Healthcare Cost Burden and Self-reported Frequency of Depressive/Anxious Feelings in Older Adults

Abstract

Many older adults are burdened by out-of-pocket healthcare costs. Using the 2018–2021 National Health Interview Survey data, we examined healthcare cost burden and its correlates and the associations between healthcare cost burden and depressive/anxious feelings. Of older adults, 11.7% reported healthcare cost burden and 18.4% daily/weekly depressive/anxious feelings. In multivariable analyses, the risk of healthcare cost burden was higher among women, racial/ethnic minorities, those with chronic illnesses, mobility impairment, and those with Medicare Part D, but lower among individuals with Medicare-Medicaid dual eligibility, Medicare Advantage, VA/military insurance, and private insurance. The risk of daily/weekly depressive/anxious feelings was higher among healthcare cost burden reporters (IRR=1.76, 95% CI=1.64–1.88), controlling for sociodemographic and health status characteristics. The COVID-19 pandemic-related delay of or lack of access to medical care was also associated with a higher risk of reporting healthcare cost burden and depression/anxiety. Policies and programs to alleviate healthcare cost burden and ensure adequate access to healthcare services, including treatment for depression/anxiety, are needed.

Introduction

Medical care and prescription drug costs in the U.S. are significantly higher than in other countries in terms of what patients pay for the same medical procedure or drug (Chow et al., 2022; International Federation of Healthcare Plans, 2022; Papanicolas et al., 2018). Although Medicare is a U.S. federal health insurance program for individuals age 65 and older and disabled individuals under age 65, Medicare beneficiaries are also obligated to pay out-of-pocket (OOP) costs--Medicare premiums, deductibles, coinsurance, and/or copayments for covered services--beyond what Medicare covers (Centers for Medicare and Medicaid Services [CMS], 2023a). Those with limited income and assets are eligible for Medicare Savings Programs, sponsored by state Medicaid agencies, that are designed to relieve low-income Medicare beneficiaries’ OOP burden by subsidizing Medicare premiums, deductibles, coinsurance, and/or copayments for covered services (National Council on Aging [NCOA], 2023). Some Medicare plans like Medicare Advantage (Medicare Part C) also have an OOP maximum to protect beneficiaries from excessive financial burden (NCOA, 2023).

Despite Medicare Savings Programs and other protective measures, many older adults with limited income and assets are burdened by OOP costs, as many of these older adults often have multiple chronic conditions requiring ongoing monitoring or treatment. According to a RAND Health study, 82% of older men and 81% of older women, compared to 47% for men and 54% for women in the 45–64 age group, in 2014 had two or more chronic physical or mental conditions (Buttorff et al., 2017). These rates are likely higher among low-income older adults as low socioeconomic status is associated with increased risk for chronic conditions, frailty, and disability (Dugravot et al., 2020; Kivimäki et al., 2020; Scott et al., 2017). The RAND study showed that outpatient visits, prescription drug fillings, emergency department visits, and inpatient hospital stays as well as OOP spending increased as the number of chronic conditions increased (Buttorff et al., 2017). An earlier study found that escalating OOP expenditures in conjunction with the increasing burden of chronic conditions were particularly noticeable among the oldest old and women, although having supplementary health insurance or Medicaid mitigated these expenses (Schoenberg et al., 2007). Studies have also shown that Medicare beneficiaries with cardiovascular disease, cancer, diabetes, and/or chronic lung disease, compared to those without these conditions, tend to incur higher OOP spending, taking away a higher proportion of annual income (Basu & Liu, 2022; Fong, 2019; Narang & Nicholas, 2017; Park et al., 2021).

A significant part of a financial burden from high OOP spending among low-income older adults with chronic conditions derives from prescription drug costs, especially high-cost specialty drugs and during the catastrophic coverage phase of the Medicare Prescription Drug Plans (Medicare Part D) (Doshi et al., 2016; Fendrick et al., 2022; Fong, 2019; Trish et al., 2016; Wang et al., 2021). The Part D Low-Income Subsidy helps people with limited income and resources pay for its premiums, deductibles, coinsurance, and other costs (CMS, 2023b). However, prescription drug cost burden is often a cause for medication underuse or non-adherence and has been found to be significantly higher among African Americans and Hispanics than among non-Hispanic Whites (Lee & Salloum, 2016; Piette et al., 2004; Weaver et al., 2010). A study found that among low-income adult patients who underused medications, two thirds also had cut back on necessities or increased debt (Heisler et al., 2005). Other studies have also found that cognitive impairment, dementia, disability/physical impairment, and fall-related injuries are common causes for high healthcare cost burden in older adults (Faridi et al., 2022; Fendrick et al., 2022; Haddard et al., 2019; Jenkins et al., 2022; Oney et al., 2022; Schoen et al., 2017).

