Availability, Use, and Barriers to Cardiac Rehabilitation in LMIC

The authors report no relationships that could be construed as a conflict of interest. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Rajesh Vedanthan is supported by the Fogarty International Center of the National Institutes of Health under Award Background: Cardiac rehabilitation (CR) is a cornerstone of secondary prevention of ischemic heart disease. It is critically important in lowand middle-income countries (LMIC), where the burden of ischemic heart disease is substantial and growing. However, the availability and utilization of CR in LMIC is not systematically known.

Cardiovascular disease (CVD) is the leading cause of death worldwide, and >80% of CVD deaths occur in lowand middle income countries (LMIC) [1,2]. In 2010, 62.5 million new cases of CVD were reported, of which 24.2 million were attributed to ischemic heart disease [3]. Agestandardized ischemic heart disease mortality is substantially higher in LMIC [1]. In addition to the global health burden, CVD will exert a significant economic burden, on the order of approximately $15 trillion over the next 20 years [3].
Given the burden of CVD globally, and in particular in LMIC, secondary prevention strategies are vitally important to stemming this growing epidemic. Over the last several decades, cardiac rehabilitation (CR) has become established as a cornerstone for a comprehensive secondary prevention strategy [4]. CR is a multidisciplinary program of supervised exercise, education, and behavior modification designed to return the patient to optimal physical, emotional, psychosocial, and vocational function. CR for patients experiencing myocardial infarction (MI), acute coronary syndrome, coronary artery bypass graft (CABG), percutaneous coronary intervention (PCI), chronic stable angina, or heart failure is a Class I recommendation from the European Society of Cardiology, American Heart Association, and American College of Cardiology [5,6].
In high-income countries, CR has been shown to reduce mortality by 25%, reduce MI by 17%, decrease recurrent hospitalizations, as well as reduce health care expenditures [5,7,8]. In a recent Cochrane meta-analysis of exercise-based CR, it was shown to reduce pooled cardiovascular mortality from 10.4% to 7.6%, and to reduce hospital admission from 30.7% to 26.1% [9]. CR is associated with improved outcomes in LMIC as well [10]. Despite strong recommendations for CR, CR services are underutilized even in high-income countries [11,12]. In LMIC, the availability and use of CR programs is not systematically known. The purpose of this study, therefore, is to perform a systematic review of peer-reviewed publications on CR in LMIC in order to characterize the availability, use, and barriers to use of CR.

METHODS
This systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and barriers to CR in !1 LMIC. Articles of any study design, systematic reviews, meta-analyses, nonsystematic reviews, and editorials were all eligible for inclusion. Articles of any language were included. Articles and abstracts that were published or in press were included, and authors of 2 studies were contacted for more detailed data than had been originally published (see Acknowledgments).
Qualitative and quantitative data were extracted and organized by region or country. Due to heterogeneity of the data, meta-analysis was not conducted. Descriptive statistics related to availability, use, and barriers were calculated when possible. When only qualitative information was available, those data were summarized.

Availability of CR
Data on the number of centers of CR by country were standardized by total population and number of deaths due to CVD as of 2012 [14]. Ideally, standardization of the number of CR programs by the number of patients indicated for CR would provide the optimal measure of CR density. However, those data are not available, so the number of CVD deaths was used as a surrogate marker to estimate patients who would be eligible for CR. Data were extracted regarding the diagnoses accepted by CR centers, the services offered, and the type of health care personnel available, as per the core components of CR as outlined by the American Association of Cardiovascular and Pulmonary Rehabilitation [15].

Use of CR
The use of CR was defined in 2 ways: 1) referral rates to CR programs by health care professionals; and 2) attendance rates at CR programs by patients. All demographic data on patients who attended CR programs were extracted when available.

Barriers to CR
The barriers to CR were categorized broadly into physician, patient, and systems factors. Qualitative and quantitative information on perspectives from patients and health care providers were extracted.

