Ventricular premature complexes (VPCs) during stress testing in the general population are commonly seen in clinical practice, but their prognostic value is not well understood. A comprehensive literature search of MEDLINE, Embase, and the Cochrane Library from January 1970 to May 2015 was conducted. Observational cohort studies on general populations evaluating the association between exercise-induced VPCs and all-cause or cardiovascular mortality were included in the analysis. Nine studies comprising 62,488 participants comparing clinical outcomes of patients with and without exercise-induced VPCs were included. The overall combined relative risks (RRs) for all-cause mortality and cardiovascular mortality in patients with exercise-induced VPCs were 1.41 (95% CI 1.23 to 1.61) and 1.86 (95% CI 1.51 to 2.30), respectively. In subgroup analysis, both frequent VPCs (RR 1.35, 95% CI 1.14 to 1.60) and infrequent VPCs (RR 1.57, 95% CI 1.13 to 2.18) were associated with an adverse outcome. VPCs during recovery were associated with an increased risk of death (RR 1.55, 95% CI 1.22 to 1.96). VPCs during exercise did not achieve statistical significance (RR 1.14, 95% CI 0.96 to 1.34), but only a few studies were included in the analysis. In conclusion, our meta-analysis suggests that exercise-induced VPCs in the general population significantly increase the risk of total mortality and cardiovascular mortality. Our study calls for further studies to assess the prognostic significance of exercise-induced VPCs and the utility of efforts to reduce the VPC burden to improve the clinical outcome.
Ventricular premature complexes (VPCs) during exercise stress testing are frequently seen in clinical practice. Studies have found that among healthy subjects without apparent heart disease, 14% to 23% were found to have exercise-related VPCs. Despite the frequent occurrence of exercise-induced VPCs in the general population, the prognostic implication of VPCs provoked by exercise stress testing remains undetermined. Traditionally, VPCs during stress testing in the general population were believed to be benign and have no significant prognostic implication. Multiple epidemiologic cohort studies have sought to identify the significance of VPCs during stress testing, but there has been conflicting evidence for prognostic implication of VPCs during exercise stress testing in this general population. To address this unanswered question, we designed a meta-analysis of observational cohort studies to assess the association of VPCs during exercise stress testing and its prognostic implication in the general population.
Methods
A comprehensive MEDLINE (PubMed), Embase, and the Cochrane Library search was performed covering the period from January 1970 to March 2015 for studies with regards to the outcome of VPCs during or after stress testing in the general population. The following MeSH keywords: “exercise test,” “exercise testing,” “exercise stress test,” “exercise stress testing,” “ventricular premature complex,” “premature ventricular depolarization,” “ventricular premature beat,” “ventricular ectopy,” “NSVT,” “nonsustained ventricular tachycardia,” or “ventricular premature complex,” and “mortality” or “cardiovascular mortality” for outcome factors were used to identify all potential relevant studies regarding this topic. Additional publications were retrieved from the bibliographies of relevant manuscripts when they were considered potentially pertinent.
We included cohort studies that fulfill the following criteria: (1) inclusion of cohort studies targeted to general populations. Studies that selected specific populations only (i.e., young adults, postmyocardial infarction patients, patients with a history of congestive heart failure) were excluded. (2) Patients underwent exercise (either treadmill or bicycle) stress electrocardiographic (ECG) testing with or without myocardial perfusion imaging or echocardiogram. Pharmacologic stress testing using dobutamine, regadenoson, adenosine, or dipyridamole was excluded from this analysis. (3) Reported adjusted relative risk (RR) or hazard ratio (HR) of effects of exercise-induced VPCs on all-cause or cardiovascular mortality. Studies with no available data for outcome measures, those with the same population as other studies (in this case, we included the first published or the more comprehensive study in the analysis) and those with data on mortality only without adjustment for other cardiovascular risk factors were excluded from the analysis.
