Acute Coronary Syndrome, Stroke, and Mortality after Community-Acquired Pneumonia: Systematic Review and Meta-Analysis

One-third of adult inpatients with community-acquired pneumonia (CAP) develop acute coronary syndrome (ACS), stroke, heart failure (HF), arrhythmias, or die. The evidence linking CAP to cardiovascular disease (CVD) events is contradictory. We aimed to systematically review the role of CAP as a CVD risk factor. We registered the protocol (CRD42022352910) and searched for six databases from inception to 31 December 2022. We included 13 observational studies, 276,109 participants, 18,298 first ACS events, 12,421 first stroke events, 119 arrhythmic events, 75 episodes of new onset or worsening HF, 3379 deaths, and 218 incident CVD events. CAP increased the odds of ACS (OR 3.02; 95% CI 1.88–4.86), stroke (OR 2.88; 95% CI 2.09–3.96), mortality (OR 3.22; 95% CI 2.42–4.27), and all CVD events (OR 3.37; 95% CI 2.51–4.53). Heterogeneity was significant (I2 = 97%, p < 0.001). Subgroup analysis found differences according to the continent of origin of the study, the follow-up length, and the sample size (I2 > 40.0%, p < 0.10). CAP is a significant risk factor for all major CVD events including ACS, stroke, and mortality. However, these findings should be taken with caution due to the substantial heterogeneity and the possible publication bias.


Introduction
Community-acquired pneumonia (CAP) and cardiovascular disease (CVD) events are the leading causes of morbidity and mortality globally [1][2][3][4][5]. CAP is one of the most common reasons for adult hospital admissions. Over one million adults in the USA are hospitalized with pneumonia annually, and about 50,000 die from this disease [4][5][6]. Similarly, these diseases are associated with a significant social burden regarding health care resource utilization and social-economic cost [5].
There seems to exist a bidirectional relationship between pneumonia and CVD [7]. On one hand, CVDs such as coronary artery disease (CAD) and stroke increase the risk of hospitalization for pneumonia [7][8][9], but the opposite could also be true. That is, pneumonia may raise the risk of acute coronary syndrome (ACS)-myocardial infarction or unstable angina-stroke, heart failure, arrhythmias, and ev en death; acutely or even years after that [7,10,11].
However, the evidence linking CAP to cardiovascular complications is contradictory and not substantial. Most published studies included a single cohort without an adequate control group. Additionally, only two meta-analyses published in full text have included some of these studies [12,13]. Therefore, we aimed to systematically review the evidence on the role of CAP or respiratory tract infections as a risk factor for cardiovascular disease (CVD) complications.

Results
We collected a total of 12,796 in the primary and 31 in the secondary search. After eliminating duplicates, 83 publications were evaluated for their titles and abstracts. Subsequently, 40 articles were analyzed in full text, of which 13 papers (nine cohort and four case-control studies) were selected for qualitative and quantitative assessment ( Figure 1). We only included full-text articles that reported the adjusted association measures-OR, RR, or HR-and a control group. The lack of a proper control group was the leading cause for the exclusion of most studies  (Supplementary Materials).
This study includes 276,109 participants, 18,298 first ACS events, 12,421 first stroke events, 119 arrhythmic events, 75 episodes of new onset or worsening HF, 3379 deaths, and 218 "incident CVD events" (Table 1). CAP increases the odds of ACS (OR 3.02; 95% The articles found were analyzed using the terms of the PECO strategy and the inclusion and exclusion criteria. In addition, relevant data from each paper were extracted and recorded in a spreadsheet: the name of authors, year and country of publication, type of study, number of patients, number of events, the measure of association, and adjusted confounders. In the meta-analysis, we pooled the adjusted odds ratios (OR), risk ratios (RR), or hazard ratios (HR) with 95% confidence intervals (95% CI) using the generic inverse variance method. Forest plots represented the quantitative synthesis. Heterogeneity among studies was assessed with Cochran's Q test and Higgins I 2 statistic. Heterogeneity was significant (p-value < 0.05, I 2 statistics >40%), then we used a random effects model. We carried out sensitivity and subgroup analyses. The risk of bias was assessed using the Newcastle-Ottawa Scale (NOS) tool [17] and publication bias was examined using a funnel plot.
We defined "all CVD events" as a composite of first ACS or first stroke event, new or worsening episodes of heart failure, arrhythmia (atrial fibrillation or paroxysmal supraventricular tachycardia), and death. Incident CVD event was defined as a composite of ACS, stroke, and fatal CHD [18]. For studies reporting OR or RR stratified into different subgroups, we considered each subgroup analysis as a separate study.

