Évaluation des performances du PMSI pour l’identification des tumeurs incidentes du système nerveux central par rapport à un registre spécialisé en Gironde, France, en 2004

Evaluation of algorithms to identify incident cancer cases by using French health administrative databases

“…Still, when we used only inpatient hospital stays, we underestimated cancer incidence by 31% for men and 15% for women. This observation was found in many estimates of site‐specific cancer incidence with the PMSI database, probably because of the increasing rate of outpatient care in cancer management (one‐third of breast cancers in the USA and half of prostate cancers in Spain). Thus, the inclusion of reimbursement data within our most accurate algorithm definition improved our cancer incidence estimation by 18.6% and 25.7% for men and women, respectively (prostate cancer [40%] and breast cancer [35%]).…”

“…Many studies have shown the importance of health administrative databases in estimating cancer incidence . Most of these were based on inpatient data to validate cancer cases .…”

“…Although 84% of our study population had at least 5 years of history in the database (argued to be largely sufficient), some cancer cases could pre‐date this history. To increase the likelihood of selecting new cancer cases, most studies have requested an additional cancer‐specific procedure code and/or excluded prevalent cancer cases . By applying both methods (algorithms 7 and 8), we produced the best SIR (1.00 [0.94–1.06]).…”

“…Algorithm 4 relied on a primary or related diagnosis of cancer or, for algorithm 4b, any hospitalization for radiotherapy. Associated diagnoses of cancer were disregarded because their use results in a high level of prevalent cases …”

“…Many studies have developed methods to identify incident cancers by using health administrative databases . However, given that no ‘date of diagnosis’ variable exists in these databases, the main difficulty remains in the ability to distinguish between prevalent and incident cases, a key issue for etiological cancer studies.…”