Using eleven years (2004-2014) of claims data from the largest US commercial health insurer, the investigators will assess the impact of switching into high-deductible health plans (HDHPs) on outcomes for patients with bipolar disorder. Patient subgroups will include patients with and without high medication cost-sharing and vulnerable populations (racial/ethnic minorities, poor, rural, major comorbidities). Interviews with patients and caregivers recruited through a major advocacy group will provide further insights into the policy issues with real-life experiences.

Bipolar disorder is a severe mental illness affecting about 3% of the U.S. population that causes personal suffering, morbidity, and premature mortality. Continuous access to medications, close monitoring, and other psychiatric care are crucial for avoiding complications of bipolar disorder such as relapse, hospitalization, and suicide. To control rising costs, payers and employers are increasingly adopting high-deductible health plans (HDHPs) with very high out-of-pocket payments. Federally-defined Health Savings Account HDHPs require full cost-sharing for all non-preventive services, including medications and specialist visits; family deductibles for HSA-HDHPs range from $2,500 to $12,700. Enrollment in HDHPs quadrupled nationally between 2006 and 2013 to 38% of all workers. Analysts expect further explosive growth because of continued health care cost pressure on families and employers. Well-informed patients in HDHPs might reduce use of unnecessary services and more expensive treatment options. However, patients might also choose to forego needed care. There is very little evidence on how particularly vulnerable patients such as those with bipolar disorder or other chronic mental illnesses fare when forced to make complex choices about spending for care under HDHPs. Given their rapid escalation, there is an urgent need to understand how vulnerable patients change their patterns of care and medication adherence under HDHPs. We will compare patients with three types of insurance: traditional plans with low or no deductible; HDHPs in which chronic medications are paid fully out-of-pocket until the deductible is met; and HDHPs where medications are subject to the same co-pays as in traditional plans.

OBJECTIVES:

Using ten years of data from the largest U.S. commercial health insurer (~70 million members in all 50 states), we will assess the impact of HDHPs on key outcomes for patients with bipolar disorder experiencing employer-mandated shifts from traditional insurance to HDHPs. Our specific aims are to evaluate: (1) changes in medication adherence, and in intensity and quality of other health care; (2) changes in adverse events; and, (3) changes in patient out-of-pocket costs. We will compare how these outcomes differ for patients in HDHPs with and without medications subject to the deductible. We will assess effects in the overall population of patients with bipolar illness and in specific vulnerable subgroups, including racial/ethnic minorities, poorer patients, rural patients, and patients with other important comorbidities.

METHODS:

We will take advantage of an ongoing natural experiment whereby employers have shifted all their employees at once from traditional insurance to HDHPs. We will use the strongest quasi-experimental, longitudinal methods available to compare the experience of patients switched by their employers into HDHPs with contemporaneous patients whose employers remain in traditional plans. A major advantage of our approach is the inclusion of only employers whose employees had no choice of insurance plans, minimizing member-level selection bias.

From preliminary data queries, we estimate a study population of ~160,000 members with bipolar disorder from 2004-2013. Our data include detailed information about insurance type, diagnoses, health services and pharmacy utilization, out-of-pocket payments, individual-level patient characteristics like income, and neighborhood-level factors like racial density. The unprecedented large sample size will allow us to answer questions about how patients from particular vulnerable subgroups respond to HDHPs, including patients who are Black or Hispanic, have low incomes, reside in rural areas, and have major comorbidities.

PATIENT OUTCOMES:

Our Aim 1 measures of the quality of bipolar treatment will include indicators of patients' access to appropriate care: prevalence and intensity of use of effective medications (antipsychotics, anticonvulsants); medication adherence; and, guideline-recommended clinical monitoring (regular outpatient mental health visits). Adverse events in Aim 2 will include psychiatric hospitalizations, which are potentially avoidable and often viewed as an indicator of suboptimal outpatient care. In Aim 3, we will assess changes under HDHPs in the co-payment amounts faced by patients for specific medical services, such as prescription fills and clinician visits, and the total burden of patient out-of-pocket costs.

PATIENT AND STAKEHOLDER ENGAGEMENT:

Our longstanding engagement with the National Alliance on Mental Illness (NAMI, the preeminent patient advocacy organization addressing issues around bipolar disorder) has shaped our study aims and our focus on measurable outcomes of particular concern to patients. We will solicit regular input from a local patient and family advisory panel (assembled with NAMI's assistance) on the refinement of methods, interpretation of study findings, shaping of recommendations, and dissemination of results. As study consultant, NAMI Medical Director Dr. Ken Duckworth will guide meetings of the patient panel and contribute perspectives from the broader community of patients and clinicians dealing with bipolar illness. Dr. Greg Simon, Director of the US Mental Health Research Network, will provide national expertise on patient experiences with serious mental illness in health plans.

ANTICIPATED IMPACT:

Our research will provide empirical data comparing how patients with bipolar illness fare under three insurance designs with vastly different requirements for cost-sharing. At a time when HDHP enrollment is exploding, the experience of patients with serious mental illnesses is largely unexamined. Advocacy groups will be able to use our findings to lobby for more patient-responsive benefit designs; policymakers will have evidence to redesign insurance benefits to better address the needs of vulnerable patients (e.g., by exempting mood stabilizing agents from deductibles).

Inclusion Criteria (for both study intervention/control groups):

[Intervention Cohort]:

• Traditional plan members with bipolar illness.
• Experience an employer-mandated switch to HSA-eligible HDHPs with full drug cost-sharing.

[Control Cohort]:

• Members with bipolar illness.
• Members whose employers offered only a traditional plan for the follow-up year.

Exclusion Criteria:

• Members age 65 years or older who could be eligible for Medicare benefits, including drug coverage through Medicare Part D.
• Members whose employer offered a choice of health plan.

Investigators will include health plan members aged 12 to 63 years at the beginning of the baseline period. Consistent with prior research,89-91 we will use medical claims data and a validated algorithm to identify members who had at least two ambulatory encounters or one hospital encounter with a diagnosis of bipolar disorder. Intervention cohorts will include traditional plan members with bipolar illness who experience an employer-mandated switch to HSA-eligible HDHPs with full drug cost-sharing or to HSA-ineligible HDHPs that subject medications only to copayments as in traditional health plans.

Per contract with our data vendor, we are unable to share individual participant data but will maintain records internally in accordance with funder guidelines.

3 years

This study was conducted retrospectively using claims data. Adverse events are not applicable to this study design.

This study had the following limitations: 1) observational and therefore subject to unmeasured confounding; 2) utilized a claims based diagnosis of bipolar disorder (not gold standard); 3) not generalizable to low-income people with bipolar disorder