“Glycometrics” – standardized metrics for inpatient glycemic control quality performance evaluation
METHODS: A cross-sectional and nationwide survey was conducted from July/2010 to January/2012. Eligible patients were ≥18 years old, had a diagnosis of diabetes and hospitalization length of stay ≥72 hours. Information on all blood glucose (BG) readings for a maximum of 20 consecutive days of hospitalization was collected by chart review. We used three analytical models: patient-day (grouped BG levels by calendar day for each patient), patient-stay (each patient’s mean BG level for the entire hospitalization) and patient-sample (all BG levels individually, without grouping). For each model we calculated the glycemic average level, the median BG, the percentage of BG measurements in range, and the percentage of hypoglycemic and hyperglycemic events.
RESULTS: Overall, 2,399 patients were surveyed in 24 hospitals located in 13 cities from all 5 regions in Brazil. Our dataset contained 71,137 BG measurements. The mean, median and percentage of BG measurements in range were similar among the three models, but hypoglycemic and hyperglycemic event rates varied considerably. Although none of the models have shown to be ideal for all functions, the patient-day model appeared to be the best tool for analysis of performance measures of the quality of inpatient glycemic control.
CONCLUSIONS: Different models are necessary for evaluating the quality of inpatients glycemic, as they can contribute to separate functions. These metrics need to be validated elsewhere to allow inter-patient and inter-hospital comparison, as well as quality improvement initiatives results assessment.