Association Between Health Insurance Literacy and Avoidance of Health Care Services Owing to Cost

Key Points Question What is the association between health insurance literacy and avoidance of health care services owing to cost? Findings In this US national survey study of 506 insured adults, 29.6% reported having delayed or foregone care because of cost. Higher health insurance literacy was associated with a lower likelihood of delayed or foregone care owing to cost for both preventive and nonpreventive care. Meaning These findings suggest that to improve appropriate use of recommended health care services, including preventive health services, clinicians and policymakers may need to adopt communication strategies that make health insurance concepts accessible to individuals regardless of health insurance literacy and improve consumers’ understanding of services exempt from out-of-pocket costs.


Introduction
Few Americans have a complete understanding of health insurance terms and details. In 2 recent national studies, only 4% to 14% of individuals were able to respond correctly to a set of questions assessing basic understanding of health insurance. 1,2 This lack of understanding was most pronounced among low socioeconomic status, racial/ethnic minority, older, or previously uninsured populations who often have high health needs necessitating use of health care services. [1][2][3][4][5] Knowledge and application of health insurance concepts (ie, health insurance literacy) have become increasingly important for all Americans as a greater number of health plans have complex costsharing features that can change annually, and many previously uninsured individuals may be newly accessing health care after coverage expansion through the Affordable Care Act (ACA). 6,7 Despite this need, there is an overall dearth of studies examining the association between individuals' health insurance literacy and their health and health care use. 8 Although there is a broad amount of literature on the outcome of general health literacy (defined as individuals' ability to understand health information needed to make health care decisions), studies of health insurance literacy are limited. 9,10 Most prior studies on this topic have focused on consumers' ability to select a health insurance plan. [11][12][13][14][15][16][17][18] Less is known about how patients navigate and use health insurance after obtaining insurance. Understanding insurance cost-sharing features, including deductibles, copayments, and coinsurance, may facilitate appropriate care-seeking and may help to prevent individuals from delaying or avoiding needed care owing to costs. In 1 study of Medicare beneficiaries, those who were less familiar with the details of their Medicare coverage were more likely to delay care because of cost, have multiple emergency department visits, and rate their overall health as poorer. 19 Other studies have found that aspects of low health insurance literacy-including lack of knowledge of drug coverage and difficulty estimating copayments-were associated with medication nonadherence 20 and delays or avoidance of outpatient care. 21 Understanding health insurance coverage may be particularly important for patients' decisions about whether to seek preventive services. Preventive care may be perceived as discretionary because it does not address acute care needs or distressing symptoms. Under the ACA, services such as cancer screenings and vaccinations that are recommended by the US Preventive Services Task Force, Advisory Committee on Immunization Practices, Health Resources and Services Administration, or National Academy of Medicine are considered exempt from out-of-pocket payments by consumers. 22 Yet many individuals may remain unaware of this cost-sharing exemption, worry about possible out-of-pocket costs, and consequently delay preventive care. 23,24 This study sought to determine the association between health insurance literacy and avoidance of health care services owing to perceived out-of-pocket costs. Specifically, we aimed to assess delay or avoidance of preventive and nonpreventive health services.

Study Design and Participants
We recruited a national, geographically diverse, nonprobability sample to participate in an online survey using Amazon's Mechanical Turk (MTurk), an online source of participants used frequently in social science research. 25,26 MTurk allows researchers to post online studies that individuals may complete for a small monetary payment. MTurk participants are more representative of the US population than in-person convenience samples, 27 and the method has been demonstrated to provide similar results for psychological and political outcomes as benchmark national samples. 28

Use of Preventive and Nonpreventive Services
Our secondary dependent variable was participants' self-reported use of selected common preventive (influenza vaccination in the past 12 months, cholesterol level check in the past 5 years) 30 and nonpreventive (emergency department visit, hospital admission) services. 29 As above, we considered these items individually and in aggregate for any preventive or any nonpreventive health services use.

