PURPOSE: Using hospitalization always has been seen as a solid measurement for exacerbation in pharmacoepidemiology, but might lead to an underestimation of disease exacerbation because of a trend towards outpatient care. The aim of this study waso quantify the incidence of different exacerbation markers in obstructive lung disease and to identify predictors for these exacerbation markers. METHODS: We conducted a cohort study using the PHARMO record linkage system, including demographic details and complete medication histories of more than two million community-dwelling residents in the Netherlands from 1985 onwards. Eligible patients were adult users of inhaled corticosteroids (ICS). Outcome parameters were hospitalization and short courses ofsystemic corticosteroids. Patients were allowed to have multiple exacerbations during follow-up. RESULTS: We identified 5327 patients. During follow-up, 8635 exacerbations occurred in 2332 patients with a trend in time towards treating exacerbations outof the hospital (p-value 0.003). Of all patients with exacerbations, 73% was not hospitalized during follow-up. Exacerbations were associated with high-dose ICS use (adjusted RR 1.4; 95% CI 1.2-1.7) and chronic systemic corticosteroid use (adjusted RR 1.9; 95%CI 1.6-2.2). CONCLUSIONS: Using hospitalization only as exacerbation marker leads to underestimating the exacerbation rate, because of exacerbation treatment out of the hospital. Patients with obstructive lung disease using chronic systemic corticosteroids or high-dose ICS use are more prone to exacerbations. This implies that these patients should be monitored carefully to prevent recurrent exacerbations which are detrimental for their prognosis and quality of life.