Prostate cancer treatment ascertained from several sources: analysis of disagreement and error

Abstract

Treatment data for prostate cancer can be obtained from a variety of sources. Each of these sources has its own strengths and weaknesses and is subject to error. In a population-based cohort of 319 prostate cancer patients, data on treatment were obtained from five sources: two patient interviews at 6 and 12 months after diagnosis, primary caregiver interviews, physician questionnaires, and medical records. Inter-reporting agreement and accuracy of reporting (compared with medical records) were assessed. Multivariate analyses examined patient, caregiver, and physician characteristics as determinants of reporting error. The agreement among different reporting methods was generally good to excellent for prostatectomy and brachytherapy (kappa range 0.70-0.90) and fair to good (kappa range 0.35-0.75) for external beam radiation and hormonal treatment. Compared with medical records, the interview- and questionnaire-based data collection methods were more accurate for prostatectomy and brachytherapy than for external beam radiation and hormonal therapy. Using medical records as the 'gold standard', patient and caregiver interviews at 6 months after the diagnosis had higher sensitivity and specificity than other reporting sources. Interviews of prostate cancer patients and caregivers are useful alternatives to medical record abstraction, particularly if carried out during, or soon after, treatment.