Study Design and Oversight
A parallel-group randomized control trial was conducted with simple random allocation to either: 1) intensive psychotherapy and case management (IPCM) within the primary care clinic, or 2) care as usual (CAU), the clinic’s usual process for managing mental health concerns including referring to community providers or providing brief, onsite behavioral health support.
Ethical trial conduct and safety were overseen and approved by the institutional review boards of the University of Minnesota, Healtheast, and the Minnesota Department of Human Services.
Pragmatic Design
The growing call for more pragmatic studies [30] has led to increased analysis of their complexity. In any study, multiple components of design exist on continuums of highly pragmatic (i.e., Does the intervention work under usual conditions?) to highly explanatory (i.e., Does the intervention work under ideal, highly controlled conditions?) [38]. Accordingly, the key instrument framing conversations about pragmatic trials, the Pragmatic-Explanatory Continuum Indicator Summary (PRECIS-2) [39], uses a wheel format to capture a trial’s scores on nine domains that affect external validity, on one end of the continuum, and isolation of precise causal mechanisms, on the other. These domains include how representative the study is of real-world conditions in participant eligibility criteria, recruitment path, care setting, organisation (expertise or resources required), flexibility in care delivery, flexibility in patient adherence, intensity of follow-up, relevance of primary outcomes to patients, and inclusiveness of the primary analyses [39]. Refugees with high trauma exposure, insecurity of basic needs and high mobility, complex physical and mental health presentations, and multiple healthcare access barriers make a compelling case for studies to be highly pragmatic in order to be applicable to primary care settings charged with their care.
The present study’s design was pragmatic on seven of nine PRECIS-2 domains: (1) few eligibility criteria: adult Karen refugees from Burma, ages 18-65, with a diagnosis of Major Depressive Disorder (MDD); (2) recruitment occurred at the time of presentation and by primary care physician referral; (3) actual primary care setting; (4) providers had high flexibility in delivering the intervention of psychotherapy and case management according to patients’ self-chosen treatment goals; (5) patients had complete flexibility in adherence to the intervention; (6) the primary outcomes were highly relevant to refugee patients, representing common presenting symptoms and areas of social functioning CVT clinicians routinely address; and (7) data analyses included all available data using intention to treat. The two domains yielding a less pragmatic score were organization (we used highly skilled refugee behavioral health providers) and follow-up (although patients in both groups were followed with the same intensity of measurement, IPCM patients received a greater number of visits than is typical of behavioral health services in a U.S. primary care setting). Overall the study was thus pragmatic except in two areas integral to our understanding of effective treatment for this vulnerable population.
Patients, Setting, and Location
Recruitment occurred on a rolling basis from May 2013 – January 2017; data collection concluded in January 2018. Primary physicians at both clinics were aware of the parameters of the study as a randomized control trial for Karen refugee patients with depression and that eligibility (including current depression diagnosis) would be determined independently by a study clinician regardless of the patient’s current or past diagnoses. Karen refugee patients were referred by their primary physician based on the presence of depression symptoms (including unremitting pain as a common manifestation of depression in refugees [3-7]) in two urban primary care clinics in St Paul, Minnesota, USA. Potential participants were invited to meet with a study clinician and a professional interpreter who explained the study and obtained informed consent. For those participants who requested time to consider whether they wanted to enroll, the study clinician made arrangements to follow up with them at a later date, usually a week later. A Consort flow chart [40] of patients through the study is presented in Fig. 1.
After obtaining informed consent, the clinician conducted an eligibility assessment comprising the major depressive episode section of the Structured Clinical Interview for DSM-IV (SCID) [41], psychosis screening questions, and the CAGE-AID [42] for substance use. Inclusion criteria were: Karen refugee, ages 18-65, meets criteria for MDD according to SCID interview (MDD criteria did not change in DSM-V). Exclusion criteria were: current enrollment in individual psychotherapy or mental health case management [43], active psychosis that study providers determined was not culturally derived or trauma-related (many patients had psychotic-like symptoms such as seeing shadows and ghosts that were normative cultural expressions of distress and these were not excluded), chemical dependency or reported problems with non-prescribed drugs or alcohol on the CAGE-AID, and acute need at the time of screening for a higher level of care than the study provided (e.g., inpatient treatment). Ineligible patients received alternative referrals, and the referring physician was informed by message in the electronic health record so that care as usual could proceed in a timely manner.
Randomized Allocation
A coin toss by a research assistant otherwise uninvolved in the study was used to determine group allocation. Outcome assessors (research staff not involved in the intervention who administered the measures) were blind to group assignment.
Intervention Group
IPCM patients received services from both a psychotherapist and a case manager for one year. Depending on patient availability, appointments were weekly or bi-weekly and lasted 45 minutes – 1 hour. A professional interpreter was utilized unless the provider was a native Karen speaker.
