Variation in Public Support for Government Action on Unexpected Medical Bills

Abstract

Context: Nearly half of the adults in the United States have received an unexpected medical bill in recent years. While government, provider, and insurance policies related to unexpected medical expenses receive attention in the media, this study focuses on variation in public support.

Methods: The study employs two multifactor survey vignette experiments to detect how different features of common health care scenarios that result in costly medical expenses influence the public's sympathy for the patient, perceived fairness of the medical costs, and demand for government action.

Findings: The results point to out-of-pocket cost, severity of the treatment, and the patient's insurance situation as important for public opinion. The public is significantly more supportive of government action when the costs are high and out of the patient's control; in contrast, respondents are generally less sympathetic toward patients described as uninsured or who seek out more costly providers.

Conclusions: The findings underscore the sensitivity of health care attitudes to framing effects, which may occur when media choose how to cover health care costs. The results also point to a potential mismatch in legislation that narrowly addresses “surprise billing,” with public support for government addressing disproportionate costs across a broader range of scenarios.

When does the public demand government action in health care? In December 2020, President Trump signed into law a version of the No Surprises Act, a bipartisan law that seeks to protect patients from unexpected and often costly “surprise medical bills” that result from unknowingly receiving care from out-of-network providers. In 2021, the Biden administration issued rules for the implementation of these protections, and the rules became active in January 2022. This federal law passed after multiple years of congressional debate and after passage of state-level laws in 32 states that sought to curb surprise billing at least in part (Adler et al. 2021; Kliff and Sanger-Katz 2021). One reason that this health care reform eventually crossed the finish line in a hyperpolarized Congress was that the public overwhelmingly supports these protections (Callaghan, Haeder, and Sylvester 2021). A Kaiser Family Foundation poll from December 2020 found that 80% of US adults favored “prohibiting health care providers from charging high out-of-network prices when patients are hospitalized or need emergency care” (KFF 2020). Surprise medical bills violate the public's fundamental sense of what is fair, a key aspect of what drives perceptions of health care services (Gollust and Lynch 2011; Lynch and Gollust 2010): why should a patient have to pay for a service they had no ability to agree to receive?

However, surprise medical bills represent just one subset of a broader category of costly and unexpected medical expenses that the public encounters on a regular basis. Patients may also receive costly medical expenses as a result of the full or partial denial of insurance coverage, because they were charged more than they expected for a procedure or visit to a provider, or because an unanticipated medical event led to a cost beyond what they had foreseen. Individuals without insurance are particularly vulnerable to costly expenses as a result of unexpected illness or injury.

To understand what types of medical expenses violate a sense of fairness, increase sympathy for patients, and drive demand for government action, this study examines how the public reacts to different health care situations involving unexpected medical expenses. The research design includes two multifactor survey vignette experiments conducted via Amazon's Mechanical Turk and Qualtrics, where respondents evaluated vignettes in which a person was described as having a sudden medical event for which they visited a hospital and received a medical bill. The vignette features varied randomly to represent a wide range of scenarios that the public might encounter personally or through depictions in media coverage related to medical bills.

While each experimental scenario is described as “sudden” and plausibly unexpected as a medical event, the scenarios varied the demographics of the patient, the out-of-pocket cost of the medical bill, the severity of the treatment required, and the respondent's insurance situation. The scenarios encompassed situations where the respondent did not have insurance, had insurance that covered part of the cost, or had insurance but the insurance did not cover the cost for a reason that was specified in the scenario. These reasons include “surprise billing” scenarios where a patient is charged because they saw an out-of-network provider, which are now covered under the No Surprises Act, as well as other situations where insurance denies part or all of the coverage—situations that may continue to arise in the current health care system because of disagreements about what insurance companies view as medically necessary or situations in which an insured patient requires care not fully covered by their chosen insurance plan.

The findings reveal the types of scenarios for which demand for government action is greatest. In particular, the size of the out-of-pocket cost is the most consistent attribute driving variation in public support. Respondents tend to be substantially more sympathetic to patients and demand government action when a medical bill is at least $10,000. This amount is well within the range of patient liability that scholars have found in studying the issue of surprise billing and unexpected medical costs (Cooper and Morton 2016; Kliff 2019).

