Medicare Fee-for-Service Spending for Diabetes: Examining Aging and Comorbidities

Diabetes Mellitus is a chronic condition affecting an estimated 25.8 million people, or 8.3% of the population of the United States in 2010, including 10.9 million adults age 65 and above [1]. Type 2 diabetes is responsible for 90-95% of diabetes cases and the majority of health spending, with prevalence increasing with age; Type 1 diabetes, formerly known as juvenile diabetes and gestational diabetes account for most of the remaining cases [2-5]. Diabetes is a leading cause of death and is often associated with comorbidities including kidney failure, lower limb amputations, adult onset blindness, obesity, hypertension, nerve damage, heart disease and stroke. Health care reform legislation in the U.S. under the Affordable Care Act (ACA) of 2010 identifies diabetes as a major population health concern, with provisions that address diabetes prevention, screening, access to wellness and prevention programs, healthcare and prescription drug coverage, enhanced surveillance and quality standards, and demonstration programs for non-institutionalized individuals with diabetes [6,7].


Introduction
Diabetes Mellitus is a chronic condition affecting an estimated 25.8 million people, or 8.3% of the population of the United States in 2010, including 10.9 million adults age 65 and above [1]. Type 2 diabetes is responsible for 90-95% of diabetes cases and the majority of health spending, with prevalence increasing with age; Type 1 diabetes, formerly known as juvenile diabetes and gestational diabetes account for most of the remaining cases [2][3][4][5]. Diabetes is a leading cause of death and is often associated with comorbidities including kidney failure, lower limb amputations, adult onset blindness, obesity, hypertension, nerve damage, heart disease and stroke. Health care reform legislation in the U.S. under the Affordable Care Act (ACA) of 2010 identifies diabetes as a major population health concern, with provisions that address diabetes prevention, screening, access to wellness and prevention programs, healthcare and prescription drug coverage, enhanced surveillance and quality standards, and demonstration programs for non-institutionalized individuals with diabetes [6,7].
The total economic cost of diagnosed diabetes in the U.S. is estimated at $245 billion in 2012, including $176 billion in medical costs and $69 billion attributed to reduced productivity [3]. This represents a 41% increase over the $174 billion in total economic costs estimated for 2007 [2]. Diabetes is associated with higher use of health care services, medications, and other supplies, with medical costs per case increasing with age [8].
Individuals 65 years of age and above account for approximately 59% of national spending on diabetes care in the U.S., most of which is paid by Medicare [9], the publicly-funded health insurance program administered by the Centers for Medicare & Medicaid Services (CMS) that covers the majority of adults age 65 and above in the U.S. [3]. Medicare Fee-for-Service (FFS) coverage has multiple parts. Medicare Part A (also known as hospital insurance) covers care in hospitals, skilled nursing facilities and other institutions for the more costly and complex treatment needs. Medicare Part B (also known as medical insurance) covers doctors' services, outpatient services, diagnosis, examinations, care, equipment and supplies relating to diabetes, as well as some preventive services for individuals at risk for diabetes and self-management training for those newly diagnosed or at risk for complications relating to their diabetes. Medicare Part D (also known as prescription drug insurance) provides prescription drug coverage and covers supplies for injecting or inhaling insulin, and is provided through private insurance companies that contract with CMS.
This report examines Medicare spending on diabetes care for fee-for-service (FFS) beneficiaries, with a focus on individuals age 65 and above (older adult population) to better understand prevalence and spending patterns. Previous studies of spending and resource use on diabetes care in the U.S. have relied on large self-reported survey data such as the Behavioral Risk Factor Surveillance System, National Medical Expenditure Survey or the Medical Expenditure Panel Survey, or from Medicare Part D health plans. These data sources enable estimates of utilization and spending, but do not represent actual hospital, medical and pharmaceutical spending that can be identified through administrative claims. For the Medicare program, funded by public dollars, this information is critical for program policy and management, especially as diabetes accounts for a growing proportion of program expenditures.

