Objective: Individuals' recollections of the number and type of health service encounters are frequently required for health economic studies. We sought to establish whether the accuracy of self-reported healthcare resource utilization is a function of the duration of the recall period and the saliency of the health service encounter.

Methods: Patient recollections of a range of community services (general practitioner visits, community midwifery visits) and hospital services (accident and emergency attendances, hospital outpatient attendances, inpatient admissions) over 4-month and 8-month time periods were obtained from women participating in a randomized controlled trial. Comparisons were made with healthcare resource utilization data extracted from medical records. Where significant differences were identified between the self-reported and medically recorded data, a multivariate linear regression model was constructed to identify the factors associated with underreporting and overreporting of healthcare resource utilization.

Results: The study revealed a tendency to underreport community service utilization, which appears to be exacerbated when the recall period is extended. A number of sociodemographic and clinical factors significantly associated with this tendency to underreport community service utilization were identified. The self-reporting of hospital service utilization over varying periods of recall was found to be more accurate.

Conclusion: It is important that economic analysts establish optimal methods for estimating resource utilization quantities within health economic analytical designs. In particular, greater emphasis should be placed on extracting information on community service utilization from medical records or routine health service information systems.