INTRODUCTION

In the wake of insurance coverage reforms and recent marked declines in employment and employer-sponsored insurance associated with the coronavirus disease 2019 (COVID-19) pandemic, US families are undergoing transitions in health insurance at a rapid pace. Such transitions may include losing health insurance coverage or switching plans from employer-sponsored insurance to Medicaid, the Affordable Care Act (ACA) Marketplace, or other types of coverage. The diversity of plans available to the American consumer underscores the importance of assessing individuals’ understanding and navigation of plan features. Health insurance literacy (HIL) encompasses the knowledge of health insurance terms and the application of health insurance concepts. Numerous survey-based studies have found that HIL is low in the general population.1,2 Moreover, lower HIL is more prevalent amongst several populations with a high risk of unmet medical needs: those of lower socioeconomic status,3 racial/ethnic minorities,4 and older adults.5

Low HIL is problematic insofar as it could lead to ineffective and inefficient use of the health care system, resulting in financial and/or medical harm. Early work in the field of HIL has shown that individuals with lower HIL had greater difficulty choosing insurance plans than those with higher HIL,6 whereas those with higher HIL were found to be more capable of making choices on coverage provisions based on their personal values and medical history.7 Moreover, custom-made decision aids to guide plan selection that provided plan education, individualized cost estimates, and assessment of priorities led to increased HIL, decision self-efficacy, and confidence in plan selection when compared to unaided navigation of the health insurance Marketplace.8 However, understanding which insurance plan to purchase does not necessarily ensure effective utilization of the health care system. As one aspect of HIL is understanding how to apply insurance concepts, utilization is an important outcome measure to study.

Longstanding work in the related field of health literacy (which is distinct from HIL in its attention to knowledge about health, rather than health insurance) has demonstrated associations with health care utilization. A 2011 systematic review including 96 primary studies found that low health literacy was associated with suboptimal utilization: specifically, low health literacy was associated with greater emergency department utilization; more hospitalizations; lower utilization of preventative health care; and decreased adherence to prescription medications.9,10

HIL is a distinct form of literacy and self-efficacy from health literacy that may affect health care utilization in different and important ways. With its focus on patients’ understanding of health insurance and health care, HIL encompasses functional knowledge of health plans’ benefits, cost-sharing, and other details that could empower appropriate use of health care. Thus, studying the association between HIL and utilization can elucidate how (or whether) this functional knowledge impacts utilization. Given the increasing number of changes to health insurance and health care in the current era, the time is ripe to review the literature on HIL as an important factor in how, when, and why people utilize health care.

METHODS

We conducted a systematic review of the literature assessing the association between HIL and health care utilization. A search and extraction protocol was developed using guidelines from the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)11 and made publicly available via PROSPERO prior to study initiation.12

Literature Search

The primary search was conducted in Medline (OVID). To develop the search strategy, the team examined relevant articles identified through team expertise and conducted exploratory searches on HIL and health care utilization in Medline. Since a MeSH term for “health insurance literacy” does not exist, the search combined literacy-related MeSH terms (e.g., health knowledge, attitudes, practice, health literacy) and keywords (e.g., confidence, attitude, understanding, ability, self-efficacy) with health insurance MeSH terms and keywords, to broadly capture the HIL literature. Additional database searches were translations of the core Medline search and included the following databases: Scopus (Elsevier), Web of Science (Clarivate), PsycINFO (EBSCOhost), CINAHL Complete (EBSCOhost), and Cochrane Central Register of Controlled Trials (Wiley). Searches were run from database inception to the date of search: August 16, 2019. See Appendix 1 for further details of search terms. Reference lists of included documents were reviewed, and citation tracking of included documents took place in Scopus. Grey literature searching included conference proceedings. There were no restrictions on dates, language, country of origin, or patient population; intended subgroup analyses included the US vs. non-US and adult vs. pediatric populations.

Study Selection

The article titles and abstracts returned by the search were reviewed by three authors (BFY, JEL, MRF) in duplicate to determine if the studies appeared to contain any data on the association between HIL and health care utilization. Inclusion criteria were based on a broad definition of HIL: the knowledge and/or application of health insurance concepts, including laws, regulations, or policies governing health insurance. Any assessments of HIL were included, from validated measures such as the Health Insurance Literacy Measure (HILM)13 to author-created measures developed for specific studies. We also included measures of health literacy broadly so that full-text articles could be screened for potential secondary assessments of HIL.

