Impact of the COVID-19 Pandemic on Medical Expenditures Among Medicare Fee-for-Service Beneficiaries Aged ≥67 Years With Diabetes Free

The coronavirus disease 2019 (COVID-19) pandemic dramatically changed how medical care was organized and services delivered during the pandemic period in the U.S. To control COVID-19 at the onset of the pandemic, local and state governments imposed restrictions on travel and nonessential services. Medical practices were forced to temporarily close their operations (1). Many patients avoided interactions with the health care system to avoid contracting the virus (2). To meet the immediate needs of those who were severely affected by COVID-19, hospitals and ambulatory care providers redirected resources to patients with COVID-19 and postponed or canceled nonurgent appointments, such as elective care or preventive visits. Ambulatory care providers also converted in-person visits to telemedicine visits when feasible (3).

Care and health services delivery for people with diabetes were even more affected than for those without diabetes because of higher hospitalization and mortality rates from COVID-19 infection (4,5). There was also an interruption in routine medical care for diabetes. Both inpatient and outpatient visits decreased, and the use of telemedicine services increased (6–8).

Older adults with diabetes were particularly vulnerable to severe COVID-19 disease. A study of 173,942 hospitalized patients with COVID-19 in 2020 found that 48.5% were aged ≥65 years and covered by Medicare, and 40.7% had diabetes (9). Compared with their younger counterparts, older people with diabetes had less access to and a lower satisfaction level with outpatient services provided by telehealth (5,10,11). These changes in the delivery of care and services in response to COVID-19 may have led to changes in health care utilization and ultimately medical expenditures for older adults, including those with diabetes.

Few population-level studies have examined how health care utilization and expenditures of people with diabetes changed related to COVID-19. Existing studies are limited to those that examined changes in unitizations of outpatient and inpatient visits or telehealth only during the initial period of the pandemic (7,8,12). These studies used data from convenience samples with a relatively small sample size (6,12). No study has examined how the pandemic affected medical expenditures or changes in all health service types simultaneously. In addition, how the pandemic affected the out-of-pocket (OOP) expenditure of patients with diabetes has not been examined.

In 2019, 27.5% of fee-for-service (FFS) beneficiaries in Medicare had diabetes (13). We used the Medicare claims database to examine how COVID-19 affected health service utilization and expenditures of beneficiaries with diabetes. Specifically, our study objectives were to 1) estimate the change in medical expenditures at both the population and individual level before and during the pandemic, 2) describe the temporal trend in medical expenditure before and during the pandemic by service type, and 3) estimate the change in OOP expenditures for Medicare beneficiaries before and during the pandemic.

We used 100% Medicare FFS administrative claims data from the Centers for Medicare & Medicaid Services Chronic Conditions Data Warehouse (14). Medicare is the federal health insurance program for people aged ≥65 years and certain younger people with disabilities and conditions. Various parts of Medicare cover different types of health services: Part A covers inpatient hospital care, skilled nursing facility (SNF) care, hospice care, and some home health care; Part B covers certain physician services, outpatient care, medical supplies, and preventive services; and Part D covers prescription drug expenditures. Our study included claims for Medicare Parts A, B, and D. We included only Medicare beneficiaries with FFS plans because expenditure data for those with Medicare Advantage plans are not available in the Medicare claims data set. We used claims spanning the period from 1 March 2019 through 28 February 2022, the most recent at the time of the study (15).

We included all beneficiaries with diabetes who had been enrolled in Medicare FFS Parts A and B continuously for 24 months. Beneficiaries were identified to have diabetes if they had at least one inpatient claim or two outpatient claims within 2 years with diabetes ICD-10-CM code E10, E11, or E13 (16,17). With the requirement of a 2-year period for identifying prevalent diabetes, the starting age of beneficiaries included in the analysis was 67 years (16). Beneficiaries who turned 67 at any time of a calendar year (i.e., 2019, 2020, 2021, 2022) were considered to be 67 years of age throughout that year in the analysis. Individuals with drug-induced diabetes, identified by ICD-10-CM code E09, were excluded from the study.

