	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
	Control	52	69.8±5.8	52.3	NR	0.6±0.2	11.0	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
56	Control	22	62.3±8.5	60.0	29.0±5.9	0.7±0.2	53.3	21.4	NR	93.3	73.2	13.3	73.3	NR	NR	NR	86.7
	SET	43	65.9±8.8	56.3	27.5±5.0	0.7±0.2	53.1	18.8	NR	90.6	78.1	18.7	78.1	NR	NR	NR	78.1
	ER	46	65.2±10.5	68.8	28.9±6.4	0.6±0.2	53.1	32.3	NR	81.3	78.1	21.9	71.9	NR	NR	NR	81.3
54a	ER+SET	48	63.9±9.0	68.7	27.0±5.1	0.7±0.1	79.2	NR	43.7	72.9	NR	NR	83.3	NR	NR	NR	91.7
	SET	45	68.5±9.4	57.8	26.9±4.5	0.7±0.1	84.4	NR	22.2	75.6	NR	NR	66.7	NR	NR	NR	88.9
54b	ER+SET	19	63.9±8.6	63.2	27.2±3.6	0.7±0.2	89.5	NR	26.3	57.9	NR	NR	57.9	NR	NR	NR	84.2
	SET	15	62.5±9.8	66.7	25.2±3.8	0.7±0.1	100.0	NR	40.0	53.3	NR	NR	80.0	NR	NR	NR	73.3
61	Control	89	67 (47–81)	67.0	NR	0.5±NR*	49.0	16.0	NR	NR	NR	NR	NR	NR	NR	NR	NR
	SET	88	67 (45–81)	66.0	NR	0.5±NR*	53.0	14.0	NR	NR	NR	NR	NR	NR	NR	NR	NR
	ER	87	66 (38–80)	64.0	NR	0.5±NR*	52.0	19.0	NR	NR	NR	NR	NR	NR	NR	NR	NR
50	ER	21	66±8.3	47.6	27.4±4.0	NR	42.8	19.0	NR	61.9	NR	NR	90.5	NR	NR	NR	95.2
	SET+ER	29	66.9±7.1	48.3	27.2±5.0	NR	37.9	10.3	NR	51.7	NR	NR	96.6	NR	NR	NR	86.2
59	ER	76	65±11.4	59.0	26±4.3	0.6±0.2	16.0	15.0	19.0	43.0	53.0	NR	NR	NR	NR	NR	NR
	SET	75	66±9.1	52.0	25±4.9	0.6±0.2	23.0	20.0	28.0	37.0	51.0	NR	NR	NR	NR	NR	NR
60	Control	11	68.1±6.8	45.0	25.2±3.2	1.2±0.1	0	0	0	9.0	NR	NR	NR	NR	NR	NR	NR
	SET	10	72.3±8.5	50.0	29.2±29.5	0.7±0.2	10.0	10.0	20.0	30.0	NR	NR	NR	NR	NR	NR	NR
51	HET	30	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
	SET	29	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
53	Control	14	71±3.7	NR	NR	0.7±0.1	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
	SET	17	72±4.1	NR	NR	0.7±0.2	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
20	SET	29	66 (58, 69)	83.0	28.7 (26, 30.4)	0.6 (0.5, 0.8)	17.0	45.0	NR	93.0	NR	31.0	69.0	NR	NR	NR	NR
	Control	35	67 (60, 76)	49.0	28.6 (24.5, 30.9)	0.6 (0.5, 0.8)	37.0	29.0	NR	74.0	NR	17.0	77.0	NR	NR	NR	NR
35	Cilostazol	89	64.5 (50–84)	94.6	NR	0.6±0.1	51.4	14.9	NR	NR	NR	NR	47.3	NR	NR	NR	87.8
	Control	87	62.9 (44–82)	89.7	NR	0.6±1	61.5	15.4	NR	NR	NR	NR	44.9	NR	NR	NR	74.4
36	Control	73	66.8±10.1	80.0	33.7±7.0	0.9±0.4	13.0	NR	NR	57.0	54.0	NR	NR	NR	NR	NR	NR
	HET	72	66.2±10.2	75.0	35.0±9.3	1.0±0.4	7.0	NR	NR	62.0	54.0	NR	NR	NR	NR	NR	NR
37	SET	11	71.3±8.5	50.0	29.2±4.1	0.7±0.1	10.0	10.0	20.0	10.0	NR	NR	NR	NR	NR	NR	NR
	Control	11	67.1±6.8	50.0	25.7±4.9	0.5±0.1	30.0	17.0	33.0	50.0	NR	NR	NR	NR	NR	NR	NR
38	Cilostazol	227	66±9.0	75.8	NR	0.7±0.2	41.4	31.7	NR	73.1	65.2	NR	NR	NR	NR	NR	NR
	Control	239	66±9.0	73.6	NR	0.7±0.4	37.7	31.4	NR	72.0	66.9	NR	NR	NR	NR	NR	NR
52	SET+ER	106	64±9.0	60.0	27.0±4.1	0.7±0.2	65.0	16.0	33.0	58.0	44.0	NR	NR	NR	NR	NR	NR