Research has shown significant associations between healthcare cost burden and depressive symptoms. Specifically, a study based on the 2011 Behavioral Risk Factor Surveillance System from 12 states and Puerto Rico showed that prevalence of depressive symptoms was significantly higher among older adults who reported cost as a barrier to seeking healthcare compared to their peers who reported it not being a barrier (Cheruvu & Chiyaka, 2019). This finding is consistent with previous studies that identified low income and limited access to healthcare services due to financial constraints as late-life depression risk factors (Areán & Reynolds, 2005; Vyas & Okereke, 2020). A study based on the 2015–2017 National Health Interview Survey also found that older adults with high OOP expenses reported elevated psychological distress along with material hardship and forgoing needed healthcare (Yabroff et al., 2019).

Healthcare cost burden is also likely a significant risk factor for late-life anxiety, as financial anxiety and worry about affording healthcare have been found to be significantly correlated with general anxiety in a sample of young and middle-aged adults (Jones et al., 2019). Depression and anxiety also tend to be co-morbid in late life (Beattie et al., 2010; Cairney et al., 2008). Yet, little research has been done on the associations between depression and/or anxiety and healthcare cost burden in older adults. Symptoms of depression/anxiety (e.g., anhedonia, excessive worrying) have negative effects on activity levels and self-care (Byers & Yaffe, 2011; Dong et al., 2020; Gulpers et al., 2016; Lenze et al., 2001), resulting in an increased need for healthcare and cost burden. Given the COVID-19 pandemic’s effect on physical and mental health outcomes and economic hardship (DiFuso et al., 2021), it is also important to examine COVID-related variables as potential correlates of healthcare cost burden and its impact on the relationship between healthcare cost burden and depression/anxiety. Compared to younger adults, older adults were more vulnerable to the physical effects of SARS-CoV-2, the virus behind the pandemic, but found to be more resilient mentally (Fields et al., 2022).

In the present study using four years (2018–2021) of U.S. nationally representative, cross-sectional survey data for individuals age 65 and older, we examined the associations between (1) healthcare cost burden reporting and sociodemographics, health status, and health insurance types, and (2) self-reported depressive/anxious feelings and healthcare cost burden. Based on previous study findings, our first hypothesis (H1a) was: Female gender, racial/ethnic minority status, chronic illnesses, mobility impairment, and low income would be associated with a higher likelihood of reporting healthcare cost burden, while Medicare-Medicaid dual eligibility and having other health insurance would be associated with a lower likelihood. There is well-established research on the multifactorial risk factors, especially biologic and social factors, for depression/anxiety in late life (Aziz & Steffens, 2013; Hellwig & Domschke, 2019). However, more research is needed to examine healthcare cost burden as a depression/anxiety risk factor. Hence, our second hypothesis (H2a) was: Healthcare cost burden reporting would be associated with a higher risk of high-frequency (i.e., daily/weekly) depressive/anxious feelings, controlling for sociodemographic factors, chronic illnesses, and mobility impairment. We then examined whether during the pandemic (2020–2021), COVID infection and COVID-related healthcare access problems would also be associated with a higher likelihood of reporting healthcare cost burden (H1b) and depression/anxiety (H2b). This study is important because both delayed and skipped healthcare services due to the cost and depression/anxiety can increase the risks of and contribute to the progression of physical, functional, and cognitive disability.

Materials and Methods

Data and sample

Data came from the 2018–2021 US National Health Interview Survey (NHIS). The annual, cross-sectional NHIS series is the principal source of information on the health, healthcare access, and health behaviors of the civilian, noninstitutionalized population residing within the 50 states and the District of Columbia at the time of the interview (National Center for Health Statistics, 2023). In this study, we used the IPUMS NHIS adult sample dataset in which NHIS variables and data were harmonized across the study years (Blewett et al., 2022), and included 31,829 (7,297 in 2018; 9295 in 2019; 6,360 in 2020; and 8,877 in 2021) sample adults age 65+. NHIS randomly selected one sample adult and one sample child from each family (prior to 2019) or household (since 2019) for face-to-face interviews throughout each year. However, due to the COVID-19 pandemic, all second quarter interviews in 2020 were done via telephone, and interviews from July to December were attempted by telephone first with follow-ups to complete them by personal visit.

Measures

Healthcare cost burden:

Respondents were asked if they, in the past 12 months, (1) delayed medical care due to the cost; (2) delayed getting mental health therapy or counseling due to the cost; (3) had problems paying or were unable to pay medical bills; (4) delayed filling prescription(s) to save money; (5) took less medication to save money; (6) skipped medication doses to save money; (7) needed but could not afford medical care due to the cost; (8) needed but could not afford prescription medicine due to the cost; and (9) needed but could not afford mental health counseling due to the cost. An affirmative answer to any of these questions was coded as an indication of healthcare cost burden (=1 vs. no reported healthcare cost burden=0).

