Current policy initiatives being implemented under the Affordable Care Act (ACA) extend health insurance to approximately four million previously uninsured persons with serious mental illness (
1) and many more individuals with other mental and substance use disorders. A provision of the ACA expands the Paul Wellstone and Pete Domenici Mental Health Parity and Addiction Equity (MHPAE) Act of 2008 to the new health insurance exchanges. The original intent of the MHPAE Act was to require group commercial insurers that offer coverage for mental health and substance use services to provide it at a level equivalent to that for other medical services. A major change under the ACA is the extension of insurance parity to the individual and small-group market and the mandate that insurers comply with an essential health benefit requirement that includes provision of coverage for mental and substance use disorders at parity (
2). In addition, state Medicaid expansions will newly insure low-income groups and substantially increase the role of government in funding mental health care (
2). These policy changes have been accompanied by heightened attention to questions about the adequacy of funding for mental health treatment in the wake of multiple mass shootings in recent years in which shooters appeared to suffer from mental disorders (
3).
Despite this flurry of attention to mental health policy, we know surprisingly little about Americans’ support for policies aimed at improving access to mental health care. A 1998 article by Hanson (
4) provided an in-depth review of public opinion about mental health insurance coverage. The author compiled and analyzed questions asked by polling firms in the period (1989–1994) leading up to the debate about universal health insurance during the Clinton presidency. The study found relatively high support for inclusion of mental health services in a mandatory benefit package. However, these data are nearly two decades old and limited in important respects. Most notably, the study relied on data collected by polling firms, which did not disclose key details about survey methodology (for example, sample size and response rate). Another study that used 1996 data found that support for increasing government spending on mental health treatment was dependent on the respondent’s group identification with persons with mental illness (
5).
Almost two decades have elapsed since these data were collected, and it is important to understand whether personal experience continues to play a role in determining attitudes about the appropriate role of government in funding mental health services in the current political environment. To fill this research gap, we conducted a national public opinion survey in January 2013 to gauge current public attitudes about support for insurance parity and increased government spending on mental health treatment. We examined how sociodemographic characteristics and political affiliation influence support for these policies. We hypothesized that personal experience with mental illness would be associated with support for both policies and that stigmatizing views about persons with mental illness would be associated with lower levels of policy support.
Methods
We conducted a national public opinion survey of adults aged 18 and over from January 2 to 14, 2013. The Web-based survey was fielded using the survey research firm GfK Knowledge Networks (GfK). GfK has recruited a probability-based online panel of 50,000 adult members, including persons living in cell phone–only households, using equal probability sampling with a sample frame of residential addresses covering 97% of U.S. households. The GfK panel used probability sampling at the first stage of recruitment when individuals are approached to participate in the panel, and the panel recruitment rate was 16.6%.
The survey was pilot-tested from December 28 to 31, 2012. The order of the survey items was randomized. To further avoid priming, respondents were asked to answer “some questions about public affairs.” A total of 1,517 respondents completed the survey. The survey completion rate, defined as the proportion of GfK panel members randomly selected for this study who completed the survey, was 70%. All analyses incorporated survey weights to produce nationally representative estimates by accounting for panel selection deviations, panel nonresponse and attrition, and survey-specific nonresponse.
Dependent variables were support for insurance parity and for increased government spending on mental health treatment. For insurance parity, we asked respondents “Do you favor or oppose requiring insurance companies to offer benefits for mental health and drug and alcohol abuse services that are equivalent to benefits for other medical services?” Response categories on a 5-point Likert scale were 1, strongly oppose; 2, somewhat oppose; 3, neither favor nor oppose; 4, somewhat favor; and 5, strongly favor. For government spending, we asked respondents “Would you like to see more or less government spending on mental health treatment?” Response categories on a 5-point Likert scale were 1, spend much less; 2, spend less; 3, spend the same as now; 4, spend more; and 5, spend much more.
Sociodemographic characteristics included respondents’ gender, age, highest level of education completed, race-ethnicity, household income, region of the country, and work status. [A table presenting unweighted and weighted data on sociodemographic characteristics of the sample and of the March 2013 Current Population Survey sample is available in an online
data supplement to this report.] As expected, characteristics of both the weighted and unweighted sample were consistent with national rates. In the weighted sample, 52% of respondents were female (N=788), 33% were nonwhite (N=398), and the mean age was 47.0±.6). Political party affiliations were Republican, Independent, or Democrat. Respondents were defined as having personal experience with mental illness or substance use if they reported that they themselves, an immediate family member, or another relative or close friend had been hospitalized, had received counseling, or had received prescription medication to treat a mental health or drug or alcohol abuse problem.