	SET	106	66±10.0	72.0	26.2±4.4	0.7±0.2	55.0	26.0	40.0	62.0	42.0	NR	NR	NR	NR	NR	NR
39	SET	31	71±1.0	89.0	NR	0.7±0.04	NR	46.0	50.0	82.0	64.0	NR	NR	NR	NR	NR	NR
	Control	30	70±1.0	92.0	NR	0.7±0.04	NR	38.0	46.0	79.0	88.0	NR	NR	NR	NR	NR	NR
25	SET	60	65±11.0	48.0	29.3±6.7	0.7±0.2	37.0	48.0	30.0	90.0	88.0	NR	NR	NR	NR	NR	NR
	HET	60	67±10.0	52.0	29.0±5.7	0.7±0.2	35.0	40.0	35.0	88.0	93.0	NR	NR	NR	NR	NR	NR
	Control	60	65±9.0	60.0	29.0±6.1	0.7±0.2	42.0	37.0	28.0	83.0	87.0	NR	NR	NR	NR	NR	NR
40	SET	106	68±8.0	86.0	27.9±4.7	0.6±0.2	46.0	26.0	NR	64.0	58.0	NR	NR	NR	NR	NR	NR
	Control	36	68±8.0	83.0	29.5±4.6	0.7±0.2	39.0	20.0	NR	64.0	60.0	NR	NR	NR	NR	NR	NR
26	HET	40	65±11.0	45.0	29.9±5.6	0.7±0.2	10.0	43.0	NR	88.0	90.0	NR	NR	NR	NR	NR	NR
	SET	40	66±12.0	45.0	29.2±7.1	0.7±0.2	10.0	43.0	NR	88.0	88.0	NR	NR	NR	NR	NR	NR
	Control	39	65±10.0	54.0	29.7±6.9	0.8±0.2	10.0	31.0	NR	79.0	85.0	NR	NR	NR	NR	NR	NR
41	Control	7	77 (56–78)	57.1	NR	NR	14.3	57.1	NR	NR	NR	NR	100.0	57.1	NR	NR	100.0
	ER	9	67 (57–76)	66.7	NR	NR	22.2	22.2	NR	NR	NR	NR	100.0	22.2	NR	NR	100.0
	SET	7	67 (58–71)	85.7	NR	NR	42.9	57.1	NR	NR	NR	NR	100.0	71.4	NR	NR	100.0
42	SET	9	66 (63–71)	77.8	NR	NR	44.4	11.1	NR	NR	NR	NR	100.0	55.6	NR	NR	100.0
	Cilostazol	9	58 (52–71)	88.9	NR	NR	33.3	33.3	NR	NR	NR	NR	100.0	33.3	NR	NR	100.0
	SET+cilostazol	7	72 (63–74)	71.4	NR	NR	28.6	14.3	NR	NR	NR	NR	100.0	57.1	NR	NR	100.0
	Control	9	67 (63.5–74)	77.8	NR	NR	33.3	22.2	NR	NR	NR	NR	88.9	55.6	NR	NR	100.0
27	SET	28 total	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
	Control		NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
43	HET	9	66±10.5	89.0	24.7±10.9	NR	33.0	11.0	11.0	22.0	56.0	NR	NR	NR	NR	NR	NR
	SET	12	69±11.8	92.0	24.9±5.3	NR	17.0	25.0	17.0	33.0	42.0	NR	NR	NR	NR	NR	NR
44	ER	35	60.2±10.0	60.0	27.5±4.9	0.7±0.2	60.0	14.3	20.0	68.6	88.6	NR	NR	NR	NR	NR	NR
	SET+ER	35	64.5±9.3	62.9	26.6±3.4	0.7±0.2	51.4	25.7	37.1	71.4	85.7	NR	NR	NR	NR	NR	NR
28	HET	18	68±7.0	67.0	25±4.0	NR	89.0	39.0	33.0	83.0	61.0	0	61.0	NR	NR	NR	94.0
	ER	9	69±7.0	100.0	27±3.0	NR	100.0	33.0	56.0	100.0	67.0	0	67.0	NR	NR	NR	100.0
45	SET	12	65.6±11.0	81.8	27.3±3.3	0.5±0.1	9.0	36.0	NR	NR	NR	NR	100.0	82.0	82.0	82.0	9.0
	Control	8	62.0±8.3	75.0	28.5±2.3	0.5±0.1	50.0	25.0	NR	NR	NR	NR	75.0	63.0	63.0	50.0	38.0
29	SET	16	69 (58–72)	50.0	31 (26–37)	0.5 (0.4–0.6)	38.0	44.0	50.0	94.0	86.0	NR	75.0	NR	NR	69.0	44.0
	HET	7	65 (55–71)	43.0	26 (20–33)	0.7 (0.6–0.8)	57.0	29.0	14.0	71.0	86.0	NR	57.0	NR	NR	57.0	29.0
30	Control	10	63.1±11.8	80.0	30.6±5.9	0.8±0.2	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
	HET	9	68.0±12.5	77.8	126.6±19.6	0.8±0.3	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
