Abstract
Context:
In 2014, Medicare hospice policy changed to shift medication payments from Part D to the hospice benefit.
Objective:
To describe patterns of medications reported to the Medicare hospice benefit after the 2014 policy change.
Methods:
Repeated cross-sectional analysis of medications reported to the Medicare hospice benefit between 2014 and 2018. Data sources include the Medicare Master Beneficiary Summary File and national revenue center claims submissions by Medicare-certified hospices. Main outcome measures by year were the mean number of overall and chronic disease medication dispensings per enrollee, the most common medications billed to the Medicare hospice program, and the proportion of medications for end-of-life, chronic diseases, and potentially inappropriate by Screening Tool of Older Persons Prescriptions in Frail criteria.
Results:
Between 2014 and 2015, the mean number of dispensings per enrollee covered by hospice temporarily increased from 12.1 (standard deviation (SD) 21.5) to 12.4 (SD 23.4) (p<.001) and returned to 12.1 (SD 23.0) in 2016, while the mean number of chronic disease medication dispensings per enrollee increased from 2.8 (SD 2.2) to 3.2. (SD 2.4) (p<.001) and remained stable. Between 2014 to 2018, the proportion of the top 100 drugs covered by hospice that were for chronic disease or met STOPPFrail criteria remained stable, while the proportion of EOL drugs declined temporarily before returning to 2014 levels.
Conclusions:
The Medicare hospice program experienced small but statistically significant increases in chronic disease dispensings per enrollee after 2014 policy changes. Whether these were clinically significant or impacted patient outcomes requires further study.
Keywords: Medicare, hospice, prescription medications, policy
INTRODUCTION
The appropriate use of medications to manage disease conditions and symptoms in hospice is critical for the delivery of high-quality care to patients approaching the end of life. Prescription drugs are essential for pain control, symptom management, and optimizing quality of life.1-3 Thus, hospice patients are prescribed an average of 10-12 medications daily,2,4-6 with the overall burden increasing while end-of-life symptom management drugs are added as death approaches.2
In the United States (US), hospice medications are paid for by two Medicare programs: the Medicare Part A Hospice Benefit and the Medicare Part D prescription drug benefit. The Part A hospice benefit pays for drugs for palliative purposes related to the patient’s terminal illness and are included in a per diem payment used to cover all clinical care, therapies, equipment, and counseling regardless of the number of services provided ($211/day in 2023 [routine home care]) with an annual cap of $32,486.92.7 When Part D began in 2006, it paid for medications for preexisting conditions not considered related to the patient’s terminal illness or related conditions, as well as chronic, stable conditions that played little role in patients’ terminal illness.8 Since that time, multiple reports from the Office of the Inspector General spotlighted instances of apparent double billing where medications that should have been covered by the hospice as part of its per diem were billed to Part D.9
In response to these concerns, Medicare policy changed in 2014 to encourage a shift of medication payments away from the outpatient Part D benefit to the Medicare hospice benefit.10,11 Before and after the 2014 policy change, hospices were required to cover all services necessary for the palliation and management of the terminal illness and related conditions, while Part D plans were required to cover drugs for treatment of conditions completely unrelated to the terminal illness and related conditions. In December 2013, the Centers for Medicare and Medicaid Services (CMS) issued a policy statement that Part D plans should reject all Part D claims for hospice enrollees until prior authorization requirements (i.e., an explanation of why the drug was not related to the terminal condition) were satisfied. On March 10, 2014, CMS clarified aspects of the earlier guidance while reiterating the expectation that Part D plans place prior authorization requirements on all drugs for hospice enrollees. CMS underscored its longstanding “general view that hospices are required to provide virtually all care that is needed for terminally ill patients” and its perspective that Part D covered medication use should be “exceptional and unusual”. A few months later in July 2014, after considering stakeholder concerns regarding operational challenges and enrollee access, CMS revised its policy to strongly encourage Part D plans to place prior authorization requirements on only four categories of prescription drugs (analgesics, antinauseants, laxatives, and anxiolytics) typically used for management of common symptoms at the end of life and very rarely appropriate for Part D payment. This policy change can be considered a possible constraint on material resources available to hospice programs, since the hospice benefit pays for drugs from a fixed per diem.7,12 However, the impact of this policy change including whether the number and types of medications paid for by hospice changed is unclear. The goal of this analysis was to describe patterns of medications reported to US Medicare hospice programs after the 2014 Medicare policy change. Based on our prior clinical experience, our hypothesis was that chronic disease medications would shift from part D to hospice and comprise an increasing proportion of the medications covered by hospice over time.
Methods
This study was reviewed and approved by the UMass Chan Medical School Institutional Review Board.
Study Design and Data Source.
