How accurate are SMR1 (Scottish Morbidity Record 1) data?

The accuracy of recording of data on SMR1 forms was reviewed for gastrointestinal diagnoses in the Greater Glasgow Health Board area and compared to previous studies. A total of 778 cases from 1987 formed the study sample and 761 (96.9%) were available for review. Recording on the SMR1 appears to have improved since 1971 and there were relatively few errors in the basic details. The crude agreement between DG1C data (primary diagnosis) and the casenote diagnosis was 560/761 (73.6%) with a Kappa statistic of 0.67. Agreement about the presence of arthritis (as a co-diagnosis) was poor so that uncritical use of SMR1 data might lead to serious underestimation of resource measures, such as length of stay, for patients with significant arthritis in addition to a gastrointestinal problem. In these data patients with arthritis found on casenote review and not recorded on the SMR1 had mean lengths of stay between 61% and 70% greater than patients without any evidence of significant arthritis.