31	SET	49	61.4±6.5	85.4	NR	0.7±0.1	87.8	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
	HET	49	60.9±5.4	79.5	NR	0.7±0.1	82.0	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
32	SET	34	63.5±7.2	90.0	27.6±3.5	0.8±0.1	83.3	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
	Control	34	62.1±6.9	83.9	27.7±2.9	0.8±0.1	77.4	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
46	Cilostazol	39	M—65 (46–86)/ F—63 (49–78)	66.7	NR	0.9 (0.8v1.0)	46.1	NR	NR	56.4	74.4	NR	NR	NR	NR	NR	NR
	Control	41	M—66 (39–78)/F—68 (46–80)	65.8	NR	0.8 (0.6–0.9)	51.2	NR	NR	61.0	73.2	NR	NR	NR	NR	NR	NR
33	SET	13	66±8	76.9	26.3±4.5	0.6±0.01	38.5	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
	HET	15	62±14	80.0	27.1±4.2	0.6±0.01	40.0	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
	Control	15	67±6	66.7	27.7±6.7	0.6±0.1	46.7	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
58	SET	20	69±8	65.0	NR	0.6±0.1	100.0	35.0	40.0	100.0	100.0	NR	100.0	65.0	20.0	90.0	55.0
	Control	20	70±11	55.0	NR	0.5±0.2	100.0	45.0	35.0	100.0	95.0	NR	95.0	65.0	25.0	85.0	35.0
47	SET	30	68±7.7	66.7	25.5±4.3	0.6±0.03	23.3	26.7	NR	63.3	40.0	NR	NR	NR	NR	NR	NR
	HET	30	68±8.9	73.3	26.5±4.5	0.6±0.04	30.0	16.7	NR	53.3	46.7	NR	NR	NR	NR	NR	NR
48	Cilostazol	133	63.1±117.6	76.7	NR	NR	50.4	23.3	NR	NR	NR	NR	NR	NR	NR	NR	NR
	Control	129	64.4±115.8	77.5	NR	NR	48.1	17.1	NR	NR	NR	NR	NR	NR	NR	NR	NR
13	Control	9	67.8±14.1	66.7	29.6±7.4	0.6±0.2	0	22.0	33.0	78.0	NR	NR	78.0	NR	NR	NR	78.0
	HET	14	69.1±7.6	71.4	27.9±3.5	0.7±0.2	0	7.0	14.0	64.0	NR	NR	100.0	NR	NR	NR	100.0
34	SET	32	76.3±3.8	81.0	NR	0.7±0.1	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
	Control	32	76.1±3.7	85.0	NR	0.7±0.1	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
49	SET	37	69 (50–85)	81.0	26.6±0.6	0.6±0.03	38.0	8.0	NR	NR	NR	NR	NR	NR	NR	NR	NR
	SET	34	66 (54–84)	79.0	28.6±0.6	0.6±0.03	24.0	18.0	NR	NR	NR	NR	NR	NR	NR	NR	NR
	Control	33	72 (56–84)	73.0	27.8±1.0	0.7±0.03	33.0	27.0	NR	NR	NR	NR	NR	NR	NR	NR	NR
22	HET	10	66.1±9.8	80.0	26.6±4.5	0.6±0.2	100.0	10.0	60.0	60.0	100	20.0	90.0	60.0	NR	NR	100.0
	Control	9	73.1±4.7	88.9	31.2±7.6	0.6±0.1	100.0	33.3	66.7	77.8	88.9	22.2	88.9	55.5	NR	NR	100.0
21	SET	31	67.0±9.3	57.1	NR	0.7±0.2	NR	38.1	47.6	57.1	42.9	NR	NR	NR	NR	NR	NR
	HET	21	67.0±7.4	80.7	NR	0.7±0.2	NR	38.7	32.3	87.1	61.3	NR	NR	NR	NR	NR	NR
23	SET	21	58.2±10.4	88.9	27.4±4.5	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
	HET	21	60.9±11.3	87.5	26.5±4.5	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR	NR
24	SET	28	65.0±10.6	67.9	29±1.0	0.6±0.2	14.3	25.0	21.4	53.6	92.9	NR	88.0	NR	NR	NR	83.0
	HET	24	65.0±9.8	66.7	28±1.0	0.6±0.2	20.8	37.5	20.8	66.7	87.5	NR	93.0	NR	NR	NR	82.0

