What can be learned about the impact of diabetes on hospital admissions from routinely recorded data?

Diabet. Med. 29, 1199‐1205 (2012) Abstract Aim  To determine whether inpatients with diabetes have different lengths of stay, day‐case listing rates or emergency readmission rates compared with those without diabetes receiving similar treatment. Method  English 2007–2008 Hospital Episode Statistics were analysed alone and after linkage to the 1.6 million people included in the 2007–2008 English National Diabetes Audit. Length of stay, day‐case listing rates and emergency readmission rates were compared between those with and without diabetes by Health Resource Group chapter. Using univariate and multivariate statistics, the potential influences of age, sex, social deprivation, type and complexity of admission were considered. Results  Of all inpatient spells in Hospital Episode Statistics, 6.2% have diabetes coded at discharge. Substantial under‐recording of diabetes has been identified—a further 3% of inpatient spells involve people with diabetes. Inpatients with recorded diabetes stay in hospital for 100% longer on average, are 50% less likely to be treated as day cases and are almost 100% more likely to be readmitted as an emergency. The adverse impact of diabetes on length of stay was similar to that for chronic obstructive pulmonary disease. Most of this adverse effect was in patients coded with diabetes in Hospital Episode Statistics. Multivariate statistical analysis showed that diabetes is independently associated with increased length of stay. The effect differed up to threefold between hospitals. Conclusions  Diabetes has an independent adverse effect on key aspects of hospital inpatient stays. If the lowest levels of adverse impact on inpatients with diabetes were more common, substantial cost savings and improved experience of care would be realized.