Consistent with theories regarding how agency and deservingness influence health care attitudes, the patient's insurance situation and the severity of the treatment also influence opinions. Specifically, those without insurance are typically viewed less favorably, particularly when the vignette suggests an individual had some agency over their insurance situation. Notably, situations that represent the more formal definition of “surprise medical billing”—in which patients receive additional costs for unknowingly seeing an out-of-network provider—are not uniquely more likely to be viewed sympathetically or receive more support for government action compared to other situations where a person is denied coverage or incurs extreme costs.

The results have implications for scholars of health care public opinion and groups working on health care reform to understand the broader conditions under which the public supports or opposes government action on medical expenses as well as for journalists who report on health care issues. In short, it matters how political elites and the media frame the issue by choosing which patients, insurance situations, and severity of costs to make salient. The findings show that the types of patient stories highlighted in descriptions of health care costs have a direct impact on public perceptions of the fairness of health care costs, sympathy for the individuals who receive these costs, and, ultimately, support for government action to curtail costly and unexpected medical expenses.

Surprise Billing

The issue of surprise medical expenses has received a great deal of attention in recent years among the health policy community and health policy journalists, as more than 30 states passed laws to address the issue. Garmon and Chartock (2016: 177) define a “surprise” medical bill as “a bill from an out-of-network provider that was not expected by the patient or that came from an out-of-network provider not chosen by the patient.” The patient could potentially be charged for the full cost of the visit or the difference in or balance of the cost charged by the physician. Before recent federal legislation, scholars estimated about one-fifth of emergency department visits resulted in such a bill (Cooper and Morton 2016; Garmon and Chartock 2016). In a data set from 2014–2015, Cooper and Morton (2016) find that the average potential additional cost of seeing an out-of-network provider was $600, with an upper “additional cost” (beyond the insurer's rate) of nearly $20,000. Even at the lower bound, any degree of unexpected medical expense could have consequences. More than 40% of insured adults between 18 and 64 years reported in February 2020 that they would not be able to pay an unexpected $500 medical bill in full at the time of services (Lopes et al. 2020).

In late 2020, Congress passed and President Trump signed the No Surprises Act into law following years of debate and lobbying from insurance companies and health care provider groups (Callaghan, Haeder, and Sylvester 2021; Hoadley and Lucia 2022; Sanger-Katz, Creswell, and Abelson 2019). The law, which took effect in January 2022, makes illegal the practice of balance billing, where patients are left holding the bag when a health provider and the patient's insurance disagree about billing. Instead, patients are to be charged the equivalent of an in-network rate, and an arbitration process should handle disputes about the payment amount between providers and insurance companies (Adler et al. 2021). Patients can only receive out-of-network charges in the event that they have given consent to receive out-of-network services and a choice of an in-network provider exists (Hoadley and Lucia 2022).

Surprise medical bills from out-of-network providers stood as an egregious example of a gap in the US health care market that can seem to violate a fundamental sense of fairness among the public. The public broadly supported federal action on surprise billing in 2019 before passage of the No Surprises Act, with support particularly strong among those who reported that they had personal experience with a surprise bill (Callaghan, Haeder, and Sylvester 2021). As Secretary of Health and Human Services Xavier Becerra said, “Getting rid of surprise billing is like removing the boogeyman from your nighttime sleep” (Kliff and Sanger-Katz 2021). Patients could receive care from and be charged by an out-of-network provider while they were quite literally unconscious and had no agency in selecting their providers.

Unexpected Medical Bills

That said, surprise medical bills are just one subset of the types of costly—and often unexpected—medical expenses that consumers can receive. As one commenter observed, “When prices are opaque, every bill, both from in-network and out-of-network providers, is a surprise” (online appendix figure A1). Individuals may receive unexpected costly medical expenses for a number of reasons that are beyond the purview of the No Surprises Act, such as when a health care provider charges more than is expected for a procedure or visit, an insurance company partially or totally denies coverage for any number of reasons beyond seeing an out-of-network provider,¹ or an unanticipated medical event leads to a cost that is outside one's typical level of household spending.² About half of the insured adults who report receiving what they view as an “unexpected medical bill” in the past two years received the bill for a reason other than seeing an out-of-network provider (KFF 2019). For example, more than 40% of adults ages 19–64 were considered to have inadequate insurance in 2020, the criteria for which include high deductibles that can lead to large out-of-pocket expenses, even if insurance covers a service (Collins, Gunja, and Aboulafia 2020).

