Objective: To estimate incidence rates of drug-related hospitalizations (DRHs) in a longitudinal population-based study with prospective event assessment.
Design: Cohort study and time-trend analysis.
Setting: All departments of internal medicine and emergency departments in the urban regions of Jena and Rostock, Germany, serving about 520,000 residents.
Participants: All patients admitted between October 1997 and March 2000. Patients with severe cutaneous reactions were excluded.
Main outcome measures: Incidence of DRH was defined by symptoms or diagnoses at admission that were very likely, likely, or possibly caused by prescription medications, according to a standardized assessment.
Results: The incidence of DRH was 9.4 admissions per 10,000 treated patients [95% confidence interval (CI) 9.0-9.9]. Rates were highest for antithrombotics with 26.9 admissions per 10,000 treated patients (95% CI 23.6, 30.1). Most frequent events were gastroduodenal lesions and bleeding (45%). Digitalis preparations showed a linearly increasing trend from 2/10,000 to 14/10,000 during ten quarters ( P<0.0001), which was exclusively attributable to digitoxin, the major source of digitalis in the study area (93%). The incidence of DRH increased with age (4/10,000 to 20/10,000). The mean length of stays in patients with DRH was 13+/-10.6 days. Cumulative direct costs for hospitalization were Euro 4 million in the two urban study areas. The annual direct costs for Germany were estimated to be Euro 400 million.
Conclusions: DRHs are a considerable public health and economic burden. A longitudinal design can observe changes in population-based incidence over time. This approach can be used for public-health planning or to evaluate outcomes of quality management programs designed to reduce drug-induced illness.