% indicates percentage; ER, endovascular revascularization; F, female; HET, home exercise therapy; IQR, inter quartile range; M, male; NR, not reported; and SET, supervised exercise therapy.

^*Mean data only, SD not reported. References showing a & b are different comparisons.

Quality Assessment

A total of 4 clinical trials including 2 trials comparing SET versus ER, ⁵² , ⁵⁹ and 1 trial comparing cilostazol versus control, ⁴⁶ and 1 trial comparing HET and SET ⁵¹ were deemed to have low risk of bias, 24 trials had moderate risk of bias,^* and 18 trials were considered to have high risk of bias.^† All included trials reported the method of randomization and the number of participants who completed the study. Twenty‐four trials used power calculations to estimate the sample size^‡ and 25 trials used intention‐to‐treat principles to analyze the results.^§ Twenty‐three trials had incomplete outcome data because of losing more than 10% of the study population.^‖ Absence of participant and/or investigator blinding was noted in 30 trials.^¶ Individual study‐level assessments are provided in Table 2.

Table 2.

Quality Assessment of All Included Trials

Study Reference	Sample Size Estimate	Intention‐to‐Treat Analysis	Mentioned the Number of Patients Who Completed the Study	Random Sequence Generation	Incomplete Outcome Data (>10% loss)	Performance Bias (Participants and Personnel Blinding)	Total Risk of Bias
20	(−)	(−)	(+)	(+)	(+)	(−)	High
50	(−)	(−)	(+)	(+)	(+)	(+)	Moderate
35	(+)	(+)	(+)	(+)	(−)	(+)	Moderate
21	(−)	(−)	(+)	(+)	(−)	(−)	High
51	(+)	(−)	(+)	(+)	(+)	(+)	Low
36	(−)	(−)	(+)	(+)	(+)	(−)	High
60	(−)	(−)	(+)	(+)	(+)	(?)	High
37	(−)	(−)	(+)	(+)	(−)	(−)	High
38	(+)	(+)	(+)	(+)	(−)	(+)	Moderate
6	(+)	(−)	(+)	(+)	(−)	(−)	High
22	(−)	(−)	(+)	(+)	(+)	(−)	High
52	(+)	(+)	(+)	(+)	(+)	(+)	Low
62	(−)	(+)	(+)	(+)	(+)	(−)	Moderate
39	(−)	(−)	(+)	(+)	(−)	(−)	High
53	(−)	(+)	(+)	(+)	(+)	(−)	Moderate
26	(−)	(+)	(+)	(+)	(−)	(−)	High
40	(+)	(+)	(+)	(+)	(−)	(−)	Moderate
25	(+)	(+)	(+)	(+)	(−)	(−)	Moderate
61	(+)	(+)	(+)	(+)	(−)	(−)	Moderate
23	(+)	(−)	(+)	(+)	(+)	(+)	Moderate
54	(+)	(+)	(+)	(+)	(−)	(−)	Moderate
41	(−)	(−)	(+)	(+)	(+)	(−)	High
42	(−)	(+)	(+)	(+)	(+)	(−)	Moderate
27	(−)	(−)	(+)	(+)	(+)	(−)	High
43	(+)	(+)	(+)	(+)	(−)	(−)	Moderate
44	(+)	(+)	(+)	(+)	(−)	(−)	Moderate
28	(−)	(+)	(+)	(+)	(−)	(−)	High
63	(+)	(+)	(+)	(+)	(+)	(−)	Moderate
29	(−)	(−)	(+)	(+)	(+)	(−)	High
30	(+)	(+)	(+)	(+)	(+)	(−)	Moderate
55	(+)	(+)	(+)	(+)	(−)	(−)	Moderate
32	(−)	(−)	(+)	(+)	(−)	(+)	High
31	(−)	(−)	(+)	(+)	(−)	(+)	High
56	(+)	(+)	(+)	(+)	(−)	(+)	Moderate
45	(+)	(−)	(+)	(+)	(−)	(−)	High
57	(+)	(−)	(+)	(+)	(+)	(−)	Moderate
46	(+)	(+)	(+)	(+)	(+)	(+)	Low
33	(+)	(+)	(+)	(+)	(+)	(−)	Moderate
58	(−)	(−)	(+)	(+)	(+)	(+)	Moderate
24	(+)	(−)	(+)	(+)	(−)	(+)	Moderate
59	(+)	(+)	(+)	(+)	(+)	(+)	Low
47	(+)	(−)	(+)	(+)	(−)	(−)	High
48	(+)	(+)	(+)	(+)	(−)	(+)	Moderate
13	(−)	(+)	(+)	(+)	(+)	(+)	Moderate
34	(−)	(+)	(+)	(+)	(−)	(−)	High
49	(−)	(+)	(+)	(+)	(+)	(−)	Moderate

