By Stacey Yue Cao

As of February 2018, about 30 million people (or 9.4% of the U.S. population) suffered from diabetes; another 84 million people aged 18 or above had prediabetic symptoms.1 Diabetes is a chronic condition; proper and timely diabetes care management is critical to ensure that people with diabetes can avoid long-term sequelae and maintain their quality of life.

To better understand gaps in managing the Medicare Fee-For-Service (FFS) population with diabetes, CareJourney partnered with the Microsoft’s Healthcare NEXT team to study national variation in health care utilization of various diabetes related treatments and interventions among the fee-for-service Medicare beneficiaries who had diabetes.

We identified 2017 Medicare FFS full enrollees aged 21 or older who had diabetes (according to ICD-10 codes) and lived in one of the 50 US states or Washington, D.C. as our base population. We used 100% sample of 2017 national Medicare fee-for-service (FFS) claims data, accessed through the CMS Virtual Research Data Center (VRDC). Out of the 61 million total of beneficiaries available in the 2017 Medicare enrollment file, about 30 million (~50% of total beneficiaries) are full Medicare FFS enrollees aged 21 or older and residing in the U.S. From this group, 6.9 million (~11%) beneficiaries also have presence of diabetes diagnoses and makes up the base population of this study. Diabetic beneficiaries with one or more days of hospice service in 2016 or 2017 were not considered in the study, as these individuals have distinct healthcare needs and end-of-life utilization patterns.

Healthcare Expenditure Among Diabetic Medicare Beneficiaries

We found that African Americans and Hispanic Medicare beneficiaries with diabetes had per-capita expenditures that were statistically significantly higher (about 30% higher) than those of white beneficiaries with diabetes. Medicare beneficiaries with diabetes who were younger than age 65 (and, therefore, enrolled in Medicare by virtue of disability) or Medicare beneficiaries with dual eligibility had per-capita overall expenditure that were statistically significantly higher (about 32-35% higher) than those without such characteristics. We reached these findings by performing two sample t-tests that compared the mean and standard deviation of per capita total part A&B spend between various demographic subpopulations with diabetes (Table 1).

Per Capita Total A&B Spend Comparison by Demographics

Table 1

N ( % of total cell 4B) ACO attributed (limited to ACOs with > 5000 benes) Not ACO (zero months) White Black Hispanic Other Female Male < Age 65 > Age 85 (including 85) Medicaid Eligible
Total Spend A&B 2017 Mean $13,642 $13,283 $13,797 $13,148 $16,963 $16,135 $12,022 $13,528 $13,769 $17,990 $14,983 $18,468
Total Spend A&B 2017 StDev $26,838 $25,554 $27,380 $25,329 $33,678 $33,123 $26,5969 $25,729 $28,015 $35,869 $23,835 $33,574

A1C Test

Hemoglobin A1C is an essential blood test to diagnose and monitor diabetes and prediabetes. Guidelines recommend at least two A1C tests per year among diabetics.2 We sought to understand the variation of A1C test across accountable care organizations (ACOs) by looking at the average A1C test frequency in a single year, adjusted on age, gender, race and hcc score.

Among the 453 ACOs with more than 5000 attributed beneficiaries in 2017, the average prevalence of diabetes was around 23%. Overall, 82 (18%) of these ACOs have an average adjusted A1C test per diabetic beneficiary of less than 2 per year, suggesting possible underutilization of proper A1C tests to manage their diabetes population. Among the remaining ACOs, average adjusted A1C test range from 2 per year to 4.6 per year. We cannot know from ACO-level aggregated statistics what prompted such distinct care patterns and if or how they are associated with health outcomes. Zooming into the provider and beneficiary level A1C data from our study will help better assess the appropriateness of A1C tests within and across ACOs and identify predictors.

Lower Extremity Amputation

Among diabetes, lower extremity amputation (LEA) is an indicator of poor long-term diabetic care management. LEA in the diabetic population can cost between $30,000 and $60,000 annually and accumulate up to half a million USD in one’s lifetime.3 Across the same 453 ACOs identified previously, we found that, on average, about 0.49% of diabetic beneficiaries had LEA. The range of average LEA rate goes from 0% to as high as 2.5% across the ACOs.

Further, we examined variation in rates of other recommended diabetes care, including annual eye exams, annual nephropathy testing, and lipid testing. In addition, we evaluated measures associated with diabetes care management, including avoidable emergency room visits, length of stay (LOS) in acute care hospitals, annual wellness visits and evaluation & management (E&M) visits with various specialties, e.g., primary care physicians, endocrinologists, ophthalmologists, nurse practitioners/physician assistants and all other specialties. Combined with different demographic, condition-based and geographic cuts, these data points will allow our users to draw actionable insights on their diabetic populations and regions, informing interventions aimed at achieving better diabetic care management and patient outcome in the long-run.

If you are interested in learning more about our findings related to the Medicare FFS diabetic population, please feel free to reach out to your Member Services representative or contact us at [email protected].

Not a CareJourney member?

Request a demo
  1. https://www.cdc.gov/diabetes/data/statistics/statistics-report.html
  2. https://www.cdc.gov/diabetes/managing/managing-blood-sugar/a1c.html
  3. Margolis D, Malay DS, Hoffstad OJ, et al. Incidence of Diabetic Foot Ulcer and Lower Extremity Amputation Among Medicare Beneficiaries, 2006 to 2008. Rockville, MD, Agency for Healthcare Research and Quality, 2010

Appendix Tables

Two Sample T-test Comparing Mean Total Cost of Care Between White and African American Diabetic Beneficiaries

The TTEST Procedure

Variable: total_spend

Table 2

Race Method N Mean Std Dev Std Err Minimum Maximum
White 5435152 $13,148 $25,329 $11 $(168,862) $4,068,970
Black 858915 $16.963 $33,678 $36 $(43,984) $3,010,837
Diff (White-Black) Pooled $(3.815) $26,623 $31
Diff (1-2) Satterthwaite $(3.815) $38

Table 3

Race Method Mean 95% CL Mean Std Dev 95% Std Dev
White 13147.9 $13,127  $13,169 $25,329 $25,314 $25,344
Black 16963 $16,892 $17,034 $33,678 $33,628 $33,729
Diff (White-Black) Pooled -3815.1  $(3,876) $(3,755) $26,623 $26,609 $26,638
Diff (1-2) Satterthwaite -3815.1  $3,889 $(3,741)

Table 4

Method Variances DF T Value Pr > |T|
Pooled Equal 6.29E+06 (123.41) <.0001
Satterthwaite Unequal 1.02E+06 (100.59) <.0001

Table 5

Equality of Variances
Method Num DF Den DF F Value Pr > F
Folded F 858914 5.44E+06 1.77 <.0001