Utilization was broadly defined as the use of and access to services for the purpose of preventing and curing health problems, promoting health and well-being, or obtaining information about one’s health.14 All measures of health care utilization that met this definition were included (e.g., outpatient visits, emergency department visits, hospitalizations, avoidance of care, medication use/adherence, vaccinations). We did not include measures of health outcomes, health expenditures, quality of care, or medication management (as opposed to use). Disagreements regarding study inclusion or exclusion were resolved through discussion with a fourth reviewer (RT).

Full texts of the articles meeting the above selection criteria were uploaded into the DistillerSR15 online tool and reviewed by two authors (BFY, MRF) to determine if the studies met the inclusion criteria and reported on the appropriate exposure (HIL) and outcome (utilization). At this phase of review (see Fig. 1), qualitative studies (e.g., reports of focus groups) in which the relationship between HIL and utilization did not comprise a key theme or major finding were excluded. Studies in which HIL was treated as the outcome and not the exposure by the authors were excluded, as this was the reverse of the study objective. After full-text review, studies with measures of general health literacy (e.g., REALM, TOFHLA, NVS) but not including HIL were excluded. Studies with self-reported utilization were included, but studies that only assessed the intention or future plans to utilize health care were excluded. Disagreements regarding study inclusion at this phase of the review were resolved through discussion with a third or fourth reviewer (AMS, RT).

Figure 1
figure 1

PRISMA diagram. Flow diagram of literature search, abstract screen, full article assessment for exclusion and inclusion criteria with most common reasons for exclusion detailed. Abbreviations: HIL, health insurance literacy.

Data Extraction and Synthesis

Data were extracted by two authors (BFY, MRF) using DistillerSR. Extracted variables included year, study type, data source, population (including sample size), country, how HIL was assessed, what type of utilization was assessed, how utilization was assessed, and the magnitude and significance of findings (Table 1). Studies were classified as a trial if there was an intervention with prospective randomization. HIL and health care utilization were considered significantly associated if p < 0.05 or if the confidence interval of the comparison measure did not include the null value. Studies that did not find a statistically significant association were included. Risk of bias review and quality assessment was conducted using principles outlined in the Appraisal Tool for Cross-Sectional Studies (AXIS), finding that the included studies met between 75 and 95% of the twenty indicators of quality (Appendix 2).16

Table 1 Patterns of Association Between Health Insurance Literacy and Utilization

Results

Study Characteristics

The initial search returned 3389 deduplicated results (Fig. 1). After reviewing the abstracts, 144 studies advanced to full-text review. Twenty-one studies ultimately met study criteria and were included in the final analysis. Of the twenty-one studies included in the final analysis (with publication dates from 2001 to 2019), three were prospective trials in which HIL was incorporated into an intervention, either as a direct objective of the study (e.g., sending participants reminders that a certain health care service is covered without any out-of-pocket costs)17,18 or as a consequence of the intervention (e.g., assigning case managers to high-risk patients to help them navigate their health insurance and health care).19 Table 1 provides a color-coded summary of results, stratified by the utilization measures; Table 2 provides more granular details on each study, including year, population, and a detailed summary of results. Two of the three interventional trials showed that increased HIL led to increased health care utilization. The remaining eighteen studies were cross-sectional analyses using surveys and/or semi-structured interviews. Most (18 of 21) of the included studies had multivariable adjustment incorporated into their analysis. Only one study was conducted outside of the USA (in Ghana),37 and only one study included a pediatric population,30 obviating the ability to perform subgroup analyses by country of origin or age of patient population. No purely qualitative studies met the final study inclusion criteria because their discussions of the relationship between HIL and utilization were ancillary, rather than constituting a major theme of the study. The most common reason for exclusion was that the study assessed health literacy, rather than HIL, or that it did not report utilization as an outcome (Fig. 1).

