Experiences With Insurance Plans and Providers Among Persons With Mental Illness

A positive patient experience with health care is an important goal of health systems, because it provides an indicator of the quality and acceptability of health services (1,2). Experience with health care received is central to the “triple aim” of better health care and widely used as a patient-centered measure of quality of care (3). Patient experiences when interacting with providers reflect how well providers communicate with patients and coordinate care to meet patients’ needs (1,2). Different from patient satisfaction, which is a subjective evaluation of the care received, patient experiences reflect specific aspects of the visit, such as provider communication (4). Patients who have positive experiences with providers have a higher level of adherence to medication and better continuity of care with providers (5) and use less health care (6). However, persons’ experience with insurance plans and the administrative processes they go through to obtain care are another important aspect of health care delivery. Difficulties in understanding benefits and navigating administrative processes can jeopardize access to and maintenance of care, which are vital for persons with mental illness to avoid acute problems.

Mental Illness and Health Care Experiences

For several reasons, mental illness may be associated with more difficulties interacting with providers and plans. Persons with mental illness have a greater number of chronic conditions than those without mental illness, leading to frequent use of health care for general medical conditions. However, physicians may be unaware of how mental illness can impair patients' social functioning (7) and may be less comfortable treating persons with mental health conditions. Persons with mental illness and general medical conditions may desire or require more time and attention from providers to explain treatment plans (8). However adverse patient-provider interactions occur, they are associated with dropping out of mental health care, which may have wide-reaching consequences for well-being, social functioning, and productivity (9). Moreover, adverse experiences that discourage care seeking among those with mental illness may complicate the care of their other chronic conditions.

Patients seeking health care may also be stymied by difficulties with insurance plans, and navigating problems with plans may be more difficult because of the mental illness. During the study period (2004–2012), use of mental health care was more restricted than use of other health care, despite existing parity laws (10), although the MHPAEA, which was enacted in 2008, introduced wider reforms than did previous laws (11). Specific administrative mechanisms to contain costs focused on behavioral health care, such as limitations on mental health visits and referral rules limiting access to specialty providers or treatments (12). Other aspects of insurance plan benefit structures, such as limitations on psychotherapy and higher copays for psychotherapy than for medication, may also negatively affect patients’ experiences with plans. The past decade has seen increased use of prescriptions and a decline in psychotherapy (13), even though combination therapy is preferred by patients with mental illness and recommended by physicians (14,15). Such regulations could have affected experiences with plans differently for persons with mental illness than for those without mental illness.

Little is known about how problems with providers and insurance plans vary by the presence of mental illness. Most research on patient experiences among persons with mental illness has focused on aspects of satisfaction, looked at mental health services, or examined specific patient populations or settings (for example, inpatients or veterans with mental illness) (16–19). Fragmented care experiences with regard to coordination of general medical and mental health care (20,21) and barriers to primary care among persons with mental illness (22–24) have been described. However, little attention has been paid to the experiences of persons with mental illness in terms of usual providers and health plans. Even though some progress has been made in implementing work on shared decision making in care for depression (25–27), research is lacking on mental illness and patient experience with regular providers and health plans. This study addressed that gap.

We addressed three research questions. Do persons with mental illness experience greater problems with their usual care providers compared with persons without mental illness? Do persons with mental illness experience greater problems with insurance plans compared with those without mental illness? For problems associated with mental illness, are rates of problems higher or lower among persons with public insurance than among those with private insurance?

Our study contributed to the literature by using symptom scales and data from a household survey, rather than data from persons with diagnosed conditions and from specific patient populations. This approach is desirable because nearly two-thirds of persons with mental disorders have no diagnosis (28). For experiences with providers, we examined experiences with the usual source of care rather than with mental health specialty care. We also used specific measures of experiences that are more actionable than measures of overall satisfaction, and we examined differences in experiences by public and private insurance.

Methods

Data for this study come from eight panels (panels 9 to 16, 2004 to 2012) of the Medical Expenditure Panel Survey (MEPS) (29), a national household survey of noninstitutionalized persons. Persons were interviewed five times over two years. The MEPS response rate varied from 58% to 68%, with sample sizes between 34,920 and 29,370. During each year, all adults age 18 and over complete a self-administered questionnaire (SAQ) that contains questions about symptoms of mental illness. Our study population was adults age 18 to 64 who were covered by insurance during the second year of the panel, responded to the SAQs in both years, and responded to at least one question about experiences with providers (among those with a usual source of care at the time of the survey) or a question about experiences with their health plan (among those who used a health service in the previous 12 months) (N=25,176). This study was a secondary analysis of public data and exempt from approval by an institutional review board.

Measures

Mental illness was a categorical indicator based on responses to two scales: serious psychological distress (SPD) and symptoms of depression (30). SPD was assessed by the Kessler-6 scale of nonspecific psychological distress (31), which asks respondents how often in the past 30 days they felt symptoms such as sadness, hopelessness, and worthlessness. Possible responses range from 1, none of the time, to 4, all of the time. A summed score of ≥13 (out of 24) indicates SPD. This criterion correctly classifies 92% of respondents with serious mental illness (31). Depression was measured with the Patient Health Questionnaire two-item depression screener (PHQ-2), which asks respondents how often over the past two weeks they had “little interest or pleasure in doing things” or were “feeling down, depressed, or hopeless.” Possible responses range from 0, not at all, to 3, nearly every day. A score greater than 3 indicates that the person should be further screened for major depression (32). Persons who met the thresholds for either or both scales (SPD or PHQ-2) were considered to have mental illness.

