Erin Craig

Recommendations for feeling better and spending less as we age: Aging in the United States through the lens of Medicare billing

March 24, 2021

Introduction

Aging in the United States is expensive, and health care costs are consistently increasing (Martin, Hartman, Lassman, & Catlin, 2020). We investigate the cost of aging through the lens of Medicare spending: we consider Medicare because it is available primarily to individuals age 65 or over (our aging population), and it is a significant cost of federal spending. In 2019, 18% of our federal government spending — $799.4 billion — went to the Medicare program (CMS Office of Enterprise Data and Analytics, 2021).

We note that money spent to CMS is not offensive: we ought to support our aging population. However, it may be useful to introspect on our spending. Specifically, we ask: on which interventions do we spend substantial amounts of money, and what morbidities are they intended to treat? The answer to this question reveals common morbidities in our aging population – and corresponding opportunities for researchers to identify new interventions. Further, we identify areas where CMS may spend money now to support decreased future spending.

Data

The Center for Medicare & Medicaid Services (CMS) annually publishes data that describes how much they have paid to individual providers for procedures performed and medications prescribed. To the best of our knowledge, this is the most comprehensive and detailed publicly available dataset describing insurance payments for our aging population. We note, however, that there are likely many age-related costs not captured in the Medicare data (Moody, et al., 2019); those insured by Medicare often have supplementary insurance, and they likely have many out-of-pocket costs that are not jointly captured in any single dataset (e.g. assistive devices, over the counter medications and home support including lawn care/home care/snow removal).

 We examine two datasets: one describes payments from Medicare Part B (medical insurance), and the other Medicare Part D (drug coverage). These data were recorded after Medicare payments were made and all claim adjustments were completed, and they have been reviewed by CMS (CMS Office of Enterprise Data and Analytics, 2021). We consider Medicare spending from 2018, as this is the most recent year for which data has been made available by CMS (and we acknowledge that, for example, spending in 2020 is likely quite different from the preceding years, due to disruptions related to COVID-19).

To preserve privacy, data are aggregated at the provider level rather than the individual patient level: it impossible to observe any individual patient’s spending, or even an average patient’s spending.

Data Limitations

We note that these two datasets do not account for the entirety of Medicare spending: in 2018, Medicare Part B spending totaled $86 billion and Part D spending totaled $62 billion.  (These are lower bounds, as CMS has removed some data to protect patient privacy.) Further, CMS cautions (CMS Office of Enterprise Data and Analytics, 2021) that these data cannot indicate the quality of care; here, we assume that a majority of providers are performing the standard of care. We believe this is a reasonable assumption; further, this assumption is necessary in order to draw conclusions from our exploratory analysis. Finally, though a majority of Medicare beneficiaries are age 65 or over, the Part B data contains information for individuals younger than 65 who receive kidney dialysis. In the Part D data, we consider patients ages 65 and up.

Where do we spend our money?

Medical Insurance: Procedures

Medicare Part B (medical insurance) spent $86 billion in 2018 across 6,070 unique HCPCS codes. We spent over half of that ($51 billion, or about 57%) on 10 code groups, listed in Table 1. Most of this spending is unsurprising: we spend a large amount on office visits ($15.2 billion), in-hospital evaluations ($7.6 billion) and emergency department evaluations ($2.1 billion). Evaluations are fairly ubiquitous and are often required in order for any other procedure to occur. Also expensive are non-orally administered medications; this is a broad category including e.g. albuterol (to support breathing), chemotherapies (to fight cancer and immune disorders), and furosemide (to reduce edema). Surprising, however, are two entries for eye health: “eye and ocular adnexa surgery” and ophthalmology ($5.2 billion); when we think of common aging disorders, we often think of cardiovascular disease, osteoarthritis, diabetes, and dementia. So why do we spend so much on our eyes?

Table 1: The HCPCS code groups on which Medicare spent the most money (2018).

Figure 1: The HCPCS code groups on which Medicare spent the most money (2018).

Medical Insurance: The Eyes Have It

We spent $2.8 billion on billing codes under the heading “eye and ocular adnexa surgery”; of this money, $2.1 billion (75%) went to cataract surgeries for 2.9 million patients. Cataracts are a result of proteins clumping together in the eye (Andley, 2009); these proteins ultimately cloud vision. The lenses of our eyes contain a protein called alpha-crystallin; this protein is thought to act as a “chaperone” that prevents other proteins from clumping together and forming cataracts. As we age, the concentration of alpha-crystallin decreases, and cataracts form (Heys, Friedrich, & Truscott, 2007). Cataract surgery is the only treatment for cataracts, and it has a high rate of success, however the most common adverse outcome requires a second surgery to restore vision (Moreau & King, 2012). And even simple surgeries can be relatively major events for fragile or isolated patients who may not have much support at home. The development of a nonsurgical intervention that prevents or delays cataract development has the potential to improve the lives for millions of Americans – and potentially, many more worldwide (particularly in countries where surgery is not possible). Moreau & King, 2012 outlines many potential opportunities for interventions, but this remains an important open area of research.