The recently passed No Surprises Act does not address these broader health care costs that affect the insured and uninsured. The uninsured may be particularly vulnerable to costly medical expenses, given that any bills will not be offset by insurance, leading to especially high out-of-pocket costs. Moreover, even within the No Surprises Act, patients may still incur costly out-of-pocket expenses for certain out-of-network services, such as ground ambulance services, and in situations where patients consciously choose to see a particular out-of-network specialist (Hoadley and Lucia 2022). While the public was supportive of government action on surprise billing (Callaghan, Haeder, and Sylvester 2021), it is less clear in which situations the public believes the government should protect patients from out-of-pocket medical costs more generally.

The Role of Public Opinion

A focus on public opinion on unexpected medical bills is important for several reasons. First, unexpected medical bills—and policies that address them—are an area of policy that has the potential to have a direct, tangible impact on the lives of American adults. In the February 2020 KFF poll, about two-thirds of insured adults reported being at least somewhat worried about being able to afford an unexpected medical bill, surpassing concerns about affording transportation, food, housing, and prescription drug costs (Lopes et al. 2020). Second, it is a topic recently on the federal-level political agenda, as the Biden administration manages the rollout of the No Surprises Act and addresses costs related to the COVID-19 pandemic. It remains unclear how the costs for testing and treatment for COVID-19 will be addressed over the long term. Ending one form of surprise billing does not put all health-cost concerns at ease.

Third, there is evidence that elites are sensitive to public opinion—or at least to negative publicity—about issues related to unexpected medical bills. In 2017, Sarah Kliff and colleagues at Vox began a project to collect medical bills received by patients across the United States after emergency department visits. Between October 2017 and May 2019, they collected more than 2,000 medical bills (Kliff 2019). In response to articles and podcasts where Kliff and others discussed specific medical bills with high emergency department costs, many providers and insurers reduced or eliminated the patient's liability in the coverage. Kliff (2019: 1458) reported that the “series has led to the cancelation of more than $60,000 in medical debt.” Kaiser Health News and NPR have continued to highlight exorbitant medical prices in the ongoing “Bill of the Month” series and provide more recent examples of situations where the patient's burden was removed only after media intervention (Appleby 2022).

Framing of Unexpected Health Care Costs

In addition to media attention leading to downstream consequences for policy making, there are strong theoretical reasons to believe that how public views on potential government action are malleable depending on the dynamics of a particular situation as covered in the media. Unexpected medical bills come in many forms, and the public may be more or less prone to call for government action depending on which of those forms are highlighted in the public discourse. This study focuses on evaluating the sensitivity of public opinion to how the issue of unexpected medical costs is framed by varying considerations related to fairness and deservingness.

The attitudes that the public expresses on a range of issues are generally based on a combination of an individual's predispositions (e.g., partisanship) and a sample of the considerations that are most salient to the individual at a given time (Zaller 1992). Health care in the United States is a highly partisan issue, with Republicans showing greater opposition to proposed interventions from the federal government, such as a public option for health insurance or the Affordable Care Act (Henderson and Hillygus 2011; Jacobs and Mettler 2018; Kriner and Reeves 2014; Schlesinger 2014). Partisanship influences opinions especially when people are asked about broad policy interventions, such as “Obamacare” or “health care reform” (Jacobs and Mettler 2018). Opinions may be more malleable to situation-specific circumstances when one is asked about more narrow policies where partisan cues are less salient (Schlesinger 2014).

Framing refers to subtle variations in the description of issues and events that can influence how people think about the issue (Druckman 2001; Gollust, Fowler, and Niederdeppe 2019). In covering a current event or issue, media professionals generally have significant discretion regarding which information to highlight, and this matters for public opinion. For example, in studying public perceptions of violent incidents, Huff and Kertzer (2018) find a difference of more than 50% in the likelihood that the public considers a particular incident reported in the news to be terrorism, just based on how the story is framed. Frames can influence how individuals come to define an issue or event, “articulate its causes,” “offer moral judgments,” or “suggest what might be done to address it” (Gollust, Fowler, and Niederdeppe 2019).

Variations in the descriptions of situations where people incur medical expenses may influence which considerations are brought to mind when an individual thinks about the health care system and whether the government should get involved. Many news stories that cover unexpected medical bills focus on “episodic” individual experiences (Iyengar 1996). For example, the Kaiser Health News and NPR “Bill of the Month” series explains specific patient medical bills. The focus on individual experiences centers perceptions of the individual's culpability or “deservingness” in attitude formation (Iyengar 1996). The choice of which individuals and types of experiences to highlight may influence support for government action.

