High Out-of-Pocket Cost Burden of Mental Health Care for Adult Outpatients in the United States

Abstract

Objective:

This report describes characteristics of patients who had high out-of-pocket (OOP) spending on mental health care relative to income.

Methods:

A sample of 8,923 U.S. adults with outpatient mental health visits was drawn from the 2018–2021 Medical Expenditure Panel Survey. Respondents who spent ≥10% of their disposable family incomes on OOP mental health visits were defined as having a high OOP cost burden.

Results:

Using weighted percentages, the authors found that 2.4% of psychiatric outpatients had a high OOP burden; among those below the federal poverty level, 12.8% had a high OOP burden. Patients with a high (vs. low) OOP burden were statistically significantly more likely to be uninsured (7.5% vs. 2.4%) or diagnosed as having a substance use disorder (8.7% vs. 2.8%) or bipolar disorder (14.5% vs. 8.0%).

Conclusions:

Despite federal policies extending the availability of insurance for mental health care, many low-income psychiatric outpatients experience high OOP cost burden.

A substantial proportion of U.S. adults experience the financial burdens of paying for health care (1). Among adults with mental disorders, financial barriers have been described as preventing patients with psychological distress from accessing mental health care (2). These financial barriers are particularly acute for individuals with a greater burden of illness, lower income, or public insurance (2), and these barriers have persisted despite enactment of the Mental Health Parity and Addiction Equity Act (2008) and the Patient Protection and Affordable Care Act (2010–2014) (3).

Considerable attention has been given to the financial burdens of individuals with psychiatric symptoms who do not access mental health care (2). Some research has further considered the financial burdens of outpatient care for patients who access mental health care (4, 5). Cost sharing for mental health care has historically been lower in comparison with other types of health care. Among survey respondents with incomes in the first quartile (i.e., lower income level) who had used mental health care in the past year, nearly 12% spent more than 20% of their family income on out-of-pocket (OOP) mental health costs (4). Low participation rates in insurance plans among U.S. mental health providers may also mean that patients who access mental health care do so outside of their insurance networks and at a high OOP cost (5). Widening gaps between in- and out-of-network costs may have disproportionate effects on individuals who use psychotherapy or require intensive outpatient treatment (5).

In this study, we examined the prevalence, sociodemographic features, and clinical characteristics of patients who accessed mental health care despite a high OOP cost burden of care. We used a nationally representative sample of patients and defined high OOP cost burden of mental health care as mental health OOP expenditures equivalent to or exceeding 10% of disposable family income (1). We hypothesized that patients with a high OOP cost burden would be disproportionately from low-income families and receive treatment for serious mental illness (e.g., bipolar disorder or schizophrenia).

METHODS

Data were drawn from the 2018–2021 public access Medical Expenditure Panel Survey (MEPS) household component files. Technical information concerning survey fielding, nonresponse, and validation is provided elsewhere (6).

The study population included adults (ages ≥18 years), aggregated at the patient level, who reported making at least one outpatient mental health visit in the survey year. Outpatient mental health visits were defined as visits with any provider associated with a qualifying diagnosis (ICD-10-CM codes F01–F03 or F10–F99) or any visit provided by a psychiatrist, social worker, psychologist, or mental health counselor (7).

Patients who reported spending 10% or more of their disposable family income on OOP mental health visits were defined as having a high OOP cost burden of mental health care (1). Although patients receiving out-of-network care may receive later reimbursement for some OOP expenditures, at this threshold, OOP expenditures may still impose a substantial financial burden on families. We simulated posttax disposable income, including cash transfers, by using family income and the National Bureau of Economic Research’s TAXSIM model, version 35 (https://taxsim.nber.org/taxsim35) (8, 9), adjusted by consumer price index to 2021 U.S. dollars (https://data.bls.gov/cgi-bin/cpicalc.pl).