Trials with 3 or more negative assessment outcomes were considered to have HIGH risk of bias. Trials with 1 to 2 negative assessment outcomes were considered to have MODERATE risk of bias and those with no negative assessment outcomes were considered to have LOW risk of bias.

Network Model

The graphical representation of all treatment arms from all included trials is shown in Figure 2. The model convergence was achieved with 100 000 iterations at all follow‐up time points. Node split analysis showed that the estimates of direct, indirect, and network assumptions were not significantly different, suggesting the absence of large inconsistencies within the network models at all follow‐up time points (Figure 2A through 2C). The network diagnostics using trace and density plots (Figures S1 through S3) and node split analysis (Figure 3A through 3C) showed that the models converged and were valid to use at all follow‐up time points. The Gelman diagnostics showed a Potential Scale Reduction Factor score of 1.000374, 1.000296, and 1.000166 at short‐term, moderate‐term, and long‐term follow‐up, respectively, suggesting the reliability of the network model.

Figure 2. Network plots summarizing all intervention arms in the included trials.

(A) Short‐term follow‐up, (B) Moderate‐term follow‐up, (C) Long‐term follow‐up. The size of the grey lines represent the number of trials and the size of the orange circles represent the number of participants in that specific arm. ER indicates endovascular revascularization; HET, home exercise therapy; and SET, supervised exercise therapy.

Figure 3. There were no significant differences in mean difference obtained from direct and indirect evidence.

(A) Short‐term follow‐up, (B) Moderate‐term follow‐up, (C) Long‐term follow‐up. This suggests that very large inconsistencies were not present within the network model. ER indicates endovascular revascularization; HET, home exercise therapy; and SET, supervised exercise therapy.

Short‐Term Follow‐Up

Data were available for the comparison of the following treatment combinations at short‐term follow‐up: cilostazol (n=451, 5 arms), HET (n=392, 15 arms), SET (n=997, 30 arms), ER (n=333, 9 arms), and SET plus ER (n=280, 6 arms) compared with controls (n=1062, 28 arms). The forest plot suggested that the largest improvement in MWD was achieved by SET plus ER (MD, 326.3 m; 95% CrI, 222.6–430.6). SET (MD, 186.8 m; 95% CrI, 136.4–237.6), and HET (MD, 89.4 m; 95% CrI: 20.9–157.7) also significantly increased MWD. Cilostazol (MD, 71.1 m; 95% CrI, −24.6 to 167.9) and ER (MD, 82.5 m; 95% CrI, −2.4 to 168.2) had no significant effect on MWD (Figure 4A).

Figure 4. Forest plots of multiple treatment strategies at different follow‐up periods.