Perceptions of deservingness have shaped public opinion in several policy areas (Jensen and Petersen 2017; van Oorschot 2000). When it comes to health, Gollust and Lynch (2011) and Lynch and Gollust (2010) find that descriptions of health and health care disparities that emphasize fairness (or a lack of fairness) and deservingness drive support for government intervention. The authors find that the public generally conceives of fairness more in terms of equal opportunity and equal treatment than in terms of equal outcomes. Scenarios where the public can attribute poor access to health care and health outcomes to the failures of an individual or specific social group are associated with less support for government involvement. In contrast, scenarios involving those who appear “deserving” or have encountered structural barriers or health difficulties beyond their control are associated with a greater willingness to demand societal support and government involvement. In addition to individual culpability, perceptions of fairness are anchored in the status quo of the health care system (Schlesinger 2014). A sense of unfairness is activated when situations violate societal norms and the perceived institutional status quo (Immergut and Schneider 2020).

In this study, I apply previous research on framing and the role of fairness in public opinion to hypothesize and test how the American public will respond to different situations that lead to unexpected and costly medical bills. These situations vary in terms of the out-of-pocket cost a patient is liable for as a result of an emergency department visit, the type of medical event and severity of treatment required, and a respondent's insurance situation. These are attributes that could appear in news stories about patient experiences in the health care system, influencing which types of considerations are brought to mind when people form and express opinions on government policies to address health care costs. The empirical expectations are as follows:

Cost

The expectation is that perceptions of fairness and demand for government action will be associated with the size of the out-of-pocket cost, as it is a concrete metric of severity. Extreme expenses may violate societal expectations of what medical care should cost. This may be particularly true when the cost is disproportionate to the treatment. For example, Kliff and colleagues reported on a $629 bill for a quick visit that required only a Band-Aid, exceeding the patient's expectations for what would be a reasonable cost (Kliff 2019).

Insurance situation

An important aspect of fairness and deservingness that comes out of previous scholarship is whether the individual contributed to (and can be assigned blame for) their situation. The US system has several cost barriers to accessing high-quality insurance coverage. In this type of system, people are less inclined to perceive this unequal access as “unfair” (Immergut and Schneider 2020). In this way, those who remain uninsured in the US system may garner less sympathy for incurring medical costs relative to those who have acquired health coverage. The empirical expectation is that those with health insurance will be viewed more sympathetically than the uninsured. Situations involving those who are insured will be associated with greater demand for government action and a sense that expenses are unfair.

Agency

In addition, I anticipate that the public will have a particularly favorable view of insured patients in cases where an insurance company denied coverage or paid less than expected, as this represents a situation outside the individual's immediate control. Likewise, I expect situations in which a patient is charged as a result of unintentional visits to out-of-network providers (surprise bills), similar to situations covered under the No Surprises Act, will generate more support for government action than situations in which insurance is functioning normally. In contrast, situations that ascribe more agency to individuals in the choice of insurance or provider will result in less support for government action to address associated medical expenses.

The public may be more predisposed to want government action to address health-related issues relative to other areas because of the distinctive character of “health needs” (Schlesinger 2014; Schlesinger and Lee 1993). For example, Jensen and Petersen (2017) argue that the public is predisposed to view sickness and physical injuries as the consequence of uncontrollable causes relative to issues of financial difficulties, such as unemployment. As a result, the public tends to be more likely to view the sick and injured as deserving of government aid. However, not all illnesses are viewed as equivalently “uncontrollable” (Schlesinger 2014) or deserving of government help (Jensen and Petersen 2017), and the public views health inequalities as less unfair when the responsibility is perceived at the individual level (Lynch and Gollust 2010).

Data and Methods

To test these expectations, two original survey experiments were designed to assess which features of a health care situation make respondents more or less supportive of patients and of government involvement. The experimental approach maximizes control over which information is presented to respondents and how it is presented to isolate the causal effects of specific features. The scenarios are designed to be realistic and similar (although more stripped down) to what a person may encounter when reading an article about someone who received a medical bill. Each scenario begins with a sudden (and plausibly unexpected) medical event. The scenarios are then designed to touch on a broad range of situations where patients receive medical bills with an out-of-pocket cost, including surprise medical bills from out-of-network providers, situations where insured patients are partially or totally denied coverage, and situations where an uninsured patient encounters an expense, among others.

