key: cord-0741205-eajn82w2
authors: Moyo, Patience; Bosco, Elliott; Bardenheier, Barbara H.; Rivera-Hernandez, Maricruz; van Aalst, Robertus; Chit, Ayman; Gravenstein, Stefan; Zullo, Andrew R.
title: Variation in influenza vaccine assessment, receipt, and refusal by the concentration of Medicare Advantage enrollees in U.S. nursing homes
date: 2022-02-11
journal: Vaccine
DOI: 10.1016/j.vaccine.2021.12.069
sha: 744e4845a1817bb0fab363511ce0b2dd6f2665ff
doc_id: 741205
cord_uid: eajn82w2

BACKGROUND: More older adults enrolled in Medicare Advantage (MA) are entering nursing homes (NHs), and MA concentration could affect vaccination rates through shifts in resident characteristics and/or payer-related influences on preventive services use. We investigated whether rates of influenza vaccination and refusal differ across NHs with varying concentrations of MA-enrolled residents. METHODS: We analyzed 2014–2015 Medicare enrollment data and Minimum Data Set clinical assessments linked to NH-level characteristics, star ratings, and county-level MA penetration rates. The independent variable was the percentage of residents enrolled in MA at admission and categorized into three equally-sized groups. We examined three NH-level outcomes including the percentages of residents assessed and appropriately considered for influenza vaccination, received influenza vaccination, and refused influenza vaccination. RESULTS: There were 936,513 long-stay residents in 12,384 NHs. Categories for the prevalence of MA enrollment in NHs were low (0% to 3.3%; n = 4131 NHs), moderate (3.4% to 18.6%; n = 4127 NHs) and high (>18.6%; n = 4126 NHs). Overall, 81.3% of long-stay residents received influenza vaccination and 14.3% refused the vaccine when offered. Adjusting for covariates, influenza vaccination rates among long-stay residents were higher in NHs with moderate (1.70 percentage points [pp], 95% confidence limits [CL]: 1.15 pp, 2.24 pp), or high (3.05 pp, 95% CL: 2.45 pp, 3.66 pp) MA versus the lowest prevalence of MA. Influenza vaccine refusal was lower in NHs with moderate (−3.10 pp, 95% CL: −3.53 pp, −2.68 pp), or high (−4.63 pp, 95% CL: −5.11 pp, −4.15 pp) MA compared with NHs with the lowest prevalence of MA. CONCLUSION: A higher concentration of long-stay NH residents enrolled in MA was associated with greater influenza vaccine receipt and lower vaccine refusal. As MA becomes a larger share of the Medicare program, and more MA beneficiaries enter NHs, decisionmakers need to consider how managed care can be leveraged to improve the delivery of preventive services like influenza vaccinations in NH settings.

Despite Medicare coverage with no out-of-pocket cost to the beneficiary, and CMS vaccination requirements for nursing homes (NHs), influenza vaccination coverage in NHs remains suboptimal [1, 2] . At 73.1%, during the 2018-2019 influenza season, vaccination coverage among adults living in NHs fell short of the Healthy People 2020 goal of 90% [3] . The coronavirus disease 2019 (COVID-19) pandemic has not only placed a spotlight on the vulnerability of NH residents to morbidity and mortality from respiratory infections, but it also has emphasized racial and socioeconomic inequities in patient care and outcomes [4, 5] . Influenza vaccination can reduce influenza severity and prevent hospitalizations to help avoid the compounded effects of influenza and COVID-19 co-circulation on the healthcare system and society [6, 7] . The widespread provision and acceptance of influenza vaccines is central to any effective strategy to mitigate the spread of influenza infection in NHs. Thus, efforts to promote vaccine uptake in NHs require careful consideration of several factors, including resident profiles, facility attributes, and type of health insurance coverage. Enrollment in Medicare Advantage (MA) is growing and is projected to increase to 51% of all Medicare beneficiaries in 2030 from 34% in 2019 [8] . Simultaneously, there is rising enrollment of racial and ethnic minorities in MA, and the health profiles of MA and Traditional Medicare (TM) enrollees are increasingly similar over time [9] [10] [11] [12] [13] [14] . The historical selection of healthier beneficiaries into MA has diminished (if not reversed) because Medicare implemented changes reducing incentives for MA plans to select enrollees with more favorable risk profiles [15] . MA is distinct from TM in that the Centers for Medicare and Medicaid Services (CMS) pay private health insurance plans on a fixed capitated fee to provide health benefits for Medicare beneficiaries. Features of MA plans such as their payment models, coordinated care, and outreach programs urging high-risk members to get vaccinated may encourage screening and preventive care use to prevent costly medical services [16] .