We constructed a stigma scale based on responses on a 5-point Likert scale to four questions: “Do you agree or disagree that locating a group home or apartment for people with mental illness in a residential neighborhood endangers local residents?” “Do you agree or disagree that people with serious mental illness are, by far, more dangerous than the general public?” “Would you be willing or unwilling to have a person with a serious mental illness work closely with you on a job?” “Would you be willing or unwilling to have a person with a serious mental illness as a neighbor?” The first two measures are related to dangerousness and come from a public opinion study conducted about attitudes toward persons with mental illness under the Robert Wood Johnson Foundation Program on Chronic Mental Illness (
6). The two measures on social distance come from items in the mental illness module of the General Social Survey (
7). The Cronbach’s alpha was .77, suggesting that the items formed a reliable scale. Therefore, we included this scaled stigma measure—ranging from 1, low stigma, to 5, high stigma—in analyses to examine how stigma attitudes were associated with support for the two policies.
We estimated ordered logit regression models to examine the associations between support for the policies and respondents’ sociodemographic characteristics, political affiliation, personal experience with mental illness, and stigmatizing attitudes. Results were consistent when we collapsed the ordinal scales into dichotomous dependent variables and reestimated the models by using logistic regression. [A table in the online
supplement presents the results of this analysis.] The study was determined to be exempt by the Johns Hopkins Institutional Review Board.
Discussion and conclusions
Findings from this nationally representative survey indicate support among Americans for two policy approaches to broadening access to treatment for persons with mental illness—insurance parity and increased government spending. Support for these policies was partly ideologically driven, with Democrats significantly more supportive than Republicans or Independents.
Public support for government spending was 10 percentage points lower than support for insurance parity. In keeping with research on attitudes about the role of government (
8), this finding is consistent with the idea that for certain subgroups of Americans, policies that involve a large government role in addressing social problems are inherently less attractive. Looking deeper, we found that respondents who supported the insurance parity policy but who did not support the increased government spending policy were significantly more likely than other respondents to identify as a Republican (p=.004). One implication is that mental health policies that appear to rely on a strong role of government will be less attractive to some. Continuing erosion of levels of trust in government among the American public (
9) speaks to the importance of developing a diverse set of policies engaging both the public and private sectors to address the weaknesses of the current U.S. mental health delivery and financing system.
We found that personal experience mattered in respondents’ attitudes toward both policies. Respondents with personal experience with mental illness—either their own or that of a close family member or friend—supported insurance parity and increased government funding for mental health treatment at significantly higher rates than those without such personal experience. The role of personal experience among respondents is consistent with the fact that the critical champions of these policies in Congress over the years, such as Senators Pete Domenici and Paul Wellstone and Representatives Patrick Kennedy and Jim Ramstad, have attributed their involvement to personal motivations (
10). Our finding is encouraging given that half of our national sample had some personal experience with mental illness—either directly themselves or through someone close to them.
Although it is not possible to ascertain the precise clinical diagnoses of the respondents or of people with whom the respondents had interacted, we would expect on the basis of disease prevalence that most had personal experience with more prevalent conditions (such as anxiety or depression) and that far fewer had personal experience with less prevalent severe and persistent mental illnesses such as schizophrenia, which occurs in about 1.1% of the U.S. adult population (
11). To the extent that these individuals did not have personal experience with more disabling mental illnesses, they may be just as susceptible as those with no personal experience to media portrayals—particularly to images linking mental illness with violence. For example, we found no difference in the proportion of respondents with personal experience of mental illness who agreed that “people with serious mental illness are, by far, more dangerous than the general population” (45%) and those who agreed with the statement but who reported no personal experience with mental illness (46%).
These findings should be considered in light of several limitations. First, for neither of the policy options did our survey instrument gauge “willingness to pay” or support relative to other policy priorities. Both questions could provide important additional information about the strength of respondents’ attitudes. Second, Web-based surveys have been criticized because of concerns about incomplete coverage and selection bias (
12). GfK attempts to minimize these issues by recruiting probability-based samples and providing Web access to those without it.
Our finding that respondents with stigmatizing attitudes toward persons with mental illness were significantly less supportive of both policies reinforces the importance of developing and evaluating antistigma efforts. This is a critical moment to refocus efforts on stigma reduction given the current political environment in which gun violence is increasingly linked to mental illness and in light of recent evidence that stigma toward those with serious mental illness—in particular, perceptions about dangerousness—is on the rise (
13–
15).