We differentiated between episodic and persistent mental illness. Many persons with poor mental health recover with no further episodes (33). For others, mental illness is a persistent condition that must be dealt with over the life course. To account for differences in persistence over time, we constructed three levels of mental illness: no mental illness at either assessment, mental illness at one assessment (episodic), and mental illness at both assessments (persistent).

We used measures of experiences with insurance plans and experience with providers from MEPS year 2. Questions about health care experiences in the MEPS are based on the Consumer Assessment of Healthcare Providers and Systems survey (34). Questions about plan experiences address five problems: finding a personal doctor or nurse; getting approval for medical care, tests, or treatment; finding or understanding plan information; calls to customer service; and paperwork. Questions about experiences with providers were asked only if the respondent had a usual source of care at the time of the survey. These questions assess whether the usual provider asks about the person's prescriptions and treatments from other doctors (yes or no), explains treatment options (yes or no), asks the person to help make treatment decisions (sometimes or never compared with usually and always), and shows respect for the person's treatment choices (sometimes or never compared with usually and always). Negative responses to the first two questions and sometimes or never to the last two questions indicate a problem. We included a binary variable for the type of care source (facility or a person).

Covariates

Health status was assessed with four measures: a dichotomized self-rated health measure indicating poor-fair health, a count of nine chronic conditions, functional limitations, and activity limitations. Demographic variables included as controls were age, race-ethnicity, education, urban location, employment status, marital status, and poverty status. Persons with both private and public coverage (4.9%) were excluded from the analysis. We included a binary measure of any change in insurance type since the previous year. We controlled for total annual out-of-pocket health expenditure during the second year of the survey. We included a variable indicating whether the survey was post-2010 to account for changes in the health care system in response to implementation of the Affordable Care Act.

Data Analysis

We conducted all analyses in Stata, version 13.0, using survey design variables (primary sampling unit and strata) developed by the survey administrators that adjust the standard errors to account for the complex sample design. We used longitudinal survey weights (and selected individuals who responded to the SAQ in both years) that were adjusted for survey nonresponse. We determined selected characteristics of the sample by mental health status. We then determined the unadjusted rate of problems with plans and providers by mental illness status. Next, we included the control variables in multivariate logistic regression models predicting each problem. Finally, we examined the marginal effect of having private insurance compared with public insurance within each mental illness category (episodic or persistent).

Results

Table 1 presents data on social status and health by mental illness status. Of note, lower education, greater poverty, unemployment, and poor health were related to having any mental illness. Persons with mental illness also had higher out-of-pocket costs for health care.

Table 2 presents unadjusted prevalence rates of problems by mental health status. Nearly one in five persons (17.8%) had problems with providers not asking about the person’s other treatments, and this did not vary by mental health status. Those with episodic and persistent mental illness were more likely than those with no mental illness to report problems with providers not explaining all options (6.9% and 8.7%, respectively, versus 5.1% for those with no mental illness), not respecting the person’s treatment choice (11.7% and 13.2% versus 9.3%), and not seeking the person's participation in decision making (18.9% and 22.2% versus 15.2%).

Rates of problems with plans were higher than rates of problems with providers, and rates of all plan problems were significantly higher for persons with mental illness compared with those with no mental illness. For persons with persistent mental illness, problems with treatment approvals were the most common (38.5%), followed by customer service (36.0%), finding information (33.4%), paperwork (29.9%), and finding a doctor (16.3%). The rank of most to least common problem was different for persons with episodic or no mental illness; the rate of problems with customer service was highest for both groups, followed by finding information, treatment approvals, and paperwork.

Table 3 shows the adjusted odds ratios (AORs) for problems with plans and providers by mental health status. Compared with the reference category of persons with no mental illness, persons with episodic mental illness were more likely (AOR=1.14, p=.046) to report that doctors did not ask about their other treatments. Odds were higher for persons with any mental illness that doctors did not explain treatment options (episodic, AOR=1.41, p=.005; persistent, AOR=1.66, p<.001). Among persons with persistent mental illness, odds were also higher that the doctor did not respect treatment choices (AOR=1.35, p=.014) and did not seek the person’s participation in decisions (AOR=1.32, p=.003).

For experiences with plans, compared with the reference category of persons with no mental illness, persons with episodic mental illness and persistent mental illness were significantly more likely to have a problem finding a provider (episodic, AOR=1.31, p=.002; persistent, AOR=1.61, (p<.001). Persons with persistent mental illness were also more likely than those with no mental illness to report problems with getting information (AOR=1.66, p=.004), getting approval for treatments (AOR=1.72, p=.002), paperwork (AOR=1.79, p=.001), and customer service (AOR=1.64, p=.003).