We conducted a repeated cross-sectional analysis for each calendar year between and including 2014 to 2018 using national Medicare administrative data sources including hospice fee-for-service (FFS) claims and the Master Beneficiary Summary File (MBSF). The hospice file contains FFS hospice claims submitted for all Medicare beneficiaries with information on the level of care provided (e.g., home health, skilled nursing), diagnoses, and dates of service. For 2014-2018, Medicare hospice claims included submission of drug information. Reporting of drug information was voluntary for the first quarter of 2014, so our analyses focus on the mandatory reporting period spanning quarter 2 of 2014 through 2018. The MBSF provides demographic information (age, sex, race, ZIP code), vital status (validated Social Security Administration date of death), and eligibility for Medicare and Medicaid. Hospice claims include Medicare Advantage enrollees.
Variables.
Medications covered by hospice were identified from the hospice claims using the national drug codes (NDCs) which are unique identifiers for a drug product (i.e., which drug product, packaging, and manufacturer). Hospice charges included the amount that hospice recorded as the total charges for the services described on the revenue code line for hospice only (not Part D drugs). Information on hospice charges is collected for CMS research purposes; these submitted charges did not affect the amount of reimbursement.13 We use MBSF data to describe enrollee age, sex, race/ethnicity, and geographic Census region (West, Midwest, South, Northeast). We used the Hospice FFS to describe the hospice admitting diagnosis and comorbidities,14 and length of stay for the first continuous enrollment in a year for each hospice enrollee.
We classified end-of-life (EOL) medications according to the International Association for Hospice and Palliative Care (IAHPC) List of essential medicines for palliative care.14,15 (Supplemental Figure 1) We also assessed the medication appropriateness at end of life using the STOPP-Frail (Screening Tool of Older Persons Prescriptions in Frail adults with limited life expectancy) criteria version 1 which corresponds to the time period of the data.16 (Supplemental Figure 2) Chronic disease management drugs medications included antiarrhythmics, antihypertensive, antidiabetic, antidementia, chronic obstructive disease-related, rheumatologic, diuretics, anti-thyroid drugs, antidepressants and antipsychotics that were not included in IAHPC essential medications for palliative care. (Supplemental Figure 3)
Data Analysis.
We used descriptive statistics to characterize hospice enrollees by each entire year studied, except 2014 which excluded January to March to reflect the policy implementation date. We limited measurement to the first observation for each enrollee between 2014 and 2018. We identified the number of unique drugs dispensed based on the generic name of each NDC identified. We enumerated the total number of drug dispensings within each year based on each revenue center line item with a NDC code matched on a generic name, since there are multiple NDCs for each active generic compound. Based on the total enrollment and total number of dispensings in each year, we estimated the total number of dispensings per enrollee. We calculated, for the top 30 drugs based on the proportion of all enrollees with hospice claims for those drugs in each year, the proportion of enrollees prescribed each drug and the mean quantity dispensed per enrollee per 30-day period (defined as number of fills divided by the number of days in hospice length of stay per 30 person-days).17 Within each year, we also enumerated the 100 most-frequently prescribed hospice medications and categorized them as EOL, chronic disease management, STOPPFrail, anti-infective, and central nervous system (CNS) and psychoactive. We use a two sample t-test to examine whether there were significant changes in the mean number of dispensings per enrollee between key study years, including the year following the policy change (i.e., between 2014 and 2015). We use a linear-by-linear test for trend to test the hypothesis that the mean number of dispensings per enrollee billed to hospice increased between 2014 to 2018. We use a Chi-square test for trend to examine whether chronic disease medications comprised a changing proportion of the annual top 100 drugs dispensed between 2014 to 2018.
RESULTS
Hospice Population
Between 2014 and 2018, the number of hospice enrollees increased in each year, with an enrollment of 1,464,405 in 2018. (Table 1) In 2014, the mean age of enrollees was 82.0 years (standard deviation (SD) 10.6) and 59.0% were female. In 2018, the mean age was 82.1 years (SD 10.5) and 57.5% were female. The proportion of patients admitted for dementia declined from 19.6% to 17.4% between 2014 and 2018, and then remained relatively stable. The proportion of patients admitted for cancer also decreased from 28.5% in 2014 to 26.7% in 2018. The proportions of other admitting diagnoses increased over the study period.
Table 1.