Research among community-dwelling beneficiaries has found higher rates of preventive services use (e.g., mammography screening, annual influenza vaccinations, and cholesterol testing) in MA compared with TM [17] [18] [19] . However, these studies did not consider NH populations and rely on data from more than a decade ago. MA plans and the characteristics of their enrollees have changed over time, as has the population of individuals receiving care in NHs [20, 21] . The composition of MA beneficiaries in a NH and its relationship with the proportion of residents vaccinated has not been characterized. Yet, this is an important lens through which to examine and address gaps in influenza vaccination coverage in NHs, especially as MA plans may be selectively contracting with NHs, such as those that are larger and are part of a chain [22] . Furthermore, the direction of relationship between MA concentration and NH vaccination rates is uncertain. The shifts in the composition of MA enrollees and the payment model that incentivizes preventive services potentially present opposing possibilities for vaccination rates in NHs. A 'financial incentives' hypothesis may suggest that an increased proportion of NH residents enrolled in MA produces higher influenza vaccination rates; participating plans that promote the health of their enrollees in NHs may emphasize preventive services such as influenza vaccinations [23, 24] . Another hypothesis informed by previous research proposes that the racial composition of NHs, based on the percentage of Black residents, contributes to individual-and facility-level variation in vaccination coverage [2, 25, 26] . Then, the proportion of non-White beneficiaries would increase in NHs with more MA enrollees lowering vaccination rates owing to disparities in care. In this context, we aimed to determine how measures of influenza vaccination offer, receipt and refusal differ among NHs with varying concentrations of residents enrolled in MA.

We conducted a national retrospective cohort study of 100% older adult Medicare beneficiaries residing in NHs during the 2014-2015 influenza season (October 1, 2014- March 31, 2015) . We selected this period to identify the study population because it overlaps with the period over which influenza vaccination is entered on the Minimum Data Set (MDS) when received from October 1-March 31. To maximize generalizability, we included all free-standing NHs in the 50 U.S. states, District of Columbia, and Puerto Rico, excluding hospital-based facilities because of significant case-mix and structural differences [27] . We analyzed long-stay (≥100 days) NH residents who were ≥ 65 years of age. The 100-day cutoff is informed by Medicare reimbursement policy covering up to 100 days of post-acute skilled nursing facility (SNF) care during each benefit period [28] . We used Medicare enrollment data combined with MDS version 3.0 clinical assessments, and facility level data. We obtained NH organizational and aggregate resident characteristics from Certification and Survey Provider Enhanced Reports (CASPER) and LTCFocus.org (LTCFocus: Long-Term Care: Facts on Care in the US) data. We used CMS's Nursing Home Compare for overall and domain-specific (staffing, quality, inspections) star rating data.

We determined a beneficiary's status of MA coverage at the time of NH admission using Medicare enrollment data. We calculated the percentage of residents in each NH who were enrolled in MA. We used the rank procedure to create a dummy variable categorizing NHs into tertiles (low, moderate, high) based on their percentage of MA enrollees.