Methods
We address prevalence and spending associated with diabetes (all types combined) among Medicare beneficiaries age 65 and above in calendar year 2010 using the 2010 Chronic Conditions PUF newly available through the CMS. These PUFs contain claims information from 100% of Medicare FFS beneficiaries (complete universe rather than a sample) for Medicare Part A (hospital insurance), Part B (physician and outpatient services) spending, and Part D (prescription drug) spending ii . Information (e.g., utilization, Medicare spending) is summarized in profiles (or groups) of beneficiaries defined by gender, age categories, dual-eligibility status, eleven chronic condition indicators (whether or not the condition exists), and length of enrollment (i.e., full year or less than full year). So, the observations in the data are not beneficiaries or claims, but groups of enrollees that have the same combinations of demographics as well as health status. For each profile, the data provide information on utilization and Medicare payments, such as average Medicare reimbursement per enrollee for inpatient services together Beneficiaries may be enrolled in one or more of these segments, and thus may be represented in one or more segment analyses in the rest of the study. Also, cost estimates for this study only include Medicare spending and exclude all other costs such as out-of-pocket expenses (e.g., co-insurance, deductible payments) and payments by third parties.

Results
The 2010 Chronic Conditions PUF provides summary information for more than 50 million Medicare FFS beneficiaries in 2010. Average Medicare reimbursements for full-year (i.e., 12 months of continuous enrollment) beneficiaries that are not dual-eligible were $3,146 and $4,056 per beneficiary in Part A and Part B, respectively. Average drug spending per beneficiary was an additional $2,048 per beneficiary.  .These segments are "disjoint" by length of enrollment in the sense that any particular enrollee can only be in the full year (i.e., 12 months) or in the less than year (i.e., 1-11 months) segment. However, even though about 90% of Part A enrollees are also enrolled in Part B, the utilization information for Part A and Part B are provided separately.
vi. Even though patients with end stage renal disease (ESRD) and disabilities, who are covered by Medicare, are also considered to be unique sub-populations, they are included in our study population. This is solely due to the fact that they are not separately available in the data.
vii. Sorting by Part B or Part D beneficiaries result in the same order.
chronic conditions. The only striking difference between Part A-B and Part D was that Alzheimer's ranked markedly higher among Part D enrollees than it did among Part A-B enrollees. Beneficiaries with only diabetes were about 5.4 (6.0)% of the Part A (Part B) beneficiaries, but another 2.4 (2.6)% had both diabetes and ischemic heart disease, and 1.1 (1.2)% had both diabetes and arthritis within the most common combinations provided in the table. Also, ischemic heart disease is the most common comorbidity of diabetes among Medicare beneficiaries.
More than 6 million (22.6%) Part A beneficiaries were diagnosed with diabetes making it the most commonly observed chronic condition among non-dual Medicare FFS beneficiaries ( Table 2). The share of diabetic beneficiaries was higher (25.0%) among Part B beneficiaries, but lower (15.6%) among Part D beneficiaries. These differences were also consistent for every gender and age category. Male beneficiaries 65 and older suffered more from diabetes compared to female beneficiaries in Part A, Part B, and Part D. For example, in all age categories (except for under 65), the percentage of beneficiaries with diabetes was at least 3-4 percentage points higher for males than females in Part A. The prevalence of diabetes increased with age initially for the aged Medicare population, but dropped for beneficiaries 85 years of age and older compared to age group 80-84 (e.g., from 28.7% to 24.2% for males and from 24.0% to 19.8% for females in Part A). This may reflect shorter life expectancies identified among individuals with diabetes, and warrants more in-depth investigation. Table 3 provides an overview of comorbidities by summarizing the top 15 combinations of chronic conditions for beneficiaries with diabetes in Part A, Part B, and Part D. Ischemic heart disease, arthritis, chronic kidney disease, and congestive heart failure appear as the major comorbidities of diabetes. Stroke\transient ischemic attack (not in the most common comorbidities), Alzheimer's, chronic obstructive pulmonary disease, cancer, and depression are not as significant comorbid chronic conditions for diabetes as the others. Only about 24% of Part A and Part B beneficiaries with diabetes did not have a comorbidity. Hence, 3 out of every 4 beneficiaries dealing with diabetes also deal with at least one more other chronic condition.   Table 2.
[2] Values in parenthesis denote the number of chronic conditions. For last two rows of the table, they denote the average number of chronic conditions. diabetic enrollee without other chronic conditions was $2,146. Average total drug costs were about $2,214. However, Medicare Part A spending increased drastically with comorbidities -2.8 fold with ischemic heart disease, 4 fold with arthritis, 5 fold with chronic kidney disease, and 8 fold with ischemic heart disease and congestive heart failure. The increases in Medicare Part B spending were also significant but not as drastic relative to Part A -1.6 fold with ischemic heart disease, 1.7 fold with arthritis, 2.6 fold with chronic kidney disease, and 2.5 fold with ischemic heart disease and congestive heart failure. Finally, among the comorbidities presented in the table, the increases in total drug costs due to comorbidities were not more than 1.9 fold (90%) which was for beneficiaries with three comorbidities: congestive heart failure, chronic kidney disease, and ischemic heart disease. On average, diabetic beneficiaries in Part A and Part B had 2.8 chronic conditions (including diabetes) with average Medicare spending of $5,741 and $5,991. Total drug costs for Part D beneficiaries were $3,119. and Part D cost for diabetic and non-diabetic beneficiaries by age (for number of beneficiaries provided in Table 2. Average Medicare Part A and Part B spending increases with age (except for beneficiaries under age 65). However, the increase in Part A spending is relatively steeper than the increase in Part B spending. Part D costs increase slightly with age for non-diabetic beneficiaries, but are flat for diabetic beneficiaries. Average Medicare spending and drug costs for diabetic beneficiaries are higher than for non-diabetic beneficiaries. For example, average Medicare spending for 65-69 year old beneficiaries with diabetes is about 3.5 and 2 times higher than for beneficiaries without diabetes for Part A and Part B, respectively. Some of this increase is explained by the differences in the average number of chronic conditions between diabetic and non-diabetic beneficiaries as shown on the right hand side of Figure 1. Interestingly, the differences in average number of chronic conditions are the largest among Part D beneficiaries, but average Medicare drug spending does not increase by age for diabetic beneficiaries even though they increase slightly for non-diabetic beneficiaries.