Characteristics of Medicare hospice enrollees between and including 2014-2018
| Hospice enrollee demographics |
2014a | 2015 | 2016 | 2017 | 2018 | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| # N = 1,047,341 |
% | # N=1,344,593 |
% | # N=1,367,186 |
% | # N=1,415,917 |
% | # N= 1,464,405 |
% | |
| Sex | ||||||||||
| % Female | 618,421 | 59.0 | 784,194 | 58.3 | 792,773 | 58.0 | 817,553 | 57.7 | 842,176 | 57.5 |
| Age (mean, SD) | 82.0 | 10.6 | 81.9 | 10.6 | 81.9 | 10.6 | 82.1 | 10.5 | 82.1 | 10.5 |
| Race/Ethnicity | ||||||||||
| White | 911,150 | 87.0 | 1,168,275 | 86.9 | 1,184,136 | 86.6 | 1, 224,422 | 86.5 | 1,262,918 | 86.2 |
| Black/African American | 85,716 | 8.2 | 109,788 | 8.2 | 112,861 | 8.3 | 116,591 | 8.3 | 121,398 | 8.3 |
| Asian | 11,434 | 1.1 | 15,919 | 1.2 | 16,775 | 1.3 | 18,390 | 1.3 | 19,562 | 1.3 |
| Unknown | 3,165 | 0.3 | 4,659 | 0.3 | 5,423 | 0.4 | 6,256 | 0.4 | 7,465 | 0.5 |
| Latinx/Hispanic | 21,493 | 2.1 | 27,121 | 2.0 | 28,395 | 2.0 | 29,227 | 2.1 | 30,639 | 2.1 |
| North American Native | 3,635 | 0.3 | 4,704 | 0.3 | 4,843 | 0.4 | 5,064 | 0.4 | 5,351 | 0.4 |
| Other | 10,746 | 1.0 | 14,126 | 1.1 | 14,751 | 1.1 | 15,964 | 1.1 | 17,072 | 1.2 |
| Primary Admitting Diagnosis | ||||||||||
| Cancer | 298,660 | 28.5 | 381,947 | 28.4 | 384,726 | 28.1 | 388,223 | 27.4 | 391,313 | 26.7 |
| Dementia | 205,753 | 19.6 | 232,247 | 17.3 | 235,622 | 17.2 | 245,515 | 17.3 | 255,161 | 17.4 |
| Circulatory/Heart | 190,686 | 18.2 | 250,581 | 18.6 | 254,149 | 18.6 | 264,509 | 18.7 | 279,151 | 19.1 |
| Stroke | 86,670 | 8.3 | 125,943 | 9.4 | 128,946 | 9.4 | 136,346 | 9.6 | 115,364 | 9.7 |
| Respiratory | 109,196 | 10.4 | 146,770 | 10.9 | 149,966 | 11.0 | 156,998 | 11.1 | 160,846 | 11.0 |
| Other | 156,364 | 14.9 | 207,099 | 15.4 | 213,777 | 15.6 | 224,326 | 15.8 | 235,719 | 16.1 |
| Length of Stay per Enrollmentb | ||||||||||
| Median LOS, IQR | 15 | 52 | 15 | 66 | 16 | 66 | 15 | 65 | 16 | 68 |
| Geographic Census region | ||||||||||
| West | 207,646 | 19.8 | 267,908 | 19.9 | 274,057 | 20.0 | 285,421 | 202 | 296,312 | 20.2 |
| Midwest | 249,537 | 23.8 | 320,827 | 23.9 | 325,920 | 23.8 | 338,298 | 23.9 | 348,777 | 23.8 |
| South | 420,785 | 40.2 | 538,377 | 40.0 | 547,991 | 40.1 | 567,587 | 40.1 | 590,203 | 40.3 |
| Northeast | 160,597 | 15.3 | 207,454 | 15.4 | 209,479 | 15.3 | 215,636 | 15.2 | 219,978 | 15.0 |
| Other | 8,770 | 0.8 | 10,015 | 0.7 | 9,723 | 0.7 | 8,961 | 0.6 | 9,116 | 0.6 |
Excluding 1st quarter (January 1- March 31 2014) before policy change.
Limited to 1st continuous admission in calendar year
Medications and quantity dispensed by year
The mean number of dispensings covered by hospice per enrollee increased from 12.1 to 12.4 between 2014 to 2015 (p<.001, t-test), and then appeared to return to 2014 levels between 2015 and 2016 (p<0.01, t-test). (Table 2) Despite these statistically significant year-by-year changes, the overall test for trend of dispensings between 2014 and 2018 was not significant (p=0.48). Similarly, there was growth in the mean number of dispensings for chronic disease medications paid for by the hospice benefit after the 2014 policy change, from 2.8 to 3.2 between 2014 to 2015 (p<.001, t-test) consistent with our hypothesis. The mean number of chronic disease medication dispensings per enrollee remained elevated after 2015 but decreased from 3.2 to 3.0 between 2017 and 2018 (p<.001, t-test). Despite these significant year-by-year changes, the overall test for trend of chronic disease dispensings between 2014 and 2018 was not significant (p=0.85).
Table 2.
