Effect of an Automated Hemoglobin A1c Reflex on Socioeconomic Disparities in Testing Frequency and Glycemic Control

Abstract

Abstract The American Diabetes Association recommends semiannual hemoglobin A1c (A1c) testing for diabetic patients with stable glycemic control and quarterly testing for patients with a change in therapy or who do not meet A1c goals. To aid in achieving this goal our laboratory performs an automated, add-on A1c test for any diabetic and hospitalized patient with no A1c ordered in the previous 60 days. The primary aim of this study was to determine the impact of an automated reflex on A1c testing frequency and glycemic control. Our second aim was to determine socioeconomic differences of reflexed A1c testing. All A1c results performed by the BJH laboratory from 5/1/2016-1/22/22 and associated patient demographics were retrieved from the laboratory information system. Patient addresses were geocoded, and a CDC-validated social vulnerability index (SVI) was assigned to each patient based on their census tract of residence. Patients were then stratified in the bottom SVI half (most vulnerable) and the upper SVI half (least vulnerable). A total of 32,861 A1c’s were performed on 17,699 patients, 6,341 of whom had >1 test. 2,952 (46.6%) patients had at least one reflexed A1c (add-on group). The median average time between A1c tests was 144.4 days in the add-on group and 209.0 days in the no add-on group (p<0.001). The average time between A1c tests for patients in the add-on group in the bottom SVI half was 143.8 days vs. 145.3 days in the top SVI half (p=0.84). In contrast, the median average time between A1c tests for patients in the no add-on group in the bottom SVI half was 215.5 days and was 199.5 days in the top SVI half (p<0.05). The number of patients with A1c in the uncontrolled range (A1c =9%) decreased by 11.5% in the add-on group compared to 1.7% in the no add-on group (p<0.001). In those with add-on testing, the proportion of patients in the bottom SVI with initial A1c =9% and a longitudinal decrease in A1c was 11.3% compared to 11.9% in the top SVI half (p=0.39). In the no add-ons group, the proportion of patients with A1c =9% increased by 3.3% in the bottom SVI and decreased by 12.1% in the top SVI (p=0.01). A socioeconomic disparity is observed in A1c testing frequency and rates of uncontrolled diabetes over time for patients without add-on testing, but not for patients with add-on testing. Automatic add-on testing for hospitalized diabetic patients may reduce socioeconomic disparities in A1c testing frequency and glycemic control.