In a multivariate regression analysis, insurance status (public or private) was significantly associated only with plan problems. Therefore, to answer the third research question about whether rates of problems vary by insurance status, we examined the marginal effect of private coverage compared with public coverage for plan problems and determined the predicted probability of problems by insurance type (Table 4). Among persons with episodic mental illness, those with private coverage were less likely than those with public coverage to have problems finding a doctor. Among persons with no mental illness, those with private coverage reported rates of problems finding information and with treatment approval that were nearly half of rates reported by those with public coverage. For each level of mental illness (episodic or persistent), those with private coverage had fewer problems with treatment approvals and customer service.

Discussion

The purpose of this study was to examine the association between mental illness and patient-reported problems in their experiences with providers and insurance plans. Overall, persons with mental illness experienced higher rates of such problems than persons without mental illness.

Some problems might result from patients with mental illness having lower capacity for administrative tasks, or the problems may truly reflect a greater administrative burden compared with others. However, our data cannot address this point. For instance, higher rates of problems with treatment approvals may be attributable to cost containment and benefit management policies, such as those encouraging the use of prior authorizations for referrals, which complicate access (35,36). Similarly, paperwork problems might result from carve-out behavioral health care plans, which have separate administrative systems for mental health and general medical care. However, because our data did not allow us to address these links, we must call for further research. In addition, although publicly insured persons reported high rates of certain problems, such as treatment approval, such differences may result from the fact that patients in this population are more likely to have coverage due to disabilities or socioeconomic vulnerability and thus may be different from patients with private insurance in other ways that we could not assess and which may affect health needs, providers seen, and plan experiences (37).

High rates of problems with finding information, customer service, and finding a doctor—overall and particularly for those with mental illness—point to opportunities for plans to explore approaches to educate enrollees about plan benefits and make information more accessible. We did not find any published studies of interventions to assess customer service. We also found very few studies on enrollees’ awareness of plan benefits, although some evidence suggests that publicly insured individuals lack knowledge about costs and coverage (38,39). Problems finding a usual source of care may also result from difficulty navigating and understanding plan information because of functional limitations related to mental illness (40,41). More speculatively, if the greater difficulty finding doctors reported by publicly insured persons reflects low reimbursement rates and financial difficulties faced by providers who serve the publicly insured population (42), then addressing these difficulties may be a good policy target for improving patient experience. Although reasons for plan problems are unclear, health plans can pursue strategies to improve patient experience before and after a clinical encounter.

In addition to problems with finding a doctor, people with mental illness reported more problems in experiences with providers. For various reasons, including a high rate of comorbid conditions, persons with mental illness are high utilizers of non–mental health care, but they often receive poor-quality primary care (20). For example, a lack of communication and coordination between their mental and general medical care providers can discourage persons with mental illness from using primary care (22,23). Persons with mental illness may also have stigma-related concerns that affect primary care seeking (24), and some research suggests that providers have negative biases toward persons with mental illness (7,35,43).

A problem with providers not seeking patient participation in treatment decisions was reported by about one in five persons overall, and rates were higher among persons with mental illness. Patients with depression may prefer that physicians make treatment decisions for them, but shared decision making can improve treatment adherence (44,45). Improving doctor-patient communication may be one way to improve adherence and facilitate shared decision making (46). Because many individuals seek mental health care in primary care settings (47), improving training for primary care providers in treating persons with mental illness may also improve patient experiences (48).

We also found that persons with mental illness experienced higher rates of problems with providers not respecting their preferred treatments, suggesting that more attention is needed to improve providers’ understanding of treatment preferences and that encourages providers to take into account a patient’s social and financial capacity (49). Problems with treatment preferences may also be understood within models of the complexity of patient conditions (50). Models of care that underscore complexity and strive to balance the treatment demands placed on patients and their capacity to meet those demands may lead to better interactions with providers (51).

We note some limitations. We identified problems in the clinical encounter, but we were not able to determine the quality of care provided. The estimated effects of mental illness on health care experiences were based on reports of persons with a usual source of care and may be biased, because persons with mental illness are less likely to have a usual source of care (52). However, the direction of the bias may be conservative: persons who are unable to find a usual source of care may also have more difficulty interacting with providers and plans. We also cannot be sure that we excluded behavioral health clinics from the analysis, because respondents were not asked to identify the type of facility beyond hospital outpatient or emergency room or nonhospital facility. In addition, recall (four to five months) may vary by mental illness status.

Conclusions

Consumers consider their experiences when making plan and provider choices (53). Our findings point to modest levels of problems in patient experiences across the public and private sectors and identify actionable factors for improvement. Although the findings are tangential to our main analysis, persons assessed in 2011 or 2012, after implementation of the Affordable Care Act, were less likely to report problems. [A table in an online supplement presents the full regression model showing the significant AOR for postimplementation.] Future research should examine trends in provider experiences among persons with mental illness and factors that shape these experiences. With expanded access to care from the Affordable Care Act, it is important to continue to monitor experiences with plans and providers among individuals with mental illness to ensure that this vulnerable population receives accessible, coordinated, and appropriate care.