Medical expenditure in our study was the total amount paid for patients’ care from all payment sources, including expenditures paid by the patient (deductibles, coinsurance, and copayments), Medicare reimbursements, and third-party payer payments. We calculated quarterly direct medical expenditures at both the population and individual level in total and by health service types (i.e., inpatient care, hospital outpatient care, Medicare Part B, SNF, home health, hospice, durable medical equipment [DME], prescription drugs). Population-level spending represented the aggregated medical expenditure among people with diabetes recorded by Medicare. Per capita total and OOP expenditures were the population-level expenditures divided by the number of people with diabetes in the same quarter. Changes in expenditures were calculated as the differences in the corresponding quarters between the prepandemic period (March 2019 to February 2020) and the 1st year (March 2020 to February 2021) or the 2nd year (March 2021 to February 2022) of the pandemic. The 1st quarter during the pandemic, for example, was defined as March 2020 to May 2020, while the 1st quarter during the prepandemic period was defined as March 2019 to May 2019. We standardized all expenditures to 2022 U.S. dollars using the medical consumer price index. Data analyses were performed using SAS Enterprise Guide 7.1 statistical software.

We conducted several subgroup analyses, as COVID-19 may have affected different subgroups of Medicare beneficiaries with diabetes differently. We calculated changes in per capita total expenditure by age-group (67–74, 75–84, ≥85 years), sex (female, male), race/ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, Asian/Pacific Islander, American Indian/Alaska Native), and Medicaid eligibility (Medicare/Medicaid dually eligible, Medicare eligible only).

Approximately 27% of Medicare FFS plan beneficiaries do not enroll in Part D plans; thus, their prescription costs were not reflected in our data. We conducted a sensitivity analysis to examine the influence of prescription plan coverage on medical expenditure by excluding the beneficiaries who did not enroll in Part D.

Demographics of Medicare beneficiaries with diabetes before and during the pandemic are presented in Table 1. There were 7.17 million beneficiaries with diabetes enrolled in the FFS plan in the year before the pandemic and 7.16 million in the 1st year of the pandemic and 6.69 million in the 2nd year of the pandemic. Compared with the prepandemic year, the number of Medicare beneficiaries with diabetes fell 6.8% by the 2nd pandemic year. The demographic distribution of the diabetes population also changed slightly during the pandemic. For example, the oldest age-group (≥85 years) accounted for 16.4% of the total before the pandemic and decreased to 16% in the 2nd pandemic year. The proportions of female, Black, and Hispanic beneficiaries and the Medicaid-eligible population also decreased slightly.

Population-level total medical expenditures decreased by 19.6% from $41.7 billion in the 1st quarter of the prepandemic year to $33.6 billion in the same quarter of the 1st pandemic year. After the initial drop, total expenditure increased slowly and returned to a level that was close to the prepandemic period ($38.6 billion) by the 4th quarter of the 1st pandemic year. In the 2nd pandemic year, the population-level total medical expenditure in all quarters remained lower than the corresponding quarters of the prepandemic period, falling to $36.8 billion in the 4th quarter (Fig. 1 and Table 2).

Expenditures decreased in all service categories in the initial quarter of the pandemic, particularly expenditures for Part B (down 32.1% to $5.4 billion), hospital outpatient (down 29.8% to $4.9 billion), hospital inpatient (down 23.0% to $10.0 billion), and home health services (down 19.4% to $1.5 billion) (Table 2). Starting from the 2nd quarter of the pandemic, expenditures in all health care service categories began to recover at different rates.

Changes in prescription expenditures differed from other services, with few changes in the 1st pandemic year compared with the same prepandemic quarter (−1.1% to 0.2%) and then rising to a level higher than the prepandemic level in the 2nd pandemic year (2.4–5.8%). Inpatient expenditures showed the greatest variation over the study period, ranging from 23% below the prepandemic level in the 1st quarter of the pandemic period to 3.2% above the prepandemic level in the 4th quarter of the 1st pandemic year. In the 2nd pandemic year, however, inpatient expenditures were more stable (−11.2% to −7.3%). Hospital outpatient expenditures dropped 29.8% in the 1st quarter of the pandemic period, then rebounded in the 2nd and 3rd quarters, but remained 11.6% lower than the prepandemic levels in the 4th quarter of the 1st pandemic year. Outpatient expenditures remained below prepandemic levels in the 2nd pandemic year, which was still 11.0% below prepandemic level in the 4th quarter.

