Prevalence of Cardioprotective Medication Use in Coronary Heart Disease Patients in South America: Systematic review and Meta-Analysis

Search strategy

This review was registered with PROSPERO (registration number CRD42020206657) and conducted in accordance with the PRISMA guidelines [15] (Supplementary File 1). We conducted a systematic search on April 28^th, 2021 on the following databases: PubMed, Embase, Cochrane, LILACS and SciELO. The search strategy contained information on the CHD diagnosis of the patients, the country where the study was performed and the most common classes of cardioprotective medications in the outpatient clinic setting. Studies published between 2000 and 2021 in English, Spanish or Portuguese and conducted in South America (Argentina, Bolivia, Brazil, Chile, Colombia, Ecuador, Guyana, Paraguay, Peru, Suriname, Uruguay and Venezuela) reporting the prevalence of cardioprotective medications in CHD patients in outpatient settings were included. Broad terms were included for the diagnosis of CHD: ‘coronary artery disease,’ ‘myocardial infarction,’ ‘ST-elevation myocardial infarction,’ ‘non-ST-elevation myocardial infarction,’ ‘acute coronary syndrome,’ ‘angina pectoris,’ ‘acute coronary syndrome,’ ‘coronary atherosclerosis,’ interventions such as ‘coronary artery bypass graft’ and ‘percutaneous coronary intervention,’ and commonly used acronyms for these terms. For details on the search strategy and PROSPERO registration see Supplementary File 2.

Eligibility criteria

The classes of cardioprotective medications taken into account were anti-platelet drugs, lipid-lowering drugs, antihypertensive agents (beta-blockers, ACE-inhibitors, ARBs, diuretics, and nitrates), oral hypoglycaemics and insulin. Intervention studies (randomized clinical trials and non-randomized interventions) and observational studies (cross-sectional, cohort and case-control studies) were included. Case reports, case series, reviews, as well as publication types other than original articles were excluded.

Study selection

The publications resulting from the search were screened by the above eligibility criteria on their titles and abstracts using the platform Rayyan CQRI [16]. Screening was conducted by two reviewers (ED, AMF). Each reviewer screened half of the articles, and an additional 10% of the articles was screened by the other reviewer to prevent interpersonal bias. The reviewers discussed discrepancies and unclear decisions until consensus was reached. The publications that fulfilled the inclusion and exclusion criteria were screened on their full-text following the same strategy.

Data extraction

Relevant data was extracted from the selected publications. Data extraction was performed using the electronic data capture system REDCap©[17] by two reviewers (ED, AMF). Each reviewer extracted data from the articles that the other reviewer had previously screened to minimize potential bias. Collected data included information on authors, publication year, publication title, name of the study, language, period in which the study was conducted, country, study design; participants characteristics including specific diagnosis like CHD, acute coronary syndrome (ACS), coronary artery bypass graft (CABG) or percutaneous coronary intervention (PCI); percentage of women, age range, mean age, socioeconomic status (including percentage of participants in the highest income and education categories as well as percentage of employment), and cardiovascular risk factors (blood pressure, body mass index, lipids and glucose levels), care setting information (type of hospital or healthcare centre, e.g. primary care, academic hospital, tertiary hospital, rehabilitation, and whether the centre was public or private), and urbanicity.

Outcome data included the prevalence of medication per medication class. In the case of drug intervention studies, we extracted data on medication prevalence at baseline. In publications with an observational design that reported medication prevalence at multiple time-points, we extracted data from the earliest time-point in order to facilitate comparison with intervention studies. If not reported directly, medication prevalence was calculated when possible.

Secondary outcome data included guideline compliance (report of compliance or non-compliance), time trends (starting year of the study) and determinants associated with use of medication in patients with established CHD (outcomes reported in stratified analysis or coefficients reported in regression models).

Quality assessment

A tool for the quality assessment of studies reporting prevalence estimates was adapted from the previous work by Zhao et al. [13], and Li et al. [18] (Supplementary File 3A). For overall risk of bias assessment, we summarized risk of bias as follows: for risk of bias in each domain (study design, study population, participation rate, participants’ characteristics and outcome) 2 points were given for low risk, 1 point for moderate risk and 0 points for high or unclear risk. Publications with a score lower than 6 out of 10 were excluded. The remaining publications were classified as: moderately low risk of bias (6–7 points), low risk of bias (8–9 points) and very low risk of bias (10 points). Reviewers ED and AMF assessed the quality of articles for which they extracted the data, and additionally they assessed the quality of 10% of the articles examined by the other reviewer. Discrepancies were discussed until consensus was achieved.