(A) Short‐term follow‐up, (B) Moderate‐term follow‐up, (C) Long‐term follow‐up. Results are expressed as mean difference (95% CrI) in meters. There was significant improvement in maximum walking distance at short‐ and moderate‐term follow‐up with a combination of SET and ER and SET alone. HET only significantly improved maximum walking distance compared with controls at short‐term follow‐up. No treatment was significantly better than controls at long‐term follow‐up. CrI indicates credible intervals; ER, endovascular revascularization; HET, home exercise therapy; and SET, supervised exercise therapy.

Moderate‐Term Follow‐Up

Data were available for the comparison of the following treatment combinations at moderate‐term follow‐up: HET (n=36, 2 arms), SET (n=499, 12 arms), ER (n=367, 7 arms), and SET plus ER (n=197, 4 arms) compared with controls (n=228, 7 arms). The forest plot suggested that the largest improvement in MWD was observed for SET plus ER (MD, 368.5 m; 95% CrI, 195.3–546.9) and SET alone (MD, 201.1 m; 95% CrI, 89.8–318.3). ER (MD, 84.2 m; 95% CrI, −35.3 to 206.4) and HET (MD, 99.4 m; 95% CrI, −174.0 to 374.9) had no significant effect on MWD (Figure 4B).

Long‐Term Follow‐Up

Data were available for the comparison of the following treatment combinations at long‐term follow‐up: SET (n=124, 4 arms), ER (n=223, 5 arms), and SET plus ER (n=104, 3 arms) compared with controls (n=150, 3 arms). The forest plots suggested that none of the treatments significantly improved MWD as compared with controls: SET plus ER (MD, 140.9 m; 95% CrI, −195.4 to 507.7), ER (MD, 122.7 m; 95% CrI, −64.8 to 347.6), and SET (MD, 29.0 m; 95% CrI, −297.1 to 384.6) (Figure 4C).

Ranking Probability

Ranking probability suggested that the SET plus ER combination was the best treatment strategy followed by SET alone during short‐term and moderate‐term follow‐up (Table S2). Initial benefits were abolished at long‐term follow‐up for all treatment strategies with available data (Figure 4).

Secondary Outcomes

QOL

A total of 29 trials reported QOL outcomes using different generic and disease‐specific questionnaires as detailed in Table S3. Generic QOL outcomes were reported in 26 trials using the 36‐Item Short Form Health Survey (SF−36), 12‐Item SF,^# or the EuroQOL‐5. ¹³ , ⁴⁴ , ⁶³ Disease‐specific QOL outcomes were reported in 25 trials using the Walking Impairment Questionnaire (WIQ),^** the Vascular QOL Questionnaire (VascuQOL), ⁶ , ⁴⁶ , ⁵² , ⁵⁵ , ⁵⁹ , ⁶² the claudication score, ⁵⁰ , ⁵⁷ or the Baltimore Activity Scale for Intermittent Claudication Questionnaire, ²⁹ the Intermittent Claudication Questionnaire, ⁵¹ and peripheral artery questionnaire ⁵⁶ respectively. The QOL outcomes are reported in Table S3.

Short‐Term Follow‐Up

Four of 6 arms reported that cilostazol significantly improved QOL as assessed with the SF‐36, VascuQOL, or WIQ by comparison with controls. ³⁸ , ⁴⁶ , ⁴⁸ Five of 11 arms reported that HET significantly improved QOL as assessed with the SF‐36 or WIQ compared with controls. ²⁵ , ²⁶ , ²⁸ , ²⁹ Eleven of 18 arms reported that SET significantly improved QOL as assessed with the Baltimore Activity Scale for Intermittent Claudication Questionnaire, SF‐36, VascuQOL, or WIQ compared with controls. ²⁵ , ²⁶ , ²⁹ , ³⁴ , ⁴⁰ , ⁴⁵ , ⁵² Seven of 9 arms reported that ER significantly improved QOL as assessed with the SF‐36, claudication score, VascuQOL, or EuroQOL by comparison with control. ⁶ , ²⁸ , ⁴⁴ , ⁵⁷ , ⁵⁹ Four of 4 arms reported that the combination of SET and ER significantly improved QOL as assessed with the SF‐36, VascuQOL, or EuroQOL compared with controls. ⁴⁴ , ⁵²

Moderate‐Term Follow‐Up

Seven of 13 arms reported that ER significantly improved QOL as assessed with the SF‐36, VascuQOL, claudication score, or peripheral artery questionnaire compared with control. ⁶ , ⁵⁶ , ⁵⁷ , ⁶²

