Long COVID Among Adults With Pre-Existing Disabilities: Evidence From the 2022 National Health Interview Survey

Literature Review

Long COVID refers to a constellation of symptoms that arise from COVID-19 infection and persist for 3 months or more. Fatigue, postexertional malaise, and cognitive dysfunction (e.g., acute confusion, brain fog, memory issues, slurring of words, and speech and language issues) are common symptoms of Long COVID (Michelen et al., 2021). Long COVID may also cause problems with hearing (Dorobisz et al., 2023; Fancello et al., 2022) or vision (Jadali & Jalil, 2022; Yusef et al., 2021). Long COVID may involve multiple organ systems and may present in a relapsing-remitting pattern or may progress or weaken over time. In some instances, the recovery timeline may exceed 35 weeks (Davis et al., 2021; Michelen et al., 2021) or have severe life-threatening consequences months or years after the initial infection (U.S. Department of Health and Human Services, 2026). Long COVID may cause disability, limiting a person’s ability to perform day-to-day activities and employment, increasing their risk of other health conditions and use of primary and short-term health care services, and having a negative association with household financial stability (Katz et al., 2023). Long COVID can also increase pre-existing mental health conditions (Hovagemyan et al., 2023).

Estimates of the percent of U.S. adults who have experienced Long COVID vary by data source and population of interest. In 2022, approximately 6%–7% of noninstitutionalized U.S. adults ever had Long COVID (Adjaye-Gbewonyo et al., 2023; Ford, 2024). Data from the 2022 National Health Interview Survey (NHIS) suggest that while nearly 7% of U.S. adults ever had Long COVID, only 3.5% were experiencing current symptoms (Adjaye-Gbewonyo et al., 2023). Overall, an estimated 15% of U.S. adults with a prior positive COVID-19 test develop symptoms of Long COVID. Much is still unknown about the biological pathways that increase the risk of Long COVID at an individual level (Krishna et al., 2023). Certain subpopulations, however, including people who are older, females, people with lower levels of education, and people with more comorbidities (e.g., hypertension, obesity), are more likely to experience Long COVID (Cohen & van der Meulen Rodgers, 2024; Durstenfeld et al., 2022; Song & Giuriato, 2023). Adults who have completed a primary vaccination series prior to infection are less likely to report Long COVID symptoms (Perlis et al., 2022).

U.S. adults with disabilities face health disparities, which may have increased their risk of getting COVID, developing Long COVID, or dying from COVID. People with disabilities have an increased risk of chronic conditions (Dixon-Ibarra & Horner-Johnson, 2014), a lack of access to culturally sensitive health care (Kennedy et al., 2017; McBride-Henry et al., 2023), and increased out-of-pocket costs for health care (Goyal et al., 2023). Adults with disabilities who are from other subpopulations that also face health disparities, including persons from racial/ethnic minority groups (Goode et al., 2014; Shenk et al., 2023; Yearby, 2018), persons who live in poverty (Goyal et al., 2023; Price et al., 2018), and persons who identify as nonstraight gender identities (Mulcahy et al., 2022; Ruedas et al., 2021), may face the multiplicative effects of the intersectionality of these disparities.

Our interest here is in expanding the recent literature about the prevalence of Long COVID among people with pre-existing disabilities. Cohen and van der Meulen Rodgers (2024) used cross-sectional data from the U.S. Census Household Pulse Survey to estimate rates of Long COVID among different subpopulations, including persons they defined as having pre-existing disabilities. While they noted that persons with pre-existing disabilities had higher rates of ever having Long COVID compared to persons with disabilities, their definition of persons with pre-existing disabilities likely overlaps with persons experiencing limitations due to Long COVID. They defined persons with pre-existing disabilities as those who responded affirmatively to having one of six limitation questions (ambulatory, cognitive, hearing, independent living, self-care, and vision), which we would argue are not distinct from symptoms experienced by persons having Long COVID.