Study 1

The first survey experiment was fielded on Amazon's Mechanical Turk (MTurk) in summer 2019, with 507 respondents. MTurk is a nonprobability convenience sample. While this MTurk sample is younger and more Democratic than the national population, the respondents still reflect a diverse sample of the adult population in the United States. MTurk samples are typically more representative than other convenience samples, such as student samples (Berinsky, Huber, and Lenz 2012), and scholars have shown that experiments conducted on MTurk replicate in other samples (Coppock 2019).

The experimental design is a multifactor or “conjoint” vignette experiment, where each respondent is shown brief descriptions of five health care scenarios.³ With 507 respondents × 5 scenarios, the total number of observations is 2,535.⁴ (See online table A1 for sample demographics and population benchmarks for comparison.) Each scenario describes a person who experiences an unexpected medical event, receives treatment, and incurs some degree of out-of-pocket expense. The age, race, gender, type of medical event, type of treatment, and cost of the treatment vary randomly in each description and independently so from one another.⁵

The scenarios also vary attributes related to the context and agency of the individual, cueing considerations related to fairness and deservingness: whether the respondent “was rushed” or “decided to go” to a hospital as well as the individual's insurance coverage. Table 1 displays all attributes and levels in the experiment. An example of a scenario is below, with each randomly varying component of the scenario placed in square brackets:

Study 2

The second study was fielded online from October 10 to October 23, 2019, through Qualtrics, an online survey firm, with 1,211 adults. Qualtrics employs quota sampling to achieve samples that are close to representative of the US adult population and uses quality checks to remove respondents who speed through survey completion. (See online table A2 for sample demographics and population benchmarks.) The experimental design on Qualtrics was very similar to the MTurk sample, with a small number of adjustments to expand on variation in a subject's insurance situation. Each respondent in the Qualtrics sample responded to six independently randomized scenarios, for a sample size of 6 × 1,211 = 7,266 observations. Table 1 summarizes the attributes and levels in each experimental design.

While the MTurk experimental design varied agency as a separate attribute, the Qualtrics study holds constant that the subject “was taken to a hospital.” Instead, the scenario varies the reasons why a subject does not have health insurance, suggesting more or less agency by the respondent in making this decision. For example, reasons include “deciding the premiums were too expensive,” “previously never felt the need to purchase health insurance,” and “recently lost health insurance after changing jobs.” The experiment also includes levels in the insurance attribute to distinguish whether a respondent chose a particular out-of-network hospital or was unknowingly visiting an out-of-network hospital or provider.

Outcomes

After every scenario in both surveys, respondents were asked their opinions about whether the patient described should pay the bill, the perceived fairness of the out-of-pocket cost, the amount of sympathy the respondent had for the patient, and whether the type of situation was something that the federal government should address.

Payment

Specifically, respondents were asked if the patient should pay the out-of-pocket cost in full, appeal the amount, or refuse to pay the cost. This is treated as a 3-point scale, from 0 = pay the cost in full to 1 = refuse to pay.

Lack of fairness

Respondents were then asked if the out-of-pocket cost was a fair outcome financially. Response options were coded on a 4-point scale, from 0 to 1: 0 = more than a fair outcome for the patient in this situation; a fair outcome for the patient in this situation; not that fair of an outcome for the patient in this situation; not fair at all to the patient in this situation = 1.

Sympathy

The question on respondent sympathy was “To what extent do you feel sympathy for the patient in this situation?” Response options were coded on a 4-point scale and included 1 = a lot of sympathy, some sympathy, a little sympathy, no sympathy = 0.

Government action

Respondents were asked “Should the federal government take action to protect patients from having to pay the out-of-pocket costs for care received in this situation, or is this not something the federal government should take action on?” Responses were coded on a 4-point scale, from definitely not = 0, probably not, probably yes, and definitely yes, the federal government should take action to address this type of situation = 1.⁶

Results

Overall, the sample is generally favorable of government action to address out-of-pocket costs across a wide range of situations. Pooling across scenarios and respondents (figure 1), in more than 70% of observations in study 1 and about two-thirds of observations in study 2, respondents indicate the government probably or definitely should take action to protect patients from the out-of-pocket costs. However, there remains variation, with respondents indicating the government should definitely not act in at least 10% of observations in each study.