Health Insurance Literacy, Health Literacy, and Numeracy
Our key independent variable was health insurance literacy, as assessed by the validated 21-item Health Insurance Literacy Measure (HILM), which is, to our knowledge, the only currently validated measure of health insurance literacy for the general population. 8 Similar to other subjective measures that have been demonstrated to be important in medical decision making (eg, subjective numeracy and health literacy), 31,32 an advantage of the HILM is that, although it is correlated with objective measures of health insurance-related knowledge and skills, it measures motivation to engage in health insurance-related behaviors (eg, information seeking, document literacy), in addition to self-reported ability and knowledge. 8 We assigned a score to each of the 21 items based on confidence in health insurance understanding and navigation (score of 1 was assigned to a response of not at all confident; 2, slightly confident; 3, moderately confident; and 4, very confident) We also report some descriptive statistics based on individuals categorized as lower HILM score vs higher HILM score. We were most interested in how participants with the lowest HILM scores compared with all others, because these would represent individuals for whom a future intervention to improve health insurance literacy would be designed. As a result, participants with HILM scores in the bottom third of the sample were designated as lower HILM (HILM scores, 0-60) and all other participants (HILM scores, 61-84) were designated as higher HILM for these basic descriptive analyses. General health literacy 32 and numeracy 33 were also assessed using previously validated measures. For general health literacy, a 3-question screening assessment was used. Responses were scored on a Likert scale, as previously validated, and the mean was estimated across the 3 items (score range, 1-5). Higher scores indicate greater perceived ability to understand health information.

JAMA Network Open | Health Policy
For numeracy, the 4-item ability subscale of the Subjective Numeracy Scale was used. Responses were scored on a Likert scale and the mean was estimated across the 4 items (score range, 1-6).
Higher scores indicate greater perceived ability and interest in using numbers.

Demographic and Health Characteristics
We assessed demographics (age, sex, race/ethnicity, educational level, income, geographic location), 34 health status, 30 chronic conditions, 35 insurance status, 34 and enrollment in a highdeductible health plan 29 (private health insurance plan with a deductible greater than $1300 for an individual or $2600 for a family), using standard items from established surveys. For income, we used the midpoint of each income category and household size to estimate participants' income as a percentage of the federal poverty level.

Statistical Analysis
We used descriptive statistics to report responses to individual survey items and bivariate and multivariable logistic regression to explore associations between health insurance literacy and our dependent variables (avoidance and use of preventive and nonpreventive health services). In multivariable analyses, we adjusted for age, sex, race/ethnicity, income, educational level, highdeductible health plan, health literacy, numeracy, and chronic health conditions. Adjusted odds ratios (aORs) and 95% CIs were calculated based on these multivariable regression results. Predicted probabilities were also assessed for selected examples by calculating marginal estimates based on multivariable regression results. We also conducted analyses that were stratified by whether the survey participant did or did not have a high-deductible health insurance plan, as those individuals may be responsible for more out-of-pocket costs before insurance starts sharing the costs of care.
Stata, version 13 (StataCorp), was used for all analyses, and a 2-sided P value <.05 was considered statistically significant.

Participant Characteristics
The survey was completed by 506 of the 511 participants who began it (participation rate, 99.0%). 36 Participants had a mean (SD) age of 34 (

Health Insurance Literacy
The mean (SD) HILM score was 63.5 ( Table 2). Because of the low numbers of participants who received each service, we did not conduct statistical analyses for individual services but considered these services in aggregate for any preventive or nonpreventive care in regression analyses.

Association of Health Insurance Literacy With Delayed or Foregone Preventive vs Nonpreventive Care
In multivariable logistic regression analyses, each 12-point increase in HILM score was associated with a lower likelihood of delayed or foregone care owing to cost for preventive care (aOR, costs before the insurance plan covers the cost (the deductible). This is defined as more than $1300 for an individual or $2600 for a family.

