"... In PAGE 4: ...mong men (RR 5.9, 95% CI, 2.8 to 12.5, p , 0.001). Results were similar among the subjects with autopsy- documented sudden cardiac death ( Table2 ). Short-term exponent also predicted nonsudden cardiac mortality but not cerebrovascular mortality (Table 2).... In PAGE 5: ... Multivariate predictors of mortality. Table2 shows the multivariate relative risks for HR variability measures ad- justed for other risk variables, such as age, gender, heart failure, angina pectoris, functional class, previous myocar- dial infarction, cardiac medication and ventricular prema- ture beats. Although the short-term exponent had weak associations to various clinical parameters (Table 3), it remained as a strong independent predictor of sudden cardiac death after adjustment for other variables in multi- variate analysis.... ..."

"... In PAGE 14: ...v. b-blockers are indicated in patients with ventricular arrythmias33 (class I, level of evidence A) ( Table11 ). SCD secondary to VF is very frequent after an acute coronary occlusion.... In PAGE 14: ...s. the untreated group. In the CAPRICORN trial in post MI patients with left ventricular dysfunction, there was a trend toward SCD reduction in the carvedilol group.66 Heart failure Patients with a history of congestive heart failure 67 or depressed left ventricular function171 show the greatest benefit from b-blockers in mortality reduction, including SCD and are indicated in all patients for the prevention of SCD (Class I, level of evidence A)35 ( Table11 ). A consis- tent contribution to the improved outcome by these drugs is related to a substantial reduction (between 40% and 55%) in SCD rates.... ..."

"... In PAGE 73: ...reduced EF, an episode of nsVT and an abnormal SAECG as predictors of SCD and non- SCD are shown in Table6 (V). The prevalence of incomplete TWA test results as well as non-diagnostic BRS and HRV results was higher among the patients who died (III).... In PAGE 73: ... The prevalence of incomplete TWA test results as well as non-diagnostic BRS and HRV results was higher among the patients who died (III). The specificity, sensitivity, and positive and negative predictive accuracies of various risk predictors in predicting sudden cardiac mortality are listed in Table6 . The positive predictive accuracy of all these variables was lower in predicting SCD (between 5 and 11%) than non-SCD (between 12 and 18%).... ..."

"... In PAGE 19: ... A further analysis of the data is done by dividing the data in certain relevant patient group subsets, and examine if the frequency of cardiac arrests in these groups relates signi cantly to air temperature. The results are reported in Table11 , and seems intuitively reasonable. It is reasonable that the frequency of cardiac arrests of non-cardiac origin is not related to... ..."

"... In PAGE 14: ...v. b-blockers are indicated in patients with ventricular arrythmias33 (class I, level of evidence A) (Table11 ). SCD secondary to VF is very frequent after an acute coronary occlusion.... In PAGE 14: ...s. the untreated group. In the CAPRICORN trial in post MI patients with left ventricular dysfunction, there was a trend toward SCD reduction in the carvedilol group.66 Heart failure Patients with a history of congestive heart failure 67 or depressed left ventricular function171 show the greatest benefit from b-blockers in mortality reduction, including SCD and are indicated in all patients for the prevention of SCD (Class I, level of evidence A)35 (Table11 ). A consis- tent contribution to the improved outcome by these drugs is related to a substantial reduction (between 40% and 55%) in SCD rates.... ..."

"... In PAGE 6: ... In particular, End-to-end Estimator functions can manage private proxy resources that replace references to shared resources. The upper half of Table6 describes compo- nents that can store-and-forward retransmissions of lost or delayed notifications, predict future values from past information already received, discard information from untrusted sources, and summarize past data so as to send only the latest information. Ultimately, summarizers could even take advantage of excess bandwidth to reduce latency further by speculating about future states as well [43].... ..."

"... In PAGE 6: ... In particular, End-to-end Estimator functions can manage private proxy resources that replace references to shared resources. The upper half of Table6 describes compo- nents that can store-and-forward retransmissions of lost or delayed notifications, predict future values from past information already received, discard information from untrusted sources, and summarize past data so as to send only the latest information. Ultimately, summarizers could even take advantage of excess bandwidth to reduce latency further by speculating about future states as well [43].... ..."

"... In PAGE 17: ...[ Table5 here] quot;Much of the maximum likelihood theory deals with the large sample (asymptotic) properties of MLE quot; (Kotz and Johnson 1985, p.342), but the quot;large sample properties of the maximum likelihood estimators still make them quite attractive for samples of 50 or more quot; (Alan J.... In PAGE 28: ...[ Table5 here] APPENDIX 4 CALCULATION OF THE MEAN AND MEDIAN RANKS (The following method of mean rank determination is after Johnson (1964, p39).) The ranks of ordered uncensored data (completed durations) are determined, then the censored and uncensored values (withdrawals) are pooled and ordered.... ..."

"... In PAGE 12: ... 0.49 mV was predictive with 95%, 90%, 94% and 91%. The con dence intervals (cf. Table1 ) are widest for speci city and negative predictive value, due to the smaller sample size of the non-survivors. 4 Discussion The non-invasive predictor of countershock success after prolonged cardiac arrest proposed here, brillation power, is independent of, and thus com- plementary to, established indicators such as mean brillation frequency or amplitude.... ..."