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 register 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 hospitalizations and outpatient oncology database) and one external (Israel Cancer Registry) database. Histologic diagnosis was confirmed by search of on-line pathology files or discharge summaries. Demographic, histologic and survival characteristics were compared among patients identified through each data source. Survival was assessed via record linkage with the Population Registry. Setting. A tertiary care university teaching hospital. Study participants. Seven hundred and twelve patients with lymphoma diagnosed between 1987 and 1992. Results. All three sources contributed independently to the total, with the majority (583 or 82%) via hospitalizations. Overlap among the sources was minimal with only 73 (10.3%) of the patients being common to all three sources. Differences (P < 0.05)in the age distribution, country of birth and population groups (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 sources (p = 0.02 for Hodgkin's disease) even after controlling for age. Conclusions. The use of multiple data sources for case ascertainment in clinical audit is justified when no patient register exists. The results indicate that use of a single internal data source would have resulted both in an underestimate of the scope of lymphoma in our institution and significant bias in terms of patient characteristics and outcome.