Association of Health Insurance Literacy With Use of Preventive vs Nonpreventive Services
We also examined participants' reported use of selected preventive (influenza vaccination, cholesterol level check) and nonpreventive (emergency department visit, hospital admission) services in the full sample of participants (eTable 3 in the Supplement). Each 12-point increase in Summary variables for preventive care and nonpreventive care were considered yes if yes was indicated for any of the services listed below in each section. Statistical analyses for aggregated delayed or foregone preventive care and nonpreventive care outcomes are represented in Table 3.
b Participant eligibility for health care services was defined as cholesterol level check (men aged Ն35 years and women aged Ն45 years; n = 133); colon cancer screening test (participants aged 50-75 years; n = 57); mammogram (women aged 21-65 years; n = 35); Papanicolaou test (women aged 21-65 years; n = 224); all other health care services (all participants; N = 506). The HILM scores dichotomized at 60 into lower HILM (n = 181) and higher HILM (n = 325). The denominator changes for each service, depending on the reporting of eligible respondents.  HILM score was associated with a higher likelihood of preventive services use (aOR, 1.57; 95% CI, 1.28-1.92), but no significant change in nonpreventive services use (aOR, 1.23; 95% CI, 0.93-1.63) ( Table 4). When we compared an individual with an HILM score of 50 with an individual with an HILM score of 75, the participant with the lower HILM score had a 53.1% predicted probability of any preventive services use compared with 74.8% for the participant with the higher HILM score. For nonpreventive services use, the predicted probability was more similar between individuals with a lower (12.1%) and higher (17.0%) HILM score. In sensitivity analyses for all regressions, we additionally examined the results continuously and the results did not change.

Discussion
In this national study of an insured sample, 150 people (29.6%) reported delaying or foregoing health care owing to perceptions of costs. Although participants were overall equally likely to avoid preventive and nonpreventive care owing to cost concerns, those with lower health insurance literacy reported significantly greater avoidance of both preventive and nonpreventive services.
Likewise, participants with lower health insurance literacy were less likely to report use of preventive services. These findings suggest that health insurance literacy is important for patients, not only while selecting a health plan, but also in health care navigation and uptake of recommended health services.
There are several possible explanations for the study's findings. First, it is likely that individuals with lower health insurance literacy may not understand many cost-sharing and cost-reduction features of their health plan, despite the ACA mandate that preventive services be covered without  Regardless of patients' health insurance literacy, these messages should aim to increase their understanding that recommended preventive services are exempt from out-of-pocket costs. Simple advertising may be used by clinicians' practices and pharmacies to draw attention to such services.
In addition, health care professionals could address costs in discussions of health care recommendations. 39,40 Limitations This study should be interpreted in the context of its limitations. First, although the sample was a national group of participants with economic and geographic diversity, it was not a nationally representative sample. The participants were younger and had higher educational attainment than the general population, which is a known limitation of MTurk samples. 41 Younger age and higher educational level could be associated with greater health insurance literacy, yet many of these individuals avoided preventive health care because of perceived costs. In addition, older adults may face even greater challenges navigating health care services owing to lower health insurance literacy and greater health care needs. 3 Second, the HILM measures confidence in understanding and using health insurance, but it does not directly measure knowledge of specific health insurance concepts.
We chose to use HILM for our independent variable as, to our knowledge, it is currently the only validated measure of health insurance literacy available. Third, we cannot definitively determine whether individuals who reported delayed or foregone care would have had a medical need for the health care service. Our analyses considering the full sample in the eligible denominator may underestimate small effects of health insurance literacy on delays of specific types of health care services.

Conclusions
To our knowledge, this study presents some of the first evidence on the importance of considering health insurance literacy in relation to health care navigation and use. We found that patients with lower health insurance literacy had greater avoidance of both nonpreventive and preventive services, despite the ACA cost-sharing exemption for recommended preventive services. Future work should examine potential drivers of foregone care among those with lower health insurance literacy as potential targets for intervention. To improve appropriate use of recommended health care services, including preventive health services, clinicians, health plans, and policymakers should adopt communication strategies that make health insurance concepts accessible to individuals regardless of health insurance literacy and improve consumers' understanding of services exempt from out-of-pocket costs.