Three computer models for the calculation of prevalence of peptic ulcer disease during long‐term treatment

SUMMARY At present the effects of maintenance treatment for peptic ulcer disease are usually calculated by using ‘life-table’ analyses. Whilst these accurately demonstrate the speed with which an initial relapse occurs they make no allowance for the fact that, in clinical practice, a relapse often responds to a further course of full-dose treatment and the patient then returns to maintenance therapy. A further compounding factor is that, in any long-term study, patients will be lost to follow-up for a variety of reasons not all related to failure of the treatment. In this paper we describe the use of ‘prevalence rates’ to better reflect the outcome of peptic ulcer management. Three ‘computer models’, which have been developed to address the problems of patients leaving the study for any reason during such a long time-period, are also described, as are the underlying assumptions made. Using the results from a long-term study of continuous treatment with cimetidine,1 the ‘prevalence rates’ of ulcer disease over 6 years were calculated. Observed relapse rates appeared to fall with time (from 2.7% for duodenal ulcer (DU) and 2.5% for gastric ulcer (GU) to 1% and 2% respectively). However, on applying the models to the data, prevalence rates tended to rise slowly with time for the first 3 years in each of the models tested. At 6 years, two of the models suggested that the prevalence rate for DU would be about 8%; this is not very different to the reported recurrence rate after surgical treatment by truncal vagotomy and pyloroplasty. It is concluded that ‘prevalence rates’ should be used to assess long-term medical treatments for ulcer disease. Similar methods could also be used to examine the medical treatment of any other disease where multiple relapses, capable of responding to re-treatment, occur. The use of models proved beneficial in compensating for patients lost during the study.