Effect of evidence-based therapy for secondary prevention of cardiovascular disease: Systematic review and meta-analysis

Background The combination pharmacotherapy of antiplatelet agents, lipid-modifiers, ACE inhibitors/ARBs and beta-blockers are recommended by international guidelines. However, data on effectiveness of the evidence-based combination pharmacotherapy (EBCP) is limited. Objectives To determine the effect of EBCP on mortality and Cardiovascular events in patients with Coronary Heart Disease (CHD) or cerebrovascular disease. Methods Publications in EMBASE and Medline up to October 2018 were searched for cohort and case-control studies on EBCP for the secondary prevention of cardiovascular disease. The main outcomes were all-cause mortality and major cardiovascular events. Meta-analyses were performed based on random effects models. Results 21 studies were included. Comparing EBCP to either monotherapy or no therapy, the pooled risk ratios were 0.60 (95% confidence interval 0.55 to 0.66) for all-cause mortality, 0.70 (0.62 to 0.79) for vascular mortality, 0.73 (0.64 to 0.83) for myocardial infarction and 0.79 (0.68 to 0.91) for cerebrovascular events. Optimal EBCP (all 4 classes of drug prescribed) had a risk ratio for all-cause mortality of 0.50 (0.40 to 0.64). This benefit became more dilute as the number of different classes of drug comprising EBCP was decreased—for 3 classes of drug prescribed the risk ratio was 0.58 (0.49 to 0.69) and for 2 classes, the risk ratio was 0.67 (0.60 to 0.76). Conclusions EBCP reduces the risk of all-cause mortality and cardiovascular events in patients with CHD or cerebrovascular disease. The different classes of drugs comprising EBCP work in an additive manner, with optimal EBCP conferring the greatest benefit.

Introduction PRISMA checklist is provided as an additional file (S1 Appendix). The study protocol was registered in the International Prospective Register of Systematic Reviews database (PROSPERO: CRD42018078069).

Systematic literature search
We performed the systematic literature search without limitations of language on EMBASE (1980 to October 2018) and Medline (1946 to October 2018). The search strategies were developed based on the PICO (population, intervention, comparator and outcome) principle [20], search terms (S2 Appendix) covering CVD (coronary heart disease (CHD) and stroke), cardiovascular drugs (lipid-modifiers, antiplatelet agents and first-line antihypertensive drugs) and terms for combination therapy. We also examined the bibliographies of some relevant reviews and articles to identify any additional studies.

Study selection
Three investigators (TTM, ZXW and LZ) independently screened studies to be included in the review using predetermined inclusion criteria. Studies were included in the systematic review if they: (i) included participants aged �18 years old with a history of coronary heart disease (MI, stable or unstable angina pectoris), stroke or transient ischaemic attack (TIA); (ii) clearly defined exposure to a combination pharmacotherapy including at least one antiplatelet agent, one lipid-modifier and one drug of ACEI/ARB, beta-blockers or other commonly used cardiovascular drugs (diuretics, calcium channel blockers, α-adrenergic blockers, aldosterone antagonist, or renin inhibitor); (iii) clearly defined the outcome of all-cause mortality, major cardiovascular events (fatal or non-fatal MI, angina, stroke or TIA); (iv) reported relative risk/ risk ratio (RR), hazard ratio (HR) or odds ratios (OR) or provided data for calculating the risk estimates.
There was no restriction on sample size or language. Conference proceedings and abstracts were excluded if there was insufficient data for determining the risk estimates and the 95% confidence intervals (CI); or if they were not cohort or case-control studies.

Appraisal of study quality
Two investigators (TTM and LQD) independently assessed the methodological quality of included observational studies reviewing the study design, implementation, loss to follow-up, exposure and outcome determination. We adapted the Newcastle-Ottawa Scale (NOS) [21] for assessing the quality of the included studies. Separate NOS criteria were used for case-control and cohort studies. Each version has eight items within three domains with a maximum of nine stars ( � ): selection (representativeness), comparability (due to design or analysis), and outcomes (assessment and follow-up). A study can receive one star for meeting each criterion, while a maximum of two stars can be given for comparability (design or analysis). Studies with one star for comparability only controlled for age and gender in the analysis whereas studies with two stars under comparability also controlled for other important variables such as body mass index, comorbidity, laboratory tests or use of other relevant drugs. A final score � seven was considered as high quality [22].

