Agreement between self-report of disease diagnoses and medical record validation in disabled older women: factors that modify agreement.

OBJECTIVES: To determine the agreement between self-report of chronic disease and validated evidence of disease using multiple ascertainment methods and to assess effects of cognition, education, age, and comorbidity.
DESIGN: Cross-sectional analysis.
SETTING: Community Baltimore, Maryland. PARTICIPANTS: One thousand two community-dwelling disabled women aged 65 and older. MEASUREMENTS: Kappa statistics were calculated to determine the relationship between self-report of 14 diseases and standardized algorithms. Analyses were stratified using Mini-Mental State Examination score, education, number of chronic diseases, and age.
RESULTS: Kappa was excellent for hip fracture (HF), Parkinson's disease (PD), diabetes mellitus (DM), cancer, stroke, and disc disease (DD); fair to good for angina pectoris, congestive heart failure, and myocardial infarction; and poor for peripheral arterial disease, spinal stenosis, osteoporosis, arthritis, and lung disease. Overall, kappa decreased with decreasing cognition and education, increasing age, and four or more diseases.
CONCLUSION: In disabled older women, self-report of physician diagnosis of HF, PD, DM, cancer, stroke, and DD appears valid. In general, increasing comorbidity and age and decreasing cognition and education do not reduce validity for diseases where agreement was excellent overall.