Partnership for Implementation of Evidence-Based Practices in Rural Primary Care

The objective of this Implementation Trial is to evaluate the implementation of evidence based practices for Bipolar Disorder and Treatment Resistant Depression in small rural Federally Qualified Health Centers. The evidence based practices were chosen and adapted by Health Center providers and patients in consultation with clinical experts. The evidence based practices include: screening for and diagnosing Bipolar Disorder, prescribing one of five mood stabilizers, on-line cognitive behavioral therapy (Beating the Blues), on-line peer support (Depression and Bipolar Support Alliance), collaborative care management and tele-psychiatry consultation. These evidence based practices were chosen based on the strength of the research evidence and the feasibility of implementation in small primary clinics serving a low income rural patient population. Providers may choose to use all, any or none of the evidence based practices with their patients. Patients may choose to adhere to all, any, or none of the treatment recommendations prescribed by their provider.

A quasi-experimental study design will be used. There are Federally Qualified Health Centers participating in the Implementation Trial. Each Federally Qualified Health Center chose one implementation clinic and one control clinic. Patients will be enrolled from both implementation and control clinics and outcomes will be compared for the three month period following enrollment. Implementation outcomes are based on the "RE-AIM" framework and include: reach, adoption, implementation-fidelity and effectiveness. Reach represents the proportion of patients who receive evidence based practices. Reach will be compiled at the patient level. Adoption represents the proportion of providers delivering evidence based practices. Provider adoption will be compiled at the primary care provider level. Implementation-Fidelity represents whether the evidence based practices are being delivered as intended. Implementation will be compiled at the patient level. Effectiveness represents the clinical improvement experienced by patients. Effectiveness will be compiled at the patient level. Data will be collected from chart review and patient surveys will be administered by an Interactive Voice Response telephone system.

There will be three samples. The Full Sample will be all patients screening positive for depression on the Patient Health Questionnaire (PHQ9) screening tool during the 9 month enrollment period. The PHQ9 is used to routinely screen patients for depression at participating clinics. The Full Sample is expected to include approximately 2,400 patients. The Bipolar Depression Sample will be patients screening positive for depression, screening positive for Bipolar Disorder (on the CIDI), and providing informed consent to collect primary data. The Treatment Resistant Depression Sample will be patients screening positive for depression, screening negative for Bipolar Disorder, currently prescribed an antidepressant, and providing informed consent to collect primary data. The purpose of the "currently prescribed an antidepressant" inclusion criterion is to identify patients who do not respond to treatment. This inclusion criterion will be determined from chart review. We expect approximately 40 patients from each of six Federally Qualified Health Centers (two clinics per system) to be eligible for the Bipolar Depression and Treatment Resistant Depression samples and to provide informed consent. Thus, the combined size of the Bipolar Depression and Treatment Resistant Depression samples is expected to be 240.

Health Center evaluation staff will test differences in patient outcomes at implementation and control sites using ordinary least squares and logistic regression analyses that will control for demographic characteristics. Compared to patients at control sites, we hypothesize that patients with a positive depression screen at implementation sites will be more likely to be screened for Bipolar Disorder than patients at control sites. With automated chart review data for 2,400 patients, we will have 84% power to detect a 5% (e.g., 25% versus 20%) difference in Bipolar Disorder screening rates. Compared to patients at control sites, we hypothesize that patients in the Bipolar Disorder Sample at implementation sites will be more likely to see receive a Bipolar diagnosis, be prescribed a mood stabilizer, engage in on-line cognitive behavioral therapy, engage in on-line peer support, receive collaborative care management, have a tele-psychiatry consultation, experience depression symptom improvement and report greater satisfaction. For the Bipolar Disorder Sample, we plan to enroll approximately 20 patients from each of six Federally Qualified Health Centers (two clinics per system) and expect at least a 75% follow-up rate (n=90), which will give us 79% power to detect a 25% (e.g., 15% versus 40%) difference in Bipolar Disorder specific outcomes (e.g., diagnosed with Bipolar Disorder). Compared to patients at control sites, we hypothesize that patients in the Treatment Resistant Depression Sample at implementation sites will be more likely to have their antidepressant prescription changed, engage in on-line cognitive behavioral therapy, engage in on-line peer support, receive collaborative care management, have a tele-psychiatry consultation, experience depression symptom improvement and report greater satisfaction. For the Treatment Resistant Depression Sample, we plan to enroll approximately 20 patients from each of six Federally Qualified Health Centers (two clinics per system) and expect at least a 75% follow-up rate (n=90), which will give us 79% power to detect a 25% (e.g., 15% versus 40%) difference in Treatment Resistant Depression specific outcomes (e.g., antidepressant prescription changed). For the combined Bipolar Disorder Sample and Treatment Resistant Depression Sample, we plan to enroll approximately 40 patients from each of six Federally Qualified Health Centers (two clinics per system) and expect at least a 75% follow-up rate (n=180), which will give us 87% power to detect a 15% (e.g., 25% versus 40%) difference in trans-diagnostic outcomes (e.g., satisfaction).