Frequency of depressive/anxious feelings:

Respondents were asked the following questions: (1) “How often do you feel depressed? Would you say daily, weekly, monthly, a few times a year, or never?”; and (2) “How often do you feel worried, nervous or anxious? Would you say daily, weekly, monthly, a few times a year, or never?” In this study, we created a dichotomous variable (daily or weekly frequency of feeling depressed or worried, nervous or anxious=1 [referred to as high frequency] vs. all other=0), as occasional (i.e., once a month or less frequent) depressive/anxious feelings are not considered clinically significant as opposed to daily or weekly frequency. For descriptive purposes only, we also reported antidepressant and anxiolytic medication intake (yes=1, no=0) at the interview time among all respondents and the level of depressive/anxious feelings among those who reported high frequency depressive/anxious feelings. Respondents were asked: “Thinking about the last time you felt depressed, how depressed did you feel?; and “Thinking about the last time you felt worried, nervous or anxious, how would you describe the level of these feelings?” The response categories were “a lot,” “a little,” or “somewhere in between.” A lot or somewhere in between for depression and/or anxiety was coded as the high/medium level versus a little.

Sociodemographic factors:

These were chronological ages 65 through 85+ (NHIS shows chronological ages up to 84 but ages 85 years or older are shown as 85+); gender (female vs. male); race/ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, Asian/Pacific Islander, American Indian/Alaska Native, multi-racial/other); marital status (married/cohabiting, widowed, divorced/separated, never married, missing); education (bachelor’s degree or higher vs. no college degree); and the ratio of family income to poverty threshold (<200%, 200%−388%, 400%−499%, 500+%, missing).

Health status:

These included: (1) Charlson Comorbidity Index score (including cancer, weak/failing kidneys, heart attack, angina pectoris, coronary heart disease, stroke, hypertension, high cholesterol, asthma, COPD, diabetes, liver disease, arthritis, dementia, and age [60–69=2; 70–79=3; 80+=4] factor; Charlson et al., 1987; Deyo et al., 1992); and (2) mobility impairment (i.e., any difficulty walking or climbing steps) (yes=1, no=0).

Health insurance:

NHIS provides data on the types of health insurance for each older-adult respondent. In this study, we examined the following: Medicare-Medicaid dual eligibility, Medicare Advantage program, Medicare Prescription Drug Plan (not counting Medicare Advantage Prescription Drug plans), VA or military (CHAMP, TRICARE) insurance, any private insurance, and no insurance.

COVID-related factors:

In the 2020 and 2021 NHIS, respondents were asked if they were informed by a health professional of having or likely having COVID-19 and tested positive for COVID infection. A positive answer to the COVID diagnosis and/or positive COVID test was coded as having had COVID infection (yes=1, no=0). Respondents were also asked if, because of the COVID-19 pandemic, they were delayed in getting medical care, did not get medical care, or did not receive homecare. An affirmative response to any of these questions was coded as having a COVID-related healthcare access problem (yes=1, no=0).

Analysis

Analyses were conducted with Stata/MP 18’s svy function to account for NHIS’s multistage, area probability sampling design. Stata’s subpop command was used for the 65+ age group to ensure that variance estimates incorporate the full sampling design. All estimates presented are weighted except for sample sizes. We used chi-square (χ²) and t tests to describe sample characteristics by healthcare cost burden and depression/anxiety frequency. We fit four generalized linear models (GLMs) for a Poisson distribution with a log link for hypothesis testing. For H1a and H1b, healthcare cost burden was the dependent variable, with sociodemographic factors, health statuses, and health insurance types as covariates. For H2a and H2b, frequency of depressive/anxious feelings was the dependent variable, with healthcare cost burden as the independent variable and sociodemographic factors and health statuses as covariates. For H1b and H2b, COVID-related factors were also entered as covariates. Survey year was excluded from all four GLMs given its lack of significance. As sensitivity analyses, we fit separate GLMs for H2a and H2b for non-Hispanic White, Black, and Hispanic older adults to examine possible racial/ethnic group differences. However, the results did not vary by race/ethnic group and, thus, were not presented. We fit GLMs rather than logistic regression models as odds ratios exaggerate true relative risk to some degree when the event (weekly/daily feelings of depression/anxiety) is a common (i.e., >10%) occurrence (Grimes & Schulz, 2008). GLM results are reported as adjusted incident risk ratios (IRRs) with 95% CIs. As a preliminary diagnostic, we used variance inflation factor (VIF), using a cut-off of 2.50 (Allison, 2012), from linear regression models to assess multicollinearity among covariates. VIF diagnostics indicated that multicollinearity was not a concern. Statistical significance was set at p<.05.

Results

Characteristics by healthcare cost burden

Table 1 shows that an average of 11.7% of US older adults reported healthcare cost burden during the four study years (11.8% in 2018; 12.8% in 2019; 12.1% in 2020; and 10.3% in 2021). Additional analyses showed no significant difference in years 2018 through 2020 (F(1.99, 2076.69)=1.27, p=0.282), but a significant decrease between 2020 and 2021 (F(1, 592)=7.01, p=0.008). Of older adults who reported healthcare cost burden, 73.2% reported financial problems with paying or inability to pay medical bills, 12.9% reported delayed payment or inability to pay mental healthcare bills, and 36.4% reported delayed, skipped, or skimped medication due to the cost.

Table 1.

