Diabetes is a complex disease associated with many potential long-term complications, such as heart disease, stroke, kidney failure, lower limb amputation, and visual impairment.1 Diabetic retinopathy (ie, damage to the small blood vessels in the retina) is one of the most common complications in patients with chronic diabetes.1-3 In fact, in an analysis of 2005-2008 data from the National Health and Nutrition Examination Survey, 28.5% of US adults aged ≥40 years with diabetes had diabetic retinopathy, and 4.4% of these adults had advanced diabetic retinopathy (ie, a vision-threatening condition).1,4 Notably, diabetic retinopathy is the main cause of new cases of blindness (ie, central visual acuity of <20/200) in US adults aged 20 to 74 years.2
Several factors have been linked to an increased risk for the development of diabetic retinopathy, including male sex, relatively longer duration of diabetes, and the use of insulin.4 Although many diabetic patients with diabetic retinopathy may be unaware of its presence or progression,3 the early signs of retinopathy can be detected on retinal screening examination, and the treatment of early retinopathy can prevent or delay blindness.2 Because of the high prevalence of diabetic retinopathy in the population with diabetes and the availability of effective treatment, the standards of medical care include screening diabetic patients for early signs of eye disease and implementing appropriate treatment to prevent blindness and other visual complications.3
Several professional organizations have developed clinical practice guidelines for diabetic retinopathy screening. Diabetic retinopathy screening guidelines established by the American Association of Clinical Endocrinologists (AACE), the American Diabetes Association (ADA), and the American Academy of Ophthalmology (AAO) recommend referring patients with type 2 diabetes to an ophthalmologist or optometrist for an annual dilated eye examination at the time of a diagnosis of diabetes.5-7 The AACE and ADA recommend dilated eye examination within 5 years of diagnosis in patients with type 1 diabetes,5,6 and the AAO recommends dilated eye examination at 5 years after diagnosis in patients with type 1 diabetes.7
Despite the long-standing availability of professional guidelines for diabetic retinopathy screening, only 55.7% of adults with diabetes in commercial HMO plans and 46.9% of those in commercial preferred provider organization (PPO) plans received screening in 2013, according to the Healthcare Effectiveness Data and Information Set (HEDIS).8,9 HEDIS is a tool used by US health plans to measure performance on important dimensions of care and service.9 Yet, there are data suggesting that patients with diabetes who receive guideline-recommended screening and care for diabetic retinopathy have better outcomes than those who do not.10,11 For example, a study assessing the effect of receiving guideline-recommended care with the onset of diabetic retinopathy and its progression in Medicare beneficiaries with diabetes found that those who received such care were diagnosed with background diabetic retinopathy earlier than those who did not, and they experienced lower rates of the onset of diminished vision or blindness.10 Furthermore, the data from a novel study that incorporated a simulation of a population of veterans with diabetes who had not yet developed diabetic retinopathy showed that vision loss associated with diabetic retinopathy increased as the screening interval was extended from 1 year to 5 years.11
The current claims-based study of commercially insured nonelderly adults with diabetes was conducted to better understand the patterns of diabetic retinopathy screening and the relationship between screening frequency and the cost of care. Our analysis, which is the first to quantify this relationship, used a novel longitudinal approach to follow patients with newly diagnosed type 2 diabetes for 10 years. Specifically, we assessed 13 years of relevant commercial claims data in patients who were newly diagnosed with diabetes and who had ≥11 years of continuous eligibility (ie, 1 year of eligibility before the year of their diabetes diagnosis and 10 consecutive years of eligibility, including the diagnosis year).
Methods and Study Design
The 2000-2013 Truven Health MarketScan commercial databases were used to identify the diabetic population. These databases contain all paid claims generated by >40 million commercially insured individuals annually, the member identification codes that allow individuals to be followed longitudinally, and standard codes for diagnoses, procedures, and drugs. Individuals who were insured by capitated plans were excluded, as were people who did not have a pharmacy benefit. HEDIS definitions were used to identify patients with diabetes, retinopathy screening, and retinopathy diagnosis.9 Gestational diabetes was specifically excluded in our selection process, which followed the HEDIS criteria.
Two distinct analyses were performed, including a 2001-2013 annual trend analysis of patients with type 1 diabetes and with type 2 diabetes and a 10-year longitudinal analysis of patients with type 2 diabetes.