We used the MDS to ascertain influenza vaccination status and reasons for vaccine nonreceipt. Although the study population included residents in a NH between October 1st -March 31st, MDS assessments can be submitted with influenza vaccination during those dates through June 30. Therefore, in line with the Nursing Home Compare influenza vaccination quality measures, we assessed all MDS assessments for eligible residents from October 1, 2014 to June 30, 2015 [29] . To reduce misclassification of vaccination status we counted beneficiaries who received an influenza vaccine outside the NH during the current influenza season as vaccinated. This consideration is particularly relevant for short-stay residents who are more likely to be vaccinated in the hospital or elsewhere compared with long-stay NH residents. We examined three vaccination measures at the NH level: 1) percentage of residents assessed and appropriately provided the seasonal influenza vaccine; 2) percentage of receipt of influenza vaccine; and 3) percentage of refusal of influenza vaccine. The percent of residents appropriately assessed and provided influenza vaccination was defined as the sum of the percent vaccinated (i.e., appropriately provided), percent offered and refused, and the percent not eligible/contraindicated (i.e., appropriately assessed). Other possible reasons for non-vaccination that we do not report due to small cell sizes are "inability to obtain influenza vaccine due to a declared shortage" if vaccine is unavailable at the NH and "none of the above."

Our analysis adjusted for NH-level variables that capture the demographic (age, sex, race/ ethnicity) composition of residents and their physical and clinical attributes (e.g., acuity index, activities of daily living scale, cognitive function scale, and comorbidities including serious mental illness and heart failure) as well as facility structural (e.g., for-profit ownership, bed count, occupancy rates, rurality, payer mix) and quality (overall star rating) characteristics. These were selected based on prior literature and substantive knowledge. The overall star rating is a composite score (ranging from 1 to 5) that takes into account a NH's performance on staffing, health inspections, and care quality measures. We included the Herfindahl-Hirschman index, which measures the concentration of NH beds in a county, as a covariate to account for variation in NH availability. Additionally, we controlled for the county-level MA penetration rate since MA markets vary substantially. MA penetration is defined as the share of Medicare beneficiaries enrolled in MA plans per county. We used MA penetration data from September 2014 which is the month prior to the start of our observation period [30] . We imputed the state average MA penetration rate for counties with missing or suppressed penetration values due to small sample sizes.

We compare the characteristics of NHs with different concentrations of residents with MA coverage. To assess the relationship between MA concentration and influenza vaccination rates, we conducted a linear regression analysis specified to account for clustering of residents within facilities and facilities within counties using the Huber-White sandwich estimator via generalized estimating equations. We specified an unstructured working correlation structure. In the model we included the dummy variable for the NH's MA concentration and the above-described covariates. Variables with a p-value < 0.05 were considered associated with the outcome.

We carried out a stability analysis to determine the robustness of the results by applying an alternate and stricter definition for MA enrollment that required MA coverage during the entire observation period instead of only at admission. This analysis provides information on the extent to which switching from MA to FFS after admission affects the results.

We additionally analyzed short-stay (<100 days) residents as they can be co-located with long-stay residents in NHs and account for an increasing share of NH residents [31] . Therefore, influenza vaccination for short-stay residents has implications for NH-wide efforts to prevent and control the spread of influenza infection. Several reasons warrant the separate analysis of long-stay and short-stay residents. First, prior research found differences in risk factors for influenza infection and outcomes between these groups [32] [33] [34] . Second, there are distinct care goals between short-and long-stay residents [35, 36] . Short-stay residents receive recuperative and rehabilitative skilled nursing care immediately following hospitalization prior to returning home. Whereas, long-stay residents predominantly receive custodial care including assistance with activities of daily living. Finally, there are potential differences in how reliably influenza vaccination status is captured in the MDS depending on the duration of a resident's NH stay. A short-stay resident is more likely to be vaccinated outside the NH and there is a possibility of undercounting if influenza vaccination status is not communicated to the NH upon admission. We present the short-stay results in the online supplementary materials.

