Selection Into Mental Health Services Among Persons With Depression

In 2016, 6.7% of adults in the United States had at least one major depressive episode. For two-thirds of this population, depression interfered significantly with the ability to manage their home, work, or social lives (1). Even though treatment rates for depression continue to increase (2–4), less than half of adults who have had an episode of major depression receive any treatment. The most common outpatient mental health services, or treatment modalities, for depression include prescription medication and group or individual counseling, either alone or in combination (5–7). Numerous studies have examined factors that predict who receives treatment for depression (8–12). While informative, this research does not explain how people come to receive different types of services once they have engaged nor does it identify the characteristics that distinguish between persons who receive different services among those in treatment. How people select into mental health services and modalities of care may also be an indicator of access. Understanding these factors might help clinicians, policy makers and administrators remove modality-specific barriers to care and increase timely entry into mental health services. In this study, we aimed to identify characteristics of patients with depression that are associated with different types of mental health services or treatment modalities (medication, counseling, or counseling and medication).

Methods

Data

Data were obtained from the 2015 and 2016 samples of the National Survey on Drug Use and Health (NSDUH). The NSDUH is an annual multistage area-probability sample survey that estimates the prevalence of substance use, mental health issues, and use of mental health services among the civilian, noninstitutionalized population of the United States (1). Response rates were 55.2% in 2015 and 53.5% in 2016. Our analytic sample consisted of adults who met the criteria of having had a major depressive episode in the past year and having received outpatient professional mental health services in the form of prescription medication and/or counseling (N=4,169). Institutional review board approval was not needed because these data are publicly available and deidentified.

Measures

Respondents were asked about outpatient treatment, counseling, and any prescription medication they received for their mental health from a doctor or other health professional in the past 12 months. We created a mutually exclusive variable for treatment modality: “medication only,” if the respondent received only prescription drugs for their symptoms; “counseling only, no medication,” if the respondent reported seeing or talking to a health professional about his or her symptoms but did not receive any prescription medications; and “medication and counseling,” if the respondent received prescription drugs and talked to a professional about his or her symptoms.

We created three mutually exclusive pathways into care, and patients were asked which best describes how they entered care: “independent,” if respondents decided on their own to receive treatment; “asked by someone,” if respondents received care because someone else thought they should; and ”ordered,” if respondents were ordered to receive treatment. Sociodemographic variables included race-ethnicity (non-Hispanic white, non-Hispanic black/African American, Latinx/Hispanic, and other), age (18–25, 26–34, 35–49, 50–64, ≥65 years), residential area (large metro, small metro, nonmetro), level of education (less than high school, high school graduate, some college, college graduate), employment status (not looking/not in labor force, unemployed, employed), and poverty level (<100% of the federal poverty level [FPL], 100%−199% FPL, and ≥200% FPL). Health status variables included self-rated health (poor/fair versus good/very good/excellent), health insurance status (insured versus uninsured), whether he or she met DSM-5 criteria for substance abuse or dependence (yes or no), probability of having a severe mental illness (0.0–0.9), and whether the respondent received inpatient mental health treatment in the past year (yes or no). All these variables could be determined from the public-use data files, including the probability of having a severe mental illness, which was computed by factoring in the occurrence of a past-year major depressive episode, responses to questions that assessed serious thoughts of suicide in the past year, and the participant’s scores on the World Health Organization Disability Assessment Schedule to assess both psychological distress and disability.

Analyses

We describe demographic, socioeconomic, and health status characteristics of respondents who received only prescription medication, those who received only counseling, and those who received both medication and counseling. We used multinomial logistic regression to identify factors associated with the relative log odds of receiving each type of treatment. In all analyses, weights were employed to represent the population from 2015 to 2016, and standard errors (SEs) were adjusted to account for the complex survey design.

Results

Characteristics of the sample of respondents who used professional outpatient mental health services for depression are presented in Table 1. As shown, 9% of respondents received prescription medication only, 22% received counseling only, and 69% received both medication and counseling.

Table 1 also shows differences in patient characteristics by treatment type. Persons who used only counseling were less likely to be white and younger and more likely to be college graduates, to be employed full-time, to live in large metropolitan regions, and to have worse self-rated health than those who used only medication. Persons who received both types of treatment were more likely to live in a large metropolitan region, to be insured, and to have greater average probability of severe mental illness compared with persons who received only medication.