The first section of results evaluates sources of this variation brought about by the randomized features of the scenarios. Treatment effects for each attribute level are calculated by computing average marginal component effects as the quantity of interest. This quantity represents the average marginal effect of a given level of an attribute (e.g., a $10,000 out-of-pocket cost) relative to a reference level (e.g., a $500 out-of-pocket cost), marginalizing over the joint distribution of the other attributes.⁷ Because all outcomes are on a 0-to-1 scale, the effects can be interpreted as percentage-point increases or decreases in movement on the scale brought about by counterfactually changing an attribute level from the reference point to a given alternative level. Figures 2 and 3 present the average marginal component effects for the full sample of respondents for each of the four main outcomes. The reference level for each attribute is represented as a point estimate at 0. To account for the sample having multiple observations per respondent, the analysis follows the predominant practice in the literature (Hainmueller, Hopkins, and Yamamoto 2014) to use linear regressions with clustered standard errors at the respondent level.

In both study 1 and study 2, across outcomes, the most consistent attribute in explaining variation in opinion is the amount of the out-of-pocket cost. Compared to a medical bill of $500, at higher costs respondents are significantly more likely to oppose a patient paying for the full out-of-pocket cost and to believe that the cost is unfair. Respondents also have significantly more sympathy for patients with higher out-of-pocket costs and are more likely to call for government action. There is also a significant difference between $1,000 and $10,000 out-of-pocket costs, but the cost effect appears to plateau somewhat after a bill reaches $10,000.

Cost is not the only consideration that matters. Respondents also tend to be sensitive to the insurance situation described in the scenario. In study 1, the reference level is a respondent who does not have insurance and pays the full cost. Compared to those without insurance, respondents are more favorable toward patients for whom insurance companies denied insurance coverage with regard to outcomes of whether the patient should pay the full cost, the perceived fairness of the situation, and amount of sympathy for the patient. Likewise, respondents have significantly more sympathy for and perceive the out-of-pocket cost as significantly less fair for patients who had insurance but saw an out-of-network provider (a typical definition of a “surprise” medical bill covered by the No Surprises Act). In study 1, there is no significant variation by insurance situation in driving support for government action. In addition, subtle variation in describing the agency of the person in selecting care—being “rushed to the hospital” versus deciding “to go to the hospital”—has no significant impact.

The scenarios in study 2 provide more nuance in describing a person's insurance situation as an alternative way to cue the agency of the person involved. Study 1 did not specify the reason an individual did not have insurance, allowing respondents to fill in this information on their own. Why someone is uninsured matters to respondents. Individuals described as not having insurance because they previously did not feel the need or felt premiums were too expensive received significantly less sympathy relative to those who had insurance that covered part of the expenses or for whom the reason for having no insurance was not described. Respondents are particularly more likely to see out-of-pocket costs as fair and less likely to support government action when a person is described as uninsured because they did not feel the need to have insurance.

Likewise, while those who unknowingly visited an out-of-network hospital or provider generated significantly higher levels of sympathy and demand for government action, similar to the findings of study 1, in study 2 those who were described as having chosen an out-of-network hospital received significantly less sympathy, and the respondents saw significantly less of a need for government action in those cases. This distinction reflects how the No Surprises Act works: those who wish to use an out-of-network provider or facility can consent to do so but will not be afforded the same financial protections. The legislation distinguishes situations where patients have more or less agency, and public opinion does the same.⁸

In study 2, the results also show an effect for situations where an insurance company has denied coverage for a medical expense not only for outcomes on perceived fairness and sympathy but also for demand for government action. The size of this effect is in line with situations where patients are unknowingly charged by out-of-network providers and facilities. This result speaks to the limitations of government interventions, such as the No Surprises Act, that only address specific situations in which patients receive unexpected medical costs. While respondents deem insurance coverage denials to be just as unfair as “surprise” charges from out-of-network providers, these situations would likely be regulated differently.

The demographic attributes in the scenario tend to matter less, on average, including negligible effects for the gender, age, and race of the person described. Respondents are somewhat responsive to the type of medical treatment, believing out-of-pocket costs to be more fair and to be the patient's responsibility when the treatment involved surgery relative to a short office visit.

