Medical visits and mortality among dementia patients during the COVID-19 pandemic compared to rates predicted from 2019

Abstract

Background: During the COVID-19 pandemic, patients with Alzheimer's disease and related dementias (ADRD) were especially vulnerable, and modes of medical care delivery shifted rapidly. This study assessed the impact of the pandemic on care for people with ADRD, examining the use of primary, emergency, and long-term care, as well as deaths due to COVID and to other causes.

Methods: Among 4.2 million beneficiaries aged 66 and older with ADRD in traditional Medicare, monthly deaths and claims for routine care (doctors' office and telehealth visits), inpatient/emergency department (ED) visits, and long-term care facility use from March or June 2020 through December 2022 are compared to monthly rates predicted from January-December 2019 using OLS and logistic/negative binomial regression. Correlation analyses examine the association between excess deaths - due to COVID and non-COVID causes - and changes in care use in the beneficiary's state of residence.

Results: Increased telehealth visits more than offset reduced office visits, with primary care visits increasing overall (by 9 percent from June 2020 onward relative to the predicted rate from 2019, p < .001). Emergency/inpatient visits declined (by 9 percent, p < .001) and long-term care facility use declined, remaining 14% below the 2019 trend from June 2020 onward (p < .001). Both COVID and non-COVID deaths rose, with 231,000 excess deaths (16% above the prediction from 2019), over 80 percent of which were attributable to COVID. Excess deaths were higher among women, non-White patients, those in rural and isolated zip codes, and those with higher social deprivation index scores. States with the largest increases in primary care visits had the lowest excess deaths (correlation -0.49).

Conclusions: Older adults with ADRD had substantial deaths above pre-pandemic projections during the COVID-19 pandemic, 80 percent of which were attributed to COVID-19. Routine care increased overall due to a dramatic increase in telehealth visits, but this was uneven across states, and mortality rates were significantly lower in states with higher than pre-pandemic visits.

Keywords: Alzheimer’s; COVID-19; Deaths; Dementia; Doctor visits; Emergency care; Health care costs; Health care use; Inpatient; Long-term care; Medicare; Mortality; Primary care; Telehealth.

Fig. 1

Monthly trends in actual and predicted utilization and spending for patients with ADRD during the COVID-19 pandemic, March 2020-December 2022. Notes: Dashed line is prediction from a regression model using monthly data from January–December 2019. Data are for patients with ADRD who are age 66 or older and enrolled that month in fee-for-service Medicare Parts A and B, in administrative data from the Centers for Medicare and Medicaid Services (CMS) for a 100% sample of beneficiaries in traditional Medicare. Observations are at the patient-month level: N = 114,638,189 person-month observations; 5,237,349 unique beneficiaries. Predicted utilization is from OLS regression models using monthly data from January–December 2019, controlling for 10-year age-gender groups, race/ethnicity (non-Hispanic Black, Hispanic, Asian/Pacific Islander, other), dual Medicare-Medicaid enrollment, Social Deprivation Index (SDI) score tercile, urban/rural/isolated areas, 24 chronic conditions, month of the year, hospital referral region (HRR), and residence in a skilled nursing facility (SNF) for rehabilitative care in the prior month (except in the long-term care facility regression analysis)

Fig. 2

Monthly trends in actual and predicted mortality rates for patients with ADRD during the COVID-19 pandemic, March 2020-December 2022. Notes: COVID EXCESS: Excess deaths attributed to COVID if there was a COVID diagnosis in an inpatient, skilled nursing facility, or hospice claim in the last 14 days of life. NON-COVID EXCESS: Excess deaths not attributed to COVID (EXCESS minus COVID EXCESS). Data are for patients with ADRD who are age 66 or older and enrolled that month in fee-for-service Medicare Parts A and B, in administrative data from the Centers for Medicare and Medicaid Services (CMS) for a 100% sample of beneficiaries in traditional Medicare. Observations are at the patient-month level: N = 114,638,189 person-month observations; 5,237,349 unique beneficiaries. Predicted deaths are from OLS regression models using monthly data from January–December 2019, controlling for 10-year age-gender groups, race/ethnicity (non-Hispanic Black, Hispanic, Asian/Pacific Islander, other), dual Medicare-Medicaid enrollment, Social Deprivation Index (SDI) score tercile, urban/rural/isolated areas, 24 chronic conditions, month of the year, hospital referral region (HRR), and residence in a skilled nursing facility (SNF) for rehabilitative care in the prior month. Appendix Figure 1b shows results are the same when predicted from a logistic regression model

