Abstract
This study proposes race/ethnicity-specific glycemic thresholds for predicting diabetic retinopathy status in white, black, and Hispanic individuals.
Diabetic retinopathy (DR) is the leading cause of acquired blindness in the US adult population and one of the major microvascular complications of type 1/2 diabetes. Epidemiologic studies have used glycemic thresholds to suggest thresholds for predicting DR status.1 While there is extensive research on racial/ethnic differences in diabetes and on a glycemic threshold for DR in general, to our knowledge, there is limited literature on race/ethnicity–specific glycemic thresholds to predict risk for DR and DR status. Using a nationally representative sample, this study proposes race/ethnicity–specific glycemic thresholds to predict DR status among US adults.
Methods
Data (2005–2008) from the National Health and Nutrition Examination Survey were analyzed for US adults (age ≥40 years). The National Health and Nutrition Examination Survey is a nationally representative sample of the noninstitutionalized US adult population; the data are deidentified and publicly available on the Centers for Disease Control and Prevention website,2 making this study was exempt from institutional review board approval. To determine DR status, retinal images were graded by ophthalmologists using 45° nonmydriatic digital images of the retina. Race/ethnicity was self-reported and categorized as Hispanic (comprising Mexican-American and other Hispanic), white, or black. Logistic regression models were fitted to predict DR status using hemoglobin A1c (HbA1c). Receiver operator curves and the Youden index were used to determine the optimal threshold for HbA1c in predicting DR status by optimizing sensitivity, specificity, and positive and negative predictive values.
Results
In total, 661 individuals (12.4%) had retinal imaging results that showed signs of DR (Table 1). Most were white men who were living above the poverty line. Black individuals had the highest proportion of DR overall (195 of 1084 [18.0%]) and of proliferative DR (12 of 195 [6.2%]), while white individuals had the lowest proportion of DR overall (291 of 3003 [9.7%]). Across all races/ethnicities, men had higher rates of DR compared with women. Across races/ethnicities and DR status, most never smoked.
Table 1. Sociodemographic and Clinical Characteristics by Race/Ethnicity and Diabetic Retinopathy Status.
Characteristic | No. (%) | ||||||
---|---|---|---|---|---|---|---|
All | Race/Ethnicity, Retinopathy Status | ||||||
Hispanica (n = 1251 [23.4%]) | Non-Hispanic Black (n = 1084 [20.3%]) | Non-Hispanic White (n = 3003 [56.3%]) | |||||
No | Yes | No | Yes | No | Yes | ||
Allb | 5338 (100.0) | 1076 (20.2) | 175 (3.3) | 889 (16.7) | 195 (3.7) | 2712 (50.8) | 291 (5.5) |
Sex | |||||||
Female | 2665 (49.9) | 567 (52.7) | 79 (45.1) | 459 (51.6) | 91 (46.7) | 1346 (49.6) | 123 (42.3) |
Male | 2673 (50.1) | 509 (47.3) | 96 (54.9) | 430 (48.4) | 104 (53.3) | 1366 (50.4) | 168 (57.7) |
Socioeconomic statusc | |||||||
Low | 1143 (21.4) | 358 (33.3) | 66 (37.7) | 198 (22.3) | 45 (23.1) | 440 (16.2) | 36 (12.4) |
High | 4195 (78.6) | 718 (66.7) | 109 (62.3) | 691 (77.7) | 150 (76.9) | 2272 (83.8) | 255 (87.6) |
Smoking status | |||||||
Current | 1068 (20.0) | 178 (16.5) | 26 (14.9) | 238 (26.8) | 37 (19.0) | 524 (19.3) | 65 (22.3) |
Former | 1763 (33.0) | 312 (29.0) | 54 (30.9) | 229 (25.8) | 65 (33.3) | 1001 (36.9) | 102 (35.1) |
Never | 2507 (47.0) | 586 (54.5) | 95 (54.3) | 422 (47.5) | 93 (47.7) | 1187 (43.8) | 124 (42.6) |
Diabetes diagnosis | |||||||
No | 4396 (84.3) | 899 (85.5) | 86 (51.2) | 728 (83.9) | 87 (45.3) | 2417 (91.3) | 179 (62.4) |
Yes | 818 (15.7) | 152 (14.5) | 82 (48.8) | 140 (16.1) | 105 (54.7) | 231 (8.7) | 108 (37.6) |
Insulin treatmentd | |||||||
No | 518 (43.7) | 93 (41.2) | 28 (29.8) | 84 (42.0) | 39 (34.5) | 235 (54.3) | 39 (32.8) |
Yes | 667 (56.3) | 133 (58.8) | 66 (70.2) | 116 (58.0) | 74 (65.5) | 198 (45.7) | 80 (67.2) |
Hypoglycemic agentsd | |||||||
No | 5134 (96.2) | 1055 (98.1) | 153 (87.4) | 867 (97.5) | 147 (75.4) | 2677 (98.7) | 235 (80.8) |
Yes | 204 (3.8) | 21 (1.9) | 22 (12.6) | 22 (2.5) | 48 (24.6) | 35 (1.3) | 56 (19.2) |
Diabetic retinopathy severity | |||||||
Mild | 537 (10.1) | NA | 136 (77.7) | NA | 141 (72.3) | NA | 260 (89.4) |
Moderate | 97 (1.8) | NA | 29 (16.6) | NA | 42 (21.5) | NA | 26 (8.9) |
Proliferative | 27 (0.5) | NA | 10 (5.7) | NA | 12 (6.2) | NA | 5 (1.7) |
Abbreviation: NA, not applicable.
Hispanic includes Mexican American or other Hispanic.
Row percentages.
