Estimating incidence of cancer from a hospital discharge reporting system

We have evaluated the reliability of hospital discharge information of the Lombardy Region, in northern Italy, in identifying incident cases of cancer. This was accomplished by linking all cancer cases automatically identified from hospital discharges to the date of the Cancer Registry of Varèse Province, part of the same Region, up to 1977. The comparison showed a consistent overestimation of rates derived from hospital data, particularly in relation to the least common sites and extreme ages. When the procedure was applied to the most common tumors and truncated ages (35-74), estimated age-specific rates corresponded quite closely to those of Cancer Registry. We concluded that estimates of cancer incidence based on hospital discharge data are reliable when applied to selected tumors and can be useful for descriptive and other epidemiological purpose.