We spent another $2.4 billion on ophthalmology; of this, $1.6 billion was directed toward “eye and medical examinations for diagnosis and treatment”. Though in aggregate this cost seems high, it seems worthwhile. Current guidelines recommend annual ophthalmology exams for those over age 65. These visits are important: opthalmologists can diagnose age-related eye disease (including e.g. cataracts and glaucoma), and they can also identify other common morbidities (like diabetes and stroke). 

Drug Coverage

Medicare Part D (drug coverage) spent $62 billion in 2018 for 1,618 unique generic drugs. We spent 25% of that (about $49 billion) on 10 drugs, listed in Table 1. The most expensive interventions in the top 10 generic drugs are blood thinners ($4.4 billion: apixaban, rivaroxaban) and diabetes medications ($3.4 billion: insulin glargine, sitagliptin).

Table 2: The 10 generic drugs on which Medicare spent the most money (2018).

Table 2: The 10 generic drugs on which Medicare spent the most money (2018).

Drug Coverage: Blood Thinners

Thrombosis (blood clots) can have lasting or fatal effects, as they increase the risk of stroke and heart attack. In people under 40, the incidence of thrombosis is less than 1 in 10,000 per year; over the age of 75, this increases to 1 in 100 per year (van Langevelde, Šrámek, & Rosendaal, 2010). So, why does this happen? As we age, it is common to experience an increase of coagulation proteins – these are proteins which cause our blood to clot (Wilkerson & Sane, 2002). However, it is not well understood why these proteins increase with age. Further, vascular wall thickness increases with age; this is also associated with blood clots (as there is less room for blood to pass through vessels).

We often treat aging individuals with blood thinners; these reduce the ability of our blood to clot. However, blood thinners are not without risk. Our risk of falling increases as we age, and when our blood has a decreased ability to clot, we risk major internal bleeds caused by otherwise minor injuries. So, the choice to prescribe blood thinners to elderly is a difficult balance of risks.

Because blood thinners are both risky and expensive, breakthroughs in the preventative treatment of blood clots will be an important contribution to society.

Drug Coverage: Insulin

Insulin controls blood sugar in people who have diabetes, where the body either cannot make its own insulin, or the body does not use insulin normally (American Society of Health-System Pharmacists, Inc., 2019). It is currently unknown how (or whether) type 1 diabetes can be prevented (Jacobsen, Haller, & Schatz, 2018); for now, the only option for those with type 1 diabetes is treatment. Risk for type 2 diabetes, however, can be modified through lifestyle changes (including healthy eating, weight loss and increased physical activity), as illustrated in four randomized trials across four countries (Gruss, et al., 2019). This research finds that lifestyle modifications are cost-effective relative to placebo. And because diabetes often comes with complications (including cardiovascular disease, impaired vision, renal disease and lower extremity amputations), diabetes prevention can vastly improve quality of life.

The Centers for Disease Control and Prevention (CDC) runs the National Diabetes Prevention Program, which aims to delay or prevent type 2 diabetes among those at risk (Centers for Disease Control and Prevention, 2019). Through randomized studies, the CDC finds that this program can cut the risk of developing type 2 diabetes by 58%, and by 71% for those over 60 years old. CMS also runs a diabetes prevention program – the Medicare Diabetes Prevention Program (MDPP) – available exclusively to patients over the age of 65 enrolled in Medicare Part B. This program is still fairly small with just 947 participants (perhaps attributable to the fact that it started enrolling patients in January 2018). It would be worthwhile for CMS to invest in identification of people at risk for type 2 diabetes to offer them enrollment into the program – they would likely save money on billing costs, and they would certainly improve quality of life for many.  It would also perhaps be worthwhile for CMS to collaborate with the CDC: lifestyle changes starting before the age of 65 (when patients become eligible for CMS) are likely more impactful than those beginning later in life.

Conclusions

Medicare spending can reveal comorbidities and procedures that are both expensive and common among our aging population. In particular, we found cataract surgeries, blood thinners, and diabetes to be notable.