Data analysis

Data analysis was conducted using R Studio [19]. Data on medication prevalence is expressed in percentages by class of medication. We reported the prevalence of each kind of medication separately. We presented separate categories for those articles reporting general classes of medications instead of specific drugs. In the case of antiplatelet drugs, we additionally showed the estimates of articles reporting the prevalence of aspirin, clopidogrel, and not-specified antiplatelet drugs combined because of the shared indication for these medications, and to be able to explore the use of this medication class in general.

Meta-analysis was performed using a mixed model from the R package ‘metafor’ for each class of medication [20]. The results were expressed as pooled prevalence with 95% confidence intervals (CI) and random effects, and displayed in forest plots by care setting. Heterogeneity was quantified with the I² test. The same statistical package was used in a sensitivity analysis to analyse potential differences in prevalence between studies conducted in Brazil and in other countries.

In order to explore time trends in medication use, mixed meta-regression models were fitted with the starting year of the study as covariate for each medication class. The reported prevalence of medication and the model prediction were plotted against the year the studies commenced in bubble plots to illustrate time-trends in medication use.

Meta-regression models were performed to discern potential factors contributing to medication use. A mixed meta-regression model was run for each class of medication, including the following covariates: the proportion of women included, time of outcome measurement since the start of the study, diagnosis of the patients included in the study, urban region, and care setting. The full models were reduced and simpler models were compared against the full models and among them with the AIC fitting statistic. Models with the lowest AIC were selected. Results were expressed as odds ratios (OR) and 95% CI.

Study selection

The search strategy resulted in 7388 publications: 2660 in LILACS, 1810 in Embase, 1538 in SciELO, 729 in Cochrane and 651 in PubMed. After removing 1606 duplicates, 5782 publications were screened on their title and abstract. 4405 publications did not fulfil the inclusion criteria and were excluded, resulting in 1377 publications eligible for full-text screening. During a full-text screening, 1218 articles were excluded (Figure 1). Of the remaining 159 publications, 86 did not reach the quality threshold during the quality assessment, and therefore 73 publications were finally included in the review.

Figure 1

Study characteristics

Table 1 describes the main characteristics of the included studies. All articles included were published between 2000 and 2020, referring to studies conducted between 1993 and 2017. Most studies were conducted in Brazil [3, 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]. Six studies were conducted in Argentina [59, 60, 61, 62, 63, 64], four in Chile [65, 66, 67, 68], four in Colombia [69, 70, 71, 72], three in Uruguay [73, 74, 75] and two were multi-country studies conducted in Argentina, Brazil, Chile and Colombia [12] and Brazil and Suriname [76]. The most common language of the articles was English (58 articles), followed by Spanish (11 articles) and Portuguese (4 articles).

In terms of study design, most publications reported on cohort studies (23 articles), cross-sectional studies (20 publications), and baseline data of randomized clinical trials (17 articles). The number of participants included in each study ranged from 20 to 2475, with a mean of 328 (SD 424). Most studies were conducted in urban areas (39 studies). Regarding the clinical setting, the majority of studies were conducted in academic or tertiary hospitals (42 articles), six in rehabilitation centres, three in primary care or community settings, two in secondary level hospitals, two in public hospitals, and 12 in other settings.

Participants’ characteristics

The most common diagnosis of the patients included was coronary heart disease (26 articles), followed by ACS (21 articles), PCI (9 articles), CABG (9 articles) and some articles included patients with more than one diagnosis [5]. Most articles included a majority of male participants. The mean percentage of female participants was 32.0% (SD 11.4%).

12 articles provided information on the socioeconomic status (SES) of participants. Educational attainment was reported by seven publications and the proportion of participants with highest educational attainment ranged from of 9.5% to 49.2%. The percentage of employed participants was reported by five articles and ranged from 33.5% to 45.0%; and the proportion of participants in the highest income category (reported in six articles) varied from 20.0% to 58.0%.

Regarding the risk factors of the study populations, 54 articles reported the prevalence of hypertension (range 45.0–96.0%) and 42 articles provided prevalence values for dyslipidaemia (36.0–96.8%). The prevalence of diabetes was reported in 56 articles (range 7.7% to 100%). The prevalence of overweight was reported in 11 of the included studies (range 28.2% to 93.5%), 5 articles reported the prevalence of obesity (range 15.0%–33.7%); and 16 articles included mean or median BMI values, ranging from 26.1 to 29.0 kg/m².

Quality assessment

The risk of bias varied by domain of the quality assessment tool: study population was the field in which more articles had a high risk of bias (13.7%), whereas most articles had low risk of bias in the fields of study design (79.5%) and participation rate (72.6%) (Figure 2). The results of the quality assessment of all included publications are displayed in Supplementary Figure 1, and of included and excluded publications in Supplementary File 3B. Supplementary file 3C details the reasons for exclusion of publications with a risk of bias score lower than six.

