BACKGROUND: Estimating the probability of pregnancy leading to delivery and the influence of clinical factors on that probability is of fundamental importance in the treatment counselling of infertile couples. A variety of statistical techniques have been used to analyse fertility data, many borrowed from survival analysis. METHODS AND RESULTS: We propose an alternative method of analysis which is based on a discrete time Markov chain approach, with states 'pregnancy (leading to a delivery)', 'not pregnant', and 'censored' and in which the transition probabilities are dependent both on the clinical characteristics of the patient and the treatment given. CONCLUSIONS: We believe that the method of analysis presented here may be preferable to standard analyses in that it better reflects the clinical situation, it is a truly discrete time analysis applied to a discrete time situation, it explicitly models the censoring process (a process which in itself provides information of interest to the physician) and can be readily extended to a variety of clinical situations.

Additional Metadata
Keywords *Markov Chains, *Reproductive Techniques, *Treatment Outcome, Age Factors, Female, Fertilization in Vitro, Humans, Infertility/*therapy, Male, Pregnancy, Regression Analysis
Persistent URL hdl.handle.net/1765/9816
Citation
McDonnell, J., Goverde, A.J., Vermeiden, J.P.W., & Rutten, F.F.H.. (2002). Multivariate Markov chain analysis of the probability of pregnancy in infertile couples undergoing assisted reproduction. Human Reproduction. Retrieved from http://hdl.handle.net/1765/9816