Table 2 Detailed Description of Included Studies

Overall, HIL was significantly associated with various types of health care utilization, with 19 of the 21 included studies showing statistically significant findings (Table 1). As Table 2 highlights, there was significant variability across studies in terms of the effect size of these relationships. Significant differences when comparing high HIL individuals to lower HIL individuals ranged from 1.5 to 36.1% absolute differences in health care utilization or medication use, 0.42 to 0.99 odds of delaying or foregoing medical care, and 1.08 to 11.86 greater odds of utilizing health services. A wide variety of utilization outcomes were measured in these studies, ranging from specific services (e.g., pneumococcal vaccination,20 mammography18) to broad measures (e.g., time since last utilization of health care21).

HIL and Preventative/Primary Care

In general, higher levels of HIL were associated with greater utilization of outpatient and preventive care services. Of the 13 studies assessing the association between HIL and utilization of primary care or other preventive services, 10 showed a positive association between higher HIL and greater use of preventive services and 3 showed no significant difference.

HIL and Avoidance of Needed Care

Of the 9 studies assessing the association between HIL and delaying or avoiding care, 8 showed that low HIL was associated with delaying or avoiding care.21,22,23,24,25,26,27,28 For example, in a study of adults with high-deductible health plans in the Kaiser Permanente System, 24% of those who mistakenly thought that their deductible applied to all office visits (when, in fact, preventive care visits had no out-of-pocket costs) said they delayed or avoided a preventive office visit because of cost, while only eight percent of those who correctly understood the cost-sharing scheme did so (OR=3.00).39 Two of these studies also demonstrated a significant association between low HIL and lower rates of medication adherence.22,23

HIL and Acute Care Utilization

Studies assessing the association between HIL and emergency department utilization (n=5) had mixed findings, with three studies showing no significant association,26,27,29 one showing higher utilization,30 and one showing lower utilization.22 Of note, these five studies used five different assessments of HIL. One study, McDonnell et al., also assessed knowledge of broader health insurance policies by measuring participants’ awareness of the federal Emergency Medical Treatment and Labor Act (EMTALA), which requires emergency departments to care for patients regardless of insurance status or ability to pay; in this study, greater knowledge about the EMTALA policy was associated with higher levels of emergency department use.30 Two of three studies assessing utilization of inpatient or surgical care showed that HIL was associated with greater utilization,22,27 and one showed no statistical significance.26

Measures of HIL Used Across Studies

HIL was measured in a variety of ways in the included studies. We identified 21 different ways of measuring HIL across the 21 studies, 19 of which were novel measures created specifically for the study. To date, there is only one validated tool that measures HIL—the Health Insurance Literacy Measure—which assesses subjective confidence and behaviors associated with selecting and utilizing health insurance.13 Published in 2014, the Health Insurance Literacy Measure was used as a measurement tool in only two of the included studies. Most other studies used measures that were created by the authors for the specific purpose of the study. For example, Kneipp et al.19 created a 20-item questionnaire about whether Medicaid covers specific services. Some studies assessed HIL very granularly; for example, Burns, Rosenberg, and Fiore were interested in how tobacco users utilized certain tobacco cessation therapies and assessed subjects’ awareness of insurance coverage for those specific therapies.31

James et al. 32 used two assessments of HIL, an objective measure developed by the Kaiser Family Foundation that assessed fact-based knowledge about insurance coverage (e.g., the definition of a deductible)2 and the Health Insurance Literacy Measure, a more subjective measure described above. The authors found that insurance knowledge (as measured by the Kaiser Family Foundation scale) was not associated with utilization patterns; however, insurance self-efficacy (as measured by the Health Insurance Literacy Measure) was positively associated with utilization. The only other study that utilized the Health Insurance Literacy Measure, Tipirneni et al., found that lower HIL was associated with greater likelihood of delayed or forgone care owing to cost and a lower likelihood of utilizing preventive, but not non-preventive, services.26

Discussion

Our systematic review found that in most studies lower HIL was associated with lower health care utilization or greater avoidance of a wide variety of health care services. However, there is a dearth of literature in this area. Several studies demonstrated that low HIL was associated with underutilization of certain high-value services, including primary care visits, other preventive care visits, and adherence to prescription medication regimens for chronic conditions. Not surprisingly, greater specificity of the HIL measurement often had stronger associations with health care utilization patterns, though this was not consistent across studies. This highlights the need for HIL researchers to determine whether it is more useful to assess HIL in a context-specific way (e.g., knowledge of coverage of a specific service) or as a general skill or behavior that will be more likely to generalize across multiple health care contexts.