1. 1. Cataracts: The development of non-surgical interventions to remove cataracts would help people in the USA and abroad (where cataract surgery is often not possible). Though there is current research in this area (Moreau & King, 2012); increasing federal funding to this research now may be worthwhile to CMS in the long run, as we could save money on future interventions (and improve quality of life for many, both in the US and globally).
   2. Blood thinners: Blood thinners increase the risk of serious or fatal bleeds for the aging population, and society would benefit from a safer alternative. Developing an alternative is challenging, as the exact mechanism that causes our blood to clot as we age is not well understood. There is current work to identify optimal individual-level management of anticoagulant therapy (Dai H., et al., 2020). We encourage researchers (and funders) to continue this work; patients will benefit from a decreased risk of adverse event, and CMS will benefit through decreased spending on adverse events.
   3. Diabetes: Type 2 diabetes is best treated through prevention. Currently, there are national programs to modify diabetes risk factors in people at higher-than-average risk of developing diabetes. Increasing enrollment in these programs (via investment into better identification of patients at risk, and perhaps also through earlier engagement) could reduce the incidence of type 2 diabetes in the US. It would be worthwhile to invest federal money in this effort across all ages (not just ages 65 and up); CMS will save money, and patients who averted type 2 diabetes may have improved quality of life.

Finally, the available data from CMS leaves much yet to be discovered. We focused on identifying interventions that are expensive; alternatively, we could have identified interventions that are common. We could also have identified interventions that are most common or expensive in each US state, to identify differences in health and health care across our country.

Bibliography

Andley, U. P. (2009). Effects of alpha-crystallin on lens cell function and cataract pathology . Current molecular medicine, 887-92.

CMS Office of Enterprise Data and Analytics. (2021, 02 19). Medicare Fee-For-Service Provider Utilization & Payment Data Part D Prescriber. Retrieved from https://www.cms.gov/files/document/part-d-prescriber-puf-methodology.pdf

CMS Office of Enterprise Data and Analytics. (2021, 02 19). Medicare Fee-For-Service Provider Utilization & Payment Data Physician and Other Supplier. Retrieved from https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/NationalHealthExpendData/NHE-Fact-Sheet

CMS Office of Enterprise Data and Analytics. (2021, 02 19). Medicare Fee-For-Service Provider Utilization & Payment Data Physician and Other Supplier. Retrieved from https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Medicare-Provider-Charge-Data/Downloads/Medicare-Physician-and-Other-Supplier-PUF-Methodology.pdf

CMS Office of Enterprise Data and Analytics. (2021, 02 19). Medicare Fee-For-Service Provider Utilization & Payment Data Physician and Other Supplier. Retrieved from https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Medicare-Provider-Charge-Data/Physician-and-Other-Supplier

Heys, K. R., Friedrich, M. G., & Truscott, R. J. (2007). Presbyopia and heat: changes associated with aging of the human lens suggest a functional role for the small heat shock protein, α‐crystallin, in maintaining lens flexibility. Aging Cell.

Martin, A. B., Hartman, M., Lassman, D., & Catlin, A. (2020). National Health Care Spending In 2019: Steady Growth For The Fourth Consecutive Year. Health Affairs (Project Hope).

Moreau, K. L., & King, J. A. (2012). Protein misfolding and aggregation in cataract disease and prospects for prevention. Trends in Molecular Medicine, 273-282.

van Langevelde, K., Šrámek, A., & Rosendaal, F. R. (2010). The Effect of Aging on Venous Valves. Arteriosclerosis, Thrombosis, and Vascular Biology, 2075-2080.

Wilkerson, W. R., & Sane, D. C. (2002). Aging and thrombosis. Seminars in Thrombosis and Hemostasis, 555-68.

Appendix

Structure: Part B Data

The Part B data released by CMS (CMS Office of Enterprise Data and Analytics, 2021) includes the amount Medicare paid to healthcare providers for each unique procedure, and the number of patients who underwent that procedure. Providers are organized by their National Provider Identifier (NPI), which is a federally issued numeric ID. Procedures are organized by their Healthcare Common Procedural Coding System (HCPCS) code, and HCPCS codes are updated and managed by CMS. HCPCS codes are both broad and specific; for example, see the codes described in Table A1. Finally, HCPCS codes are grouped into categories; for example, HCPCS codes numbered 99201 through 99215 all refer to “in-office or other outpatient services” (e.g. a 30-minute check-up in your doctor’s office).

Table A1: Example HCPCS codes

Concretely, the Medicare Part B data is structured as in Table A2. In this snippet of the data, we see the provider with NPI 1003000126 performed procedure 99217 a total of 68 times across 67 patients and received $3,936 from Medicare in total for this service. That same provider also performed procedures 99218 and 99233 and earned $1,459 and $23,433 respectively

Table A2: Example rows from the Medicare Part B dataset.

Structure: Part D Data

The Part D data released by CMS (CMS Office of Enterprise Data and Analytics, 2021) includes the amount Medicare paid for drug prescriptions written by each unique procedure, the number of patients who received that prescription, and the number of standardized 30-day prescriptions that were filled. As with the Part B data, providers are organized by their NPI. Drugs are described by their drug name (in the case of trademarked drugs) and their generic name. We show the structure of this data in Table A3. 

Table A3: Example rows from the Medicare Part D dataset.