Data extraction and management
Authors TTM and LQD independently completed the data extraction form which was crossmatched to ensure consistency and accuracy. Details of the study duration and design, sample size and participant characteristics, study setting and data source, intervention(s) and outcome (s) definitions, covariates from each of the included studies were extracted. Risk estimates in the form of RR, OR or HR and their corresponding 95% CIs were used as a measure of the association between intervention and outcome. For each study, we extracted the risk estimates adjusted for the most number of confounding variables. For studies without an adjusted result, the crude results were used for analysis.

Data analysis
The risk estimates of each observational study were pooled in the meta-analysis to obtain the pooled RR. When a single study presented several risk estimates (i.e., separate estimates for the combination of four and three drugs), we adjusted the pooled estimates for within-study correlation. The inverse variance method with random effects models was used to calculate the pooled RRs and 95% CIs [23].
Heterogeneity was assessed using the Cochran Q test and Higgins' I 2 statistic [19]. Galbraith plot and subgroup analyses were carried out to investigate potential sources of heterogeneity and conduct sensitivity analyses. Galbraith plot evaluates the weight of each study on the meta-analysis by estimating the average RR and its contribution to the Q test [24]. In sensitivity analyses, we excluded studies with high weight shown by Galbraith plot and repeated the random-effects meta-analysis. Subgroup analyses were conducted to identify study-level heterogeneous factors, which included design (prospective cohort study, retrospective cohort study and case-control study), diagnosis of CVD (CHD, acute coronary syndrome (ACS), MI and stroke), age (<65 years, 65-75 years and >75 years), length of follow-up (<1 year, 1 year and >1 year), study regions (Europe, Asia, North America, multi-regions) and different treatment groups. All statistical analyses were performed using STATA version 15.0 and Revman version 5.3.

Results from systematic literature search
A total of 10,970 records were exported from the literature research. Titles and abstracts were screened and the full texts of 56 articles were further reviewed. 21 studies met the inclusion criteria for this systematic review, involving 117,881 participants with CVD.  Table 1 summarize the characteristics of the included studies. All studies were published in English and from 2005 onwards: twelve were prospective cohort studies, six were retrospective cohort studies [13][14][15][16][17][25][26][27][28][29][30][31][32][33][34][35][36][37], and three were case-control studies [38][39][40]. Twenty observational studies included were considered as high quality according to their NOS score � seven (S2 and S3 Tables). The study of Timoteo et al. [32] was excluded due to the low quality with a NOS score of five.

Mortality
We included seven cohort and two case-control studies that provided results from combinations of EBCP and compared the risk of all-cause mortality with none or one component of EBCP in the primary meta-analysis (Fig 2). All the included studies presented a potential benefit of combination therapy with a lower risk of all-cause mortality. The pooled RRs of cohort and case-control studies were 0.55 (95% CI 0.47-0.64) and 0.68 (95% CI 0.62-0.75) respectively. Overall, the use of combination therapy reduced the risk of all-cause mortality by 40% (95% CI 34%-45%). In the study of Tay et al. [28], the outcomes were examined between younger patients (age < 75 years) and elderly patients separately. Younger patients benefited more from combination therapy than elderly individuals.