Figure 2

Prevalence of medication

The prevalence of medication for each study as well as the pooled prevalence estimate per medication is displayed in Figures 3, 4, 5 and summarized in Table 2. The prevalence of beta-blockers was reported in 53 studies, with a pooled estimate of 73.4% (95%CI 66.8% – 79.1%) (Figure 3). 44 articles reported the prevalence of ACEI/ARB use, with a pooled estimate of 55.8% (95%CI 49.7% – 61.8%) (Figure 3). The overall prevalence of antiplatelet drugs (including aspirin, clopidogrel, and articles that didn’t specify the antiplatelet drug) was retrieved from 51 studies, and the pooled prevalence estimate was 84.6% (95%CI 79.6% – 88.5%) (Figure 4). The prevalence of aspirin specifically was retrieved from 44 studies and their pooled estimate was 85.1% (95%CI 79.7% – 89.3%) (Figure 4). The prevalence of statins was reported in 50 articles and the estimated pooled prevalence was 78.9% (95%CI 71.2% – 84.9%) (Figure 5). Total heterogeneity in the meta-analysis models high, ranging from 97.8% (ACEI/ARBs model) to 99.0% (antiplatelet model). No significant differences were observed in the prevalence of any medication classes between Brazil and other countries.

Figure 3

Figure 4

Figure 5

The pooled prevalence estimates of medications and medication classes reported by fewer articles is displayed in supplementary Figures 3, 4, 5 and summarized in Table 2. This includes antihypertensive drugs (without specification), diuretics, nitrates, antiplatelet drugs (without specification), calcium channel blockers, clopidogrel, dual antiplatelet therapy, lipid-lowering drugs (without specification), high-intensity statins, and fibrates.

Time trends

The prevalence of beta-blockers, ACEI/ARBs and statins use significantly increased with time. The use of all antiplatelet drugs and aspirin in particular remained relatively stable over time and the association between use of these medications and year of the study was not significant (Figure 6). The changes in use for other classes of medications were not significant, and they are shown in Supplementary Figure 6. There were too few observations in the lipid-lowering drugs (without specification), high-intensity statins, and fibrates to analyse time-trends.

Figure 6

Guideline compliance

From the publications included, 19 articles reported whether the prevalence of medication use was adequate. Half of them reported that medication use was low [59, 62, 76, 77, 78] or insufficient compared to guideline recommendations [2, 12, 48, 61, 79]. Other articles reported that cardioprotective medication use was adequate or high [32, 60, 66, 67, 73, 75], or in line with guideline recommendations [38, 65, 80].

Determinants of medication use

Summary of main findings

The current systematic review shows large variation in the use of cardioprotective medication among CHD patients, ranging from 55.8% for the use of ACEI/ARB drugs to 85.0% for the use of aspirin. A similar number of studies reported suboptimal and adequate guideline compliance. Time-trend analysis for the period 1993 to 2017 showed an increase in the use cardioprotective medication, with the exception of all antiplatelet drugs and aspirin. Use of beta-blockers, statins, overall antiplatelet drugs and aspirin in community settings was lower compared to academic and tertiary centres. The use of antiplatelet drugs in public centres and of aspirin in rehabilitation centres was also lower compared with tertiary centres.

Prevalence of medication

The prevalence of cardioprotective medication that we observe in South America varies per medication class and shows a general underuse of medications. We observe differences in the prevalence of medication use reported in Europe and North America [8, 9, 82]. When comparing prevalence estimates found in this review, we observed that the prevalence of antiplatelet drugs and beta-blockers was higher than the estimates found for the PURE study [8] (55.4% of antiplatelet use and 45.4% of beta-blocker use in Europe and Canada) and a systematic review by Naderi et al. (65% of antiplatelet use and 62% of beta-blocker use) [9]. The prevalence of ACEI/ARBs we observed was higher than reported in the PURE Study (46.8% in Europe and North America), but lower than in the review by Naderi et al. (70%) [9]. Prevalence estimates from the international EUROASPIRE IV [82] registry were higher than the ones observed in our review for all medication classes (93.8% for antiplatelets, 82.6% for beta-blockers,58.9% for ACE inhibitors and 27% for insulin) except oral hypoglycaemics (oral sulphonylurea 24.9%) and lipid lowering drugs (fibrates 1.8%). The prevalence of statin use we observed was higher than reported in the PURE study (56.7% in Europe and North America), similar to the prevalence estimate described in a systematic review (76%) [9] and lower than the estimate from EUROASPIRE IV (85%) [82].

