Using a Stochastic Model to Investigate Time to Absorption Distributions

This paper demonstrates that, after being tested for goodness of fit, Markov chain models may be used in epidemiological investigations to compute the time to absorption distributions useful in designing follow-up studies. Also illustrated is the use of these models to compare distinct cohorts with respect to stochastic movement among states. Using data from a historical prospective study conducted in Skopje, Yugoslavia, we employ a Markov chain model of women's reproductive paths to compare the stochastic behavior of first pregnancy aborters and first pregnancy deliverers. We estimate the time to next delivery distribution within each of these cohorts, constructs useful in determining the required sample size for future studies of the effects of induced abortion.