In the trend analysis, diabetic patients aged 18 to 64 years were identified for each index year between 2001 and 2013. Patients with type 1 diabetes were defined as those who had (1) no prescription for drugs for the treatment of noninsulin diabetes; (2) ≥1 prescriptions for insulin; and (3) an International Classification of Diseases, Ninth Revision, Clinical Modification diagnosis code of 250.xx coded to 5 digits, with the fifth digit being a 1 or a 3 more than 50% of the time. Patients with type 2 diabetes were defined as those who did not meet the type 1 diabetes definition, as stated above.
The total diabetic population (denominator) for each calendar year was required to have ≥1 months of eligibility in the calendar year and eligibility in all months of the previous year to distinguish newly diagnosed diabetes from existing diabetes. The annual prevalence of diabetes, the diabetic retinopathy screening rates, the prevalence of diabetic retinopathy, and the allowed cost per member per month (PMPM) were calculated. The allowed cost PMPM included payer reimbursement plus member cost-sharing.
The index year was defined as the year in which the individual was identified with type 2 diabetes. The 10-year longitudinal analysis consisted of newly diagnosed patients with type 2 diabetes who were aged 18 to 64 years and were eligible in the year before each of the 4 index years (2001-2004) and had ≥10 years of continuous eligibility after the index year. Patients with newly diagnosed diabetes were identified as those not meeting HEDIS diabetes identification criteria in the year before the index year. The analysis was limited to patients with type 2 diabetes, because retinal screening of patients with type 1 diabetes need not begin until 5 years after diagnosis.5-7 Data from the 4 index years of the patients with diabetes were combined, and the cost trends were matched to the 2004 index year cohort using the annual diabetes population cost trends that were calculated in the trend analysis.
The longitudinal population was segmented into the number of years that each individual met the HEDIS diabetic retinopathy screening criteria (ie, 0, 1-4, 5-7, and 8-10 years) over the 10-year period. The relationship between the number of years with retinopathy screening and the PMPM allowed costs were analyzed after adjustment for age and sex, the severity of diabetes in the year after diagnosis (diabetes drug therapy proxy classification), and the US Census geographic region. The diabetes drug utilization was identified using Wolters Kluwer Medi-Span National Drug Codes for diabetes. T-tests were performed to compare the difference in the mean incremental cost from year 1 to year 10 in the 4 retinopathy screening cohorts.
In the trend analysis, the prevalence of diabetes among patients aged 18 to 64 years increased from 3.93% in 2001 to 5.08% in 2013 (P <.001). The annual retinal screening rate rose from 26.27% in 2001 to 29.58% in 2013 (P <.001). The average total unadjusted allowed cost of care for each individual with diabetes rose from approximately $822 PMPM in 2001 to approximately $1395 PMPM in 2013. Table 1 shows the annual trend in key metrics from 2001 to 2013.
To understand the impact of screening on costs, we examined the relationship between increased adherence for retinal screening over a 10-year period (0 years, 1-4 years, 5-7 years, and 8-10 years) and allowed claim costs in patients with newly diagnosed type 2 diabetes. Figure 1 shows the distribution of the patients with diabetes by years with screening since diagnosis.
Table 2 provides the study population’s characteristics for each of the 4 screening frequency cohorts. In the 0-year screening cohort, the highest number of patients was in the East South Central region (37% total). In the 8- to 10-year screening cohort, the highest number of patients (44% of total) was in the South Atlantic region. Census region adjustments were made for every year, depending on the regions relevant to participating members for that year.
We examined the relationship between age, sex, and drug class utilization, as well as the average allowed claims costs in year 10 after diagnosis to determine the need for risk adjustment among the retinal screening cohorts. We found a statistically significant positive correlation between the age at diagnosis of patients with type 2 diabetes and the PMPM claims cost in year 10 (P <.001; Figure 2).
The highest average allowed claims cost in year 10 was found in women aged 18 to 29 years at diagnosis, likely because of the cost of obstetrical care. Figure 3 shows the relationship between the severity of diabetes as measured by drug class utilization and the PMPM cost claims in year 10. After adjusting for age, sex, drug class utilization, and geographic region, we identified the no-screening cohort to have the lowest allowed PMPM cost claims in year 10 after diagnosis.