RESULTS
Our search yielded a total of 5,805 citations, of which 34 satisfied full inclusion and exclusion criteria. Figure 1 summarizes the number of articles involved at each stage of the systematic review process. Online Appendix 2 summarizes the reasons for exclusion of articles at the abstract and full-text levels of screening. At the full-text screening level, the majority of articles were excluded due to a lack of information on availability, use, or barriers to CR. A total of 34 articles were included in this review, with information on 31 LMIC-2 in Africa, 7  summarized in Table 1, with references provided in Online Appendix 3.

Availability of CR
The total number of CR centers in LMIC was reported from 1986 to 2013 and varied considerably across countries ( Figure 2, see Online Appendix 3 for exact results). The lowest density of CR programs was reported in Algeria, where only 1 CR program was functioning in 2008. The highest absolute number of CR centers was 51 and was reported in Serbia. The highest density of CR centers was reported in Lithuania, with 348 CR centers per 100,000 annual CVD deaths, and the lowest density was found in Algeria, with 1 CR center per 100,000 annual CVD deaths.
In addition to this quantitative information, qualitative information on CR availability was reported in editorials, review articles, and surveys of health professionals (references [7e9] cited in Online Appendix 3). Overall, India and China were reported to have low availability of dedicated CR programs. In Indonesia, the first CR program began in 1978, and subsequently expanded to many cities (exact number not specified), with CR being provided by both community-based organizations as well as the National Heart Organization. The indications accepted for CR, the CR services offered, and the type of personnel available varied significantly across countries ( Figure 3, Online Appendix 4). The most commonly accepted indications for CR were following MI, PCI, or CABG. Exercise training and physical activity counseling were offered in the vast majority of programs. Overall, a broader array of services was offered in Latin American and Caribbean countries than in Africa, Asia, and Europe. The personnel involved in CR were heterogeneous across countries; however, the majority of programs involved a cardiologist.

Use of CR
Referral rates to CR were available for 6 European countries and for Mexico from 2013 and 2009, respectively. There was a high degree of variability in CR referral rates, from 5% in Mexico to 90% in Lithuania ( Figure 4). Most countries had a referral rate <40% (references [16,31] cited in Online Appendix 3).
Those countries reporting CR attendance, defined by the number of CR attendees as a fraction of number referred to CR, are summarized in Figure 5 (full results for CR attendance rates are given in Online Appendix 5). Overall, CR attendance was reported in studies published from 1984 to 2012; among patients referred for CR, CR attendance rates ranged from 32% in Bulgaria to 96% in Lithuania. Among patients eligible for CR, CR attendance rates were 40% in Belarus, 90% in Lithuania, and 10% in Romania. In Johannesburg, a 50% completion rate of CR was observed at a single center between 1986 and 1990. In Russia, the rate of CR attendance among patients who were referred was 12% in 1984 and was subsequently reported as 42% in 2011. In Iran, the rate of CR attendance among all patients with PCI and CABG was 4%; however, there was 87% attendance among patients who were referred to CR. Additionally, the first and only Algerian CR center as of 2008 reported a population of 158 patients enrolled per year (reference [1] cited in Online Appendix 3).
Information regarding the characteristics of patients referred for CR was available from Brazil, Iran, Pakistan, and Romania and is summarized in Table 2. In all studies, males made up >65% of patients who were referred for CR. In Iran, Pakistan, and Romania, those who did not attend CR were more likely to be male, whereas in Brazil, those who did not attend CR were more likely to be female. CR participants had higher levels of education; their employment status was variable. In Iran and Romania, a greater proportion of those who did not attend CR were smokers, whereas in Pakistan, a greater proportion of CR attendees were smokers. The indications for CR were variable.

Barriers to CR
Barriers to CR were reported in surveys of CR programs, and, in 1 article, by patients who had dropped out of CR programs ( Table 3). The most commonly reported barrier to CR was the lack of physician referral to CR, cited by 45% to 93% of CR programs and health professionals. Patient-related factors cited were affordability, particularly due to lack of insurance coverage; transportation difficulties, primarily driven by long distances to CR centers; an unwillingness to attend CR; and competing priorities on patients' time. The most frequently reported systems factor was a lack of personnel and resources.