Medication dispensings billed to hospice by year, 2014-2018
| 2014a | 2015 | 2016 | 2017 | 2018 | |
|---|---|---|---|---|---|
| N = 1,047,341 | N=1,344,593 | N=1,367,186 | N=1,415,917 | N=1,464,405 | |
| Total Number of Dispensings | 12,643,663 | 16,616,258 | 16,506,295 | 17,162,756 | 16,760,994 |
| Mean Dispensings per enrollee (SD) | 12.1 (21.5) | 12.4 (23.4) | 12.1 (23.0) | 12.1 (23.5) | 11.5 (22.7) |
| Total # dispensings for top 100 drugs in that year | 11,184,283 | 14,815,039 | 14,693,879 | 15,312,911 | 14,967,189 |
| Chronic Disease Medicationb dispensings (of top 100 drugs) | 2,972,411 | 4,304,029 | 4,387,873 | 4,577,900 | 4,353,082 |
| Mean Chronic Disease Dispensings (of Top 100 drugs) per enrollee (SD) | 2.8 (SD 2.2) | 3.2 (SD 2.4) | 3.2 (SD 1.9) | 3.2 (SD 2.3) | 3.0 (SD 1.9) |
Excluding 1st quarter (January 1- March 31 2014) before policy change.
See Supplemental Figure 3 for List of Chronic Disease Medications in the Top 100 Medications Reported to the Medicare Hospice Benefit Between 2014 to 2018
Most common medications covered by hospice, by year
Lorazepam and morphine were the most common medications billed to hospice across all years. Table 3 also shows the proportion of enrollees who used each of the 30 most commonly drugs used in each year. The proportion of patients who received these drugs under hospice increased slightly 2014 to 2015, and then remained relatively stable. For some drugs, there was an increase in the proportion of users from 2014 to 2015 followed by a decline towards baseline (e.g., furosemide). Less commonly there was a sustained increase (e.g., gabapentin) or decrease (e.g., atropine) over the observation period. Medications commonly used for the management of EOL symptoms remained among the top 10 medications billed to hospice in each year.
Table 3.
The most common medications billed to Medicare hospice benefit, by % enrollees prescribed drug in a year (2014-2018)
| 2014a | 2015 | 2016 | 2017 | 2018 | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Generic Drug Name |
Quantity Dispensed Per Patientb Per 30 Days |
% Enrollees |
Generic Drug Name |
Quantity Dispensed Per Patienta Per 30 Days |
% Enrollees |
Generic Drug Name |
Quantity Dispensed Per Patienta Per 30 Days |
% Enrollees |
Generic Drug Name |
Quantity Dispensed Per Patienta Per 30 Days |
% Enrollees |
Generic Drug Name |
Quantity Dispensed Per Patienta Per 30 Days |
% Enrollees |
| Lorazepam | 2.4 | 45.8% | Lorazepam | 2.5 | 49.4% | Lorazepam | 2.5 | 49.5% | Lorazepam | 2.5 | 50.4% | Lorazepam | 2.6 | 47.8% |
| Morphine Sulfate | 2.5 | 43.2% | Morphine Sulfate | 2.6 | 47.1% | Morphine Sulfate | 2.6 | 46.9% | Morphine Sulfate | 2.6 | 47.6% | Morphine Sulfate | 2.6 | 44.9% |
| Atropine Sulfate | 2.3 | 24.6% | Atropine Sulfate | 2.3 | 19.0% | Hyoscyamine Sulfate | 2.3 | 20.1% | Hyoscyamine Sulfate | 2.3 | 23.3% | Hyoscyamine Sulfate | 2.3 | 22.6% |
| Haloperidol Lactate | 2.1 | 14.8% | Haloperidol Lactate | 2.3 | 17.3% | Haloperidol Lactate | 2.3 | 17.1% | Haloperidol Lactate | 2.3 | 17.6% | Haloperidol Lactate | 2.3 | 16.2% |
| Furosemide | 1.7 | 12.2% | Hyoscyamine Sulfate | 2.3 | 16.1% | Atropine Sulfate | 2.3 | 12.3% | Prochlorperazine Maleate | 2.1 | 12.3% | Prochlorperazine Maleate | 2.1 | 10.9% |
| Hydrocodone/Aceta minophen | 1.7 | 11.8% | Prochlorperazine Maleate | 2.1 | 12.9% | Prochlorperazine Maleate | 2.1 | 12.3% | Furosemide | 1.8 | 10.9% | Furosemide | 1.8 | 10.2% |
| Prochlorperazine Maleate | 2.0 | 11.2% | Furosemide | 1.9 | 11.7% | Furosemide | 1.8 | 11.3% | Atropine Sulfate | 2.3 | 9.7% | Ipratropium/Albuterol Sulfate | 1.7 | 8.8% |