Another theme of our systematic review was that HIL may enable cost-conscious navigation of the health care system, and low HIL could be a barrier to effective care navigation. For example, eight of the nine studies that assessed delayed or forgone care found that lower HIL was associated with avoidance of needed care. This suggests that HIL is a key mediator of effective navigation of the many layers of the US health care system.

Our analysis of studies assessing HIL and health care utilization dovetails with other literature assessing the relationship between HIL and navigation of health insurance plan selection and affordability. Studies of factors influencing how people select insurance plans have shown that a lack of cost transparency33 and a perceived lack of reliable information about how to distinguish different insurance plans34 are significant barriers towards selecting a plan. One study found that individuals over the age of 50 with concerns about the affordability of health insurance were more likely to delay or avoid care.35 Thus, improving HIL may enable individuals to exhibit cost-conscious care navigation on two fronts: choosing an appropriately tailored insurance plan and accessing and utilizing the health care system after obtaining insurance. Ideally, this would be facilitated by laws regulating how insurance plans communicate to consumers about the details of their benefits and cost-sharing. The ACA included provisions to improve the transparency of health plans’ cost-sharing, but the implementation of these provisions has been thus far delayed.42

Our findings also beg the question: to improve the use of high-value care, should health care professionals aim to raise the level of HIL among all patients or focus on communicating health insurance terms and concepts in plain language? The intervention studies included in this review focused on the unique details of specific health insurance plans and specific covered services (e.g., education on Medicare’s coverage of the influenza vaccine), which may not translate to broader use of high-value care (e.g., use of primary care vs. the emergency department). In addition, changes in the availability and type of insurance, medication formularies, in-network providers, and out-of-pocket costs of various services may make the implementation of any HIL intervention to encourage specific types of high-value care challenging and short-lived. Further studies could delineate whether HIL performs similarly across all populations and insurance models, or if specific populations may benefit from targeted interventions aimed at increasing specific aspects of HIL.42

As noted earlier, despite the importance of the topic, our review found that research about HIL is limited, both by the number of studies conducted to date and by the number of validated tools available to assess HIL. The substantial majority of studies in the systematic review relied upon a HIL assessment tool that was created by the authors specifically for that study, making comparisons across HIL studies challenging. Several of the reported measures were narrow both in their assessment of HIL (e.g., knowledge of out-of-pocket costs for a specific therapy) and in their examination of health care utilization (e.g., utilization of that specific therapy). A recent study that was published after the date of our search compared two other HIL scales (likelihood of utilization vs. confidence in utilization) and found divergent results in individuals’ delaying care or reporting burdensome medical bills.41 This study highlights the difficulty of developing an assessment tool that captures both the broad concept of HIL and the specific applicability to health insurance plan selection and health care navigation. Even the validated Health Insurance Literacy Measure tool focuses on measuring confidence in HIL (a subjective measure) but lacks a knowledge-based/objective assessment. Development and validation of new comprehensive measures of HIL are needed to advance this emerging and valuable field. Despite these differences in measurement, 19 of the 21 studies reported a statistically significant outcome, although there may be a publication bias that has led other studies with negative or null results to remain unpublished.

Our systematic review was conducted with a pre-specified protocol, comprehensive review of the literature, and coding my multiple reviewers. However, our review was primarily focused on biomedical and health care databases; we did not conduct a search with other types of databases such as those used in legal literature.

CONCLUSION

In summary, the literature addressing the association between HIL and health care utilization is limited and lacks standardized measures to assess HIL. This review of the current literature suggests that low HIL is a barrier to effective utilization of important health care services such as primary care, preventive services, medication adherence, and minimizing delays or avoidance of care for urgent needs. Thus, improving HIL and increasing plain language communication of health insurance plan features at the point of care navigation may be effective strategies to encourage more effective and cost-conscious utilization.