Two of 4 arms reported that HET significantly improved QOL as assessed with the SF‐36, WIQ, or Intermittent Claudication Questionnaire compared with control. ⁴³ , ⁵¹ Ten of 16 arms reported that SET significantly improved QOL assessed by the SF‐36, SF‐12, WIQ, peripheral artery questionnaire, Intermittent Claudication Questionnaire, or VascuQOL compared with controls. ⁴³ , ⁵¹ , ⁵² , ⁵⁶ , ⁶² Two of 4 arms reported that the combination of SET and ER significantly improved QOL assessed with the SF‐36 or VascuQOL compared with controls. ⁵²

Long‐Term Follow‐Up

Six out of 11 arms reported significant improvement in certain domains of QOL in the ER group using the SF‐36, claudication score, VascuQOL, and WIQ questionnaires. ⁵⁷ , ⁶² , ⁶³ One arm reported that HET did not significantly improve QOL as assessed with the WIQ questionnaire. ⁴³ Two of 7 arms reported that SET significantly improved QOL as assessed with the SF‐36 and VascuQOL questionnaires. ⁵⁴ None of the 4 arms testing SET and ER combined reported any significant improvement of QOL as assessed with the SF‐36 and VascuQOL questionnaires. ⁵⁴ , ⁵⁵

Adverse Events

A total of 28 clinical trials reported adverse events as defined within the methods section^†† (Table S4). None of the trials reported any significant differences in SAEs between groups. Twenty‐one patients had an myocardial infarction as reported in 9 trials. ²⁴ , ²⁵ , ²⁶ , ⁴⁰ , ⁵⁵ , ⁵⁶ , ⁵⁸ , ⁵⁹ , ⁶³ Thirty patients were reported to have had a stroke in 10 trials.^‡‡ Events requiring hospital admissions were not reported in any of the included trials. Twelve trials reported lower limb revascularization procedures performed in 158 patients during the study follow‐up period.^§§ Three trials reported other vascular procedures performed in 6 patients. ⁴⁰ , ⁵⁴ , ⁵⁹ Fifteen patients were reported to have required an amputation in 8 trials. ⁶ , ⁴⁰ , ⁵² , ⁵³ , ⁵⁵ , ⁶¹ , ⁶² , ⁶³ Twenty‐one trials reported 149 deaths during the follow‐up period.^‖‖

DISCUSSION

The results of this NMA, suggested that during short‐term follow‐up ER plus SET, SET alone, and HET alone all significantly improved MWD in people with intermittent claudication. During moderate‐term follow‐up only ER plus SET and SET alone significantly improved MWD. During follow‐up of 2 years or more, none of these treatments significantly improved MWD. These analyses were limited by the small number of participants included within the available trials meaning the analyses were not adequately powered to identify moderate treatment effects.

Cilostazol did not significantly improve MWD during short‐term follow‐up and no trials testing longer‐term use were identified. The largest benefit was achieved for ER plus SET. The findings agree with a prior NMA which reported ER plus SET was the most effective intervention, although this study did not consider the length of follow‐up. ⁶⁴ Unlike the prior NMA the current report found that ER alone did not significantly improve MWD at any time point. This difference is likely reflective of the separation of outcome data by time of follow‐up in the current study. The prior NMA reported all outcome data together irrespective of follow‐up time. The effect of ER alone on MWD during short‐term follow‐up in the current study (MD, 82.5 m; 95% CrI, −2.4 to 168.2) was very similar to that reported in the prior NMA (MD, 85 m; 95% CrI, 4–170]). The additional findings of the current NMA are important because lack of benefit as early as 2 years after treatment may lead to many participants feeling that having the treatment is not worthwhile. The findings differ from a prior Cochrane review that included a conventional MA and reported cilostazol was effective in improving MWD. ⁶⁵ This disparity may be because of the much smaller number of comparator groups included in the Cochrane review, which reported traditional meta‐analyses. The Cochrane review also included unpublished data and trials reported before 2000, which were not included within this meta‐analysis. This difference in included data likely contributed to the disparate findings of the meta‐analyses. NMA includes analyses of all included trials including those not compared head to head or directly. NMA assumes that it is reliable to compare between participant group included in different trials and the consistency analysis and other model diagnostics performed in this NMA suggested that the indirect comparisons did not introduce any important biases or large inconsistencies.

