Annual health care costs among Medicare Beneficiaries with essential tremor

Abstract

Introduction:

Essential tremor (ET) is one of the most common neurological disorders, affecting an estimated 2.2% of the entire US population. Despite its high prevalence and associated morbidity and mortality, there are no published data on the medical costs associated with ET care.

Methods:

This is a retrospective secondary data analysis using the 5% Medicare claims data from 2016 to study age-eligible Medicare beneficiaries diagnosed with ET (diagnostic code G250) relative to a propensity score-matched group of comparison beneficiaries without ET (27,081 in each arm). Comparisons were matched within age strata and on the full set of Charlson comorbidity indicators, race, and sex. We examined encounter-level costs (amounts paid) and total annual costs of care (in constant $2021 dollars) adjusting for age, sex, race/ethnicity, provider specialty, setting, and the most common comorbidities, using a generalized linear model.

Results:

The final sample included 54,162 total beneficiaries, with an average age of 75, 65% female and 94% Non-Hispanic White. On average, Medicare beneficiaries with at least one outpatient or physician office visit with an ET diagnosis have $1068 (95% CI: $981, $1154) in additional direct medical care expenditures per year relative to statistically similar comparison beneficiaries of the same age. Across the population, we predicted aggregated additional spending attributable to ET among Medicare beneficiaries between $1.5 billion and $5.4 billion per year.

Conclusion:

The estimated direct medical costs among Medicare beneficiaries with an ET diagnosis aggregated to the population-level are non-trivial. These data begin to fill a gap in knowledge.

1. Introduction

Essential tremor (ET), which is among the most prevalent neurological disorders, has a pooled prevalence of 1.33% (all ages), with a 74% increase in prevalence with each decade of age, reaching values of 5.79% among those age 65 and older and values greater than 20% in those in their 90s [1]. Although in the past the condition was sometimes labeled “benign,” this term is misleading. The tremor is usually progressive, producing disabilities with basic daily activities such as eating and writing. Aside from tremor, patients may exhibit cognitive impairment compared to similarly-aged controls as well as other neurological issues (e.g., mild gait ataxia). Despite its high prevalence and associated functional disability, there are no published data on the medical costs associated with ET care. Understanding the direct medical costs is an important first step in describing the burden of disease to help policy-makers and researchers prioritize efforts to reduce burden.

To fill this gap in the literature, we conducted a cost of illness study using a nationally representative sample of Medicare beneficiaries aged 65 or older covered by traditional (fee-for-service - FFS) Medicare to estimate a total annual cost of care among beneficiaries with an ET diagnosis relative to a propensity score matched-comparison group of beneficiaries without ET. We examined the extent to which encounter-level costs varied by care setting, provider specialty, presence of common comorbidities, and demographic features (race/ethnicity, age, and sex).

2. Patients and methods

2.1. Study design

This is a retrospective secondary data analysis of the excess costs of care for ET using the 5% Centers for Medicare and Medicaid Services (CMS) Medicare carrier (physician office), inpatient, and outpatient encounters from 2016 (dates of service from January 1 to December 31). Prescription medication claims were not included in this analysis as not all beneficiaries have a Part D plan (and hence would not be represented). Roughly 64% of fee-for-service beneficiaries were covered by a Part D plan in 2016 [2]. However, we calculated annual costs for the most frequently prescribed medications for individuals with ET as described below.

The inclusion criteria for the analytic sample were: (1) age ≥65 years; (2) continuous enrollment in Medicare fee-for-service (FFS) Parts A and B for the full calendar year of 2016; (3) Medicare beneficiary status was “aged without end stage renal disease (ESRD)”; and (4) at least one outpatient or carrier claim with a diagnosis of ET (G250). Up to 25 diagnosis codes are available for each encounter; a beneficiary was defined as having ET if any of the 25 diagnosis codes in any outpatient or physician office visit claim was ET. Any beneficiaries with related neurological conditions (G21, G23, G24, G25) that are sometimes conflated with ET were excluded. See Supplemental Table 1 for sample construction details.

2.2. Measures

Costs.

Costs for inpatient stays were calculated as Medicare claim payment amounts including pass-through per diem payments (daily pass-through amount multiplied by the number of covered days per stay. All payment amounts were converted to constant 2021 U S. dollars using the Consumer Price Index for medical care [3]. The main cost outcome variables were (1) encounter-level costs by setting and provider specialty among non-inpatient encounters; (2) total annual costs of care by setting patient characteristics. Population-level prescription medication costs are approximated as described below.

