Background: Endophthalmitis is a severe condition that requires hospitalization with at least day care. Information on the incidence rate, costs and consequences of endophthalmitis is scarce.
Objective: To estimate the number of patients with endophthalmitis hospitalized in France, as well as the average costs and hospital budget consequences.
Methods: French Programme de Médicalisation des Systèmes d'Information (PMSI) data for 2006, derived from the official DRG classification, were analysed. Data were extracted concerning the following primary diagnoses: 'purulent endophthalmitis', 'other endophthalmitis' and 'endophthalmitis associated with another disease'. Two durations of hospitalization were compared: the actual duration and a weighted DRG duration. The cost of hospitalization was weighted by the average DRG cost + daily hospital costs x the difference between the actual and weighted DRG days in hospital. All costs are presented in euro, year 2007 values.
Results: A total of 1518 patients (mean age 68.7 years; 47.1% male) experienced 1725 hospitalizations for endophthalmitis, including 1416 cases (82.1%) admitted to public hospitals. The majority of patients (79.1%) were classified by DRG codes that did not specify endophthalmitis (DRG 02M03Z). Most patients (1342) were given a drug injection and 510 underwent vitrectomy. Four patients died in hospital and 75 were transferred to other hospitals. The actual duration of hospitalization for endophthalmitis in public hospitals was 8.1 days (mean), whereas the average weighted DRG duration was 5.1 days, which underestimated the actual duration by 3 days. The average hospital cost was 3688 euro per patient, totalling 6,361,119 euro per annum for all public and private hospitalizations in France, including 223,723 euro as day care. If hospital funding was wholly based on DRG tariffs, the budget for endophthalmitis would be severely underestimated. The DRG inclusion of 'severe acute ocular infections' as a proxy for endophthalmitis dramatically underestimated its true cost by approximately 30%.
Conclusion: For health economic evaluations, it is inappropriate to use DRG classifications as proxies for endophthalmitis. Expressed more generally, hospitalization cost analyses should not be based on any specific DRG, but always on the clinically relevant primary diagnosis. The PMSI clustering algorithm underestimates the hospital budgets required for endophthalmitis. Lastly, the PMSI (exhaustively reporting all hospitalizations) is best suited to capturing yearly endophthalmitis incidence rates, average costs and national health expenditure.