Patient demographic variables included sex, age, race-ethnicity (Hispanic, White, Black, and Asian), education level, family income, income by federal poverty level (FPL) (≤100% FPL, 100%–200% FPL, 200%–399% FPL, or ≥400% FPL), and health insurance status (Medicare, Medicaid, private insurance, or uninsured). Clinical variables included 6-item Kessler Psychological Distress Scale (K6) scores of ≥13 (range 0–24, with higher scores indicating more serious psychological distress) (10), Elixhauser comorbidity index scores (modified to exclude depression and psychosis) (11), and psychiatric diagnosis. Diagnostic variables were coded by using the ICD-10-CM as follows: mood disorder (F30.X, F31.X, F32.X, F33.X, F34.0, F34.1, F34.8, F34.9, F39.X), depression (F32.X, F33.X, F34.1), bipolar disorder (F30.X, F31.X, F34.0), anxiety disorders (F40.X, F41.X, F93.0, F94.0), psychotic disorders (F20.X, F21.X, F23.X, F25.X, F29.X), attention-deficit hyperactivity disorder (F90.X), adjustment disorders (F43.X), dementias (F01.X, F02.X, F03.X), substance use disorders (F1X), and developmental disorders (F7X, F8X).

We conducted analyses with Stata, version 16.1, and pooled MEPS linkage file weights to account for complex survey design, nonresponse, and the process of post-stratification (6). Survey-weighted means and 95% confidence intervals are reported by OOP burden group (i.e., low vs. high). We tested between-group differences by using logistic regression for binary outcomes and ordinary least squares for continuous outcomes; p values correspond to the regression F statistic. We performed multiple hypothesis testing corrections by using the Benjamini-Hochberg procedure (12). This study was granted exemption status by the Columbia University Medical Center Institutional Review Board.

RESULTS

The sample consisted of 8,923 patients with outpatient mental health visits, 228 (2.4% [weighted]) of whom had a high OOP burden (Table 1). Patients with a high OOP burden were younger than those with a low OOP burden (mean age=40.19 years vs. 44.72 years, respectively; p=0.001). Patients with a high OOP burden (vs. low) were also statistically significantly less likely to be married (13.2% vs. 41.4%, p<0.001) and more likely to be living alone (55.4% vs. 28.5%, p<0.001).

TABLE 1.

Characteristics and service use of U.S. adults (N=8,923) with one or more outpatient mental health visits, by level of out-of-pocket burden^a

	Low burden (N = 8,695)		High burden (N=228)
Characteristic	%	95% CI	%	95% CI	p
Weighted share, overall	97.6	97.2–97.9	2.4	2.1–2.8
Weighted share, ≤100% FPL	87.2	84.8–89.6	12.8	10.4–15.2
Male sex	36.8	35.1–38.5	42.8	34.0–51.7	.169
Age (mean years)	44.72	44.03–45.41	40.19	37.48–42.90	.001b
Race-ethnicity
Hispanic	10.9	9.7–12.1	15.2	8.9–21.5	.134
White	86.7	85.4–88.0	83.6	77.5–89.6	.241
Black	10.7	9.5–11.9	11.3	6.8–15.8	.770
Asian	3.6	2.8–4.4	5.0	1.4–8.6	.359
High school graduate	87.5	86.2–88.7	83.3	77.3–89.3	.116
College graduate	39.8	37.5–42.1	34.9	26.5–43.4	.280
Married	41.4	39.5–43.3	13.2	7.9–18.6	<.001b
Living alone	28.5	26.9–30.1	55.4	47.6–63.1	<.001b
Family income (mean $)	89,957	86,533–93,381	17,757	10,412–25,102	<.001b
≤100% FPL	12.3	11.3–13.4	72.4	64.9–79.9	<.001b
Private insurance	66.4	64.2–68.6	52.5	44.3–60.6	.001b
Medicaid	24.7	22.9–26.6	33.4	26.2–40.6	.009b
Medicare	22.9	21.4–24.4	23.1	17.0–29.2	.960
No insurance	2.4	1.9–2.9	7.5	3.3–11.7	<.001b
K6 score ≥13	33.7	32.1–35.2	42.8	35.5–50.1	.012b
Elixhauser comorbidity index score (mean)c	.68	.65–.72	.65	.50–.80	.646
Total N of psychiatric diagnoses (mean)	1.71	1.67–1.75	1.86	1.70–2.03	.067
Anxiety disorder	47.2	45.6–48.9	48.1	39.8–56.5	.833
Mood disorder	46.4	44.3–48.5	58.1	49.9–66.3	.007b
Major depression	41.5	39.5–43.4	49.1	40.6–57.5	.075
Bipolar disorder	8.0	7.0–9.0	14.5	8.7–20.3	.008b
Psychotic disorder	2.0	1.6–2.4	.1	−0.1 to .3	.003b
ADHD	11.5	10.3–12.8	9.3	4.8–13.9	.392
Adjustment disorder	13.9	12.7–15.1	19.4	12.7–26.0	.068
Substance use disorder	2.8	2.2–3.4	8.7	4.3–13.1	<.001b
Dementia	1.7	1.3–2.0	3.2	0.2–6.3	.198
Developmental disorder	1.4	0.9–1.8	2.2	−0.5 to 4.8	.468
Annual mental health out-of-pocket expenditure (mean $)	328	297–359	3,670	2,595–4,745	<.001b
Annual mental health total expenditure (mean $)	1,804	1,654–1,953	6,736	5,192–8,280	<.001b
Per-visit out-of-pocket expenditure (mean $)	38	36–41	117	81–153	<.001b
Any inpatient use	16.9	15.8–17.9	15.3	9.2–21.5	.640
Any emergency department use	17.9	16.8–19.0	15.6	9.6–21.7	.488
N of outpatient visits (mean)	19.8	19.0–20.6	43.2	32.8–53.5	<.001b