Two investigators independently conducted data extraction and bibliography evaluation. Discrepancies with regards to any of the extracted data were resolved by thorough discussion and consensus. Studies that did not meet the selection criteria and duplicate studies were excluded from the final analysis.
The methodological quality of the included studies was assessed based on the standardized set of predefined criteria described by Proper et al. The assessment tool was evaluated and validated to assess the methodological quality of prospective cohort studies. The assessment criteria consisted of 15 items, including the study population and participation, study attrition, data collection, and data analyses. This was used to evaluate each study included in the analysis. Positive quality criteria of 8 items or more were required for the study to be included in our analysis. A study was considered high quality if the number of positive items on the validity/precision (V/P) criteria was ≥5 of 9.
Stress-induced VPCs were defined as any ventricular premature complexes or nonsustained ventricular tachycardia (NSVT) during or after stress testing. The definition of frequent VPCs varied among included studies. For subgroup analysis, we defined frequent VPCs as those occurring more than 5 beats in 1 minute or VPC burden >10% during the exercise or recovery phase of the stress testing. Patients who had VPC burdens >10% or NSVT were classified as a high VPC burden group and subgroup analysis was performed separately.
Adjusted RR or HR of all-cause or cardiovascular mortality with 95% upper and lower CI was used to measure the effect size for the meta-analysis. All-cause mortality was used for the analysis as the primary outcome when one study reported both all-cause and cardiovascular mortality. When multiple RRs based on the frequency and timing of VPCs were reported, we tried to use the overall RR if available. If those data were not available, outcomes for the VPCs during the recovery phase and frequent VPCs were used for the meta-analysis.
Heterogeneity among individual studies was assessed using the Higgins I 2 and Q statistic. Both the fixed-effect model and random-effect model were used to calculate the pooled odds ratio or RR for the overall effect. The fixed-effect model was used when substantial heterogeneity was not found (i.e., I 2 ≤50%), and the random-effect model was used when substantial heterogeneity was found (i.e., I 2 >50%). Begg’s funnel plot and Egger’s test were used to assess publication bias. The Stata SE version 12.0 software package (StataCorp, College Station, Texas) was used for the statistical analysis. We followed the Meta-analysis of Observational Studies in Epidemiology (MOOSE) guideline for the meta-analysis of observational studies.
Results
A total of 1,329 study titles were identified by the initial literature search. Among these, 21 full-text articles evaluating the prognostic value of stress-induced VPCs were reviewed, and 9 studies met the eligible criteria for the systematic review. Of the 21 full-text articles reviewed, 8 did not have appropriate outcome data, 2 were review articles, one was performed on a different patient population (young athletes only) and one was based on the same cohort as another included in the study, and therefore were excluded from the analysis. This yielded a total of 9 studies comprising 62,488 participants included in our final analysis ( Figure 1 ).
The main characteristics of the studies included in the final analysis are summarized in Table 1 . Prevalence of exercise-induced VPCs varied significantly among studies included in the analysis from 3.7% to 55%. Among the 9 included studies, 6 studies attempted to exclude patients with an underlying coronary artery disease (CAD) or congestive heart failure [CHF] based on the history at baseline. All selected studies provided risk assessments adjusted for multiple clinical variables by various statistical methods, including Cox regression analysis or propensity-matched analysis. All included studies were high quality based on the quality assessment tool using the validity/precision (V/P) criteria, with a score of at least 7 ( Table 2 ). There was no evidence of publication bias among studies that had either all-cause mortality or cardiovascular mortality as the primary outcome (Egger linear regression test, p for bias = 0.81 and 0.46, respectively). Begg’s funnel plot did not identify a considerable asymmetry ( Figure 2 ).