Characteristics of older adults by self-reported healthcare cost burden, 2018–2021

N (%)	No Healthcare cost burden 28,420 (88.3%)	Healthcare cost burden 3,409 (11.7%)	p
Types of healthcare cost burden1 (%)
Had problems with paying or was unable to pay medical bills	n/a	73.2
Delayed or was unable to pay mental healthcare bills	n/a	12.9
Delayed, skipped, or skimped medication due to the cost	n/a	36.4
Survey year			<0.001
2018	25.6	25.8
2019	26.0	28.8
2020	20.1	20.9
2021	28.3	24.5
Age, M (SE)	73.9 (0.05)	72.5 (0.13)	0.104
Sex (%)			<0.001
Female	54.2	60.9
Male	45.8	39.1
Race/ethnicity (%)			<0.001
Non-Hispanic White	77.0	65.5
Non-Hispanic Black	8.1	16.1
Hispanic	8.4	12.8
Asian/Pacific Islander	4.9	3.2
American Indian/Alaska Native	0.5	0.6
Multi-racial/other	1.1	1.8
BA/BS or higher degree (%)	20.7	14.2	<0.001
Marital status (%)			<0.001
Married/co-habiting	59.6	50.7
Widowed	12.0	18.9
Divorced/separated	21.2	22.2
Never married	4.7	5.2
Missing	2.5	3.0
Family income to poverty line (%)			<0.001
<200%	25.5	48.5
200%−399%	30.8	31.3
400%−499%	11.6	7.5
500+%	29.2	11.3
Missing	3.0	1.4
Depression/anxiety frequency			<0.001
Daily or weekly	16.1	36.0
Monthly, a few times a year, or never	83.9	64.0
Charlson Comorbidity Index2 score, M (SE)	5.67 (0.02)	6.42 (0.05)	<0.001
Cancer (%)	25.8	25.1	0.508
Weak/failing kidneys (%)	3.7	6.7	<0.001
Any heart disease (%)	17.2	25.5	<0.001
Stroke (%)	7.4	12.3	<0.001
Diabetes (%)	19.0	32.0	<0.001
COPD (%)	9.1	18.4	<0.001
Dementia (%)	2.8	3.8	0.005
Mobility impairment (%)	36.8	57.6	<0.001
Health insurance			<0.001
Medicare-Medicaid dual eligible	6.8	9.0	<0.001
Medicare Advantage3	29.6	32.3	0.007
Medicare Part D4	44.1	48.7	<0.001
VA/military insurance	5.6	3.1	<0.001
Private insurance	41.2	29.4	<0.001
No health insurance	0.5	1.8	<0.001
Had COVID in 2020 or 2021 (% of N=15,237)	5.2	6.4	0.163
COVID-related healthcare delay or lack of access (% of N=15,237)	14.8	26.4	<0.001

Pearson’s χ² tests for categorical variables and t-tests for age and Charlson Comorbidity Index scores were used to examine significant differences between healthcare cost burden reporters and non-reporters.

¹The total percentage is more than 100% because many respondents reported more than one type of healthcare cost burden.

²Including cancer, weak/failing kidneys, heart attack, angina pectoris, coronary heart disease, stroke, hypertension, high cholesterol, asthma, COPD, diabetes, liver disease, arthritis, dementia, and age (60–69; 70–79, and 80+) score.

³The percentages refer to those among individuals with Medicare Part B or individuals who did not know or refused if they had Part B.

⁴The percentages refer Medicare Prescription Drug Plan, not counting Medicare Advantage Prescription Drug plans.

Compared to non-reporters, healthcare cost burden reporters included higher proportions of women, Blacks or Hispanics, widow/ers, and those with low socioeconomic status in terms of education and income. Additional analyses showed that healthcare cost burden reporters were 10.2% of men, 13.0% of women, 10.1% of Non-Hispanic Whites, 20.7% of Blacks, 16.8% of Hispanics, 8.0% of Asians/Pacific Islanders, 14.5% of American Indians/Alaska Natives, and 17.5% of multiracial/other older adults. Compared to 16.1% of non-reporters, 36.0% of healthcare cost burden reporters reported high-frequency depressive/anxious feelings. They also had significantly higher Charlson Comorbidity Index score than non-reporters. Although reporters did not differ from non-reporters on the history of cancer diagnosis, they included significantly higher proportions of those with kidney and heart diseases, stroke, diabetes, COPD, dementia, and mobility impairment. In 2020 and 2021, more reporters reported delayed or lack of access to medical care. Compared to non-reporters, reporters included higher proportions of the Medicare-Medicaid dual eligible and those with Medicare Advantage and Medicare Part D, but lower proportions of those with VA/military and private insurances. Please note that the NHIS percentages for Medicare Advantage and Medicare Part D are lower than the percentages from the general population of Medicare beneficiaries. This is likely due to the fact that (1) NHIS sample did not include nursing home residents; (2) self-reported survey data tend to undercount public program enrollment when compared to administrative enrollment counts; (3) Medicare Advantage in NHIS is for those with Medicare Part B (or those who did not know or refused if they had Part B); and (4) Medicare Prescription Drug Plan did not include Medicare Advantage Prescription Drug plans (email communications with IPUMS NHIS Medicare Part D specialists, 5/20/2023).

