Association of episodic physical and sexual activity with triggering of acute cardiac events: systematic review and meta-analysis

Abstract

Context: Evidence has suggested that physical and sexual activity might be triggers of acute cardiac events.

Objective: To assess the effect of episodic physical and sexual activity on acute cardiac events using data from case-crossover studies.

Data sources: MEDLINE and EMBASE (through February 2, 2011) and Web of Science (through October 6, 2010).

Study selection: Case-crossover studies investigating the association between episodic physical or sexual activity and myocardial infarction (MI) or sudden cardiac death (SCD).

Data extraction: Two reviewers extracted descriptive and quantitative information from each study. We calculated summary relative risks (RRs) using random-effects meta-analysis and absolute event rates based on US data for the incidence of MI and SCD. We used the Fisher P value synthesis method to test whether habitual physical activity levels modify the triggering effect and meta-regression to quantify the interaction between habitual levels of physical activity and the triggering effect.

Results: We identified 10 studies investigating episodic physical activity, 3 studies investigating sexual activity, and 1 study investigating both exposures. The outcomes of interest were MI (10 studies), acute coronary syndrome (1 study), and SCD (3 studies). Episodic physical and sexual activity were associated with an increase in the risk of MI (RR = 3.45; 95% confidence interval [CI], 2.33-5.13, and RR = 2.70; 95% CI, 1.48-4.91, respectively). Episodic physical activity was associated with SCD (RR = 4.98; 95% CI, 1.47-16.91). The effect of triggers on the absolute rate of events was limited because exposure to physical and sexual activity is infrequent and their effect is transient; the absolute risk increase associated with 1 hour of additional physical or sexual activity per week was estimated as 2 to 3 per 10,000 person-years for MI and 1 per 10,000 person-years for SCD. Habitual activity levels significantly affected the association of episodic physical activity and MI (P < .001), episodic physical activity and SCD (P < .001), and sexual activity and MI (P = .04); in all cases, individuals with lower habitual activity levels had an increased RR for the triggering effect. For every additional time per week an individual was habitually exposed to physical activity, the RR for MI decreased by approximately 45%, and the RR for SCD decreased by 30%.

Conclusion: Acute cardiac events were significantly associated with episodic physical and sexual activity; this association was attenuated among persons with high levels of habitual physical activity.

Figure 1. Search Strategy Flowchart

One publication provided results from 2 separate populations, which were considered as independent strata in our analyses. ISI indicates Institute of Scientific Information.

Figure 2. Forest Plot of Studies Assessing the Association of Episodic Physical and Sexual Activity With Myocardial Infarction and Sudden Cardiac Death

Summary results for each exposure-outcome subgroup are presented separately. Within each subgroup, studies are arranged by the point estimate of relative risk. Values greater than 1 indicate that exposure is associated with increased risk of the outcome. Squares are proportional to the weight of each study in the meta-analysis. CI indicates confidence interval.

Figure 3. Meta-regression Graph of Triggering Relative Risks for Myocardial Infarction and Sudden Cardiac Death Over Habitual Physical Activity Levels

Meta-regression graph of relative risks (RRs) for myocardial infarction is based on 3 studies that used the same scale (weekly frequency) to report habitual activity levels^,,; graph for sudden cardiac death is based on 2 studies that used the same scale.^, Subgroup estimates are depicted as circles proportional to their precision (inverse of the variance of the log[RR]). The solid line indicates fitted values by random-effects meta-regression. rRR indicates relative RR calculated from the meta-regression.