Abstract
BACKGROUND: Hospital discharge data have been used to study trends in Lower Extremity Amputation (LEA) rates in people with and without diabetes. The aim of this study was to assess the reliability of routine hospital discharge data in the Republic of Ireland (RoI) for this purpose by determining the level of agreement between hospital discharge data and medical records for both the occurrence of LEA and diagnosis of diabetes. METHODS: Two concordance studies between hospital discharge data (HIPE) and medical records were performed. To determine the level of agreement for LEA occurrence, HIPE records were compared to theatre logbooks in 9 hospitals utilising HIPE over a two-year period in a defined study area. To determine the level of agreement for diabetes diagnosis, HIPE records were compared to laboratory records in each of the 4 largest hospitals utilising HIPE over a one week period in the same study area. The proportions of positive and negative agreement and Cohen's kappa statistic of agreement were calculated. RESULTS: During a two-year study period in 9 hospitals, 216 LEAs were recorded in both data sources. Sixteen LEAs were recorded in medical records alone and 25 LEAs were recorded in hospital discharge records alone. The proportion of positive agreement was 0.91 (95% CI 0.88-0.94), the proportion of negative agreement was 0.99 (95% CI 0.98-0.99) and the kappa statistic was 0.91 (95% CI 0.88-0.94). During a one-week study period in 4 hospitals, 49 patients with diabetes and 716 patients without diabetes were recorded in both data sources. Eighteen patients had diabetes in medical records alone and 2 patients had diabetes in hospital discharge records alone. The proportion of positive agreement was 0.83 (95% CI 0.76-0.9), the proportion of negative agreement was 0.99 (95% CI 0.98-0.99) and the kappa statistic was 0.82 (95% CI 0.75-0.89). CONCLUSIONS: This study detected high levels of agreement between hospital discharge data and medical records for LEA and diabetes in a defined study area. Based on these findings, we suggest that HIPE is sufficiently reliable to monitor trends in LEAs in people with and without diabetes in the RoI.