Diabetes Screening Rates and Practices in Patients With Sickle Cell Disease

Diabetes mellitus (DM) and sickle cell disease (SCD) both disproportionately impact Black people (1); however, minimal data defining DM prevalence in patients with SCD exist, and there is no consensus on how to best screen for DM in these patients. The prevalence of DM was previously thought to be lower in those with SCD, but recent studies suggest rates similar to those of the U.S. Black population (1). One challenge is that hemoglobin A_1c (HbA_1c) may be falsely low in individuals with SCD (1) due to the shortened erythrocyte life span, presence of hemoglobin variants, and use of blood transfusions. While prior studies demonstrated likely underdiagnosis of DM in those with sickle cell trait (2), no similar literature exists on SCD. We examined DM screening rates and practices in adults with SCD at our institution, with the goal of informing a future prospective study that will compare diagnostic strategies in this population.

This was a single-center retrospective chart review of adult patients (≥18 years of age) with SCD who received longitudinal care, defined as an outpatient visit at Boston Medical Center (BMC) between 1 June 2016 and 31 May 2019. This study period was chosen to avoid potential bias from the coronavirus disease 2019 pandemic. SCD genotypes (i.e., HbSS, HbSC, and HbSβ^0or+) were defined by hemoglobin electrophoresis. People with sickle cell trait were excluded. The study was determined to be exempt from review for human subject research by the Boston University/BMC institutional review board.

Demographics, clinical histories, and laboratory data were deidentified and collected by a single investigator (S.S.), with monthly review for consistency. The primary outcome was percentage screened for DM, defined as measurement of HbA_1c, fructosamine, or both. While fructosamine is not recommended in screening guidelines, it can be used as a surrogate marker of glycemic control when there are potential limitations to HbA_1c. It was included because the goal was to assess rates of attempted DM screening. The retrospective nature of the study did not allow ascertainment of fasting status for blood glucose levels. No participants underwent oral glucose tolerance testing. Secondary outcomes included the number who qualified for screening per the 2019 American Diabetes Association (ADA) guidelines (the most recently available during the study period) (3), screening method(s), and frequencies of DM and prediabetes diagnoses. DM and prediabetes were defined by diagnosis documentation in at least one clinician note during the study period (ICD-10 coding or free text). Fructosamine levels were correlated with HbA_1c levels using the formula HbA_1c = 0.017 × fructosamine (µmol/L) + 1.61.

A total of 222 patients met study inclusion criteria; 148/222 (66.7%) had HbSS disease, 123/222 (55.4%) were female, and 188/222 (84.7%) were Black. The mean age was 36.1 ± 12.6 years. Screening occurred in 28.8% of all patients and in 42.9% who qualified per ADA guidelines; 58/64 (90.6%) were screened with HbA_1c alone (Table 1). While 5/222 (2.3%) were diagnosed with DM, 2/63 (3.2%) had HbA_1c consistent with DM but were not diagnosed. Three of the 222 patients (1.4%) were diagnosed with prediabetes; an additional 6/63 (7.9%) had HbA_1c consistent with prediabetes without a diagnosis. Twenty-three of the 222 (10.4%) patients had HbA_1c ≥5.7% (39 mmol/mol) at least once, a correlating fructosamine level, or a random serum glucose level >200 mg/dL, suggesting DM or prediabetes diagnosis.

Our results demonstrate low screening rates for DM in patients with SCD, slightly lower than that observed in the U.S. population (46.2%) (4). Of patients screened, >90% were screened with HbA_1c alone, suggesting potential for underdiagnosis. Our data support underdiagnosis in this population even with positive testing, as 10.4% of patients had laboratory data consistent with prediabetes or DM but only 3.7% carried a diagnosis.

Limitations include use of retrospective data from a small cohort at a single site. Screening practices at BMC, an urban safety net hospital with a sickle cell center, may not be representative of practices at other sites. However, we expect that low screening rates and high rates of using HbA_1c alone are not unique and may be amplified in settings with less SCD experience. Data were collected via chart review, making variation in documentation and clinical practice a potential limitation. Additionally, lack of fasting data or documentation of symptoms limited our ability to assess other DM screening modalities.

In summary, we observed low rates of DM screening in patients with SCD and that DM and prediabetes diagnoses were overlooked even when screening occurred. This is significant, as missed DM diagnoses may contribute to vascular complications, morbidity, and mortality in an already high-risk population. Further studies are needed to investigate diagnostic strategies for DM in patients with SCD and guide screening to improve clinical outcomes.