Results
We collected a total of 12,796 in the primary and 31 in the secondary search. After eliminating duplicates, 83 publications were evaluated for their titles and abstracts. Subsequently, 40 articles were analyzed in full text, of which 13 papers (nine cohort and four case-control studies) were selected for qualitative and quantitative assessment ( Figure 1). We only included full-text articles that reported the adjusted association measures-OR, RR, or HR-and a control group. The lack of a proper control group was the leading cause for the exclusion of most studies    51-4.53). However, heterogeneity was significant (I 2 = 97%, p < 0.001). The sensitivity analysis-with outliers excluded-did not significantly affect the overall estimate. In the subgroup analysis, we found statistically significant differences according to the continent of origin of the study (I 2 = 78.2%, p < 0.10), the length of follow-up (I 2 = 89.4%, p < 0.10), and the sample size (I 2 = 75.1%, p < 0.10). Conversely, we did not find statistically significant differences between subgroups according to the study design (I 2 = 0%, p < 0.93) (Figure 2A-G). Due to the limited data available, it was impossible to perform subgroup analysis according to other variables (i.e., the participants' sex, the interval from exposition to the outcome, or cardiovascular risk levels). Similarly, we did not perform meta-regression analyses due to the limited number of studies.
All of the studies included had a low risk of bias (Table 2). However, the funnel plot suggested publication bias ( Figure 3).
Corrales-Medina et al. performed a meta-analysis to determine the incidence of ma cardiac complications in CAP patients. They searched Medline, Scopus, and EMBASE
Corrales-Medina et al. performed a meta-analysis to determine the incidence of major cardiac complications in CAP patients. They searched Medline, Scopus, and EMBASE for
Corrales-Medina et al. performed a meta-analysis to determine the incidence of major cardiac complications in CAP patients. They searched Medline, Scopus, and EMBASE for observational studies of adults with CAP reporting the following: overall cardiac complications, incident HF, ACS, or incident cardiac arrhythmias occurring within 30 days of CAP diagnosis. They found 25 articles that met the eligibility and minimum quality criteria. Seventeen articles (68%) reported cohorts of CAP inpatients. In this group, the pooled incidence rates for overall cardiac complications (six cohorts, 2119 patients), incident HF (eights cohorts, 4215 patients), ACS (six cohorts, 2657 patients), and incident cardiac arrhythmias (six cohorts, 2596 patients) were 17.7% (95% CI 13.9-22.2, 14.1% (95% CI 9.3-20.6), 5.3% (95% CI 3.2-8.6), and 4.7% (95% CI 2.4-8.9), respectively. One article reported cardiac complications in CAP outpatients, four in low-risk (not severely ill) inpatients, and three in high-risk inpatients. The incidences for all outcomes except the overall cardiac complications were lower in the two former groups and higher in the latter. One additional study reported on CAP outpatients and low-risk inpatients without discriminating between these groups. Twelve studies (48%) asserted the evaluation of cardiac complications in their methods, but only six (24%) defined them. Only three studies, all examining ACS, carried out risk factor analysis for these events. No study analyzed the association between cardiac complications and other medical complications or their impact on other CAP outcomes. Nevertheless, the authors concluded that major cardiac complications occur in a substantial proportion of patients with CAP [12].
Tralhão et al. undertook a meta-analysis to report the incidence of overall complications, ACS, new or worsening HF, new or worsening arrhythmias, and acute stroke as well as short-term mortality outcomes. In addition, they reviewed the interplay between the two conditions (pneumonia and CVD complications). They included 39 observational studies involving 92,188 patients, divided by setting (inpatients versus outpatients) and clinical severity (low risk versus high risk). They reported that the overall cardiac complications occurred in 13.9%, ACS in 4.5%, HF in 9.2%, arrhythmias in 7.2%, and stroke in 0.71% of the pooled inpatients. Furthermore, the meta-regression analysis suggested that overall and individual cardiac complication incidence decreased. After adjusting for confounders, cardiovascular events after CAP independently increased the risk for short-term mortality (range of OR: 1. 39-5.49). The authors' findings highlighted the need for effective, large, trial-based, preventive, and therapeutic interventions in this patient population [13].
Baskaran et al. performed a meta-analysis, searching for observational studies to summarize the literature on the incidence of ACS in adults with CAP. They looked for Medline and Embase and reported that 103 studies met the inclusion criteria. The authors included 26 studies (n = 66,347 patients), most of them of good quality. This meta-analysis showed that the pooled incidence of ACS in-hospital and 30 days after CAP was 3.2% (95% CI 2.4-4.0%; n = 17 studies) and 3.5% (95% CI 2.8-4.2%; n = 25 studies), respectively. Sensitivity analysis excluding studies with selected cohorts (elderly, predominantly male, or ICU admissions only) showed a higher pooled incidence for in-hospital ACS (4%, 95% CI 2.7-5.3%, n = 13 studies) but unchanged for 30-day incidence. These researchers concluded that their meta-analysis showed a small but significant risk of ACS in patients with CAP. This study was published in 2020 only in abstract form [79].
It is worth noting that most of the primary studies published to date and included in these three previous meta-analyses [12,13,79] had significant limitations. For example, most of them did not have an adequate control group or did not control for potential confounders (Supplementary Materials), all of which may affect the real effect of the exposition or intervention [80][81][82][83]; therefore, we excluded more than 50 of these studies in our systematic review . Furthermore, the meta-analyses by Corrales-Medina et al. [12], Tralhão et al. [13], and Baskaran et al. [79] are "meta-analyses of proportion". A "proportional meta-analysis" differs significantly in its methodology from a "traditional meta-analysis". We address these aspects later. Furthermore, the study by Baskaran et al. was published only in abstract form [79], so it is impossible to know what studies were included. Despite the limitations of the three meta-analyses described previously, their conclusions were concordant with our study [12,13,79].
