Population access to mental health services appears to have declined in the past decade, partly as a result of limited health insurance coverage and a growing number of uninsured Americans (1). Failure to obtain needed mental health care appears to be closely associated with a lack of or inadequate insurance coverage and with out-of-pocket costs (2,3). Unmet need for mental health care can result in higher prevalence of chronic mental and general medical illnesses and excess mortality (4–6). Effective policy planning to address access disparities requires determining the current prevalence of unmet need, identifying groups at relatively high risk for unmet need, and tracking the impact of policy reforms over time (7).

Andersen and Aday’s widely cited behavioral model (8–10) posits that access to health care is in large part a function of individual attributes classified into three groups of factors: predisposing (sociodemographic attributes), enabling (socioeconomic status and insurance coverage), and need (health and disability status). Analyses of data from the National Survey of Drug Use and Health (1), the National Comorbidity Survey (11), and the National Health Interview Survey (NHIS) (3,12–14) show that unmet need for primary care or mental health services was higher among adults without health insurance (1,3,11,12,14), working-age adults (1), and adults with a psychiatric diagnosis (13) and multiple functional limitations (14) or impairments (1).

The Affordable Care Act (ACA) of 2010, in conjunction with the Mental Health Parity and Addiction Equity Act of 2008 (MHPAEA), is intended to reduce the number of uninsured persons and improve access to mental health services. The objectives of this study were to assess recent national trends in mental health care access, identify population risk factors associated with unmet need for mental health services, and establish baseline data in order to track the impact of the ACA and the MHPAEA.

The NHIS is a continuing probability survey of households conducted by the National Center for Health Statistics (NCHS). It is widely regarded by the research community as the most comprehensive and current source of population health data in the United States (15). Each year, the NCHS selects a new representative panel of households for face-to-face interviews. The Family Core Survey collects basic descriptive data for all family members. One adult and one child from each family are then randomly selected for more detailed questions in the Sample Adult Core and the Sample Child Core Survey. This study used publicly available data from both sample core surveys for the 1997 to 2010 panels (N=543,632). For subpopulation comparisons, only data from the most recent 2010 panel was used (N=38,434).

In the NHIS Sample Adult and Child Core Surveys, the subsection on health care access and utilization included the following item: “During the past 12 months, was there any time when you needed mental health care or counseling but didn't get it because you couldn't afford it?” This variable (0, no access problem; 1, could not afford mental health services) was the basis for all trend and group comparisons conducted for this study.

There are several threats to validity inherent in this measure of mental health access. First, it requires individuals to retrospectively recall perceived need and financial circumstances, introducing possible recall bias. There is also no external validation for mental health status (for example, clinical assessments) or help-seeking behavior (for example, insurance claims). Changes in population rates of unmet need over the study period may reflect changes in incidence of mental health problems, changes in social attitudes toward disclosure of mental health problems, or changes in income or insurance coverage. However, these potential confounds affect most survey-based research investigating human behavior and are not unique to the questions at hand.

Our longitudinal comparisons were stratified by health insurance coverage (uninsured, private insurance, Medicare, and Medicaid). Relative rates of unmet mental health needs in 2010 were calculated for specific sociodemographic (age, gender, marital status, race, and ethnicity), socioeconomic (income and health insurance coverage), and health (self-assessed health status and activity limitation) groups. Risk factors associated with unmet mental health need were then entered simultaneously in a logistic regression model. In the model, health insurance was collapsed into a single categorical variable (public, private, or none).

All NHIS data were weighted to be generalizable to the U.S. population. The NCHS weights are based on estimates from the 1990 or 2000 Decennial Census. To address concerns regarding sampling error in complex household surveys, the SUDAAN Crosstabs procedure was used to generate standard errors (SEs) for prevalence estimates. Following a protocol established by the NCHS, prevalence estimates with relative SEs >30% were collapsed in summary tables. Simple group comparisons were tested with a Wald chi square test, and final strength of association measures were tested with adjusted odds ratios (AORs).

Unmet need for mental health services increased from 4.3 million in 1997 to 7.2 million in 2011, with the bulk of unmet need concentrated in the working-age population (18–64 years). Rates of unmet need for mental health care were approximately five times higher for uninsured respondents than for privately insured respondents. [Figures illustrating these trends are available in an online data supplement to this report.]

Table 1 presents data on the relative rate of unmet need for mental health services among selected risk groups in 2010 and provides AORs and 95% confidence intervals (CIs) for these groups. Compared with seniors, both children (AOR=3.0) and working-age adults (AOR=8.3) were more likely to report unmet need. Women (AOR=1.9) were more likely than men to report unmet need. Compared with whites, nonwhites (AOR=.7) were slightly less likely to report unmet need. Compared with non-Latinos, Latinos (AOR=.8) were also slightly less likely to report unmet need. In addition, persons with lower incomes (AOR=1.5) were more likely to report unmet need than those with higher incomes. Compared with individuals who had private insurance coverage, those who had no coverage (AOR=3.7) were more likely to report unmet need. Finally, fair or poor health (AOR=2.5) and limitations in daily activities (AOR=2.7) were associated with unmet need.

These results show that the number of Americans who reported unmet need for mental health services increased markedly from 1997 to 2010. Cross-sectional analyses of 2010 data indicated that unmet need for mental health services was significantly higher among working-age adults, women, unmarried persons, those with no health insurance, those with low incomes, and those in fair or poor health. These findings are consistent with previous research examining risk factors related to underutilization of mental health services (12).

In all study years, the relative rate of unmet need and the total number of persons with unmet need was highest among the uninsured, but the disparity between insured and uninsured populations grew more pronounced from 1997 to 2010. If the ACA is fully implemented in 2014, it is anticipated that most uninsured Americans will be enrolled in expanded Medicaid programs or in private plans. In addition, mental health and substance abuse parity requirements will be enforced for all private plans participating in health insurance exchanges, which should increase access to care by further reducing out-of-pocket cost barriers.

If previously uninsured individuals begin to seek mental health services or substance abuse treatment, provider capacity will become an even more urgent concern. There is a pressing need to examine how these policy initiatives will affect local access issues. In rural areas, where prevalence rates of mental health and substance abuse disorders are comparable to rates in urban and suburban areas, profound mental health provider shortages already exist. Additional research and program development in this area are top priorities.

If the ACA and MHPAEA realize their objectives of reducing the number of uninsured persons and improving access to mental health services, we should see a reduction in perceived unmet need for mental health services in the years after implementation. Despite the limitations noted in regard to the NHIS measure of mental health access, this longitudinal data will allow us to track whether access to mental health services improves.

The analysis was conducted with funding from the Program of Excellence in Rural Mental Health and Substance Abuse Treatment, Washington Life Sciences Discovery Fund (grant LSDF 08-02; Dr. Roll, principal investigator) and the National Institute on Disability and Rehabilitation Research (grant H133G070055; Dr. Kennedy, principal investigator). These agencies had no role in study design; in the collection, analysis, and interpretation of data; in the writing of the report; or in the decision to submit the paper for publication.

The authors report no competing interests.