Data preparation and analyses were conducted using SAS version 9.4 (SAS Institute, Inc., Cary, NC). The Brown University Institutional Review Board approved this study.

From a national total cohort of 1,690,642 Medicare beneficiaries ≥ 65 years of age, we identified 936,513 long-stay residents living in 12,384 unique Medicare-certified NHs between October 1, 2014 and March 31, 2015 (Table 1) . At the resident level, the overall prevalence of MA enrollment at the time of NH admission was 21.4% among long-stay residents. When NHs were classified into three groups by their prevalence of residents enrolled in MA, the groups were classified at the following thresholds: low MA concentration (0% to 3.3%), moderate MA concentration (3.4% to 18.6%), and high MA concentration (>18.6%). The range for the prevalence of MA in NHs within the highest MA concentration category was 18.64% to 100%. There were 13 NHs, representing 0.3% of 4126 NHs in the highest concentration category, with an MA prevalence more than 90%.

Resident and facility characteristics varied by the prevalence of MA beneficiaries in NHs. As the prevalence of MA-enrolled residents increased, the beneficiaries tended to be older in age, and more racially and ethnically diverse. NHs with the highest prevalence of MA enrollees were more often larger, part of a chain system, and located in urban settings. The resident acuity index varied minimally across categories of MA prevalence. However, NHs with increasing MA prevalence had residents with more limitations in activities of daily living, greater cognitive impairment, but lower levels of serious mental illness than NHs with lower MA prevalence. The majority (89.2%) of NHs with low MA prevalence had a high overall star rating of 4 or 5 compared with about half of NHs in the other MA categories who met the same ratings. The prevalence of MA in NHs was higher for facilities located in counties with greater MA penetration rates.

On average, 96.9% of long-stay residents in NHs with a low prevalence of MA-enrolled residents were assessed and appropriately considered for influenza vaccination compared with 94.7% of residents in NHs with the highest prevalence of MA. While the unadjusted rates of influenza vaccine receipt were similar, vaccine refusal decreased as the prevalence of MA enrollees in a NH increased (Fig. 1) . NH variables that were positively associated with higher rates of appropriate assessment and provision and influenza vaccination included mean age, occupancy rate, high NH quality star rating, percent with serious mental illness, percent paying with Medicaid, and the Herfindahl-Hirschman index. In contrast, for profit and chain ownership and increasing percent of Black residents were associated with decreased assessment and appropriate provision, and influenza vaccination. See Table 3 for the covariate estimates and the corresponding 95% confidence limits.

Changing the definition of MA enrollment yielded substantively similar results to the main analysis. There appeared to be a clearer dose response relationship when MA was defined on the basis of enrollment throughout the entire observation period rather than at the time of admission. See supplementary Table S1.

See supplementary materials (Tables S2-S4 and Figure S1 ) for a summary of the short-stay results.

This study investigated influenza vaccination receipt and nonreceipt among older adults in NHs, and their variation on the basis of the concentration of residents enrolled in MA. We found that although nearly all long-stay residents were assessed and appropriately considered for influenza vaccination (95.5%), influenza vaccine receipt was lower (81.3%) largely due to high refusal rates (13.4%) when the vaccine was offered. Additionally, although crude estimates were similar, in adjusted models we found that as the concentration of MA enrollees increased so did receipt of influenza vaccination among long-stay residents.