| Acetaminophen | 2.1 | 10.8% | Hydrocodone/Acetaminophen | 1.7 | 10.3% | Hydrocodone/Acetaminophen | 1.7 | 9.7% | Hydrocodone/Acetaminophen | 1.7 | 9.1% | Hydrocodone/Acetaminophen | 1.7 | 8.2% |
| Haloperidol | 2.0 | 8.8% | Haloperidol | 2.1 | 8.9% | Haloperidol | 2.1 | 8.6% | Ipratropium/Albuterol Sulfate | 1.8 | 9.0% | Atropine Sulfate | 2.3 | 8.0% |
| Bisacodyl | 2.0 | 8.6% | Ipratropium/Albuterol Sulfate | 1.8 | 8.6% | Ipratropium/Albuterol Sulfate | 1.7 | 8.4% | Haloperidol | 2.1 | 8.3% | Haloperidol | 2.1 | 6.9% |
| Ipratropium/Albuterol Sulfate | 1.7 | 7.6% | Acetaminophen | 2.3 | 6.8% | Acetaminophen | 2.3 | 7.8% | Acetaminophen | 2.3 | 6.9% | Acetaminophen | 2.3 | 5.8% |
| Fentanyl | 2.3 | 6.8% | Albuterol Sulfate | 1.6 | 6.7% | Albuterol Sulfate | 1.5 | 6.3% | Albuterol Sulfate | 1.5 | 5.9% | Tramadol Hcl | 1.4 | 5.7% |
| Albuterol Sulfate | 1.5 | 6.7% | Fentanyl | 2.4 | 6.6% | Fentanyl | 2.4 | 6.2% | Fentanyl | 2.4 | 5.8% | Ondansetron | 1.9 | 5.5% |
| Omeprazole | 1.6 | 6.5% | Omeprazole | 1.8 | 6.3% | Bisacodyl | 2.1 | 6.0% | Tramadol Hcl | 1.4 | 5.5% | Albuterol Sulfate | 1.5 | 5.4% |
| Potassium Chloride | 1.6 | 5.7% | Oxycodone Hcl | 2.3 | 5.5% | Omeprazole | 1.7 | 5.8% | Oxycodone Hcl | 2.3 | 5.5% | Fentanyl | 2.4 | 5.3% |
| Oxycodone Hcl | 2.2 | 5.2% | Scopolamine | 3.0 | 5.4% | Oxycodone Hcl | 2.3 | 5.6% | Bisacodyl | 2.3 | 5.3% | Oxycodone Hcl | 2.3 | 5.1% |
| Sennosides/Docusate Sodium | 1.6 | 5.1% | Potassium Chloride | 1.7 | 5.3% | Tramadol Hcl | 1.4 | 5.2% | Omeprazole | 1.8 | 5.2% | Gabapentin | 1.8 | 4.9% |
| Alprazolam | 1.8 | 5.0% | Bisacodyl | 2.1 | 5.2% | Potassium Chloride | 1.6 | 5.1% | Ondansetron | 2.0 | 5.1% | Quetiapine Fumarate | 1.9 | 4.8% |
| Dexamethasone | 1.9 | 4.8% | Nystatin | 1.2 | 5.1% | Nystatin | 1.1 | 5.0% | Gabapentin | 1.8 | 4.9% | Bisacodyl | 2.3 | 4.7% |
| Tramadol Hcl | 1.4 | 4.4% | Dexamethasone | 2.0 | 5.0% | Dexamethasone | 2.0 | 4.8% | Potassium Chloride | 1.7 | 4.9% | Omeprazole | 1.7 | 4.6% |
| Lactulose | 1.3 | 4.3% | Alprazolam | 1.9 | 4.8% | Gabapentin | 1.8 | 4.7% | Nystatin | 1.1 | 4.8% | Trazodone Hcl | 1.7 | 4.5% |
| Prednisone | 1.8 | 4.3% | Tramadol Hcl | 1.4 | 4.8% | Promethazine Hcl | 1.7 | 4.6% | Dexamethasone | 2.0 | 4.7% | Potassium Chloride | 1.6 | 4.5% |
| Gabapentin | 1.7 | 4.2% | Promethazine Hcl | 1.7 | 4.7% | Alprazolam | 1.9 | 4.5% | Quetiapine Fumarate | 1.9 | 4.6% | Nystatin | 1.1 | 4.5% |
| Trazodone Hcl | 1.6 | 4.1% | Prochlorperazine | 2.1 | 4.4% | Ondansetron | 2.0 | 4.4% | Trazodone Hcl | 1.6 | 4.5% | Ondansetron Hcl | 1.4 | 4.4% |
| Quetiapine Fumarate | 1.8 | 3.8% | Gabapentin | 1.8 | 4.3% | Trazodone Hcl | 1.6 | 4.4% | Promethazine Hcl | 1.8 | 4.3% | Dexamethasone | 2.0 | 4.3% |
| Temazepam | 1.7 | 3.7% | Prednisone | 1.9 | 4.3% | Polyethylene Glycol 3350 | 1.1 | 4.3% | Polyethylene Glycol 3350 | 1.1 | 4.3% | Polyethylene Glycol 3350 | 1.1 | 4.1% |
| Mirtazapine | 1.6 | 3.7% | Lactulose | 1.4 | 4.3% | Quetiapine Fumarate | 1.9 | 4.3% | Alprazolam | 1.9 | 4.3% | Prednisone | 1.8 | 4.0% |
| Citalopram Hydrobromide | 1.6 | 3.4% | Trazodone Hcl | 1.6 | 4.2% | Prednisone | 1.8 | 4.2% | Prednisone | 1.8 | 4.2% | Promethazine Hcl | 1.7 | 3.8% |
| Sertraline Hcl | 1.7 | 2.9% | Polyethylene Glycol 3350 | 1.1 | 4.2% | Scopolamine | 3.0 | 4.1% | Ondansetron Hcl | 1.4 | 3.9% | Alprazolam | 1.9 | 3.7% |
| Metoprolol Tartrate | 1.7 | 3.0% | Quetiapine Fumarate | 1.9 | 4.0% | Lactulose | 1.4 | 3.9% | Lactulose | 1.4 | 3.8% | Mirtazapine | 1.7 | 3.6% |
Excluding 1st quarter (January 1- March 31 2014) before policy change.
