MARKOV - A COMPUTER-PROGRAM FOR MULTISTATE MARKOV-MODELS WITH COVARIABLES

This paper discusses a computer program, called MARKOV, designed to fi t a multi-state Markov model with covariables, with a particular empha sis on the analysis of survival data. The Markov model consists of k - 1 transient disease states and one absorbing state. The exact transit ion times are not observed, except in situations such as death. Baseli ne transition intensities and regression coefficients are estimated vi a the method of maximum likelihood using a quasi-Newton optimization a lgorithm. The program's output includes the parameter estimates, the s tandard error of the estimates, the matrix of the correlation of the e stimates and minus two times the log-likelihood function, evaluated at the initial values and at the maximum likelihood estimates. Optionall y, survival curves can be generated from each transient state, for one or more combination of covariable values and simple tests about the p arameters. The program is illustrated by using a four-state model to d etermine factors influencing diabetic retinopathy in young subjects wi th insulin-dependent diabetes mellitus.