Is More Always Better? a Real-World Data Analysis of Hemoglobin A1C Data in Patients with Diabetes

OBJECTIVES: Laboratory data derived from real-world data are necessary for analyses where test results are indicators of disease severity, progression, or treatment plans. Research assessing the validity and reliability of laboratory data and its impact on outcomes is lacking. This study assesses the impact of completeness of hemoglobin A1C (HbA_1c) lab results among patients with diabetes and the association between HbA_1c and hospitalization.

METHODS: Adult patients with a diagnosis of diabetes (earliest = index) from 2018 to 2020 were selected from the Merative^TM MarketScan® administrative claims databases with laboratory data available. Patients were required to have one year of continuous enrollment pre and two years post index (baseline and follow-up) and have at least four HbA_1c tests during follow-up. Patients with matching lab results (HbA_1c value within 45 days of the test) were identified. Descriptive analysis and logistic regression models were conducted to assess the impact of missingness of matched lab results and the association between HbA_1c value and hospitalization while adjusting for baseline patient characteristics.

RESULTS: Out of 89,936 eligible patients (mean age 57.6, 13.4% type 1 diabetes, 75.8% commercially-insured), 52.9% had at least one matched HbA_1c result, and 20.2% had complete matched results. During follow-up, 19.3% of patients had at least one hospitalization. A patient’s first HbA_1c value was strongly correlated with future HbA_1c values (r=0.69, p<0.001). In the logistic regression conducted among those with at least one matched HbA_1c result, a patient’s first HbA_1c was significantly associated with hospitalization (OR=1.03, p<0.001). However, adding an additional HbA_1c test result did not improve model fit (p=0.868).

CONCLUSIONS: Study findings suggest having a single lab result is highly predictive of hospitalization; incremental gain of additional results available is marginal. Leveraging additional sources of laboratory data to increase the number of patients with at least one matching lab result can enhance robustness of an analysis.