^aData were from the Medical Expenditure Panel Survey, 2018–2021. Weighted share is the survey-weighted percentage of patients in the low and high out-of-pocket mental health care burden groups. The next row of data represents the survey-weighted percentage in each group of patients (N=1,829; N=1,645 in the low-burden group and N=184 in the high-burden group) with family incomes ≤100% FPL. FPL, federal poverty level; K6, 6-item Kessler Psychological Distress Scale.

^bIndicates statistical significance at the level of 0.05 after Benjamini-Hochberg correction for multiple testing. The p values correspond to the F statistic associated with the test for differences between groups.

^cElixhauser comorbidity index weighted averages represent scores with depression and psychosis removed. Possible scores range from −19 to 89, with higher scores indicating greater disease burden.

Clinically, patients with a high OOP burden (vs. low) were more likely to experience serious psychological distress (K6 score ≥13: 42.8% vs. 33.7%, p=0.012). Patients with a high OOP burden were also statistically significantly more likely to have been treated for a substance use disorder (8.7% vs. 2.8%, p<0.001) or bipolar disorder (14.5% vs. 8.0%, p=0.008). These patients were significantly less likely to have been treated for a psychotic disorder (0.1% vs. 2.0%, p=0.003). The groups did not significantly differ by the mean number of distinct psychiatric diagnoses.

Reported annual family incomes of patients with a high OOP burden were significantly lower than those of patients with a low burden ($17,757 vs. $89,957, p<0.001), and patients with a high burden were more likely to be below the FPL (72.4% vs. 12.3%, p<0.001). Patients with a high OOP burden were more likely to have Medicaid (33.4% vs. 24.7%, p=0.009) and more likely to be uninsured (7.5% vs. 2.4%, p<0.001), compared with those with a low OOP burden. Race-ethnicity, educational attainment, and sex did not significantly differ between the groups. Among patients with incomes ≤100% FPL, 12.8% had a high OOP burden.

Total mean annual mental health OOP spending for patients in the high OOP burden group was $3,670, compared with $328 in the low OOP burden group (p<0.001). Total mental health care expenditures for the high OOP burden group were $6,736, compared with $1,804 in the low OOP burden group (p<0.001). The mean number of outpatient visits in the high OOP burden group was 43.2, compared with 19.8 in the low OOP burden group (p<0.001). Average per-visit OOP costs were $117 in the high OOP burden group, compared with $38 in the low OOP burden group (p<0.001). Inpatient and emergency department use did not significantly differ between the groups.

