Patients' self-report of diseases in the Medicare Health Outcomes Survey based on comparisons with linked survey and medical data from the Veterans Health Administration.

We evaluated the patient self-report questions about disease from the Medicare Health Outcomes Survey (HOS), using linked Veterans Health Administration (VA) data for patients who are eligible for both Medicare and Veterans Affairs (VA) care to estimate their utility as measures of illness burden. Patients were classified for 12 diseases on the basis of HOS question responses and these were compared with classifications based on similar questions from a VA survey or diagnostic codes from VA medical records. Agreement between classifications based on the 2 surveys was good with over 75% of patients affirming the disease in the HOS also affirming it in the VA survey for most diseases. HOS disease status also agreed well with VA-based disease status using diagnostic codes for most diseases, with reasonably good specificity (70%-94%) and sensitivity (65%-85%). The relatively poor measures of agreement for some of the conditions could be related to differences in question wording and other factors. These findings varied only slightly by education, age, and race. Furthermore, independent decrements in health status, derived from the SF-36 associated with each disease based on the survey questions, were similar in the 2 surveys. These results suggest that patients can provide reasonably good reports of their morbidity in survey questions and that patient self-report questions about disease can be used reliably in case-mix adjustments and in stratifications of patients by diseases.