Study objective: Control group characteristics as comorbidity and chronic psychosocial problems may play an important part in study outcomes. A primary care data base was used to quantify the effects of varying the case mix of participants.
Design: Historical cohort study.
Setting: Data were collected from 1967-1996 in four Dutch general practices performing the Continuous Morbidity Registration Nijmegen.
Patients and controls: All newly diagnosed type 2 diabetic patients in the period 1967-1989 fulfilling the WHO criteria (n = 265); for each type 2 diabetic patient a control was matched for practice, sex, age, and social class; from these controls subgroups were selected based on the absence of different types of morbidity; these subgroups were also matched for practice, sex, age, and social class.
Main results: The relative risk of mortality in type 2 diabetic patients in comparison with various subsets of controls ranged from 1.33 (95% CI 0.97, 1.81) to 2.16 (95% CI 1.46, 3.20).
Conclusion: Control group characteristics as comorbidity and chronic psychosocial problems turned out to influence the risk estimation in a cohort study. General practice data enhance the study of these aspects.