Covariates.

Encounter-level costs were adjusted by setting (inpatient, outpatient, and physician office), provider specialty (3 categories: primary care provider, neurologist, and all other specialists following CMS’ definitions of these categories [4]), patient age (continuous), sex, race/ethnicity (categorical: non-Hispanic White, non-Hispanic African American/Black, and all other), and a set of indicators for having one of the Charlson comorbidity conditions [5]. We note that provider specialty is not reported for inpatient claims. Annual costs of care were adjusted by patient age, sex, race/ethnicity, state of residence, and comorbidities.

2.3. Statistical analyses

A greedy propensity score matching (PSM) approach was used to identify beneficiaries without ET but who were statistically similar to ET beneficiaries on observable characteristics. We matched within each year of age (e.g. 34 strata from ages 65 to 98 on sex, race/ethnicity (Hispanic, non-Hispanic Black and non-Hispanic white) and 16 indicators for each of the conditions used in the Charlson comorbidity index [5].

We calculated descriptive statistics and conducted balancing tests of the full analytic sample and by whether the beneficiary had an ET diagnosis during the year or not. A generalized linear model (GLM) with a log link function assuming a gamma distribution was used to estimate total annual costs across all encounters and settings during the year adjusting for age, sex, race/ethnicity, comorbidities, and the number of visits. The GLM addresses the skewed cost distribution, avoids transformation/re-transformation biases, and allows us to produce doubly robust estimates.

We approximated the annual, population-level costs of prescription medications by multiplying percentage of ET patients using the most common medications [6] times the average annual cost [14].

Finally, we aggregated our direct medical and prescription medication costs for ET patients to the population level using a lower and upper bound estimate of the number of Medicare beneficiaries with ET. Previous studies have estimated about 7 million individuals in the U.S. have essential tremor [7] and that the prevalence increases (exponentially) with age. One study reporting age-based prevalence rates suggested that approximately 60% of all cases of ET are among those age 65+ [8], which would imply about 4.2 million individuals age-eligible for Medicare with ET. Applying the age-based prevalence rates from that study to the population counts of Medicare beneficiaries in each age group, however, suggests only about 1.2 million beneficiaries with ET. Thus, we used these two values (1.2 million and 4.2 million) as the lower and upper estimates of the number of Medicare beneficiaries in our population-level estimates of costs of care. We calculated bootstrapped confidence intervals using the estimated range of direct medical costs of care.

In addition, we conducted sensitivity checks using different PSM approaches (fewer covariates, requiring a smaller radius for matching). Results are similar to those presented here. Two-sided statistical tests were conducted with p-values of less than 0.05 considered statistically significant. Stata SE, Version 16 was used for all estimation after the matched sample was created. The UTSW IRB deemed this secondary, de-identified study as exempt.

3. Results

The final analytic sample included 27,081 Medicare beneficiaries with at least one claim with an ET diagnosis and 27,081 propensity score matched comparison beneficiaries. The sample excluded any beneficiaries with diagnoses that are related and sometimes confused with ET (see appendix for more details). In Supplemental Table 2, we report descriptive statistics. Beneficiaries with an ET diagnosis are statistically more likely to have liver disease (0.52% vs. 0.39%), even after matching within the age stratum. They also have more of all types of visits and greater annual expenditures. Notably, conditional on having an inpatient visit, there is no statistically significant difference in the inpatient expenditures. However, only 1% of the PSM comparison patients have an inpatient visit relative to about 5% of ET patients.

In Table 1, we present the results from the GLM estimates adjusting for age, sex, race/ethnicity, and indicators for the chronic conditions. Across all setting types, ET patients have $1068 (95% CI: $981, $1154) additional healthcare payments relative to statistically similar patients without ET. We estimated the annual costs of prescription medications for ET patients ranges from about $284 million to $995 million (see Supplemental Table 4).

Table 1.

Adjusted means and confidence intervals (CI).