Hall et al. (2024) used data from the National Survey on Health and Disability (NSHD) to estimate rates of Long COVID among people with pre-existing disabilities. They compared these rates to findings about the general population from the U.S. Census’ Household Pulse Survey (HPS). The NSHD, an online survey that only includes people with disabilities, has the benefit of gathering detailed data about the date of disability onset so that the data can be used to identify people who had disabilities prior to the COVID-19 pandemic. Hall et al. (2024) estimated (from the NSHD) that 10.4% of people with pre-existing disabilities and (from the HPS) and 7.5% of the general population had Long COVID symptoms in 2022. However, the NSHD survey is not nationally representative, is weighted to represent nonelderly adults, and does not include people without disabilities as a possible reference group. Comparing results from that survey to results from the general population surveyed in the HPS (which includes people with and without disabilities) can bias the findings.

As Long COVID can cause disability, it can be difficult to identify, in cross-sectional population-level data, people who had pre-existing disabilities and then were impacted by Long COVID. Many nationally representative surveys that are fielded by federal agencies, for example, use a set of six questions to identify people with disabilities. The NHIS, for example, uses six disability constructs that are included in the Washington Group – Short Set (WG-SS) to measure limitations in ambulation, cognition, communication, hearing, self-care, and vision. Persons responding as having some or a lot of difficulty in these areas are commonly denoted as having a disability. As shown in Table 1, however, Long COVID can cause limitations in each of these areas. Other research attempts to measure the population with disabilities using cross-sectional data, and these questions may be conflating these two populations (Nguyen et al., 2024).

OPEN IN VIEWER

Table 1.

Crosswalk of Long COVID Symptoms With Washington Group – Short Set (WG-SS) Survey Questions.

Limitation	Experienced by people with Long COVID?
Ambulatory	Yes
	(e.g., fatigue, muscle weakness)
	(Davis et al., 2021; Gheorghita et al., 2024;
	Michelen et al., 2021)
Cognitive	Yes
	(e.g., acute confusion, brain fog, memory issues)
	(Davis et al., 2021; Gheorghita et al., 2024;
	Michelen et al., 2021)
Communication	Yes
	(e.g., slurring of words, speech/language issues)
	(Davis et al., 2021)
Hearing	Yes
	(e.g., hearing loss)
	(Dorobisz et al., 2023; Fancello et al., 2022)
Self-care	Yes
	(e.g., limits daily activities)
	(Gheorghita et al., 2024; Katz et al., 2023)
Visual	Yes
	(e.g., corneal nerve fiber loss, vision deterioration)
	(Jadali & Jalil, 2022; Yusef et al., 2021)

We take a different approach in our cross-sectional analysis by using a nationally representative household survey, the NHIS, that includes adults with and without disabilities and by crafting four different unique definitions of disability that can be used to identify persons who had disabilities prior to the pandemic. We estimate the prevalence of ever having Long COVID for four groups of persons with disabilities: (1) Working-age Medicare beneficiaries, (2) Single working-age adults without children who are receiving Social Security Disability Insurance (SSDI) and/or Supplemental Security Income (SSI), (3) People who reported that they had a disability prior to age 22, and (4) People with a Veteran’s Administration (VA) disability rating.

Each of these groups represents people who likely had disabilities before the pandemic. These populations interact with a range of different service systems, which may all need information about how Long COVID is impacting their service populations. Such information can inform policy and practice that can better support the needs of people with disabilities who have Long COVID by increasing access to necessary health care and by considering how programs might need to be adapted for people who are now experiencing Long COVID symptoms. We further explain our rationale and limitations for identifying these groups in “Measures” section.

Data

Conducted by the U.S. Centers for Disease Control and Prevention, the NHIS is a cross-sectional, nationally representative, household interview survey that collects data on health and health care experiences from the civilian noninstitutionalized population residing within the 50 states and the District of Columbia. Each year, the NHIS gathers information from approximately 27,000 adults and over 9,000 children across roughly 35,000 households. For this study, we use data from the 2022 adult sample NHIS file (unweighted n = 27,640), as this was the first year that the NHIS included questions about Long COVID. We restricted our data set to adults aged 18 and older. Data are publicly available and include appropriate sample weights. Cases with missing data were dropped from our analyses. All analyses were conducted in Stata and/or R.