Sensitivity analysis
The heterogeneity of the primary meta-analysis was high, with I 2 = 87.9% (p < 0.001) (Fig 2). In the Galbraith plot (S3 Fig), the study of Tay et al. [28] induced the highest heterogeneity, followed by Hippisley et al. [38], Kirchmay et al. [39] and Yan et al. [30] We repeated the primary meta-analysis with the random-effects model after excluding each of the four studies (S4 Table). Tay et al. [28] was shown to be the largest contributor to heterogeneity. When omitting the study, I 2 decreased to 72.1% though the pooled RR did not change remarkably (0.68, 95% CI 0.67, 0.72). We undertook subgroup analyses to examine the potential sources of heterogeneity related to age, study regions, different diagnoses, length of follow-up and study designs on the EBCP's effect on all-cause mortality (S5 Table). The results show significant differences between subgroups in age (P < 0.05), region (P < 0.01), follow-up duration (P < 0.05) and study type (P < 0.05), indicating the four covariates were likely to be associated with heterogeneity. Conversely, diagnosis of CVD, did not affect heterogeneity of the primary meta-analysis (P = 0.16). The results of subgroups by age show that younger patients may benefit more from reductions in all-cause mortality from EBCP than elderly individuals, with RRs of 0. 44  control study: 0.68) were also presented to be related to the heterogeneity. In addition, we performed another sensitivity analysis within studies which had the reference group of 0 EBCP drug (S4 Fig). The results showed no significant different from the primary meta-analysis (Fig 2).

Discussion
Our meta-analysis of observational studies assessed the effects of EBCP with antiplatelet drugs, ACEIs/ARBs, beta-blockers, and lipid-modifiers on mortality and major cardiovascular events in CVD patients. The results show a benefit for EBCP, suggesting an overall decrease in the risk of all-cause mortality (approximately by 40%) and cardiovascular events (25%-30%) compared to either monotherapy or no therapy.
In this systematic review, we examined the effects of increasing the number of components of EBCP. The results show that each additional component of EBCP could confer a potential 8-9% survival benefit of patients with CVD with a median follow-up of one year. When weighting the impact of each component, we found that antiplatelet agents made the greatest contribution to the beneficial effects of combination therapy on survival in patients with CHD. Excluding antiplatelet drugs from optimal EBCP decreased the beneficial effects by 27%. Our results are supported by a meta-analysis of 193 RCTs. Based on 9,605 deaths, the study reported that antiplatelet therapy produced a significant 15% reduction in vascular deaths (P<0.0001) and about one-sixth of all-cause mortality (P<0.0001). The study also provided strong evidence of benefit from antiplatelet therapy to major cardiovascular events (non-fatal MI, non-fatal stroke or vascular death) [41]. Effect of evidence-based therapy for secondary prevention of cardiovascular disease Therefore, unless there are contraindications, antiplatelet agents should be considered as the first component of EBCP in the secondary prevention of CHD.
The evidence available from the literature for beta blockers and statin therapy is equally as strong. A meta-analysis of 147 RCTs suggested that beta-blockers could reduce CHD events by 29% (RR 0.71, 95% CI 0.66, 0.78). Additional RCT studies have also shown that beta-blockers play an important role in reducing mortality and morbidity for up to a year after an MI [42]. A meta-analysis of 14 RCTs of statins also demonstrated that statins could reduce the risk of allcause mortality by 12% and major vascular events by 21% [43]. Thus, beta-blockers and statins count as valuable components of the optimal EBCP for CVD.
In our systematic review, we found a more modest effect for ACEI/ARB as part of EBCP. In particular two studies have also shown that the inclusion of ACEI/ARB in combination with statins, antiplatelet agents and beta-blockers was associated with a lower risk of mortality [17,38].
In this systematic review, we found some research gaps in terms of EBCP in secondary prevention of CVD.
Firstly, most studies included in the systematic review are based on CHD patients. Only the study of Park et al. [27] was conducted in stroke patients. There is a paucity of evidence for the benefit of EBCP in reducing the mortality risk in stroke patients, even though stroke represents a significant proportion of all cardiovascular disease. Whilst co-morbidities and risk factors cluster together, there is still a lack of data regarding any potential mortality benefit of ACEI and beta-blocker in post-stroke patients who otherwise do not have an indication for their prescription. This should be a priority area for further research.
Secondly, even though we did not limit any other conditions co-existing with CVD in the study population, we could not find any studies specifically evaluating the effects of EBCP for Effect of evidence-based therapy for secondary prevention of cardiovascular disease secondary prevention of CVD in patients with comorbidities. Most of the studies included in our review adjusted the risk estimates with comorbidities. Thus we were unable to identify if the results are applicable equally in the presence of other conditions. Comorbidities are highly prevalent in patients with CVD. A Dutch nationwide study found the percentage of patients with comorbidity were 40% and 32% in coronary heart disease and cerebrovascular disease, respectively [44]. In the context of clinical and functional heterogeneity, CVD patients with different co-conditions may have different responses to pharmacotherapy. In addition, interactions between cardiovascular drugs and treatment for comorbidities also need attention. For example, some nonsteroidal anti-inflammatory drugs like ibuprofen and naproxen are known to interfere with the antiplatelet effects of aspirin [45,46] as well as affect renal function and hence handling of all components of EBCP, in particular ACEIs and ARBs.
Thirdly, most studies included in this systematic review only focused on the combination of aspirin, clopidogrel, beta-blockers, ACEIs/ARBs and statins, observational evidence for the combination of some other commonly used drugs is lacking. This may in part be due to a lack of mortality benefit for many of these drugs tested in randomized trials (e.g. diuretics, CCBs [47], and fibrates [48]), a lack of conclusive evidence of benefits for some drugs on the secondary prevention of CVD (e.g. spironolactone and eplerenone [49]), but may also be due to a lack of follow-up time for newer medications that have come to market e.g. sacubitril/ valsartan combination.
Finally, the length of follow-up in most of the included studies was less than one year, and only effects of drugs in discharge were examined without considering other important longterm effects. These include the possibility of sequential drug exchange or poor drug adherence. Effect of evidence-based therapy for secondary prevention of cardiovascular disease Only the study of Bezin et al. [16] reported the cumulated use of cardiovascular drugs, showing a persistent benefit of combination therapy and additionally reductive effects on the occurrence of major adverse cardiac events or mortality when increasing the number of components.