Part B expenditures have remained relatively stable since the 2nd quarter of the pandemic (−7.4% to −1.0%). SNF expenditures were stable for the 1st three quarters of the pandemic period (−2.9% to 0.1%) but then followed a similar trend to inpatient expenditures since the 4th quarter of the 1st pandemic year. Similarly, home health expenditures fell sharply in the two 4th quarters of the pandemic years after having rebounded a bit in the preceding quarters. Hospice is the only service whose expenditures showed a nearly monotonically decreasing trend since the pandemic began (Table 2). Expenditure on DME during the pandemic period was higher than that of the corresponding prepandemic quarter for all quarters except the 1st quarter. OOP expenditures at the population level fell by 19.8% in the early pandemic period and were 7.4–10.2% lower than the prepandemic level thereafter (Fig. 1).

Despite its initial drop (−17.6%) in the 1st quarter of the pandemic, per capita total medical expenditure rebounded starting in the 2nd quarter of the pandemic (−3.1%) and reached a level that was higher than the prepandemic level since the 3rd quarter of the pandemic. The per capita total medical expenditure reached to $6,518 in the 1st quarter of the 2nd pandemic year and then fell slightly since then (Fig. 1 and Table 2).

Per capita medical expenditure by services showed similar trends to that of population-level expenditures, but with a much wider fluctuation. Hospice expenditures were the most stable (−0.9% to 2.8%). Prescription expenditures were consistently higher during the pandemic, rising 8.5–13.0% higher than prepandemic levels ($1,221–$1,252) in the 2nd pandemic year.

Per capita expenditure for inpatient care fluctuated. It dropped to 21.1% below the prepandemic level in the 1st quarter of the pandemic, then increased to 6.9% above the prepandemic level in the 4th quarter of the 1st pandemic year ($1,592–$1,999), and then fell to 4.9% below prepandemic levels in the 1st quarter of the 2nd pandemic year. Hospital outpatient per capita expenditures were lower than prepandemic levels in the 1st pandemic year (−28.1% to −3.5% [$775–$999]), but returned to near or above prepandemic levels in the 2nd year (−3.0% to 3.7% [$961–$1,096]). Part B expenditures returned to near-prepandemic levels in the 3rd quarter of the 1st pandemic year (−0.4% [$1,210]) and remained higher in the 2nd year (1.0–5.8% [$1,056–$1,285]). Home health expenditures fluctuated widely, ranging from 17.4% below prepandemic levels in the 1st quarter to 1.8% above the prepandemic level in the 3rd quarter of the 1st pandemic year. DME expenditures dropped in the 1st quarter of the 1st pandemic year and then became higher in all quarters of the 2nd pandemic year (8.3–15.4% [$104–$118]) (Fig. 2).

The per capita OOP expenditure was 17.8% lower than prepandemic levels at the start of the pandemic but gradually recovered to 0.6% below prepandemic levels by the 2nd quarter of the 2nd pandemic year. It then slightly decreased to 1.8% below prepandemic levels from the 2nd to the 4th quarter of the 2nd pandemic year. Throughout the study period, per capita OOP expenditures remained below prepandemic levels.

Patterns of changes in per capita expenditure by subgroup (Supplementary Figs. 1–4) were similar to that of the overall population. The magnitude of changes by age-group seemed to increase by age. By sex, women experienced less change in per capita expenditure than men.

By race/ethnicity, changes after the initial quarter of the pandemic for non-Hispanic Black beneficiaries seemed to fluctuate less compared with other racial/ethnic groups. The American Indian/Alaska Native group appeared to have seen the largest percentage increases over time, while Asian/Pacific Islander beneficiaries had a large increase only in the 2nd pandemic year.