The results next assess how the cost and insurance situation attributes described in the scenario work together in creating demand for government action, focusing on study 2, where the insurance situations are described in more detail. Following Leeper, Hobolt, and Tilley (2019), figure 4 displays the marginal means of demand for government action for a given cost and insurance combination.⁹ Overall, the two attributes primarily work in an additive way. As the out-of-pocket cost described grows larger from the lower to higher cost descriptions, demand for government action tends also to grow slightly larger within each insurance situation. While this is generally true, the public remains most skeptical of government action in cases where a person displayed explicit agency by choosing an out-of-network provider or by being uninsured because they did not feel the need for insurance. In addition, as the cost grows toward $20,000, respondents become significantly more in favor of government action in cases where insurance has covered part of the out-of-pocket expense. Even in those cases where insurance provides some coverage, which are typical of how patients share costs with insurance within the current health care system, at high costs the public demands government action at similar rates as in cases where government has enacted protections against costs incurred as the result of a person unknowingly visiting an out-of-network doctor.

Respondent Partisanship and Demographics

As discussed in the theoretical section, political opinions are generally a mix of salient considerations, which may be influenced by framing, as well as underlying respondent characteristics, such as partisanship, which has had a major influence on broader health care attitudes. While the previous analyses focus on framing through the attributes varied in the vignettes, in supplemental analyses in online appendix tables A5 and A6 the outcomes are regressed on the scenario attributes and on a set of respondent characteristics to assess how support for the patient and government action varies across respondents.

The results show that older respondents and those with higher incomes tend to be less supportive of government action, while those who have a previous experience of being charged more than they expected by a health care provider are more supportive of government action.¹⁰ Overall, however, respondent partisanship is the most consistent predictor of support for government action, with Democrats being significantly more supportive of government action than Republicans, marginalizing over the distribution of scenarios. Democrats are also more likely to view the patient as sympathetic and the cost as unfair, and to oppose the patient paying the bill.

One possible reason for the baseline lower support for government intervention among Republicans in these scenarios is that the question asked about the federal government. If Republicans prefer more decentralized government involvement in health care, the lower support may be reflective of this preference and not a general preference against government action of any type. After the experiment was completed in study 2, respondents were also asked separately their opinions on whether the government should pass a law on surprise medical billing.¹¹ The No Surprises Act had not been passed at the time of the survey. Among Democrats, 82% said that the federal government should pass a law, while 16% indicated it should be left to individual states. Among Republicans, 57% indicated the federal government should pass a law, and 31% indicated it should be left to states. Thus, it is the case that Republicans are more likely than Democrats to prefer to leave the decision to states in at least one subset of unexpected billing scenarios, but this does not fully explain the partisan gap in support for government action. Among Republicans, 12% indicated neither the federal nor state governments should act, compared to just 2% of Democrats.

Subsequent analyses further examine the extent to which the partisan gap varies across scenarios in the experiments by interacting the insurance vignette attribute from study 2 with respondent partisanship. Figure 5 displays marginal means for Democrats and Republicans (including leaners) for each outcome. Across scenarios, Democrats typically perceive the out-of-pocket cost as less fair than Republicans, are more likely to oppose the patient paying the full cost, and are more likely to have sympathy for the patient and support government action.

Within the experimental design, there tend to be few interactive relationships between respondent partisanship and the insurance situation, with some exceptions. The partisan gap is smaller in cases where a person chose an out-of-network hospital. Compared to this situation, the partisan gap is significantly larger in scenarios where a person was uninsured because of a job change or no specified reason as well as in cases where a person unknowingly saw an out-of-network doctor. There, Democrats are particularly more likely than Republicans to desire federal government action. Overall, the mean for Republicans on government action is still typically around or above 0.5, suggesting that Republicans generally tend to be weakly supportive of government action. Figure 6 similarly examines partisan differences by the cost of the scenario. Here, the partisan gap remains relatively stable. Both partisan groups show stronger calls for government action at higher levels of out-of-pocket costs, demonstrating the malleability of both partisan subgroups' opinions to variations in the scenarios highlighted.

Conclusion

The experimental results suggest the public is generally supportive of government action to address unexpected medical costs, and the degree of support is sensitive to small changes in the features of the health care scenarios that lead to these unexpected medical expenses. The results have several implications for the character of public opinion, media coverage of unexpected medical bills, and policy making.