Defined as number of fills divided by the number of days in hospice length of stay per 30 person-days
Categories and appropriateness of medications covered by hospice
There were a total of 2,435 distinct drugs dispensed across all years. The number of unique drugs across years varied: 2014 – 1884; 2015 – 1901; 2016 – 1906; 2017 – 1816; 2018 – 1805. The top 100 drugs accounted for 88-89% of all drug dispensings in each year. Table 4 shows categorization of the 100 most common medications billed to the Medicare hospice benefit in each year, 2014-2018. In 2014, of the top 100 medications submitted to Medicare in that year, 30 were EOL drugs, 42 were chronic disease drugs, 7 were anti-infectives, and 15 met STOPP Frail criteria for potentially inappropriate medications at EOL. These were roughly the same in 2018: 29 were EOL drugs, 43 were chronic disease drugs, 7 were anti-infectives, and 16 met STOPPFrail criteria for potentially inappropriate medications at EOL. There appeared to be a temporary decline in the number of the top 100 medications classified as EOL medications from 2014 (30) to 2015 (26). However, there was no statistically significant change in the proportion of the 100 most common medications billed to hospice that were classified as a chronic disease management medication (p=0.85, Chi-square test of trend).
Table 4.
Categorization of the 100 most common medications billed to Medicare hospice benefit, 2014-2018
| Categorya | 2014 | 2015 | 2016 | 2017 | 2018 |
|---|---|---|---|---|---|
| End-of-Life Essentialb (%) | 30 | 26 | 28 | 29 | 29 |
| Chronic Disease Managementc (%) | 42 | 43 | 44 | 44 | 43 |
| Anti-infectives (%) | 7 | 8 | 6 | 6 | 7 |
| Potential STOPPFraild (%) | 15 | 15 | 15 | 15 | 16 |
Categories not mutually exclusive; columns do not sum to 100
IAHPC Criteria (International Association of Hospice and Palliative Care)14
Chronic disease management medications included antiarrhythmics, antihypertensive, antidiabetic, antidementia, chronic obstructive disease-related, rheumatologic, diuretics, anti-thyroid drugs, antidepressants and antipsychotics that were not included in IAHPC essential medications for palliative care.
STOPPFrail (Screening Tool of Older Persons Prescriptions in Frail)16
DISCUSSION
In this study, we observed a small but statistically significant increase in the mean number of chronic disease medication dispensings per enrollee billed to hospice in the year after the Medicare policy change. There was also a temporary increase in overall dispensings per enrollee billed to hospice, but with a restoration to the level that preceded the policy change in the subsequent years. While it is difficult to fully explain these dynamics based on only hospice data and without Part D, these findings could be interpreted to mean that the 2014 policy change didn't have much, if any, effect in the short or long-term. This could imply that the quick revision that was made by CMS to focus the policy on symptom management medications took effect before hospices had implemented changes to prescribing and billing practices. However, the absence of a shift in symptom management medication prescribing over a longer time period suggests the mechanism by which CMS sought to precipitate the change (i.e., instituting prior authorizations) was largely ineffective. Ineffectiveness may be due to a number of reasons including prior authorizations may not have been fully implemented or were easily bypassed, weakly enforced, or produced insufficient deterrence; further mixed methods research is needed to explore such explanations. Alternatively, the small transient changes we observed could also mean that hospices made modest changes in prescribing in the short term as an immediate reaction to the new policy, particularly by shifting some chronic disease medications unrelated to the terminal diagnosis to hospice.
Prior research examining changes in hospice medication use following the Medicare policy change is scarce. One cross-sectional study quantified the prevalence of various medications used during 2014-2016 using combined Part D and hospice drug information, but did not examine annual trends.18 The prevalence of use reported in the prior study for specific EOL medications including morphine (53%) and lorazepam (56%) were higher than we observed in our study based on hospice claims, likely reflecting ongoing billing for such medications to Part D.