Fig. 3

Excess deaths by sociodemographic factors and medical conditions for patients with ADRD during the COVID-19 pandemic, June 2020-December 2022. Notes: Dual eligibility for Medicare and Medicaid is used as a proxy for being a low-income beneficiary. Social Deprivation Index (SDI) scores are assigned by the Centers for Medicare and Medicaid Services (CMS) based on ZIP code by linking to 2014-2018 American Community Survey data. The SDI score ranges from 0-100, with a higher SDI score associated with the lower socioeconomic status of the zip code area. We group the SDI into terciles to allow for non-linear effects on utilization and outcomes. Federal Rural-Urban Commuting Area (RUCA) codes from 2010 are assigned by CMS based on census tract. Race/ethnicity are categorized using the Research Triangle Institute race and ethnicity indicator, which identifies more beneficiaries as Hispanic based on surname COVID EXCESS: Excess deaths attributed to COVID if there was a COVID diagnosis in an inpatient, skilled nursing facility, or hospice claim in the last 14 days of life. NON-COVID EXCESS: Excess deaths not attributed to COVID (EXCESS minus COVID EXCESS). Data are for patients with ADRD who are age 66 or older and enrolled that month in fee-for-service Medicare Parts A and B, in administrative data from CMS for a 100% sample of beneficiaries in traditional Medicare. Observations are at the patient-month level: N = 114,638,189 person-month observations; 5,237,349 unique beneficiaries. Predicted deaths are from OLS regression models using monthly data from January–December 2019, controlling for 10-year age-gender groups, race/ethnicity (non-Hispanic Black, Hispanic, Asian/Pacific Islander, other), dual Medicare-Medicaid enrollment, Social Deprivation Index (SDI) score tercile, urban/rural/isolated areas, 24 chronic conditions, month of the year, hospital referral region (HRR), and residence in a skilled nursing facility (SNF) for rehabilitative care in the prior month. Chronic Conditions are from the Centers for Medicare and Medicaid Services (CMS) Data Warehouse (CCW)

Fig. 4

Geographic variation in actual versus predicted routine and acute care utilization and in mortality for patients with ADRD during COVID-19 pandemic, June 2020-December 2022. Note: Data are for patients with ADRD who are age 66 or older and enrolled that month in fee-for-service Medicare Parts A and B, in administrative data from the Centers for Medicare and Medicaid Services (CMS) for a 100% sample of beneficiaries in traditional Medicare. Observations are at the patient-month level: N = 114,638,189 person-month observations; 5,237,349 unique beneficiaries. Predicted utilization and deaths are from OLS regression models controlling for 10-year age-gender groups, race/ethnicity (non-Hispanic Black, Hispanic, Asian/Pacific Islander, other), dual Medicare-Medicaid enrollment, Social Deprivation Index (SDI) score tercile, urban/rural/isolated areas, 24 chronic conditions, month of the year, hospital referral region (HRR), and residence in a skilled nursing facility (SNF) for rehabilitative care in the prior month

Fig. 5

Change in actual versus predicted office or telehealth related to excess mortality for patients with ADRD during the COVID-19 pandemic, June 2020-December 2022. Note: Data are for patients with ADRD who are age 66 or older and enrolled that month in fee-for-service Medicare Parts A and B, in administrative data from the Centers for Medicare and Medicaid Services (CMS) for a 100% sample of beneficiaries in traditional Medicare. Observations are at the patient-month level: N = 114,638,189 person-month observations; 5,237,349 unique beneficiaries. Predicted utilization and deaths are from OLS regression models controlling for 10-year age-gender groups, race/ethnicity (non-Hispanic Black, Hispanic, Asian/Pacific Islander, other), dual Medicare-Medicaid enrollment, Social Deprivation Index (SDI) score tercile, urban/rural/isolated areas, 24 chronic conditions, month of the year, hospital referral region (HRR), and residence in a skilled nursing facility (SNF) for rehabilitative care in the prior month