Characteristics by frequency of feelings of depression/anxiety

Table 2 shows that 18.4% of all older adults reported experiencing high-frequency depressive/anxious feelings in the past year, without significant difference during the four years (F(2.99, 3221.14)=2.19, p=0.087). Of these older adults, 9.6% reported depression only, 54.1% anxiety only, and 38.3% both depression and anxiety. Compared to 6.0% and 6.5% of those with no/low-frequency depression/anxiety, 30.8% and 35.7% of those with high-frequency depressive/anxious feelings were taking antidepressant and anxiolytic medication, respectively, at the time of the interview. Over two thirds of those with high-frequency depressive/anxious feelings experienced the feelings at high/medium level. Those who had high-frequency depressive/anxious feelings also had significantly higher Charlson Comorbidity Index score and included higher proportions of individuals reporting mobility impairment and delayed or lack of access to healthcare due to the COVID pandemic. They also included a higher proportion of Medicare-Medicaid dual eligible individuals, women, non-Hispanic Whites, widowed or divorced/separated individuals, and those with income<200% of poverty threshold, but a lower proportion of those with VA/military insurance.

Table 2.

Characteristics of older adults by depression/anxiety frequency, 2018–2021

N (%)	None-to-low frequency1 25,910 (81.6%)	High frequency2 5,919 (18.4%)	p
Depression or anxiety (%)			n/a
Depression only		9.6
Anxiety only		54.1
Both depression and anxiety		38.3
Level of depression/anxiety (%)			n/a
A little		31.8
A lot or between a lot and a little		68.2
Antidepressant medication intake (%)	6.0	30.8	<0.001
Antianxiety medication intake (%)	6.5	35.7	<0.001
Healthcare cost burden (%)
Had problems with paying or was unable to pay medical bills	6.9	16.1	<0.001
Delayed or was unable to pay mental healthcare bills	0.5	4.0	<0.001
Delayed, skipped, or skimped medication because of cost	3.1	9.4	<0.001
Any of the above	9.2	22.9	<0.001
Survey year			0.087
2018	26.0	24.2
2019	26.1	27.1
2020	20.2	19.9
2021	27.7	28.8
Age, M (SE)	73.8 (0.05)	73.6 (0.10)	0.087
Sex (%)			<0.001
Female	53.0	63.9
Male	47.0	36.1
Race/ethnicity (%)			<0.001
Non-Hispanic White	75.1	77.8
Non-Hispanic Black	9.3	7.8
Hispanic	8.8	9.4
Asian/Pacific Islander	5.1	3.3
American Indian/Alaska Native	0.5	0.6
Multi-racial/other	1.2	1.2
BA/BS or higher degree (%)	20.7	14.2	<0.001
Marital status (%)			<0.001
Married/co-habiting	59.7	53.2
Widowed	20.6	24.4
Divorced/separated	12.3	15.5
Never married	4.6	5.3
Missing	2.8	1.6
Family income to poverty line (%)			<0.001
<200%	26.7	34.7
200%−399%	30.8	31.1
400%−499%	11.4	9.6
500+%	28.2	22.3
Missing	2.9	2.3
Charlson Comorbidity Index score3, M (SE)	5.61 (0.02)	6.34 (0.04)	<0.001
Mobility impairment (%)	35.0	57.9	<0.001
Health insurance
Medicare-Medicaid dual eligible	6.4	9.6	<0.001
Medicare Advantage4	30.0	29.5	0.574
Medicare Part D5	44.5	45.2	0.417
VA/military insurance	5.5	4.5	0.011
Private insurance	40.1	37.2	<0.001
No health insurance	0.6	0.9	0.175
Had COVID in 2020 or 2021 (% of N=15,237)	5.2	6.0	0.155
COVID-related healthcare delay or lack of access (% of N=15,237)	14.3	24.1	<0.001

Pearson’s χ² tests for categorical variables and t-tests for age and Charlson Comorbidity Index scores were used to examine significant differences between the none-to-low frequency group and the high frequency group.

¹None-to-low-frequency=Never, once a month, or a few times a year

²High frequency=Daily or weekly

³Including cancer, weak/failing kidneys, heart attack, angina pectoris, coronary heart disease, stroke, hypertension, high cholesterol, asthma, COPD, diabetes, liver disease, arthritis, dementia, and age (60–69; 70–79, and 80+) score.

⁴The percentages refer to those among individuals with Medicare part B or individuals who did not know or refused if they had part B.

⁵The percentages refer Medicare Prescription Drug Plan, not counting Medicare Advantage Prescription Drug plans.