Our analysis of national hospice claims data over 5 years suggests that the 2014 Medicare policy change either had little effect or may have contributed to a small shift of some medication billing away from Part D to the Medicare hospice per diem. This change was most measurable for chronic disease medications covered by Medicare hospice programs. Whether this change was clinically meaningful or had an impact on patient quality of life or symptom management requires further study. Interestingly, chronic disease medications and STOPP-Frail medications together comprised the majority of the 100 most common medications billed to the Medicare hospice benefit. This represents an opportunity for education about appropriate medication use in hospice, regardless of payor. Palliative care and hospice clinicians are in a position to review the role and limited benefits of such medications for patients in hospice, with the goal of trying to align prescribed medications with those that optimize symptom control and minimize unnecessary burden.19
One policy implication is to consider reinstating medication reporting by hospice programs in order to track current prescribing patterns of drugs billed to hospice. This is important because of the emergence of expensive chronic disease medications, both small molecule drugs and biologics,20 that were not widely available when Medicare implemented the 2014 policy. Biologics, which treat conditions such as anemia and diabetes, make up just 2% of all prescriptions, account for 40% of US pharmaceutical sales and 70% of the growth in drug spending.21,22 Further analysis with contemporary combined hospice-Part D claims could help inform the design of a Medicare policy carve out, separate from the hospice benefit, by estimating the cost to cover all medications for all hospice enrollees. A carve out in insurance refers to a specific benefit that is separated from the overall insurance contract and often managed by a different entity, often for specialized areas like behavioral health or pharmacy.23
Our findings must be considered in light of certain limitations. First, these findings are only based on medications reported to hospice. Because medications are also covered by Part D claims, a complete picture of the quality and quantity of prescribing for hospice patients awaits analyses based on a combined hospice and Part D dataset. Second, there was left censoring of data prior to April 1, 2014 because we limited inclusion to the post-policy change period. This analytic decision affected longitudinal variables such as length of stay, but does not affect the overall patterns of use, which are cross-sectional. Mitigating these limitations is the lack of research focused on characterizing the effect of Medicare hospice policy on quality of care delivery.
It remains to be seen whether the creation of a Medicare policy carve out is possible. In the meantime, further analyses using quasi-experimental methods and with linked hospice-Part D dispensing data are needed to understand the extent to which the 2014 policy achieved intended declines in Part D payments, whether such declines were distributed evenly across drug classes, and what if any downstream effects changes may have had on care quality and outcomes. Further, at the hospice level, it will be important to examine whether there is any evidence of “cherry-picking” enrollees based on medication regimen complexity to avoid those with high drug costs. Taken together, there is still much work to do to understand the impact of the 2014 Medicare hospice policy change.
Supplementary Material
Key Message.
This descriptive study was a preliminary examination of patterns of medications paid for by the Medicare hospice program after national policy changes meant to drive payment of medications away from the Medicare Part D program to the hospice per diem.
ACKNOWLEDGEMENTS
This work was supported by the National Institutes of Health [grant numbers K24AG068300; R01AG068450]. Dr. Alcuksy was supported by R01AG068450.
Footnotes
Prior presentations: An earlier version of this analysis was presented at the 2022 International Conference for Pharmacoepidemiology, Copenhagen, Denmark.
DISCLOSURES
Dr. Tjia is a consultant to CVS Health and CVS Omnicare for work unrelated to this study. Other authors have nothing to disclose.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
REFERENCES
- 1.Stevenson J, Abernethy AP, Miller C, Currow DC. Managing comorbidities in patients at the end of life. BMJ. 2004;329(7471):909–912. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Currow DC, Stevenson JP, Abernethy AP, Plummer J, Shelby-James TM. Prescribing in palliative care as death approaches. J Am Geriatr Soc. 2007;55(4):590–595. [DOI] [PubMed] [Google Scholar]
- 3.Cadogan CA, Murphy M, Boland M, Bennett K, McLean S, Hughes C. Prescribing practices, patterns, and potential harms in patients receiving palliative care: A systematic scoping review. Explor Res Clin Soc Pharm. 2021;3:100050. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.McNeil MJ, Kamal AH, Kutner JS, Ritchie CS, Abernethy AP. The Burden of Polypharmacy in Patients Near the End of Life. J Pain Symptom Manage. 2016;51(2):178–183.e172. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Dwyer LL, Lau DT, Shega JW. Medications That Older Adults in Hospice Care in the United States Take, 2007. J Am Geriatr Soc. 2015;63(11):2282–2289. [DOI] [PubMed] [Google Scholar]