	Comparison Patients		ET Patients		Difference
	Adjusted Mean	95% CI	Adjusted Mean	95% CI		95% CI
Total Payments	$1742	($1692, $1793)	$2810	($2730, $2890)	$1068	($981, $1154)
By Setting:
Outpatient	$959	($931, $987)	$1290	($1254, $1327)	$332	($288, $375)
Carrier	$587	($574, $599)	$732	($716, $747)	$145	($126, $164)
Inpatient	$165	($135, $196)	$765	($701, $828)	$599	($529, $668)

Notes: Adjusted means obtained after two-part GLM, adjusting for age, sex, race/ethnicity, and indicators for the chronic conditions shown in Table 1.

In Fig. 1 we show the range of total population-level costs of care including the average annual direct medical costs and prescription medication costs assuming a lower and upper bound estimate of the number of Medicare beneficiaries with ET. The total cost of treating ET among Medicare beneficiaries ranges from almost $1.5 billion to $5.4 billion, depending on which estimate of prevalence of beneficiaries with ET that we use.

Fig. 1.

Aggregate population costs of essential tremor for medicare beneficiaries (age 65+).

4. Discussion

There are no published data on the medical costs associated with ET care. Here, we fill that gap in knowledge.

In our sample of Medicare FFS beneficiaries, we found that, on average, ET patients are significantly more likely to be female and White relative to the average Medicare beneficiary (see Supplemental Table 3).

Importantly, those with an ET diagnosis in our sample were more likely to have all the Charlson comorbidity conditions. For example, 32% of our sample of ET beneficiaries have chronic obstructive pulmonary disorder relative to only 19% of the comparison sample. Without adjusting for this difference in disease profiles, we would estimate a larger difference in the medical expenditures of the two groups which may not be entirely attributable to ET. However, these differences are also noteworthy in and of themselves and are worthy of future study. There are few large-scale studies of the rates of comorbidity in ET vs. those without. Two are relevant and their observations are in line with our data. First, in a study of more than 5 million US veterans, ET patients were more likely than those without ET to have numerous comorbidities [9]. Second, a more recent study [10] using a large US administrative claims dataset similarly noted that ET patients had a higher number of comorbidities than non-ET patients (5.3 vs. 4.0).

After comparing Medicare beneficiaries diagnosed with ET with a smaller and more statistically similar comparison group, we found the average adjusted annual direct medical care expenditures at $2,810, which was $1068 greater than the expenditures among statistically similar adults of the same age without ET. There are significant differences across all settings with ET beneficiaries more likely to have an encounter in each setting type.

Our estimates suggest an aggregated population-level burden between $1.5 and $5.4 billion per year. It is important to compare these estimates to those from other neurological disorders. Studies on the economic burden of PD have estimated annual direct medical care expenditures between $11,900 and $17,400 per patient, for an aggregated population-level burden between $7.4 to $8.6 billion (in 2020$) [11, 12]. Although we are not making direct comparisons across these patient populations, our annual cost estimates of annual health care costs per ET patient are significantly lower than for patients with PD. However, ET is more prevalent so the total economic burden estimates may be closer, at least by some estimates (e.g. our highest estimate of $5.4 billion versus $7.4 billion). These data begin to frame some aspects of the health care burden associated with this disease. To our knowledge, there are no other data in ET to which we can compare this.

The proportion of our sample that was white was higher than that of the US population. We assessed whether annual ET-related care costs differed between non-Hispanic whites and blacks, but did not find a difference. Hence, it is unlikely that the racial composition of our sample influenced the results we report here.

One additional caveat about our estimates is that they are likely to miss important other indirect or non-medical costs, which include costs related to lost work or productivity, disability or reduced quality of life, informal and formal costs of caregiving, and other costs such as transportation. One study on quality of life among ET patients found that ET patients report greater interference with their ability to write, talk with others, eat, and drink, creating greater feelings of embarrassment relative to those with PD [13]. Such burdens are not incorporated into our analysis, but can have substantial impacts. In addition, our findings are drawn from a nationally representative sample of adults, age 65 or older who are covered by FFS Medicare who have an ET diagnosis during the year only and may not be generalizable to younger ET patients or those covered by Medicare Advantage plans.

In summary, the direct medical care expenditures attributable to ET amount to nearly $1100 per patient per year which totals $1.5 to $5.4 billion in aggregate. More research is needed to quantify other aspects of economic burden among these patients. In particular, the costs associated with disability, functional limitations, increases in mental health conditions and cognitive decline need to be incorporated to have a more comprehensive assessment of the economic burden of ET.