Discussion
These analyses show actual treated prevalence and hospital and medical spending (and prescription drug spending) for individuals in the Medicare FFS population using claims information available from newly available CMS public use files. Findings are generally consistent with trends identified in surveys and research that show diabetes as a common chronic condition, especially among older adults, with increases in prevalence and associated spending as the population ages, and shorter life expectancy [1][2][3][4][5]. The findings also show overall increases in the prevalence of chronic conditions with advancing age.
People with diabetes account for approximately a quarter of Medicare FFS beneficiaries in 2010 with a larger share among Part A beneficiaries and a significantly lower share among Part D beneficiaries ( Table 2). Most beneficiaries with diabetes had other chronic conditions, which averaged 2.8 conditions (with diabetes). Spending was modest for the 5-6% of Part A and Part B beneficiaries with only diabetes and no other chronic conditions, but increased substantially for the majority of beneficiaries with diabetes who also had other chronic conditions. For example, annual spending associated with hospital care (Part A) increased by about 3 fold ($1,502 vs. $537), on average, if a diabetic beneficiary also had ischemic heart disease ( Table 3) which is the most common comorbidity for beneficiaries with diabetes.
Not surprisingly, hospitalization accounts for the largest portion of overall resource use and spending for diabetes in this and other studies [2,3,10]. According to the ADA, diabetes contributes to higher rates of hospital admissions and longer average length of stay regardless of the reason for admissions (and controlling for other factors that affect hospital length of stay) [2].
As this study and others have found, health care spending was substantially greater for individuals with diabetes than for individuals without diabetes, and health resource utilization and associated spending increases with age [2,3,11]. This trend is clear for the population 65 years of age and over. However, beneficiaries under age 65 had higher, off-trend spending than the population age 65 and over, likely reflecting their eligibility and enrollment as a result of their disability status.
These analyses also confirm reports showing that diabetes often occurs along with other chronic conditions, and that comorbidities increase with advancing age [10,12]. Some of these conditions, such as ischemic heart disease, are related to diabetes, others are comorbidities independent of diabetes, such as osteoarthritis or Alzheimer's disease. The most common comorbidities identified in this study include: ischemic heart disease, arthritis, chronic kidney disease, osteoporosis and depression. Medicare spending increased across age groups as the number of chronic conditions increased. This finding is consistent with research by Struijs and colleagues who reported that the number of comorbidities is a strong predictor for the volume of medical care utilization for patients with diabetes [10].
The presence of comorbidities such as those in this study can complicate diabetes management, including patients' abilities to manage their self-care, and increase overall costs of care [13,14]. As Piette and Kerr [13] and others [15,16] point out, conditions such as depression and arthritis may impair patient functioning and present barriers to lifestyle change and regimen adherence, while conditions unrelated to diabetes such as emphysema and chronic low back pain can have a more debilitating impact on patients' health status than diabetes, and disabling conditions such as advanced heart failure and dementia may make diabetes self-care goals impossible to reach. These conditions also compete for patients' time and financial resources.
Complications of diabetes, particularly cardiovascular diseases, neurological diseases, and renal conditions, also lead to increased use of health resources and spending [2,10]. A 2012 ADA study found medical spending increases from 25 percent in emergency departments to 45 percent in hospital inpatient facilities were due to treating complications of diabetes [3]. Other studies have found that people with uncontrolled diabetes or with diabetes complications incur diabetes costs two to eight times more than people with controlled or non-advanced diabetes. [17,18] Diabetes also increases the cost of treating general conditions that are not directly related to diabetes [3,10].
Effective population health tools and interventions-including diabetes education, self-management and self-care programs-are available and are expected to increase availability under the ACA that can improve diabetes management and prevent complications and, potentially, prevent or delay the development of Type 2 diabetes and its frequently occurring comorbid sequellae in persons at high risk. But despite medical and public health advances, it is not clear which interventions work best with which populations.

