Comparative evaluation of quality of diabetes care within a population using discrete HbA1c thresholds versus longitudinal trends within a defined time period.

Abstract

HbA1c is a quality measurement for a population's diabetes care. Currently, discrete values are used to judge the care quality. However, an improved trend may have been achieved without reaching a discrete threshold. We aimed to compare the quality of glycaemic care using discrete thresholds versus longitudinal trends of a patient's HbA1c. We derived a study population of 4062 with two consecutive HbA1c's within 1-6 months. We used current discrete thresholds and then used these to define the longitudinal thresholds (the change in HbA1c between the first and second measurement). We found that even with discrete thresholds, overall care was improving with 61% (Excellent care), 13% (Good care), 13% (Average care) and 13% (Poor care) turning into 68%, 12%, 11% and 9%, respectively (χ² 3335, p<0.0001). Using longitudinal trends shows a greater improvement of care with the original values achieving 74%, 7%, 9%, 7% and 4%, respectively (χ² 4111.3, p<0.0001). There was an additional 28% (Good care), 15% (Average care) and 12% (Poor and Very poor care) who with longitudinal trends improved to an excellent category without being identified as such. Our study highlights the need to consider longitudinal trends when measuring quality of diabetes care.