DISCUSSION AND CONCLUSIONS

Overall, a small subset of patients experienced a significant OOP burden associated with mental health care expenses. Of patients with family incomes below the FPL who used outpatient mental health care, one in eight had a significant OOP mental health care burden. Although the two groups did not differ by sex, race-ethnicity, or educational attainment, patients in the high OOP burden group had a greater number of outpatient mental health visits, higher total annual mental health care costs, higher average per-visit OOP costs, and lower family incomes. Patients with a high OOP mental health care burden were more likely to be younger, to be experiencing serious psychological distress, and to have a diagnosis of bipolar disorder or substance use disorder compared with other psychiatric outpatients. It is surprising to note that only a small percentage of patients in the high OOP burden group reported being uninsured. More than one-third of patients in the high OOP burden group were covered by Medicaid, and over half had private health insurance.

These findings highlight that patients with the highest risk for acute mental health care use and those at risk for high OOP burden were distinct groups. Patients with high OOP burden had service use patterns dominated by outpatient visits rather than acute care visits. Low family incomes combined with private insurance and a diagnosis of a mental disorder contributed to elevated risk for high OOP mental health care burden. Patients in the high OOP burden group were younger on average and may have received financial support from outside their immediate household. Higher per-visit OOP spending in the high OOP burden group may reflect that patients in this group sought out-of-network care at higher rates or may be enrolled in less generous health insurance plans. Lower incomes may have also constrained patients to selecting high-deductible plans.

Patients with substance use or bipolar disorders were overrepresented in the high OOP burden group and may be more likely to engage in higher-intensity outpatient treatments. For example, evidence-based intensive outpatient treatment programs for substance use disorders may consist of multiple weekly visits for a period of 3–6 months (13). Although these intensive outpatient treatment models are established in behavioral health care, they may have few equivalents in general medical practice, exacerbating existing difficulties with the implementation of parity legislation.

This analysis had several limitations. First, the MEPS relies on participant recall and diaries that may underestimate mental health care use. However, information provided by participants is supplemented by a medical provider survey to increase the validity of the responses. Second, we focused on patients with one or more outpatient mental health visits in the survey year. We did not include the larger percentage of individuals with mental disorders who did not seek care, may have unsuccessfully sought mental health care, or may have deferred mental health care for nonfinancial reasons (e.g., stigma, fear of employer retribution, or patient preferences). Third, we were unable to study plan features such as deductible amount or out-of-network coverage. Furthermore, OOP cost burden did not reflect net burden among patients who sought out-of-network coverage and subsequently received reimbursement for visits. Fourth, because of data limitations, we set the threshold for high OOP burden at ≥10% of disposable family income being spent on mental health care and described the small group of patients who met this threshold; however, we acknowledge that many other patients who experienced high OOP mental health care burden may not be included in this definition. Finally, we were limited in our ability to describe the context, symptoms, and type of care sought by patients in the high OOP burden group, along with their circumstances, history of illness or risk, or other features of their illness that may help to explain their high willingness to pay for mental health care even when it is associated with financial hardship.

Despite policy changes extending the availability of insurance for mental health care, our study showed that many low-income adults in the United States who used outpatient mental health care experienced a high OOP burden. Although outpatient care is often framed as a lower-cost alternative to inpatient care in mental health, the share of costs borne by patients in the form of OOP spending may differ in outpatient mental health care versus general medical care. Evidence-based, high-intensity outpatient treatment may still be out of reach for many patients.

HIGHLIGHTS.

• Despite high rates of insurance, one in eight patients with family incomes below the federal poverty level and at least one outpatient mental health visit experienced a high out-of-pocket cost burden of mental health care.

• Patients with a high out-of-pocket burden disproportionately used outpatient mental health services rather than acute care services, and they were more likely to be diagnosed as having a substance use or bipolar disorder compared with other outpatients.

• Despite health insurance expansion and parity legislation, intensive outpatient treatments in mental health care may remain too costly for many patients.