- 6.Zueger PM, Holmes HM, Calip GS, Qato DM, Pickard AS, Lee TA. Older Medicare Beneficiaries Frequently Continue Medications with Limited Benefit Following Hospice Admission. J Gen Intern Med. 2019;34(10):2029–2037. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Centers for Medicare and Medicaid Services. Part D Payment for Drugs for Beneficiaries Enrolled in Hospice—Request for Comments 2013. December 06, 2013. Report No.: HHS-0938-2013-F-3444. Available at: https://www.hhs.gov/guidance/document/part-d-payment-drugs-beneficiaries-enrolled-hospice-request-comments. Accessed on: November 3, 2023. [Google Scholar]
- 8.Centers for Medicare and Medicaid Services. Update to Hospice Payment Rates, Hospice Cap, Hospice Wage Index and Hospice Pricer for Fy 2020. Center for Medicare and Medicaid Services; 2019. Available at: https://www.cms.gov/Outreach-and-Education/Medicare-Learning-Network-MLN/MLNMattersArticles/downloads/MM11411.pdf. Accessed on: November 3, 2023. [Google Scholar]
- 9.Office of the Inspector General. Medicare Could Be Paying Twice for Prescription Drugs for Beneficiaries in Hospice. Washington, D.C.: Department of Health and Human Services; 2012. June 2012. Report No.: A-06-10-0059. Available at: https://oig.hhs.gov/oas/reports/region6/61000059.pdf. Accessed on: November 3, 2023. [Google Scholar]
- 10.Larrick A, Wilson L. Part D Payment for Drugs for Beneficiaries Enrolled in Medicare Hospice. Baltimore, MD: Centers for Medicare and Medicaid Services; 2014. Available at: https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/Hospice/Downloads/2014-PartD-Hospice-Guidance-Revised-Memo.pdf. Accessed on: November 3, 2023. [Google Scholar]
- 11.Larrick Chavez-Valdez A, Wilson L. Update on Part D Payment Responsibility for Drugs for Beneficiaries Enrolled in Medicare Hospice. In: Department of Health and Human Services, ed. Baltimore, MD. 2016. [Google Scholar]
- 12.Office of the Inspsector General. Medicare Part D Is Still Paying Millions for Drugs Already Paid for under the Part a Hospice Benefit. Washington, D.C.: Department of Health and Human Services; 2019. August 2019. Report No.: A-06-17-08004. Available at: https://oig.hhs.gov/oas/reports/region6/61708004.asp. Accessed on: November 3, 2023. [Google Scholar]
- 13.Department of Health and Human Services. CMS Manual System Pub 100-04 Medicare Claims Processing. 2017. Available at: https://www.cms.gov/Regulations-and-Guidance/Guidance/Manuals/Internet-Only-Manuals-IOMs-Items/CMS018912. Accessed on: November 4, 2024.
- 14.De Lima L. International Association for Hospice and Palliative Care list of essential medicines for palliative care. Ann Oncol. 2007;18(2):395–399. [DOI] [PubMed] [Google Scholar]
- 15.Harman SM, Walling AM. Palliative care: The last hours and days of life. UpToDate. 2023. Available at: https://www.uptodate.com/contents/palliative-care-the-last-hours-and-days-of-life. Accessed on: November 4, 2024.
- 16.Lavan AH, Gallagher P, Parsons C, O'Mahony D. STOPPFrail (Screening Tool of Older Persons Prescriptions in Frail adults with limited life expectancy): consensus validation. Age Ageing. 2017;46(4):600–607. [DOI] [PubMed] [Google Scholar]
- 17.Gerlach LB, Powell V, Zhang L, Bynum JPW, Maust DT. The Most Common Medications Billed to the Medicare Hospice Benefit Among Hospice Beneficiaries in the U.S. J Pain Symptom Manage. 2021;62(3):e3–e6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Gerlach LB, Kales HC, Kim HM, et al. Prevalence of psychotropic and opioid prescribing among hospice beneficiaries in the United States, 2014-2016. J Am Geriatr Soc. 2021;69(6):1479–1489. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Tjia J, Karakida M, Alcusky M, Furuno JP. Deprescribing in Palliative Care. Expert Rev Clin Pharmacol. 2023;16(5):411–421. doi: 10.1080/17512433.2023.2197592. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Makurvet F. Biologics versus small molecules: Drug costs and patient access. Med Drug Disc. 2020; 9. 100075. 10.1016/j.medidd.2020.100075. [DOI] [Google Scholar]
- 21.Aitken M, Kleinrock M. Medicine Use and Spending in the U.S. A Review of 2017 and Outlook to 2022. IQVIA Institute for Human Data Science. https://www.iqvia.com/-/media/iqvia/pdfs/institute-reports/medicine-use-and-spending-in-the-us-a-review-of-2017-and-outlook-to-2022.pdf. Accessed on 27 February 2025. [Google Scholar]
- 22.Gottlieb S. Remarks from FDA Commissioner Scott Gottlieb, M.D., as prepared for delivery at the Brookings Institution on the release of the FDA’s Biosimilars Action Plan. U.S. Food and Drug Administration, July 18, 2018, Available at: https://www.fda.gov/news-events/press-announcements/remarks-fda-commissioner-scott-gottlieb-md-prepared-delivery brookings-institution-release-fdas Accessed on 27 February 2025. [Google Scholar]
- 23.Blumenthal D, Buntin MB. Carve outs: definition, experience, and choice among candidate conditions. Am J Manag Care. 1998;4 Suppl:SP45–57. [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.