Limitations
These analyses have several limitations. They profile prevalence and spending for FFS Medicare beneficiaries, but do not examine whether and to what extent these patterns resemble prevalence, spending and chronic conditions in beneficiaries enrolled in non-FFS settings such as Medicare Advantage managed care plans. Data to enable such a comparison are not available, as CMS does not collect or release comparable data from these plans.
The CMS data used for these analyses do not collect information on out-of-pocket spending, although out-of-pocket spending is considerable for individuals with diabetes. These costs can be a barrier to condition management and regiment adherence, especially for those beneficiaries who do not have Medicare Supplemental insurance. Outof-pocket spending for medication incurred by people with diabetes are higher than for individuals with most other chronic conditions [19]. Other out-of-pocket spending may include co-pays and transportation related to medical care, and purchases of healthy food to maintain diabetic dietary requirements.
The data contain information on only eleven chronic conditions. Hence, we cannot observe and control for other diseases (e.g., obesity, hypertension) that might affect the health of Medicare beneficiaries. Estimates for total and average number of chronic conditions may underestimate actual values and establish lower bounds. However, if these unknown chronic conditions have a random distribution over Medicare enrollees in the data, then our estimates for Medicare spending and drug costs may not be biased in any direction.
Study data do not include information on Medicare beneficiaries' race and ethnicity, so this study does not address prevalence and spending or profile chronic conditions across racial and ethnic groups. Blacks, Hispanics, and Native Americans have higher reported prevalence of diabetes than White populations [4]. These populations are also more likely to report having chronic conditions and lower incomes and Blacks, in particular, are less likely to report having Medicare supplemental policies to help cover associated care costs [9].

Conclusion
Medicare FFS spending for diabetes-related care for older adults is substantial, with management and treatment often complicated by comorbid conditions and complications that further fuel spending for patients and the Medicare program. This research report provides prevalence and spending trends from 2010 that represent actual hospital, medical and pharmaceutical spending that can be identified through Medicare administrative claims. This information is critical for program policy and management as Medicare implements new ACA initiatives and continues to focus on prevention and treatment to curb increases in diabetes and related program expenditures for older adults in the FFS population.