Despite the proven efficacy of pharmacotherapy and psychotherapy for treatment of depression, many depressed primary care patients still do not receive adequate treatment. Moreover, African Americans and other ethnic minorities have lower rates of guideline-concordant treatment for depression than whites. Studies have identified patient-clinician communication and patient cultural beliefs, attitudes, and preferences as potential targets for interventions to improve outcomes and reduce disparities in care. Current state-of-the-art interventions still struggle to achieve one-year remission rates over 65% for patients with major depression [53
]. While these interventions result in clinical improvement for ethnic minority patients, disparities in receipt of guideline-concordant care and functional outcomes are not eliminated [19
This study compares the effectiveness of a patient-centered, culturally targeted adaptation with the standard conceptually-based QI intervention that was proven to be helpful for improving care for depression. The adapted intervention incorporates several successful features of previous QI interventions while also introducing novel elements. Both interventions use multifaceted approaches that include educational and epidemiological strategies to target intrinsic motivation and rational decision-making among health professionals [54
]. The patient-centered adaptation aims to enhance clinicians' participatory decision-making skills and uses marketing strategies to adjust educational products and care management services to the needs of African Americans [54
Limitations of the study should be discussed. The cluster design presents unique challenges with regard to comparability of groups (at the cluster and individual patient level), allocation concealment, maintenance of ITT principles, empty clusters, and participant switches from one cluster to another [55
]. Several strategies have been recommended to overcome recruitment challenges: monitoring cluster guardian (e.g
., practice leader) adherence to the study protocol before randomization of the cluster; including at least one patient participant before randomizing a cluster; and using recruiters who are blinded to participant allocation status. In this study, there was loss to follow-up among randomized clinicians before patient recruitment, which led to empty clusters (e.g
., clinicians who did not contribute patients to the study) and failure to reach the recruitment target among patients, which may compromise the study's statistical power to detect differences in the primary outcome. Due to the staff resources available for recruitment and time frame needed for delivery of the clinician interventions, it was not feasible to delay randomization of clinicians until the first patient in each cluster was recruited. However, recruiters in this study were blinded to the allocation status of each cluster and its patients. With regard to analysis challenges, empty clusters and patients who switch from one cluster to another may lead to violation of the ITT principle. Recommendations include taking into account both clusters and individuals that withdraw or are lost to follow-up by using ad hoc
missing data methods; keeping clusters as they were randomized, and using adjustment or propensity-score methods to deal with potential imbalances in both cluster and individual characteristics. In addition to these strategies, we will conduct sensitivity analyses to examine the potential impact of deviation from assumptions underlying such analytic approaches on the robustness of the study findings. Another design constraint is the limited ability to identify which aspects of the multi component interventions are most effective; we intend to conduct post-hoc
analyses to determine whether the number of intervention contacts, use of intervention materials, and rating of intervention effectiveness by clinicians and patients are related to study outcomes.
Limitations of the interventions include the lack of booster exposures for clinicians and the reliance on telephone contacts and limited focus on enhancing access to psychotherapy for patients. These limitations were dictated by concerns about interference of the clinician interventions with clinical care and productivity; lack of office space for interventionists; and heterogeneous access to mental health specialists among patients with different health insurance coverage. Data collection and follow-up limitations include the reliance on self-report measures of process and health outcomes and relatively short follow-up time frame for assessment of functional outcomes from patients.
Still, the BRIDGE study addresses several limitations of previous studies. It includes under-resourced community-based practices and clinicians and African-American patients who have not been well-represented in previous clinical trials to improve depression care. Moreover, the patient-centered and culturally tailored approach to depression care is a relatively new one that has not been tested in most previous studies. The investigators found establishing strong relationships and aligning priorities with practice leadership early in the design of the study and throughout its implementation enhanced the success of recruitment at particular sites. Additionally, drawing on the strengths of site-specific administrative staff and electronic medical resources, when available, and adapting to the needs of clinicians and practices increased the efficiency and effectiveness of patient recruitment and acceptability and delivery of the interventions. The BRIDGE study will provide knowledge about how to recruit and enroll clinicians and patients into community-based participatory research programs; how to change clinicians' behaviors with regard to depression care in actual practice; and how to improve acceptability of treatments, quality of care, and depression outcomes for African Americans and other ethnic minorities with depression.