Strengths and limitations
In the absence of RCTs, we did the systematic review of observational studies. Our study has several strengths. Firstly, we undertook extensive analysis in exploring potential variables that could affect the effects of secondary prevention for CVD, hence providing clinicians with an evidence base for their decision-making. Secondly, our results are robust and consistent, as shown by our extensive analyses by using influence analysis, subgroup analysis and sensitivity analysis.
There are some limitations in the current study. Firstly, the results of some subgroup analyses were not credible enough because only one study was included. Secondly, differences in study designs, exclusion criteria, control groups selection, duration of follow-up, exposure and outcome definitions, including covariates and analyses models can affect the accuracy of pooled estimates for both crude and adjusted RRs. Thirdly, several studies reported the estimated effect sizes with HRs and ORs instead of RRs, and the exact statistical method was not clearly described. We were not able to exclude the influence on results by combining these three types of estimates in the meta-analysis. The variability between studies was unavoidable, and the study conclusions should be evaluated alongside the reported heterogeneity. Nevertheless, we conducted sensitivity analyses to examine the impact of heterogeneity between studies and assessed of the potential causes of heterogeneity. In addition, as studies included in each Effect of evidence-based therapy for secondary prevention of cardiovascular disease meta-analysis were less than ten, we did not examine the publication bias [50,51]. Considering all included studies reported a positive effect of combination therapy only with a difference in the extent, therefore we think that important publication bias due to a preferential publication of large studies with positive findings has not occurred.

Conclusions
In conclusion, our systematic review and meta-analysis suggest that in patients with CVD, EBCP can reduce the risk of all-cause mortality by approximately 40% and major cardiovascular events by 25%-30%. Antiplatelet agents, beta-blockers and statins could be considered as stable components of combination therapy in secondary prevention of CHD.   Effect of evidence-based therapy for secondary prevention of cardiovascular disease S3