Correlates of healthcare cost burden: GLM results

The second column of Table 3 shows that the risk of healthcare cost burden reporting was higher among women, Blacks (IRR=1.49, 95% CI=1.35–1.65), Hispanics (IRR=1.27, 95% CI=1.12–1.44), or multiracial people, and divorced/separated individuals, but lower with advancing ages and those with income ≥200% (compared to income <200%) of poverty threshold. The risk of healthcare cost burden reporting was also higher among those with higher Charlson Comorbidity Index score (IRR=1.12, 95% CI=1.11–1.14), mobility impairment, (IRR=1.65, 95% CI=1.52–1.79), and Medicare Part D (IRR=1.14, 95% CI=1.04–1.20). However, the risk was lower among those who had Medicare-Medicaid dual eligibility (IRR=0.55, 95% CI=0.48–0.64), Medicare Advantage (IRR=0.85, 95% CI=0.78–0.93), VA/military insurance (IRR=0.58, 95% CI=0.46–0.73), or a private insurance (IRR=0.72, 95% CI=0.66–0.80). The third column of Table 3 shows that in 2020 and 2021 (i.e., during the COVID-19 pandemic), the risk of healthcare cost burden reporting was higher among those who also reported delayed or lack of access to medical care (IRR=1.66, 95% CI=1.47–1.88); however, Medicare Advantage was no longer a significant factor. These findings largely support H1a and H1b.

Table 3.

Correlates of healthcare cost burden (vs. no burden) among older adults: GLM results

	Healthcare cost burden vs. no burden, 2018–2021 IRR (95% CI)	Healthcare cost burden vs. no burden, 2020–2021 IRR (95% CI)
Age	0.94 (0.93–0.94)***	0.93 (0.92–0.94)***
Female vs. Male	1.18 (1.09–1.28)***	1.13 (1.00–1.27)ǂ
Race/ethnicity vs. Non-Hispanic White
Non-Hispanic Black	1.49 (1.35–1.65)***	1.43 (1.22–1.67)***
Hispanic	1.27 (1.12–1.44)***	1.23 (1.01–1.50)*
Asian/Pacific Islander	0.80 (0.63–1.02)	0.81 (0.57–1.16)
American Indian/Alaska Native	0.90 (0.57–1.45)	0.98 (0.42–1.28)
Multi-racial/Other	1.37 (1.06–1.78)*	1.28 (0.88–1.86)
Marital status vs. Married/cohabiting
Widowed	0.99 (0.90–1.10)	1.01 (0.87–1.17)
Divorced/separated	1.14 (1.03–1.25)*	1.04 (0.91–1.19)
Never married	0.85 (0.73–1.00)	0.72 (0.58–0.90)**
Missing	1.08 (0.84–1.38)	1.21 (0.90–1.61)
BA/BS degree vs. No degree	1.09 (0.98–1.21)	1.25 (1.06–1.49)**
Income to poverty line vs. <200%
200–399%	0.66 (0.61–0.72)***	0.73 (0.65–0.83)***
400–499%	0.47 (0.40–0.56)***	0.51 (0.41–0.64)***
500+%	0.31 (0.27–0.36)***	0.34 (0.27–0.41)***
Missing	0.35 (0.24–0.50)***	n/a
Charlson Comorbidity Index score	1.12 (1.11–1.14***)	1.13 (1.10–1.16)***
Mobility impairment	1.65 (1.52–1.79)***	1.73 (1.55–1.94)***
Medicare-Medicaid dual eligibility vs. No dual eligibility	0.55 (0.48–0.64)***	0.59 (0.46–0.75)***
Medicare Advantage vs. No Medicare Advantage	0.85 (0.78–0.93)***	0.90 (0.79–1.03)
Medicare Part D vs. No Medicare Part D	1.12 (1.04–1.20)**	1.18 (1.05–1.31)**
VA/military insurance vs. No VA/military insurance	0.58 (0.46–0.73)***	0.58 (0.42–0.82)**
Private insurance vs. No private insurance	0.72 (0.66–0.80)***	0.71 (0.61–0.82)***
Had COVID vs. Did not have COVID		1.02 (0.80–1.31)
Delayed or lack of access to medical care during COVID vs. No delay or lack of access		1.66 (1.47–1.88)***
N	31,829	15,237

^*p<0.05;

^**p<0.01;

^***p<0.001

Associations of frequency of depressive/anxious feelings with healthcare cost burden: GLM results

The second column of Table 4 shows that the risk of high-frequency depressive/anxious feelings was higher among those who reported healthcare cost burden (IRR=1.76, 95% CI=1.64–1.88), controlling for all covariates included in the model. Of the covariates, age, being Black or Asian/pacific Islander, and having income ≥400% of poverty threshold were associated with a lower risk; however, Charlson Comorbidity Index score (IRR=1.09, 95% CI=1.07–1.10), mobility impairment (IRR=1.77, 95% CI=1.67–1.88), female gender (IRR=1.35, 95% CI=1.27–1.43), and divorced/separated state (IRR=1.09, 95% CI=1.01–1.17) were associated with a higher risk of reporting high-frequency depressive/anxious feelings. The third column of Table 4 shows that in 2020 and 2021, the risk of high-frequency depressive/anxious feelings was higher among those who reported delayed or lack of access to medical care (IRR=1.40, 95% CI=1.29–1.52) while healthcare cost burden remained a significant factor (IRR=1.67, 95% CI=1.52–1.84). These findings support H2a and H2b.

