THE CONTRIBUTION OF MULTIPLE DATA SOURCES TO A CLINICAL AUDIT OF LYMPHOMA IN A TEACHING HOSPITAL

Objective. To assess the independent contribution of data sources used for case ascertainment in a clinical audit of the outcome of lymphoma patients in our institution. Design. In the absence of a unified regi ster of lymphoma patients diagnosed and/or treated in our institution, we used an elective approach to case ascertainment. Sources included two internal databases (computerized discharge data for hospitalizatio ns and outpatient oncology database) and one external (Israel Cancer R egistry) database. Histologic diagnosis was confirmed by search of on- line pathology files or discharge summaries. Demographic, histologic a nd survival characteristics were compared among patients identified th rough each data source. Survival was assessed via record linkage with the Population Registry.Setting. A tertiary care university teaching h ospital. Study participants. Seven hundred and twelve patients with ly mphoma diagnosed between 1987 and 1992. Results. All three sources con tributed independently to the total, with the majority (583 or 82%) vi a hospitalizations. Overlap among the sources was minimal with only 73 (10.3%) of the patients being common to all three sources. Difference s (P < 0.05)in the age distribution, country of birth and population g roups (defined by religion) as well as histologic diagnoses among the sources were noted. In addition, survival of those ascertained via the outpatient database was higher than those ascertained from other sour ces (p = 0.02 for Hodgkin's disease) even after controlling for age. C onclusions. The use of multiple data sources for case ascertainment in clinical audit is justified when no patient register exists. The resu lts indicate that use of a single internal data source would have resu lted both in an underestimate of the scope of lymphoma in our institut ion and significant bias in terms of patient characteristics and outco me.