However, direct comparison with these studies is challenging because they were conducted in different contexts, regions and time periods and other definitions of medication use. The PURE study was conducted entirely in community settings in high-, low- and middle-income countries and regions. The review by Naderi et al. [9] included studies from high income countries in Europe, North America and Australia, and their definition of medication use was limited to prescription refills. EUROASPIRE IV included a majority of secondary and tertiary level centers and was conducted from 2012–2013, while the present review also included research from community settings and studies that started since 1993 [82].

Guideline compliance

Most publication included in this review reported that the prevalence of medication use is suboptimal, while others articles find it to be in compliance with guidelines. Despite some publications considering treatment rates high or adequate, we still find that a notable proportion of the patients do not receive guideline-recommended medications: for example, one third of CHD patients were not receiving beta-blockers and almost one fourth were not receiving statins, although these medications are recommended by guidelines. This low use of antihypertensives may be explained by individuals having adequate blood pressure levels or contraindications, despite guidelines recommending these drugs for all CHD patients. Statins, however, are generally well-tolerated drugs and they are recommended to all CHD patients regardless of cholesterol levels. Therefore, the fact that a substantial proportion of patients does not use them may respond to factors other than possible contraindications. Challenges to adhere to guidelines identified by clinicians include difficulties to change usual practice, time pressure and case complexity among others [83].

Articles considering the prescription rates adequate still noted that medication use decreased with time after diagnosis [59, 75], indicating that there is room to improve medication adherence and secondary prevention of CHD [59]. It is noteworthy that some publications report that achievement of cardiovascular risk factor targets was inadequate despite high levels of medication use [59, 66, 67, 75, 80], which may be attributed to the use of suboptimal doses [59].

Time trends

We observed increased use of most cardioprotective medications. These trends are in line with large surveys conducted in Europe that report an increase of the use of cardioprotective medications from 1999 and 2004 to 2013 [84, 85]. These changes may be attributed in part to the implementation of evidence-based clinical guidelines and public health policies in many South American countries. Mendis et al. [77] previously highlighted the lack of clinical guidelines as a potential factor contributing to the low treatment rates in the PREMISE study. Guideline recommendations have changed in the last decades. For example, the target LDL cholesterol level recommended in guidelines by scientific societies has become lower, from 100 mg/dl in guidelines from 2008 and 2001 [86, 87], to 75mg/dl [88], and finally to 70mg/dl in the most recent guidelines [89, 90, 91, 92]. The recommendation to prescribe statins to CHD patients changed accordingly, and the most recent recommendations from scientific societies and clinical guidelines recommend statin use in CHD patients regardless of their cholesterol level. These changes may promote a higher intake of statin use, which is in line with research showing a decrease in cholesterol levels globally and also in the South American region [93].

The gradual investment and unfolding of public healthcare systems with wide coverage, such as the Sistema Unico de Saúde (SUS) in Brazil, has promoted the use of medication by a growing primary care network and provision of drugs free of charge [48]. Although there are still barriers to medication access, the growing coverage of public healthcare systems has allowed more CHD patients to access recommended medications.

Determinants of medication use

Strengths

This review summarizes evidence of a large number of studies from South America published in English, Spanish and Portuguese between 2000 and 2021. The comprehensive search included any article that contained data on any of the cardioprotective medication assessed in secondary prevention of CHD: anti-platelet drugs, lipid-lowering drugs, antihypertensive agents (beta-blockers, ACE-inhibitors, ARBs, diuretics, and nitrates), oral hypoglycaemics and insulin. We included articles from regional and international databases, and from a variety of settings (public and private centres, academic/tertiary centres and community settings, and urban and rural areas). In light of the results of this review, which show that care setting may influence the estimation of medication use prevalence, it is especially valuable that our review includes studies conducted in various settings.

In total, this review pooled the prevalence of cardioprotective medication use of 23,938 participants. Most of the articles included in this review had a low or moderate risk of bias, indicating low risk of bias for the pooled estimate.

Limitations

This study aimed to summarize evidence on the prevalence of cardioprotective medication use for secondary prevention of CHD in South America. However, most studies were conducted in a limited number of countries and predominantly in Brazil, the largest and most populous country in South America. Our sensitivity analysis showed no prevalence differences between Brazil and the rest of the included countries, though no information was available for Bolivia, Ecuador, Guyana, Paraguay, Peru or Venezuela. As a result, the prevalence estimates of our study may not be generalizable for these countries. Furthermore, we were not able to run meta-regression for several medication classes because there were few observations. We pooled prevalence estimates from the included articles, but the heterogeneity in the from the meta-analysis models was high. We thus presented the results by subgroups and undertook meta-regression models to identify factors associated with medication use, which slightly reduced the heterogeneity of estimates.