1 July 1987 in Circulation
Impairment of left ventricular function is the major predictor of mortality after acute myocardial infarction, but it is not known whether this is best described by ejection fraction or by end-systolic or end-diastolic volume. We measured volumes, ejection fractions, and severity of coronary arterial occlusions and stenoses in 605 male patients under 60 years of age at 1 to 2 months after a first (n = 443) or recurrent (n = 162) myocardial infarction and followed these patients for a mean of 78 months for survivors (range 15 to 165 months). There were 101 cardiac deaths, 71 (70%) of which were sudden (instantaneous or found dead). Multivariate analysis with log rank testing and the Cox proportional hazards model showed that end-systolic volume (chi 2 = 82.9) had greater predictive value for survival than end-diastolic volume (chi 2 = 59.0) or ejection fraction (chi 2 = 46.6), whereas stepwise analysis showed that once the relationship between survival and end-systolic volume had been fitted, there was no additional significant predictive information in either end-diastolic volume or ejection fraction. Severity of coronary occlusions and stenoses showed additional prediction of only borderline significance (p = .04 in one analysis), but continued cigarette smoking did remain an independent risk factor after stepwise analysis. For a subset of patients (n = 200) who had taken part in a randomized trial of coronary artery surgery after recovery from infarction, surgical "intention to treat" showed no predictive value.(ABSTRACT TRUNCATED AT 250 WORDS)
1 September 1996 in Journal of the Royal Statistical Society Series B: Statistical Methodology
1 February 1984 in The American Journal of Cardiology
1 July 1999 in Journal of the Royal Statistical Society Series B: Statistical Methodology
1 March 1997 in Biometrika
We consider fitting categorical regression models to data obtained by either stratified or nonstratified case-control, or response selective, sampling from a finite population with known population totals in each response category. With certain models, such as the logistic with appropriate constant terms, a method variously known as conditional maximum likelihood (Breslow & Cain, 1988) or pseudo-conditional likelihood (Wild, 1991), which involves the prospective fitting of a pseudo-model, results in maximum likelihood estimates of case-control data. We extend these results by showing the maximum likelihood estimates for any model can be found by iterating this process with a simple updating of offset parameters. Attention is also paid to estimation of the asymptotic covariance matrix. One benefit of the results of this paper is the ability to obtain maximum likelihood estimates of the parameters of logistic models for stratified case-control studies, compare Breslow & Cain (1988), Scott & Wild (1991), using an ordinary logistic regression program, even when the stratum constants are modelled.