| Study | Year | Country | Number of Subjects | Population | Mean Age | Male | Follow Up (years) | Type of Stress Test | Clinical Endpoints | Definition of Frequent VPCs | Timing of VPCs | RR (95% CI) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Jouven et al. | 2000 | France | 6101 | Asymptomatic men without known cardiovascular disease | Unspecified | 100% | 23 | Bicycle Exercise | Cardiovascular mortality | > 10% of all ventricular depolarizations | Stress | 2.53 (1.65 – 3.88) |
| Myers et al. | 2002 | USA | 2534 | Subjects without known cardiovascular disease | 55.5 | 100% | 6.2 | GXT | All-cause mortality | > 10% of all ventricular depolarizations | Stress and recovery | 1.14 (0.64 – 2.01) |
| Mora et al. | 2003 | USA | 2994 | Asymptomatic women without known cardiovascular disease | Unspecified | 0 | 20.3 | GXT | All-cause mortality | > 10% of all ventricular depolarizations | Stress and recovery | 1.19 (0.9 – 1.58) |
| Frolkis et al. | 2003 | USA | 29244 | Subjects without known CHF | 56 | 70% | 5.3 | GXT | All-cause mortality | ≥ 7 per minute | Recovery | 1.5 (1.1 – 1.9) |
| Stress | 1.1 (0.9 – 1.3) | |||||||||||
| Morshedi-Meibodi et al. | 2004 | USA | 2885 | Subjects without known cardiovascular disease | 43 | 48% | 15 | GXT | All-cause mortality | > 0.22/min | Stress and recovery | 1.66 (1.14 -2.4) |
| Beckerman et al. | 2005 | USA | 5754 | Subjects without atrial fibrillation | 64 | 100% | 6 | GXT | Cardiovascular mortality | > 10% of all ventricular depolarizations | Stress and recovery | 1.6 (1.1 – 2.1) |
| Dewey et al. | 2008 | USA | 1847 | Subjects without known CHF | 56 | 95% | 5.4 | GXT | All-cause mortality | > 0.43 per minute | Recovery | 1.71 (1.07 – 2.73) |
| Stress | 1.34 (0.88 – 1.04) | |||||||||||
| Marine et al. | 2013 | USA | 2099 | Subjects without known cardiovascular disease | 52 | 52% | 13.5 | GXT | All-cause mortality | Presence of ≥ 3 consecutive beats | Stress and recovery | 1.3 (0.89 – 1.9) |
| Dhoble et al. | 2014 | USA | 9030 | Asymptomatic women without known cardiovascular disease | 49.3 | 58% | 8.1 | GXT | All-cause mortality | > 5 per minute | Unspecified | 1.51 (1 – 2.15) |
| Study | Study Population and Participation | Study attrition | ||||||
|---|---|---|---|---|---|---|---|---|
| Adequate description of source population | Adequate description of sampling frame, recruitment methods, period of recruitment and place of recruitment (setting and geographic location) | Participation rate at baseline at least 80%, or if the nonresponse was not selective (show that baseline study sample does not significantly differ from population of eligible subjects) | Adequate description of baseline study sample (i.e., individuals entering the study) for key characteristics | Provision of the exact n at each follow-up measurement | Provision of exact information on follow-up duration | Response at short-term follow-up (up to 12 months) was at least 80% of the n at baseline and response at long-term follow-up was at least 70% of the n at baseline | Information on not selective nonresponse during follow-up measurements | |
| Jouven et al. | Yes | Yes | Yes | Yes | No | Yes | Yes | Yes |
| Myers et al. | No | No | Not presented | Yes | No | Yes | Yes | No |
| Mora et al. | Yes | Yes | Not presented | Yes | No | Yes | Yes | No |
| Frolkis et al. | Yes | Yes | Yes | Yes | No | Yes | Yes | No |
| Morshedi-Meibodi et al. | Yes | Yes | Yes | Yes | No | Yes | Yes | No |
| Beckerman et al. | Yes | Yes | Not presented | Yes | No | No | Not presented | No |
| Dewey et al. | Yes | Yes | Yes | Yes | No | Yes | Yes | No |
| Marine et al. | Yes | Yes | Yes | Yes | No | Yes | Yes | No |
| Dhoble et al. | Yes | Yes | Yes | Yes | No | Yes | Yes | No |