DISCUSSION
In this systematic review, we found substantial geographic variability in the availability, use, and barriers to CR in LMIC. In general, the availability of CR was low relative to the burden of CVD in LMIC. Referral rates for CR were also variable and were <40% in a majority of countries. Attendance rates varied across geography and were also suboptimal. Among those referred for CR, higher attendance was generally correlated with certain patient characteristicshigher educational background and nonsmoking status. The most commonly cited barrier to CR in LMIC was lack of physician referral.
The overall goals for rehabilitative care among cardiac patients in LMIC were laid out by the World Health Organization Expert Committee on Rehabilitation after CVD in 1993. These included availability of CR to all patients with CVD, awareness of health care providers and the public for the need for CR, patient and provider education on CR, and integration of CR into the health system, with the type of program matched to the needs of the community [16]. Progress toward these goals has proven to be suboptimal in LMIC, as evidenced by the findings of this  jg SCIENCE systematic review, including the low density of CR programs in the majority of LMIC, low referral rates, and low rates of participation by patients in CR.
A recent review of the global availability of CR reported CR availability in 68% of high-income countries, 28% of middle-income countries, and 8% of low-income countries [17]. Based on 2003 estimates [18], in the United States, there were 721 CR programs per 100,000 CVD deaths, which is significantly higher than the highest estimate in LMIC from our study, found in Lithuania. Overall, we found significant geographic variability in the amount of published data describing the availability of CR services in LMIC. The most comprehensive data are available from Latin America/Caribbean and Europe, whereas updated publications are lacking in Asia and Africa.
Use of CR worldwide, even in high-income countries, is suboptimal; both CR referral and participation in CR can be improved. In the United States, referral to CR following MI was established as a hospital performance measure in 2007; however, CR referral remains suboptimal [19,20]. Data from Medicare beneficiaries in the United States show that only 13.9% of MI survivors and 31.0% of patients following CABG participate in CR [21]. A 2012 Canadian study of ischemic heart disease patients referred for CR found a 49.3% completion rate of CR [22].
Our data from LMIC are generally consistent with those figures from high-income countries. For example, in Iran, the rate of CR attendance among all patients with PCI and CABG was only 3.6%, despite nearly 86.5% attendance among patients who were referred to CR, likely reflecting inadequate referral to CR [23,24]. Most other  LMIC in our study had low referral and attendance rates, with notable exceptions such as Colombia, Latvia, and Lithuania, which were characterized by high referral and participation rates. Literature on completion rates of CR in LMIC were lacking in the retrieved articles, and consistency of CR attendance needs further evaluation in LMIC; a recent review indicates that CR dropout rates have been reported at over 80% in Iran [25].
Part of the heterogeneity in CR utilization may be explained by health systems in each country. As with in high-income countries, health care utilization is linked closely to cost and payment structures. CR referral rates in the United States are highest among patients with private insurance carriers [19]. In Lithuania, where CR availability and utilization were exceptionally high, health care is a state-funded industry with referral to CR embedded into care metrics [26]. In contrast, many parts of South Asia, Southeast Asia, and Africa rely on out-of-pocket payment systems. In addition to payment systems for CR, access to cardiac interventions leading to CR referral are also heterogeneous among LMIC, which would also lead to variability in CR availability and use as a secondary prevention strategy [2].
Efforts to improve on the suboptimal availability and use of CR in LMIC require a careful analysis of the barriers to implementation and utilization of this service, particularly given the resource scarcity to make changes in health care delivery. Barriers to CR in high-income countries include lack of referral, low knowledge and insight into services, beliefs about heart disease, negative views of the health care system, financial and work constraints, long distance to services, hours of operation, lack of family  jg SCIENCE support, and medical comorbidities. Female sex, older age, lower socioeconomic status, and low educational attainment have been correlated with poor CR attendance [11]. Whereas our results were generally consistent with barriers reported from high-income countries, we found additional patient-related barriers to CR specific to LMIC, including lack of resources and poor affordability of CR programs. Conversely, medical comorbidities were not reported as a barrier to CR in LMIC in our review.
In the United States, solutions to overcome barriers to CR have been proposed; these target patient-centeredness of CR programs, systematization of CR referral, improved accessibility, and data-driven performance improvement [27,28]. In LMIC, the solutions to improve utilization of CR must be tailored to the unique barriers faced by resource scarcity, and novel strategies for the implementation of secondary prevention programs may be necessary. A recent consensus statement on CR delivery in low-resource settings outlined the essential components of CR, and low cost alternatives to CR delivery methods practiced in high-income countries. These include improved risk stratification of patients to enable decreased supervision of patients at low risk for ischemic/arrhythmic events and the use of low-cost exercises and equipment [29].
Based on the barriers to CR in LMIC identified in this review, we have proposed solutions for this setting in Figure 6, based on several studies evaluating novel CR strategies. A recent Cochrane systematic review and metaanalysis, which included studies from Iran and Turkey, concluded that home-based CR was as effective as centerbased CR [30]. Subsequently, another study of homebased CR in Xi'an, China, found that home-based CR, in comparison with usual care (i.e., no CR), improved healthrelated quality of life [31]. Indeed, the use of innovative non-center-based CR in LMIC may improve adherence and overall outcomes after cardiovascular events. In addition, in India, yoga-based CR has been shown to improve left ventricular ejection fraction, blood glucose control, and body mass index, in comparison with standard CR [32]. Mobile technology is emerging as a tool for increasing patient-centeredness of CR and addressing the issue of distances and time [33]. A virtual world-based CR program with features of social interactivity and active learning is currently being evaluated for feasibility and effectiveness; this would additionally address issues of motivation [34].
Completion of CR programs has been shown to be associated with a lower risk of death, lower risk of MI, all-cause hospitalization, and cardiac hospitalization [22]. In LMIC, CR has also been demonstrated to improve CVD risk factors [25]. In high-income countries, CR has been   . Cost-effectiveness and affordability analyses of CR programs, and associated efforts to develop novel CR delivery methods in LMIC, must take into account the cost savings from preventing hospitalization, as well as increasing productivity. Indeed, novel CR delivery methods may likely be a cost-effective and potentially affordable approach to improving access to CR in LMIC.