Socioeconomic status was determined by considering household income to their appropriate poverty threshold by the poverty guidelines, specific to family size, state, and year.
Currently taking medication.
Among all races/ethnicities, the optimal predictive HbA1c threshold was 6.0% (Table 2). For black and Hispanic persons, the optimal HbA1c thresholds were 6.5% and 6.4%, respectively. The optimal HbA1c threshold for white people was 6.0%. The 6.5% threshold for black individuals had the highest specificity (85.0%) and sensitivity (52.3%) compared with those for white and Hispanic individuals. All 3 race/ethnicity-specific thresholds were significant predictors of DR status based on the generated odds ratios.
Table 2. Diagnostic Values for Optimal Threshold for Glycemic Indices for Diabetic Retinopathy.
Characteristic | All Estimate (95% CI) | Estimate (95% CI) | ||
---|---|---|---|---|
Hispanica | Non-Hispanic Black | Non-Hispanic White | ||
Hemoglobin A1c, % | 6.0 | 6.4 | 6.5 | 6.0 |
OR (95% CI)b | 4.4 (3.8-5.3) | 5.5 (3.9-7.8) | 6.2 (4.5-8.7) | 4.2 (3.2-5.4) |
Sensitivity | 55.2 (51.4-59.0) | 49.7 (42.3-57.1) | 52.3 (45.3-59.3) | 45.0 (39.3-50.7) |
Specificity | 78.3 (77.1-79.5) | 84.9 (82.7-87.0) | 85.0 (82.7-87.4) | 83.6 (82.2-85.0) |
Positive predictive value | 26.2 (24.1-28.8) | 34.8 (28.9-40.7) | 43.4 (37.1-49.7) | 22.7 (19.3-26.1) |
Negative predictive value | 92.5 (91.7-93.3) | 91.2 (89.5-93.0) | 89.1 (87.0-91.2) | 93.4 (92.4-94.4) |
Abbreviation: OR, odds ratio.
Hispanic includes Mexican American or other Hispanic.
All ORs are significant at the 5% significance level (ie, none of the 95% confidence intervals include 1).
Discussion
This study proposes race/ethnicity–specific glycemic thresholds for predicting DR status in white, black, and Hispanic individuals. Colagiuri et al1 proposed an HbA1c threshold of 6.4% for predicting DR in the US population, which was most similar to our proposed thresholds for black and Hispanic individuals.1 While past studies have proposed glycemic thresholds for predicting DR, race/ethnicity–specific thresholds are needed because of racial/ethnic differences in complications of diabetes. Past population-based studies have identified racial/ethnic differences in glycemic levels for diabetes diagnoses, which suggests a need for race/ethnicity-specific thresholds in predicting diabetes-associated complications.3
With diabetes becoming more prevalent, it is important to identify biomarkers and thresholds for diabetic complications. Epidemiological studies have supported HbA1c as a significant predictor of DR across all races/ethnicities and ages.4 Interestingly, black and Hispanic individuals had a higher proportion of overall DR prevalence and more severe proliferative DR compared with white individuals, yet the optimal glycemic threshold was lowest for white individuals.
Routine ophthalmologic follow-up compliance typically is poor in some cohorts of people with diabetes.5 Health care professionals might use these glycemic thresholds as a tool to emphasize the importance of routine ophthalmologic follow-up. Importantly, all 3 race/ethnicity–specific glycemic thresholds are less than the recommended 7.0% for people with diabetes.6 This finding suggests that adequate glycemic control does not guarantee protection from diabetic complications, such as DR. These thresholds might provide health professionals with a more individualized approach to diabetes-related patient care. In summary, this study demonstrates race/ethnicity-specific glycemic thresholds for DR in US adults. Future longitudinal studies are necessary to further validate these thresholds and to investigate the racial differences in DR.
References
- 1.Colagiuri S, Lee CM, Wong TY, Balkau B, Shaw JE, Borch-Johnsen K; DETECT-2 Collaboration Writing Group . Glycemic thresholds for diabetes-specific retinopathy: implications for diagnostic criteria for diabetes. Diabetes Care. 2011;34(1):145-150. doi: 10.2337/dc10-1206 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Centers for Disease Control and Prevention NHANES questionnaires, datasets, and related documentation. https://wwwn.cdc.gov/nchs/nhanes/default.aspx. Accessed July 15, 2018.
- 3.Davidson MB, Schriger DL. Effect of age and race/ethnicity on HbA1c levels in people without known diabetes mellitus: implications for the diagnosis of diabetes. Diabetes Res Clin Pract. 2010;87(3):415-421. doi: 10.1016/j.diabres.2009.12.013 [DOI] [PubMed] [Google Scholar]
- 4.Semeraro F, Parrinello G, Cancarini A, et al. . Predicting the risk of diabetic retinopathy in type 2 diabetic patients. J Diabetes Complications. 2011;25(5):292-297. doi: 10.1016/j.jdiacomp.2010.12.002 [DOI] [PubMed] [Google Scholar]
- 5.Keenum Z, McGwin G Jr, Witherspoon CD, Haller JA, Clark ME, Owsley C. Patients’ adherence to recommended follow-up eye care after diabetic retinopathy screening in a publicly funded county clinic and factors associated with follow-up eye care use. JAMA Ophthalmol. 2016;134(11):1221-1228. doi: 10.1001/jamaophthalmol.2016.3081 [DOI] [PubMed] [Google Scholar]
- 6.Association AD; American Diabetes Association . 5. glycemic targets. Diabetes Care. 2016;39(suppl 1):S39-S46. doi: 10.2337/dc16-S008 [DOI] [PubMed] [Google Scholar]