Measures

Within the NHIS, the measurement of Long COVID is a sequential process. First, to determine if a person ever had COVID-19, the NHIS asks individuals if they either underwent testing or were told by a doctor or other health professional that they likely had coronavirus or COVID-19. Next, if individuals answer yes to either of these questions, they are asked about how severe their symptoms were. If a respondent states that they had mild, moderate, or severe symptoms of COVID-19, they are asked: “Did you experience any symptoms lasting 3 months or longer that were not present prior to contracting COVID-19?” Survey interviewers could provide the following language if needed:

Long term symptoms may include tiredness or fatigue, difficulty thinking, concentrating, forgetfulness or memory problems, sometimes referred to as ‘brain fog’, difficulty breathing or shortness of breath, joint or muscle pain, fast-beating or pounding heart (also known as heart palpitations), chest pain, dizziness on standing, depression, anxiety, mood changes.

As stated earlier, we cannot rely on standard disability measures within the NHIS to identify people with pre-existing disabilities who also ever had Long COVID. Instead, we created four different measures of disability that were unrelated to COVID. Our first measure is persons who were working-age Medicare beneficiaries (persons age 18–65 who were receiving Medicare). These are likely people who qualified for Medicare due to receiving disability benefits from the Social Security Administration (SSA), which requires that a beneficiary have a 2-year waiting period before receiving Medicare, and so the disabling condition would have occurred well before the pandemic. Our second measure is single working-age adults without children who were receiving either SSDI or SSI due to disability. For both programs, the application and award process can take a year or more. Restricting this second measure to single adults without children ensured that we were capturing benefit receipt of the single adult (rather than other family members), as benefit receipt is only captured at the family level within the NHIS. Focusing on single adults implies that we are capturing a single adult with a disability that was severe enough to receive SSA disability benefits. Third, we identified a group of people who stated that they had a disabling condition that began before age 22 and restricted analyses using this measure to persons age 25 and older (in 2022) to ensure that these people had a disability prior to the start of the pandemic in 2020. This ensures that the disability or limitations the respondent was referring to were not caused by COVID-19. Fourth, we created a measure of people who had a VA disability rating, which would identify people with a service-connected disability. As of 2022, Long COVID was not considered a service-connected disability.

As a comparison in our initial set of descriptive tables, we include estimates of the proportion (unweighted and weighted) of adults who had disabilities according to the WG-SS questions included in the NHIS. We measured this using the reported level of difficulty respondents report in six functioning domains: cognition (remembering or concentrating), communication (understanding or being understood by others), hearing, mobility (walking or climbing stairs), seeing, and self-care (washing all over, dressing). Persons who responded that they experienced “a lot of difficulty” or “cannot do at all” to at least one limitation question were considered to have a WG-SS disability.

We also included sociodemographic characteristics in our final analyses as covariates. Age was measured as an interval measure. Race was measured as non-Hispanic white or not. Sex was measured as male or female. Gender was measured as straight or not. Educational attainment was measured as having at least a high school education or not. Metropolitan status was measured as living in a large central, large fringe, medium, or small metro area or not. The region was defined as Northeast, North Central/Midwest, South, or West. Overweight was measured as overweight or obese, according to self-reported Body Mass Index categories included in the NHIS. The usual source of health care was measured as having one or more usual sources of care or not. Hypertension was measured as ever having been told that a respondent had hypertension or not. COVID vaccine receipt was measured as having had at least one COVID vaccine.

Analytical Approach

We first present unweighted and weighted sample sizes for the population that ever had Long COVID and the different disability measures. Next, we present a cross-tabulation of the disability measures with ever having Long COVID, testing for differences between groups using chi-square. Finally, we conduct sequential regressions to estimate the association of disability status with the odds of ever having Long COVID. Please note that these analyses do not consider mortality and are conditional on people surviving COVID. The first regression estimates the risk of ever having COVID based on age, race, sex, gender, educational attainment, metropolitan status, region, being overweight, having a usual source of health care, having hypertension, and receipt of any COVID vaccine. The resulting estimates are included as a variable in the second logit, which estimates the odds of Long COVID, controlling for risk of ever having COVID, disability status, and the sociodemographic characteristics included in the first regression.