Study limitations
As a systematic review of the peer-reviewed published reports, publication bias is the foremost limitation of our results. This is most likely to affect our results on CR availability; it is likely that more CR programs exist in LMIC than we have reported, and our results do not comprise a global index of CR programs. However, our goal was to assess the status of the published data on this topic, and we would argue that critical assessment of CR programs, including summary of peer-reviewed publications, is a necessary step toward dialogue, expansion, and improvement of the available resources.
A second limitation is that we chose to include articles since 1980, and the oldest article meeting inclusion criteria was from 1984. Given the rapid pace of health care innovation and overall health system development, we recognize that these older data may not be relevant to the present-day situation in LMIC. However, the majority of included studies were published since 2000; therefore, we feel that our results generally represent the current status of the published data.
Finally, in this systematic review, there were varied definitions of the patient population, outcomes, and comparator groups in the data, making it unsuitable for aggregation in the form of a meta-analysis. Additionally, many of the data points collected did not include measures of imprecision, for which reason we were not able to report on this.

CONCLUSIONS
The available published data reveals a wide variability in the availability of CR in LMIC. There is overall insufficient availability of CR services in LMIC to meet the growing burden of CVD in these areas. Where CR is available, utilization rates of the services are suboptimal. Whereas many barriers to CR in LMIC are also present in highincome countries, there are some notable differences reflecting lower resources, health system deficiencies, and affordability. This study also illustrates the geographic disparity of publications on the availability, use, and barriers to CR in LMIC. Whereas this issue has been studied extensively in Europe and South America, there is a substantial deficiency in published data from Asia and Africa. Accurate enumeration of CR programs is the first step to identifying key areas for policy interventions and program development; thus, there is a significant need for systematic data gathering in these regions.
Interventions to quickly increase implementation and use of this cornerstone of secondary CVD prevention must be targeted to fit the needs of local populations and may need to deviate from the accepted practices in high-income countries. It is imperative on the global health community to incorporate novel CR delivery models into efforts directed at secondary prevention of CVD, in line with the World Health Organization's target of reducing mortality from noncommunicable diseases by 25% [36].