Figure 4 shows the relationship between the screening rate in patients with type 2 diabetes and the adjusted allowed PMPM cost claims in year 10. In patients with type 2 diabetes who had retinopathy screening in at least 1 of the 10 years after their initial diagnosis, this relationship was not statistically significant (P >.315).
After adjusting for age, sex, drug class severity, and geographic region, we analyzed the correlation between retinal screening frequency over the 10 longitudinal years and the PMPM increase in allowed cost claims from year 1 to year 10. The average adjusted allowed PMPM costs were shown to increase steadily from year 0 to year 10 in each cohort (Figure 5). The incremental mean cost increase from year 1 to year 10 was $503 in the 0-year screening cohort, $688 in the 1- to 4-year screening cohort, $705 in the 5- to 7-year screening cohort, and $720 in the 8- to 10-year screening cohort.
The 0-year screening cohort was shown to have the lowest costs in the year before receiving a diagnoses of diabetes of all the screening cohorts (Table 3). The cost differences between screening cohorts were statistically significant, with the exception of the differences between the 5- to 7-year and 8- to 10-year cohorts.
The increase in the average PMPM cost from year 1 to year 10 was found to be statistically significant in the 0-year screening cohort compared with each of the 3 other screening cohorts; however, the difference between the mean increase in PMPM costs between the 1- to 4-, 5- to 7-, and 8- to 10-year screening cohorts was not statistically significant (Table 4).
Our analysis confirms the increase in diabetes prevalence over time that has been noted elsewhere. The Centers for Disease Control and Prevention, for example, reported that the age-adjusted US prevalence of diagnosed diabetes increased from 4.7% in 2001 to 6.4% in 2011.12 In our study of commercially insured nonelderly individuals, we found that the prevalence of diabetes increased from 3.93% in 2001 to 5.33% in 2011.
The current analysis also showed a statistically significant increase in the screening rates for diabetic retinopathy, from 26.27% in 2001 to 29.58% in 2013. As a benchmark, 55.7% of adults with diabetes in commercial HMO plans and 46.9% of those in commercial PPO plans received screening for diabetic retinopathy in 2013, according to the HEDIS criteria.8,9
In those health plan populations, diabetic retinopathy screening was identified on the basis of claims data plus medical records information, and a normal result on a diabetic retinopathy screening in 2012 meant that an absence of screening in 2013 counted as screening. Regardless, the absolute screening rates for diabetic retinopathy are low, relative to the accepted standards of care.6,7
In the longitudinal analysis, we observed that <30% of patients with type 2 diabetes received regular screening in the 10-year period after their diagnosis. Diabetic retinopathy screening should be promoted, because many US patients with diabetes do not receive the care that is necessary to prevent visual impairment and blindness.13
On a demographically adjusted basis in patients with newly diagnosed type 2 diabetes, the average increase in PMPM cost over the 10-year period since the diagnosis year was not statistically significant in the 3 cohorts that received screening and was lower in the cohort that had no screening.
We acknowledge several limitations in our analysis based on data from a commercially paid claims database. First, any retinal screening that was not identifiable through a medical claim (eg, that was performed without appropriate coding, without any claim submitted, or with a claim submitted to vision insurance) was not captured.
Second, comorbid conditions, which are common in patients with diabetes, were not specifically identified as explanatory variables, and they might have accounted for some of the cost differences among the cohorts.
Third, the analysis excluded diabetic patients who died in the 10-year period of the study and patients who were aged >55 years in the year of diagnosis.
Fourth, because the database we used contained claims only from a commercial population, the findings may not apply to populations that are insured by Medicare or Medicaid.
Fifth, changes in the treatment of retinopathy may have biased the analysis toward higher costs during the later years in the longitudinal analysis.
Finally, using the number of drug classes prescribed in the year of diagnosis as a proxy may not have accurately captured the severity of the patients’ diabetes.
