Comparing the congruency of self-report and provider records of depressed elders' service use by provider type

Background: An accurate accounting of service use is necessary to understand use patterns and outcomes. Yet such an accounting remains challenging, in part because of the reliability and validity of the collection method and sources. Objectives: This study describes 2 methods of data collection: self-report and the retrieval of provider records. We report on the effort, yield, and challenges of retrieving records. Then, we compare the congruency and completeness of 2 methods: self-report and provider records. Finally, we examine the impact of various patients' characteristics on congruency rates. Method: Our sample of depressed older participants was recruited from an inpatient geropsychiatry unit before they were discharged into the community. We interviewed participants at 3 points during a 6-month period. Provider records were obtained across provider type, based on self-report and snowballing technique. We calculated congruency rates and examined completeness of either data source on 91 participants with completed provider records. Using logistic regression, we examined the differences in congruency by provider type as well as factors related to the congruency. Results: The record retrieval process is labor-intensive and challenging. We found that congruency rates were statistically higher for pharmacy and hospital providers and lower for physicians. We also found higher counts of service use, higher depression levels, and being married were significantly related with lower congruency between self-report of service use and provider records. Discussion: Although we found relatively high congruency rates between self-report and service records, the choice of methods depends on the purpose of the research and breadth of provider types.