Table 4.

Associations of high frequency (weekly/daily) feelings of depression/anxiety with healthcare cost burden: Results from generalized linear models

	High-frequency vs. no/low-frequency feelings in 2018–2021 IRR (95% CI)	High-frequency vs. no/low-frequency feelings in 2020 & 2021 IRR (95% CI)
Healthcare cost burden vs. no burden	1.76 (1.64–1.88)***	1.67 (1.52–1.84)***
Age	0.97 (0.97–0.98)***	0.97 (0.97–0.98)***
Female vs. Male	1.35 (1.27–1.43)***	1.30 (1.20–1.41)***
Race/ethnicity vs. Non-Hispanic White
Non-Hispanic Black	0.69 (0.62–0.76)***	0.69 (0.58–0.80)***
Hispanic	0.95 (0.86–1.05)	0.92 (0.79–1.07)
Asian/Pacific Islander	0.74 (0.62–0.88)**	0.74 (0.58–0.95)*
American Indian/Alaska Native	1.03 (0.78–1.34)	0.75(0.40–1.41)
Multi-racial/Other	0.80 (0.64–0.99)*	0.62 (0.43–0.89)**
Marital status vs. Married/cohabiting
Widowed	1.06 (0.98–1.13)	1.07 (0.97–1.19)
Divorced/separated	1.09 (1.01–1.17)*	1.17 (1.05–1.29)**
Never married	1.11 (0.99–1.24)	1.16 (0.99–1.34)
Missing	0.64 (0.50–0.82)***	0.59 (0.44–0.79)**
BA/BS degree vs. No degree	1.00 (0.94–1.08)	1.06 (0.95–1.18)
Income to poverty line vs. <200%
200–399%	0.95 (0.88–1.01)	0.97 (0.89–1.06)
400–499%	0.89 (0.80–0.98)*	0.93 (0.81–1.07)
500%	0.92 (0.85–0.99)*	0.94 (0.84–1.05)
Missing	0.85 (0.71–1.03)	n/a
Charlson Comorbidity Index score	1.09 (1.07–1.10)***	1.09 (1.07–1.11)***
Mobility impairment vs. No mobility impairment	1.77 (1.67–1.88)***	1.62 (1.48–1.77)***
Had COVID vs. Did not have COVID		1.04 (0.87–1.23)
Delayed or lack of access to medical care during COVID vs. No delay or lack of access		1.40 (1.29–1.52)***
N	31,829	15,237

^*p<0.05;

^**p<0.01;

^***p<0.001

Discussion

In this study, we examined correlates of healthcare cost burden, followed by associations of frequency of depressive/anxious feelings with healthcare cost burden. Of all older adults, 11.7% reported healthcare cost burden. The proportion of healthcare cost burden reporters did not change from 2018 to 2020, but a significant decrease was shown between 2020 and 2021, likely a reflection of reductions in emergency department visits and hospital admissions among older adults during the initial phase of the pandemic (Smulowitz et al., 2021). Regarding depressive/anxious feelings, 18.4% reported high frequency (i.e., daily/weekly); however, among those with healthcare cost burden, 36.0% reported high frequency. Of those who had high-frequency depressive/anxious feelings, two thirds reported depression and over 90% reported anxiety, showing that anxiety, more so than depression, was highly prevalent among older adults reporting healthcare cost burden. The high rates of high-frequency depression/anxiety and the high depression/anxiety severity level among healthcare cost burden reporters are concerning.

Our multivariable analyses showed that the risk of healthcare cost burden was higher among women and Black and Hispanic older adults, which was consistent with previous study findings of prescription drug cost burden and related medication underuse or non-adherence being significantly higher among these older adults (Lee & Salloum, 2016). Multivariable analysis results also showed that, like VA/military and private insurances, Medicare-Medicaid dual eligibility and Medicare Advantage were associated with a lower risk of healthcare cost burden, but Medicare Part D, along with chronic illnesses and mobility impairment, is associated with a higher risk. The latter finding is not surprising given strong evidence on how the Part D “donut hole” coverage gap affects OOP costs and utilization (Park & Martin, 2017). In 2023, under a provision in the Inflation Reduction Act, Part D enrollees will pay no more than $35 per month for covered insulin products in all Part D plans. At the same time, Part D enrollees are responsible for increased OOP costs for the deductible and in the initial coverage phase and must pay more OOP before qualifying for catastrophic coverage in 2023 (KFF, 2022). It is notable that higher proportions of healthcare cost burden reporters than non-reporters had Medicare-Medicaid dual eligibility, Medicare Advantage, and Medicare Part D; however, dual eligibility and Advantage programs were associated with a lower risk of burden reporting whereas Part D was associated with a higher risk.