Our finding that NHs with a greater share of MA enrollees have higher influenza vaccination coverage rates among long-stay residents are consistent with perspectives that MA plans promote preventive care use. Given that nearly all the attention on MA efforts to improve preventive care use has targeted community-dwelling beneficiaries, the extent to which MA plans conduct health promotion efforts in NHs is unknown. Individual MA plans conduct care coordination and health promotion efforts for their beneficiaries with varying rigor and success. As such, MA beneficiaries may not experience these benefits uniformly as MA plans are not created equal [24] . The processes that MA plans have in place for outreach and education for providers and patients in NH settings deserve attention in efforts to increase vaccination rates. The importance of addressing this knowledge gap is magnified by the growing enrollment in MA [8] , expensive costs of post-acute and long-term care [37] , and high risk of morbidity and mortality due to respiratory infections in NH residents and older adults generally [38, 39] . The COVID-19 pandemic adds further imperative to explore levers (e.g., care coordination and initiatives to promote preventive care) at the MA plan level to improve NH influenza vaccination coverage. While improving uptake of the annual influenza vaccine is a perennial challenge [2] , the availability of a vaccine for COVID-19 means that it will be even more critical to ensure high vaccination rates among NH residents -a population that has experienced disproportionately high rates of COVID-19 cases and deaths [40] .

Since the composition of residents in a NH often includes a mix of post-acute short-stay and long-stay residents [41] , effective influenza mitigation strategies should also target improving the assessment and appropriate provision of the vaccine to short-stay residents [34] . This may require NHs to maintain vaccine supplies over a longer period during influenza season. Doing so could create a low-barrier opportunity for NHs to improve their influenza vaccination performance by extending their efforts to offer and vaccinate short-stay residents. Such targeted efforts could be especially beneficial for NHs with large proportions of short-stay residents. In addition, our results suggest actions to improve overall NH Compare star ratings (targeting 4 or 5 starts) could contribute to better vaccination rates. While the quality domain of star ratings includes NH vaccination coverage, this is unlikely to fully explain the strong independent associations of the overall star rating with vaccination rates in multivariable analyses.

This study has limitations. First, this is a cohort study focusing on a single influenza season (2014) (2015) . Nonetheless, the findings provide foundational evidence that point to the relevance of further investigation through longitudinal and more recent data. Second, we relied on facility-level resident acuity and comorbidity measures from the CASPER database rather than resident-level MDS clinical assessments. However, by using CASPER variables we avoided making assumptions that would be required to handle missing data particularly for short-stay residents who more frequently have missing information on MDS-derived variables. In addition, our findings may not generalize to beneficiaries younger than 65 years, residing in the community, or with insurance coverage other than Medicare.

In conclusion, this study found that higher concentration of MA beneficiaries in NHs is associated with increased rates of influenza vaccination receipt among long-stay residents after adjusting for covariates. Vaccine refusal when offered was lower as the prevalence of long-stay MA beneficiaries increased. As the MA program continues to grow and more MAenrolled beneficiaries enter NHs, concerted efforts by MA plans and NHs will be essential to improve influenza vaccination rates and reduce vaccine refusals. This importance is magnified in the COVID-19 era when mitigating the transmission of respiratory infections is of critical importance for the health of NH residents and staff.

Refer to Web version on PubMed Central for supplementary material.

This work was supported by a grant to Brown University from Sanofi Pasteur (project identifier: 005705). The authors' institution retained the right to publish and publicly present all results. Sanofi Pasteur was not involved in establishing the scope of the study, creating the initial protocol, designing the study, or performing analysis, but was involved in suggesting edits to the final study protocol and reviewing the final manuscript. Incorporation of any edits suggested by Sanofi Pasteur was not compulsory. Dr. Zullo is also supported, in part, by grants from the National Institute on Aging (R21AG061632 and R01AG065722) and National Institute of General Medical Sciences (U54GM115677).

Unadjusted vaccination receipt and non-receipt by Medicare Advantage concentration among long-stay residents, 2014-2015. Moyo Table 3 Estimates from multivariable regression models of vaccination rates among long-stay nursing home residents, 2014-2015. Vaccine. Author manuscript; available in PMC 2022 March 12.