The Medicaid-eligible population with diabetes had a slight decrease in the 1st quarter of the pandemic and remained higher than the prepandemic level in per capita expenditure throughout the pandemic period. The per capita expenditure was 8.3% higher than the prepandemic level in the 3rd quarter of the pandemic and remained 7.4% higher at the end of the study period.

Results of the sensitivity analysis are presented in Supplementary Figs. 1 and 2 and Supplementary Table 1. After excluding Medicare FFS beneficiaries who were not enrolled in Part D plans, the percentage change compared with the prepandemic level among Part D enrollees followed a trend similar to all enrollees, except that the change among Part D enrollees was 0–2% above the change among all Medicare FFS beneficiaries. The quarterly per capita expenditure increase was higher when including only Part D enrollees for all health service categories, with the largest difference in prescription expenditure. The quarterly per capita total expenditure increased by $600–$652 compared with the original results, while the quarterly per capita prescription expenditure increased by $392–$442 compared with the original prescription expenditure.

The COVID-19 pandemic affected how medical care was delivered in the U.S. How these changes affect health care utilization and expenditures remains largely unknown. To our knowledge, this study is the first to comprehensively describe the medical expenditure trend before and during the pandemic in Medicare beneficiaries with diabetes.

The sharp decrease in medical expenditures in the early phase of the pandemic was due to falls in health service utilization. In the 1st quarter of the COVID-19 pandemic, health care utilization fell in all types of services except for home health compared with the same quarter in the prepandemic period. Total numbers of inpatient and outpatient visits fell by 28% and >30%, respectively (Supplementary Table 2). These results are also in line with findings from previous studies. Among U.S. adults aged ≥60 years with diabetes, 17% reported that their health care was disrupted in the early period of the pandemic (2). Both inpatient and outpatient visits fell (6). These decreases in health care utilization translated to a lower medical expenditure.

The decrease in the number of Medicare beneficiaries with diabetes also contributed to the fall in total expenditure at the population level during the pandemic period. The total number of Medicare beneficiaries had been steadily increasing in each quarter of the prepandemic year (Supplementary Table 3). In comparison, the number of Medicare FFS beneficiaries with diabetes dropped from 6.58 to 6.04 million from the last quarter of the prepandemic period to the last quarter of the pandemic period (Supplementary Table 3). More beneficiaries with diabetes died during the pandemic compared with prepandemic period (Supplementary Table 3). Medicare estimated that a 2.9% savings resulted from the deferral of care and excessive deaths during the pandemic (18).

The reduction in medical spending at the population level for the Medicare program during the pandemic does not necessarily imply a corresponding decrease in overall medical expenditures, as our study did not account for federal expenditure on COVID-19. The Department of Health and Human Services has provided >$150 billion through the Provider Relief Fund (PRF) as of early 2022 to fund hospitals and other health care providers during the pandemic (19). PRF payments were made to eligible providers who diagnose, test, or care for individuals with COVID-19 and have health care–related expenses and lost revenues attributable to COVID-19. More than $50 billion of PRF was allocated to Medicare providers and may not have been comprehensively reflected in FFS claims (19). In addition to the funds that directly went to Medicare providers, the PRF has allocated >$10 billion to research, production, and distribution of COVID-19 vaccines, which also is not reflected in insurance claims. The medical expenditure for the Medicare program may fall because of the decrease in medical expenditures at the population level, but the increases in spending related to COVID-19 could have exceeded this decrease.

Our findings show that per capita total medical expenditures were above the prepandemic level since the 3rd quarter of the pandemic. Upon further examination of the number of utilization and expenditure per utilization by type of health services, we observed that there was little change in number of utilization for hospice and DME (Supplementary Table 2). The number of outpatient visits had been fluctuating. The home health services utilization exhibited an increase, while the number of inpatient visits and number of prescription medications showed a small decrease (Supplementary Table 2). However, spending per unit of service for inpatient services, SNFs, and prescriptions increased substantially (Supplementary Table 2). Prescription drug spending made up nearly 20% of all medical expenditures for Medicare beneficiaries with diabetes, both before and since the pandemic. The increase in per capita prescription expenditures contributed to the per capita total expenditure increase among Medicare beneficiaries with diabetes. As there was only a small decrease in prescription fills during the pandemic period, the increase in spending per prescription seems responsible for the higher per capita expenditure on prescription drugs. Reasons for a higher spending per prescription could include increased use of newer and more expensive disease treatment options that emerged within the market in recent years, such as sodium–glucose cotransporter 2 inhibitors and glucagon-like peptide 1 agonists for cardiovascular and kidney disease risk reduction (20), and the use of newer, more expensive insulin in the past 2 years (21,22).