| Study | Data collection | Data analyses | Score | |||||
|---|---|---|---|---|---|---|---|---|
| Adequate description of measurement and definition of frequent or exercise induced VPC | VPC was assessed at a time prior to the measurement of the health outcome | Adequate measurement of the health outcome: objective measurement of the health outcome done by trained personnel by means of standardized protocols of acceptable quality and not by self-report | The statistical model used was appropriate | The number of cases was at least 10 times the number of the independent variables | Presentation of point estimates and measures of variability (CI or SE) | No selective reporting of results (yes for no selective reporting, no for presence of selective reporting) | ||
| Jouven et al. | Yes | Yes | Yes | Yes | Yes | Yes | Yes | 14 |
| Myers et al. | Yes | Yes | Yes | Yes | Yes | Yes | Yes | 10 |
| Mora et al. | Yes | Yes | Yes | Yes | Yes | Yes | Yes | 12 |
| Frolkis et al. | Yes | Yes | Yes | Yes | Yes | Yes | Yes | 13 |
| Morshedi-Meibodi et al. | Yes | Yes | Yes | Yes | Yes | Yes | Yes | 13 |
| Beckerman et al. | Yes | Yes | Yes | Yes | Yes | Yes | Yes | 10 |
| Dewey et al. | Yes | Yes | Yes | Yes | Yes | Yes | Yes | 13 |
| Marine et al. | Yes | Yes | Yes | Yes | Yes | Yes | Yes | 13 |
| Dhoble et al. | No | Yes | Yes | Yes | Yes | Yes | Yes | 12 |
Meta-analysis of 7 studies reporting the all-cause mortality demonstrated that exercise-induced VPCs are associated with an increased risk of all-cause mortality (RR 1.41, 95% CI 1.23 to 1.61; Figure 3 ). In addition, there was a substantial increase in the risk of cardiovascular mortality in subjects who had exercise-induced VPCs based on 4 studies that had cardiovascular mortality as the primary outcome (RR 1.86, 95% CI 1.51 to 2.30; Figure 4 ).
All included studies reported data on frequent exercise-induced VPCs and risk of death. Among these, 4 studies used a definition of frequent VPCs that met the predefined criteria for the pooled analysis and had all-cause mortality as the primary outcome, and 2 studies had cardiovascular mortality as the primary outcome. Frequent exercise-induced VPCs were found to be positively associated with an increased risk for all-cause mortality (RR 1.35, 95% CI 1.14 to 1.6) and cardiovascular mortality (RR 1.97, 95% CI 1.26 to 3.08). Only 3 studies had the outcome of infrequent exercise-induced VPCs, and pooled outcome of those studies demonstrated that infrequent exercise-induced VPCs were associated with a significantly increased risk of all-cause mortality as well (RR 1.57, 95% CI 1.13 to 2.18).
Only 3 of the 9 studies reported data on the specific timing of exercise-induced VPCs, and 2 of those studies had all-cause mortality as the primary outcome. The overall combined RR of all-cause mortality in relation to VPCs during recovery was 1.55 (95% CI 1.22 to 1.96), whereas the overall RR in relation to VPCs during exercise was 1.14 (95% CI 0.96 to 1.34), which did not reach statistical significance.
Six studies attempted to exclude patients with an underlying CAD or CHF based on the history at baseline, and 5 of those studies reported all-cause mortality as the primary outcome. The pooled adjusted RR for all-cause mortality comparing subjects with exercise-induced VPCs to those without was 1.36 (95% CI 1.15 to 1.6).
Table 3 summarizes the results of other subgroup meta-analyses on subsets of patients. Subgroup analyses were performed based on mean follow-up period, and high VPC burden as well as timing and frequency of exercise-induced VPCs as previously described. Overall, all subgroups had a significantly increased risk of death compared with the subjects who did not have exercise-induced VPCs.
