On the basis of this research, we selected 6 large studies (Table 1).1, 2, 10, 11, 12?and?13 Two other studies8?and?9 Cefaloridine
that initially did not fulfill all inclusion criteria were added because authors send us their raw data that were not available in their original published articles. Finally, in order to gather more data on patients with ICD, 2 further studies by Sacher et al17 and Sarkozy et al18 were selected that analyzed only patients with ICD in detail (Table 2). The article by Sarkozy was admitted, although it referred to only 47 patients. The homogeneity of the available studies was tested by using a heterogeneity test. In addition, the Begg test was used to evaluate any predominant effect. When performing cumulative analyses, we excluded any studies that also had been part of multicenter investigations in order to avoid the double counting of patients. Event rates (ICD-FVA and SD) were expressed per 1000 patient-years of follow-up, with 95% confidence intervals. Follow-up durations that were expressed as median values were changed to means and variances by using the method described by Hozo et al.19 The cumulative analysis of homogeneous studies Raf inhibitor
was done by calculating the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of risk factors, expressed as both percentages and proportions. When considering each risk factor (eg, spontaneous type 1 ECG pattern), we classified the presence and absence of this risk factor in patients with events as true positive and false negative, respectively. Conversely, in patients without events, the presence and absence of this risk factor was classified as false positive and true negative, respectively. All analyses were performed by means of the StataSE 12.0 statistical software (StataCorp, College Station, TX). Two sample-proportion tests were used to calculate statistical differences between group percentages. Event rates were compared between groups by means of incidence rate (incidence density or person-time) data. In all statistical tests, a value of P < .05 was considered see more
statistically significant (when not specified, the P value should be considered 2-tailed). We asked all authors of the included studies to check their respective data in our text and tables and, if possible, to provide further data. The authors found only minimal errors in our analysis of data pooled from their original articles. All errors were corrected. The original articles by Eckardt et al,7 Takagi et al,8 Kamakura et al,9 Giustetto et al,10 and Probst et al,11 also included patients with previous cardiac arrest.