After 2 years follow‐up, none of the treatments was found to be effective at improving MWD, in line with findings from a recent randomized trial. ⁶ These findings emphasize the poor durability of the available treatments for intermittent claudication, highlighting the need for new therapies. It should be noted, however, that the number of patients for whom data were available at long‐term follow‐up was limited and the findings may in part reflect the need for further trials testing long‐term outcomes. Intensive medical management is important in people with intermittent claudication in order to reduce the risk of major cardiovascular events, such as myocardial infarction and stroke. There is now good evidence that intensive medical management can also reduce the risk of major adverse limb events. The recent COMPASS (Cardiovascular Outcomes for People Using Anticoagulation Strategies) trial found that low‐dose rivaroxaban (2.5 mg twice daily) plus aspirin significantly lowered the incidence of major adverse limb events as compared with aspirin alone. ⁶⁶ , ⁶⁷ Intensive low‐density lipoprotein lowering has also been reported to have similar benefits in patients with peripheral artery disease. ⁶⁸ Observational studies also report a significant reduction in limb loss and mortality in high‐intensity statin users compared with low‐intensity statin users. ⁶⁹ Greater focus on optimizing medical management in people with intermittent claudication may improve QOL and reduce major adverse limb events.

Important considerations in selecting treatments for intermittent claudication include QOL and adverse events. Because of the inconsistent and limited reporting of these outcomes, however, it was not possible to meta‐analyze these outcomes. Among the included trials there was evidence in some trials that all 5 treatment combination were able to improve QOL at short‐term follow‐up, although findings were inconsistent. Findings were similar at moderate‐term follow‐up, except that no trials testing cilostazol were identified and where available most trials suggested that QOL was not improved at long‐term follow‐up.

Adverse events were infrequently reported in the included trials and may not be representative of outcomes in clinical practice. Little information was reported on perioperative complications following ER, drug‐related adverse events, and requirement for amputation. In addition, because of the inconsistent and poor reporting of adverse events, the minimum data requirement for meta‐analyses focused on serious adverse events and mortality was not met. Large trials are needed to examine the effect of exercise on mortality and cardiovascular events in people with peripheral artery disease. These are important considerations when selecting the very different treatment strategies available and need more consistent and complete reporting in future trials.

A number of limitations of this NMA should be acknowledged, including the limited number of trials, particularly testing outcome over the long term and examining the effects of cilostazol and HET. Further trials examining long‐term outcomes, in particular, are required to draw reliable conclusions. Limited and inconsistent information about the medical management of participants was reported from the included trials. Most authors did not report even the frequency of statin or antiplatelet medication prescription. Importantly, NMA assumes overall similarity between studies, and interstudy differences of the included population could not be addressed. Variation in medical management between the included trials may have contributed to the heterogeneous outcomes reported. Furthermore, NMA has a number of inherent limitations as it models evidence from both direct and indirect comparisons and therefore should be interpreted cautiously. ⁷⁰ The ranking probability approach shows only the relative ranks but not the absolute differences between the different treatment strategies. ⁷¹ One limitation of the current NMA was that trials published before 2000 were excluded. This was felt to be required in order for the comparison to be contemporary owing to the marked evolution of management approaches, particularly ER, and trial methods over time. Furthermore, there was substantial variation in the HET programs tested. ¹⁴ Most SET programs tested involved regular treadmill walking. Because the primary outcome was MWD in a treadmill test, it is likely this NMA has overestimated the benefit of SET owing to an established training to the outcome measure phenomenon. ⁷² Furthermore, detailed SAEs were not reported in most trials. These data are required to appropriately interpret the value of each treatment strategy.

Conclusions

In conclusion, this NMA suggests that ER plus SET and SET alone are effective at improving MWD in the short and moderate term (ie, under 2 years). None of the treatments, however, was effective at improving MWD during long‐term follow‐up. More durable therapies are needed for intermittent claudication.

Sources of Funding

Funding from the National Health and Medical Research Council (1063476 and 1022752), James Cook University, The Townsville Hospital and Health Services Study, Education and Research Trust Fund, and Queensland Government supported this work.

Disclosures

Professor Golledge holds a Practitioner Fellowship from the National Health and Medical Research Council (1117061) and a Senior Clinical Research Fellowship from the Queensland Government, Australia. The remaining authors have no disclosures to report.

Supporting information

Data S1

Tables S1–S4

Figures S1–S3

(J Am Heart Assoc. 2021;10:e019672. DOI: 10.1161/JAHA.120.019672.)