Our key finding is that healthcare cost burden reporters were 1.7 times more likely to report high-frequency depressive/anxious feelings. Previous studies (Cheruvu & Chiyaka, 2019; Yabroff et al., 2019) showed significant associations of depressive symptoms and psychological distress with healthcare cost burden among older adults. Those who struggle to afford healthcare expenses, especially those with multiple chronic conditions, most likely experience heightened stress, depressed mood, and worry, which in turn can contributed to deterioration of health problems and the need for more frequency healthcare use and added healthcare cost burden. Research has also shown that older adults with depression and anxiety use healthcare services at a higher rate than those without (Buczak-Stec et al., 2017; Porensky et al., 2009). Through negative thinking patterns and heightened sensitivity, depression/anxiety may also have amplified perceived healthcare cost burden.

Black older adults were more likely than their non-Hispanic White peers to report healthcare cost burden, but they were less likely to report depressive/anxious feelings. It is possible that Black older adults were more reluctant than their non-Hispanic White peers to self-report depressive/anxious feelings for cultural reasons and stigma (Conner et al., 2010). A study based on the Centers for Medicare and Medicaid Services Health Outcomes Survey (of Medicare Advantage plan enrollees) showed that after adjusting for covariates, odds of screening positively for depression were higher among Black, Hispanic, and other racial/ethnic minority older adults than among their non-Hispanic White peers (Hooker et al., 2019). This suggests that actual prevalence of depression among minoritized older adults may be higher than the rates based on their self-reports.

Delayed or lack of access to medical care during the COVID-19 pandemic was also a significant factor for reporting healthcare cost burden and high-frequency depressive/anxious feelings, which is not surprising given the pandemic’s impact on physical and mental healthcare delivery (Bhome et al., 2021; Patt et al., 2020). Along with the fear of COVID infection and physical distancing, delayed or lack of access to care likely engendered depression and worry/anxiety about not receiving timely care, potential worsening of health problems, and associated costs, although COVID infection per se was not a significant factor for either healthcare cost burden or depression/anxiety.

The study had some limitations. First, cross-sectional data could show only correlation, not causation. More research with longitudinal data is needed to examine potential bi-directional relationships between depression/anxiety and healthcare cost burden. Second, both healthcare cost burden and depressive/anxiety feelings were self-reported and not based on any validated measures and subjected to recall bias. While the questions about healthcare cost burden were detailed and thorough, the extent of OOP costs was not queried, barring any analysis of how the degree of healthcare cost burden may be associated with depression/anxiety. Depression/anxiety frequency was assessed with a single item. A more detailed measure is needed. Third, though causes of late-life depression/anxiety are multifactorial, social and psychological risk factors (e.g., social isolation and loneliness) were not entered in the model as NHIS did not provide data on them. Fourth, although both 2020 and 2021 NHIS data were used to examine the COVID-19 pandemic’s influence, the 2020 data likely captured only partial COVID-19 effects as the time frame of the healthcare cost burden questions were within the preceding 12 months. Additionally, potential effects of different data collection methods (face-to-face vs. over telephone) in the 2020 NHIS were not factored in.

Both healthcare cost burden and depression/anxiety in late life have significant negative physical, mental, and cognitive health consequences. Foregoing/delaying medical care and/or skipping needed medications along with untreated/inadequately treated depression/anxiety not only increase morbidity and mortality risks but may also accelerate cognitive decline in older adults (Hellwig & Domschke, 2019). On a policy level, our findings underscore the importance of addressing healthcare affordability in policies aimed at improving both physical and mental health of older adults. Policymakers should consider strategies such as expanding access to affordable healthcare coverage, implementing programs to reduce OOP expenses, and facilitating access to mental health treatments. Along with expansion and improvement of Medicare subsidies, especially those for Part D programs, the price of healthcare services needs to be adjusted. A largely for-profit healthcare system and pharmaceutical industry, hospital consolidation, inadequate healthcare industry regulation, and lack of a national healthcare system are culprits for the high healthcare cost in the US (Galvani et al., 2020). According to a study by Galvani et al. (2020), a single-payer, universal healthcare system is likely to lead to a 13% savings in national health-care expenditure, equivalent to more than US $450 billion annually (based on the value of the US$ in 2017). The entire system could be funded with less financial outlay than is incurred by employers and households paying for health-care premiums combined with existing government allocations (Galvani et al., 2020). While support for a single-payer healthcare system has been rising, its political prospect is dim given massive special interest group opposition and other contextual factors (Brown, 2019).

Clinical implications of the findings are that healthcare and social service providers should recognize the interconnection between financial strain associated with physical health problems and mental health in older adults. In particular, social workers for older adults who delay or lack access to healthcare services due to cost burden should assess their depression/anxiety and assist those experiencing depression/anxiety to be able to access pharmacotherapy and/or evidence-based psychosocial interventions. These treatments should be affordable for all older adults regardless of their income levels. Innovative, low-cost psychosocial interventions that are scalable and sustainable for all older adults with depression/anxiety are needed. Care management services to assist low-income older adults’ access to both financial aid services and psychosocial interventions for depression/anxiety are also needed.