The findings point first to the importance of the amount of the out-of-pocket cost in shaping public opinion. Cost represents the clearest indicator of the severity of the impact of the medical bill on a person, and the public is responsive to this indicator. Respondents are also more willing to accept out-of-pocket costs for situations that require significant medical treatment than for those that require only minor treatment. When the cost appears greatly disproportionate, respondents are more likely to demand government action. The experimental results show that just that $500 difference between the levels can influence perceptions of the situation and calls for government action. Meanwhile, the $10,000 and $20,000 levels—where support for government action is greatest—still represent a range of realistic levels of out-of-pocket costs a patient may be liable for paying. These results suggest that highlighting the prevalence of more extreme and disproportionate levels of out-of-pocket costs in stories of unexpected medical bills has the potential to result in significantly higher levels of support for government action from the public across partisan subgroups.

The context of the expense also matters. Situations where a patient receives a medical bill for unknowingly seeing an out-of-network provider are also typically among those more likely to garner sympathy and demand for government action, but these “surprise billing” situations do not appear uniquely more likely to result in demand for government action compared to other situations where the patient is insured but denied coverage. While the survey was administered before passage of the No Surprises Act, the law centers on addressing visits to out-of-network providers more narrowly and does not address the wider range of unexpected costs studied in the survey, where patients are denied insurance coverage or otherwise left with a cost perceived as unreasonably high. The findings speak to a potential gap between the legislation that has passed and public support for government action, pointing to possible avenues for expanding health cost protections where patients encounter especially large medical bills as a result of disputes with insurance companies, high deductibles, or services required outside one's plan.

Beyond cost, features suggesting more or less agency of the individual also matter. Situations where the patient is described as being uninsured are associated with opinion that is less sympathetic toward the patient and less demanding of government action, especially when the lack of insurance appears to be a choice. These results are aligned with studies that show deservingness matters for demand for societal or governmental support: uninsured patients are on average seen as slightly less sympathetic than insured patients in similar scenarios. Opinions about fairness are anchored in the status quo of the health care system (Immergut and Schneider 2020). In a status quo environment where individuals must overcome cost barriers and purchase health insurance coverage, a “failure to do so” garners less sympathy. These findings suggest that media stories that highlight costs the uninsured face are less likely to lead to demand for government action relative to stories that highlight the costs that those who are insured may still confront in the current health care system. Future research can also examine if the public distinguishes between those with private insurance and those with Medicaid or other forms of insurance through the government.

The experiments do have limitations. First, the vignettes differ from episodic media accounts of unexpected and costly medical bills in that they do not include as much description of the patient's personal background or details of the event. As a result, the vignettes may not elicit as much of an emotional reaction or be as memorable as a full-length news story of someone's experience (Gollust, Fowler, and Niederdeppe 2019; Iyengar 1996), which could influence levels of sympathy for those in the scenarios. Second, the experimental features included only represent a subset of the universe of health situations that lead to unexpected medical expenses. In particular, the surveys were conducted before the COVID-19 pandemic, although cost concerns remain well into the pandemic. In October 2021, polling showed that 46% of adults found it difficult to pay out-of-pocket costs for medical care not covered by insurance (Montero et al. 2022). Future research may examine how costs brought about by COVID-19 may uniquely influence perceptions of fairness and support for government action, particularly given the polarization of pandemic-related attitudes.

In addition, while the experiment measures which features are associated with demand for government action, it is less clear if these features also explain exactly how the public wants the government to act, making the policy making implications less straightforward. For example, with regard to surprise medical bills, before passage of the No Surprises Act a Kaiser Family Foundation April 2019 survey noted that the public was split on whether insurers, health care providers, or both should be responsible for paying the “balance” of medical bills (KFF 2019). Callaghan and colleagues (2021) similarly found less consensus on specific policy solutions for surprise billing. After the law's passage and implementation, the public still lacks knowledge about the status quo, with more than three quarters of US adults indicating that they had heard little or nothing at all about the No Surprises Act in March 2022 (KFF 2022).

Even if the public is unified in support of government action on a broad range of health care costs, a lack of consensus on how the government should act and a lack of knowledge about the policy status quo could still weaken the public's voice and their ability to hold policy makers accountable, including the urgency elites feel to respond to public attitudes on the issues.

Acknowledgments

The author is grateful for the valuable feedback from the anonymous reviewers and editor during the review process as well as the feedback from participants at the annual meeting of the American Political Science Association.

Notes

References