The pathophysiologic mechanisms by which pneumonia can trigger CVD are probably diverse and involve several ways: (1) systemic and coronary artery inflammation increases cardiovascular risk; (2) infection and inflammation promote platelet activation and thrombosis; (3) changes in nitric oxide (NO) synthase and cyclooxygenase (COX) lead to endothelial dysfunction; (4) pneumonia impairs myocardial contractility, oxygen demand, and delivery; and additionally (5), the microorganisms may have a direct effect on cardiovascular risk [84][85][86][87][88][89]. However, most of the mechanisms above-mentioned were also observed in ischemic heart disease patients, particularly those with ACS. Thus, these findings may also affect pneumonia development after ACS, reflecting a reverse or bidirectional association [8,23,84,88].
To the best of our knowledge, this is the first "non-proportional" meta-analysis on CAP and CVD that reports ORs instead of proportions as a measure of effect size. A proportional meta-analysis is a data synthesis method that allows one to calculate a pooled, overall proportion from several individual proportions for a certain event instead of estimating an effect size, as conducted in a "traditional meta-analysis" [90,91]. That is, a proportional meta-analysis can include dichotomous data reported as a percentage. Proportional meta-analysis is encouraged when conducting systematic reviews of prevalence, incidence, and maybe, interventions and therapies where appropriate [92]. However, for dichotomous outcomes, the Cochrane Handbook does not recommend using proportional meta-analysis [14].
Furthermore, the appropriateness of conducting a proportional meta-analysis is controversial, as the individual studies contributing to such a meta-analysis commonly have been conducted in different contexts. These studies' prevalence and cumulative incidence estimates reflect unique population characteristics [93]. This methodology raises concerns when proportional meta-analysis assumes homogeneity, and an average estimate across different populations may be of little clinical use [92,94]. However, Corrales-Medina et al. [12] and Tralhão et al. [13]. reported that they used stratification (mainly according to treatment setting and clinical severity) to minimize the influence heterogeneity in estimating the effect sizes.
This work has some limitations. (1) Although we performed subgroup analysis according to the continent of origin, the type of study design, the sample size, and the length of follow-up, we could not perform subgroup analyses according to other important variables such as age, sex, treatment setting, clinical severity, and timing after CAP because of the lack of data. (2) It is possible that a meta-regression analysis could further explain the origin of the heterogeneity, although we did not perform this analysis due to limited data. (3) We cannot rule out a possible publication bias against negative studies that did not find a significant association between CAP and CVD complications. (4) The fact that the different continent of origin explains the heterogeneity found could reflect the different forms of diagnosis, treatment, and prognosis, both for pneumonia and cardiovascular complications in each of these countries.
Overall, if the heterogeneity is high, the conclusions of the meta-analysis may be less generalizable. Nevertheless, this does not necessarily mean that the results are incorrect or do not have clinical importance. Instead, it is crucial to interpret the meta-analysis results considering the heterogeneity and possible explanations for the differences between the studies included [44,[95][96][97]. In addition, there is always clinical and methodological diversity in a meta-analysis, so statistical heterogeneity is inevitable. Since systematic reviews bring together studies that are diverse both clinically and methodologically, heterogeneity in their results is to be expected [95][96][97]. In our study, heterogeneity was significant in most of the outcomes. However, we addressed this heterogeneity by implementing the recommendations of the Cochrane Handbook: (1) We verified the data correctness; (2) we explored the potential causes of heterogeneity; (3) we performed a meta-analysis with a random effects model, and (4) we performed sensitivity analyses [14].
We highlight some of the strengths of our work: (1) Our search strategy was thorough and complete; (2) we included the odds ratios instead of proportion as the effect measure; consequently, this is the first "traditional" meta-analysis, instead of proportional metaanalysis on this topic; (3) we included primary studies that specifically examined the association between pneumonia and CVD complications; (4) we excluded studies that reported a single cohort without a control group; and (5) we only included studies that reported the adjusted effect sizes. Therefore, our results are more robust than any other meta-analysis reported before.

Conclusions
In conclusion, our study shows that pneumonia should be considered as a new risk factor for cardiovascular complications. Furthermore, our findings support the hypothesis that inflammation triggered by acute and chronic infections such as pneumonia is crucial in the pathogenesis of atherosclerosis and cardiovascular complications. However, this conclusion should be taken with caution due to the limitations of our study. Institutional Review Board Statement: Ethical review and approval were waived since this was a secondary study.
Informed Consent Statement: Patient consent was waived since this was a secondary study.

Conflicts of Interest:
The authors declare no conflict of interest.