Apparent fluctuations were observed in inpatient expenditure at both the population and per capita level in the 4th quarter of the 1st pandemic year and the 1st quarter of the 2nd pandemic year. Results from examining the number of utilizations and expenditures per unitization showed a surge in both the number of hospitalization cases and spending per hospitalization case (Supplementary Table 2). Our estimation was consistent with the surveillance data from the Centers for Disease Control and Prevention, where the count of hospitalized cases of COVID-19 hit a peak during the last quarter of the 1st pandemic year and dropped sharply in the next quarter (23). The phenomenon that diabetes worsens COVID-19 outcomes (and COVID-19 is particularly harmful to those with diabetes) could possibly explain the surge in spending per hospitalization case for the last quarter of the 1st pandemic year (24).

The sensitivity results showed the magnitude of the impact of Part D enrollment on the results of the analysis. Except for the prescription expenditure, Part D enrollees had a higher expenditure in other health service categories. The potential explanations for this phenomenon may stem from two factors. First, it is conceivable that individuals who did not enroll in Part D plans were healthier compared with Part D enrollees. As a result, the nonenrollees’ medical expenditures were lower. Alternatively, it is plausible that these individuals had other private insurance plans that covered the specific medical expenditure, which is beyond the scope of our data access. The subgroup results highlight sex and racial/ethnic disparities in health care expenditures during the pandemic. Per capita expenditures for males remained higher than prepandemic levels, while per capita expenditures for females remained stable or lower. This disparity may reflect differences in health care seeking behavior or health care services accessibility. Additionally, different racial/ethnic groups experienced varying trends in health care expenditures. American Indian/Alaska Native populations had the largest percentage increases over time, and the non-Hispanic Black population with diabetes experienced steadier expenditures. This information can be used to develop targeted policies to support people from some racial and ethnic groups and address potential health care expenditure barriers to access and treatment. Medicaid-eligible populations had higher per capita expenditures throughout the pandemic period. Further research is needed to monitor and understand the nature of this phenomenon.

Our study had several limitations. We assessed the medical expenditure trend 2 years into the pandemic, which may not capture incremental disease progression or longer-term effects. The medical expenditure observed in the study does not represent all the medical expenditures paid for the Medicare population, as the expenditures associated with tests and vaccines are not included in claims data. Our study included only Medicare FFS beneficiaries as data on health care utilization and expenditures were not available for beneficiaries enrolled in Medicare Advantage plans. Thus, our calculated total expenditures did not reflect the total Medicare spending for all beneficiaries with diabetes. In addition, our estimated per capita expenditure may not be generalizable to beneficiaries enrolled in Medicare Advantage plans. FFS and Medicare Advantage beneficiaries may differ in health profiles and utilization of health care services. We did not specifically examine the expenditure associated with telehealth use. The telehealth expenditure component was part of the hospital outpatient and Part B carrier expenditure in our analysis. Since the beginning of COVID-19 pandemic, Medicare has expanded the coverage of telehealth and offers health care providers equal reimbursement as for in-person visits (25). Future studies could examine how this policy offsets the decrease of in-person visits and affects the quality of diabetes care.

In conclusion, our study is one of the first to describe how the COVID-19 pandemic affected health services utilization and medical spending in older Medicare FFS beneficiaries with diabetes. We found that the pandemic had different impacts on medical expenditures at the population and individual level and by health service type and subgroup of the covered population.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. Y.W. designed the study, analyzed the data, and wrote the manuscript. P.Z. designed the study, analyzed the data, and reviewed and edited the manuscript. X.Z., D.R., and G.I. reviewed and edited the manuscript. Y.W. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.