Our findings show that the rate of annual diabetic retinopathy screening has increased only slightly in the past 12 years, approaching 30% in 2013, which is well below the recommendations for screening at a minimum of every 2 years. We observed that <30% of patients with type 2 diabetes receive regular retinopathy screening. The increase in healthcare expenditures over a 10-year period after a diagnosis of type 2 diabetes was not found to be significantly different between patients with different retinopathy screening rates, and the increase in healthcare expenditure over a 10-year period since diagnosis was found to be lower in unscreened patients than in screened patients. Data from previous studies suggest that the screening and treatment of diabetic retinopathy improves health outcomes. Diabetic retinopathy screening should continue to be promoted as a high-value service, because many US patients with diabetes do not yet receive the care that is necessary to prevent visual impairment and blindness.
Mikele Bunce, PhD, provided critical review and input during the revision of this manuscript.
Funding for this study was provided by Genentech, Inc.
Author Disclosure Statement
Dr Weisman, Dr Turpcu, and Ms Rajput are employees of Genentech. Mr Dave is an employee of Point B Management Consulting, and is a past employee of Genentech. Ms Fitch, Mr Engel, and Dr Blumen reported no conflicts of interest.
Ms Fitch is Principal and Healthcare Consultant, Milliman, Inc, New York, NY; Dr Weisman is Medical Director, Genentech, Inc, South San Francisco, CA; Mr Engel is Associate Actuary, Milliman, Inc, New York, NY; Dr Turpcu is Principal Health Economist, Genentech, Inc, South San Francisco, CA; Dr Blumen is Principal and Healthcare Consultant, Milliman, Inc, New York, NY; Ms Rajput is Health Economist, Genentech, Inc, South San Francisco, CA; Mr Dave is Senior Associate, Point B Management Consulting, San Francisco, CA.
1. Centers for Disease Control and Prevention. National diabetes statistics report: estimates of diabetes and its burden in the United States, 2014. 2014. www.cdc.gov/diabetes/pubs/statsreport14/national-diabetes-report-web.pdf. Accessed February 19, 2015.
2. Rosberger DF. Diabetic retinopathy: current concepts and emerging therapy. Endocrinol Metab Clin North Am. 2013;42:721-745.
3. Ding J, Wong TY. Current epidemiology of diabetic retinopathty and diabetic macular edema. Curr Diab Rep. 2012;12:346-354.
4. Zhang X, Saaddine JB, Chou C-F, et al. Prevalence of diabetic retinopathy in the United States, 2005-2008. JAMA. 2010;304:649-656.
5. Handelsman Y, Bloomgarden ZT, Grunberger G, et al. American Association of Clinical Endocrinologists and American College of Endocrinology—Clinical Practice Guidelines for Developing a Diabetes Mellitus Comprehensive Care Plan—2015. Endocr Pract. 2015;21(suppl 1):1-87.
6. American Diabetes Association. Microvascular complications and foot care. Diabetes Care. 2015;38(suppl 1):S58-S66.
7. American Academy of Ophthalmology Retina/Vitreous Panel. Preferred Practice Pattern guidelines: diabetic retinopathy. 2014. www.aao.org/preferred-practice-pattern/diabetic-retinopathy-ppp--2014. Accessed August 4, 2015.
8. National Committee for Quality Assurance. The State of Health Care Quality 2014. October 2014. www.ncqa.org/Portals/0/Newsroom/SOHC/2013/SOHC-web_version_report.pdf. Accessed August 13, 2015.
9. National Committee for Quality Assurance. What is HEDIS? www.ncqa.org/HEDISQualityMeasurement/WhatisHEDIS.aspx. Accessed August 13, 2015.
10. Sloan FA, Grossman DS, Lee PP. Effects of receipt of guideline-recommended care on onset of diabetic retinopathy and its progression. Ophthalmology. 2009;116:1515-1521.
11. Day TE, Ravi N, Xian H, Brugh A. Sensitivity of diabetic retinopathy associated vision loss to screening interval in an agent-based/discrete event simulation model. Comput Biol Med. 2014;47:7-12.
12. Centers for Disease Control and Prevention. Diabetes Public Health Resource: Crude and age-adjusted rate per 100 of civilian, noninstitutionalized population with diagnosed diabetes, United States, 1980-2011. Updated September 5, 2014. www.cdc.gov/diabetes/statistics/prev/national/figage.htm. Accesssed February 19, 2015.
13. Bressler NM, Varma R, Doan QV, et al. Underuse of the health care system by persons with diabetes mellitus and diabetic macular edema in the United States. JAMA Ophthalmol. 2014;132:168-173.