Table 2 shows factors associated with treatment modality. The relative odds (RO) of receiving counseling and of receiving both medication and counseling compared with receiving prescription medication alone were greater among persons who were ordered into care than among those who sought mental health care independently, with analyses controlling for socioeconomic and health characteristics. Latinx/Hispanic respondents had greater relative odds of receiving counseling than medication compared with non-Hispanic whites (RO=3.37, SE=1.60). Persons ages 50 years and older were more likely to receive medication than counseling alone compared with those between the ages of 18 and 25. Similarly, compared with living in a large metropolitan area, living in a nonmetropolitan area was associated with lower adjusted odds of counseling relative to medication (RO=0.43, SE=0.10). Odds of receiving both medication and counseling increased with greater probability of having a severe mental illness (RO=1.21, SE=0.19).

Discussion

Respondents’ severity of mental illness, pathway into care, race-ethnicity, residential area, and age were associated with treatment modality. Studies based on randomized controlled trials have suggested that among persons with severe mental illnesses, combined treatment is more efficacious (13). In an observational study such as ours, the finding that a higher probability of having a severe mental illness increased the likelihood of receiving both medication and counseling is reassuring. Being ordered into care was associated with receiving both medication and counseling. People might be ordered into mental health service modalities that specifically require frequent interaction with providers, given that coercion, although ethically questionable, expedites engagement with services (14). Latinx/Hispanic respondents were more likely to receive counseling than medications, consistent with lower rates of antidepressant use among this population in general (15, 16). Latinx/Hispanic people also tend to use professional mental health services for long-term coping and when they trust the interpersonal skills of the mental health service provider (17). These conditions might more easily be met in counseling than with medication. Findings regarding place of residence are consistent with prior research suggesting that people in rural areas are more likely to use medications (18) because of the shortage of mental health professionals and the elevated levels of stigma associated with mental illness in these areas (18, 19). The finding that adults age 50 years and older are less likely than persons 18 to 25 to use counseling is also consistent with the steady increase in antidepressant use among adults and the decline in use of psychotherapy, especially among persons age 35 and older (3, 5, 20, 21).

These findings should be interpreted in light of certain limitations. First, use of mental health services, including different treatment modalities, might depend on factors beyond the patient socioeconomic, demographic, and health characteristics that were included in this analysis. Although we assessed comorbid substance use disorder, assessing other mood and anxiety disorders as well as stress-related conditions that co-occur with depression would have strengthened the analyses. Characteristics of mental health systems and health insurance plans and the availability and types of providers may also be factors shaping patients’ use of different types of mental health services, but this information was not available in the data. Second, although nonresponse weights included in the analyses adjusted for nonresponse to the NSDUH, some of the bias introduced at different stages of the multistage sampling may not have been corrected. In addition, persons with severe mental health problems are less likely than those without severe mental illnesses to be captured in population-based surveys such as the NSDUH. For example, people at risk of severe mental illnesses, such as those who are homeless and/or reside in shelters, are incarcerated, or are otherwise institutionalized, are not included in the NSDUH. Finally, we did not include persons who were surveyed and treated with medication, counseling, or therapy in an inpatient setting only.

Conclusions

How a person enters care, severity of symptoms, race-ethnicity, age, and place of residence influence the specific type of mental health service a patient receives for depression. Understanding how health and sociodemographic characteristics influence selection into different kinds of services and treatment modalities might help improve psychiatric services, thereby reducing the burden of untreated depression. The literature on patterns of use of psychiatric services has helped researchers and clinicians to predict individuals’ probability of seeking care. Findings from this study suggest that factors such as how people enter care (independently or via coercion) are associated with the kind of services they receive. We broadened the scope of “access to treatment” by exploring selection into treatment. There may be benefits to increasing concurrent access to both modalities of mental health services, especially for people who are more likely to be ordered into care. Finally, beyond assessing patients’ selection into treatment, research should also identify which treatment approaches might work for whom. Research that explores the connection between selection into mental health services, treatment efficacy, and patient-reported outcomes of treatment is warranted.