Trends in Influenza and Pneumococcal Vaccination Among US Nursing Home Residents, 2006-2014

Persistence of Racial Inequities in Receipt of Influenza Vaccination Among Nursing Home Residents in the United States

Influenza Season Nursing Home Vaccination Coverage Dashboard

COVID-19 Preparedness in Nursing Homes in the Midst of the Pandemic

COVID-19 Preparedness in US Home Health Care Agencies

Flu Vaccination Urged During COVID-19 Pandemic

Influenza immunization and COVID-19

Kaiser Family Foundation. A dozen facts about Medicare Advantage in 2020

Dx for a careful approach to moving dual-eligible beneficiaries into managed care plans

Medicare Advantage Enrollees More Likely To Enter Lower-Quality Nursing Homes Compared To Fee-For-Service Enrollees

How successful is Medicare Advantage?

How Much Favorable Selection Is Left in Medicare Advantage?

High-Cost Patients Had Substantial Rates Of Leaving Medicare Advantage And Joining Traditional Medicare

Adjusted Mortality Rates Are Lower For Medicare Advantage Than Traditional Medicare, But The Rates Converge Over Time

Steps to reduce favorable risk selection in medicare advantage largely succeeded, boding well for health insurance exchanges

Anthem encourages high-risk members to get a flu shot

Racial and ethnic differences in use of mammography between Medicare Advantage and traditional Medicare

The effect of Medicare Advantage enrollment on mammographic screening

Preventive services among Medicare beneficiaries with supplemental coverage versus HMO enrollees, medicaid recipients, and elders with no additional coverage

Thirty-Year Trends in Nursing Home Composition and Quality Since the Passage of the Omnibus Reconciliation Act

Effects of Caps on Cost Sharing for Skilled Nursing Facility Services in Medicare Advantage Plans

Medicare Advantage enrollees' use of nursing homes: trends and nursing home characteristics

Managed care for long-stay nursing home residents: an evaluation of Institutional Special Needs Plans

Variation in Screening Mammography Rates Among Medicare Advantage Plans

Racial inequities in receipt of influenza vaccination among long-term care residents within and between facilities in Michigan

Racial inequities in receipt of influenza vaccination among nursing home residents in the United States

Cost and case-mix differences between hospital-based and freestanding nursing homes

Medicare coverage of skilled nursing facility care

Centers for Medicare and Medicaid Services Nursing Home Quality Measures

Medicare Advantage State/ County Penetration

Trends in Post-Acute Care Use Among Medicare Beneficiaries

Long-term Care Facility Variation in the Incidence of Pneumonia and Influenza

Geographic Variation in Pneumonia and Influenza in Long-Term Care Facilities: A National Study

Risk factors for pneumonia and influenza hospitalizations in long-term care facility residents: a retrospective cohort study

Nursing home care in the USA

Short-and long-term residents of nursing homes

Medicare postacute care payment reforms have potential to improve efficiency of care, but may need changes to cut costs

Influenzaand RSV-associated hospitalizations among adults

The burden of respiratory infections among older adults in long-term care: a systematic review

Nursing Home Care in Crisis in the Wake of COVID-19

Nursing Homes That Increased The Proportion Of Medicare Days Saw Gains In Quality Outcomes For Long-Stay Residents

Residents with schizophrenia and/or bipolar disorder, mean (sd) % 9

Herfindal index of nursing home beds in county, mean (sd)

MA penetration in NH county, mean (sd) %

Nursing Home Quality Characteristics High overall Nursing Home Compare rating, 4-5 stars

Overall star rating (range 1-5 stars), mean (sd)

Inspection/survey rating, mean (sd)

LTCFocus: Long-Term Care: Facts on Care in the US, ADL: Activities of Daily Living -on a scale of 0 to 28, higher scores indicate more limitations in these activities, CFS: cognitive function score -lower scores indicate less cognitive impairment. * The Herfindal index is a measure of the size of any entity in relation to the industry in which those entities operate (i.e., market concentration)

We thank Kevin McConeghy for assistance in preparing the original project data.