CVT psychotherapists providing the intervention included 3 doctoral-level psychologists and 1 master-level clinical social worker. CVT case managers included 4 master-level social workers and 1 bachelor-level social worker. Additional training for clinical staff included: completion of a 10-module blended learning course created by CVT called Fundamentals of Providing Services to Torture Survivors [44]; Adult Mental Health Targeted Case Management training provided by the state of Minnesota [45]; individual clinical supervision conducted by senior CVT psychotherapists; participation in monthly psychological consultation and clinical social work group supervision with other CVT providers, and up to 60 hours of yearly continuing education training per individual provider.
Consistent with pragmatic randomized trial design to examine real-world practice with refugees [26, 29], providers delivering psychotherapy and case management tailored appropriate trauma and depression interventions to individual patients. Case management’s function was to help patients gain access to medical, social, educational, vocational and other necessary services connected to their mental health needs [46]. Case management interventions focused on re-establishing safety and stabilization [33, 34, 47]; facilitating communication, problem-solving and understanding between patients and medical providers [48]; and increasing skill in navigating health and community systems in resettlement [48]. Each patient and his/her case manager developed and worked from an Individual and Community Support Plan (ICSP) [46] that prioritized 3-5 goals, stated in the patient’s words (e.g., “I want to work to help my family with bills”; “I want to become U.S. citizen”). Core components of the case management are described in Table 1.
Psychotherapy functioned to increase patients’ coping skills and understanding of their symptoms, as well as to alleviate these symptoms and their impact. Psychotherapists taught mind-body awareness and relaxation skills calibrated to survivors of severe trauma and catastrophic losses. They provided education about the connections between trauma/stress and symptoms, use of medications and normative expectations for the doctor-patient relationship in Western culture (e.g., medications are not shared; dosage is not changed safely without consultation with one’s doctor; patient is expected to raise concerns proactively rather than waiting to be asked, etc.), and compensatory strategies for patients with significant memory or concentration impairments to take their medications accurately and follow behavioral aspects of their health care plan. Psychotherapists applied evidence-based treatments for PTSD and depression tested on refugee populations, including Narrative Exposure Therapy and Cognitive Behavior Therapy [3, 28]; they also utilised components of other psychoeducational approaches and trauma-focused treatments, such as Sensorimotor Psychotherapy [49], and patient-centered methods such as Motivational Interviewing [50]. Where psychological assessment and diagnoses had implications for needed services or benefits, psychotherapists advocated within medical, legal, and social service systems on behalf of individual patient needs related to mental health symptoms (e.g., completing waiver forms for the U.S. civics exam and English language requirement for U.S. citizenship, etc.). Common components of the psychotherapy are summarized in Table 1.
At the team level, CVT’s approach emphasized active interdisciplinary coordination and a relational focus anchored in cultural humility [51] to address survivors’ priorities as the primary architects of their healing and work together to co-construct meaning and behavioral change. CVT providers communicated frequently with one another and with patients’ primary care providers to address overarching themes and challenges in a patient’s care. CVT providers scanned their assessments and case notes into patients’ Electronic Medical Records (EMRs). CVT clinicians also communicated with physicians via the EMR and reviewed their patients’ EMRs with read-only access. When possible, clinicians communicated in-person with physicians spontaneously between patient sessions, in warm handoffs with patients, and in planned case consult meetings with physicians.
Trauma and loss were understood to have ongoing community-based sociopolitical and historical dimensions rather than being conceptualized as discrete past events that happened to individuals. Treatment was responsive to the instability and ongoing acculturation stressors in the lives of refugee patients requiring responses to multiple unplanned interruptions, including financial, housing, employment, family, and health-related crises. A Karen coordinator provided repeated, active follow-up to remind clients of appointments and overcome transportation and language barriers. Interventions addressed symptoms recognized by conventional biomedical culture and Karen idioms of distress identified by patients to their CVT providers. Common approaches used with the intervention group are described in a published toolkit for serving refugees in primary care settings [52].
Table 1. Functions and Components of Psychotherapy and Case Management Intervention
_____________________________________________________________________________________________
Case Management
Function: assist patients to gain access to medical, social, educational, vocational and other necessary services connected to their mental health needs
Components:
- Assessing patients’ needs and goals and impact of mental illness, and incorporating patients’ strengths and progress toward goals
- Planning goals and goal-related steps, updating the individual and community support plan, finding new resources
- Referring and linking to resources, supports and services
- Coordinating with medical providers, community resources and natural supports identified by each patient as important to his or her recovery process
- Monitoring the effectiveness of the resources, supports and services being utilized, especially with respect to refugees navigating health and community systems in resettlement
- Discussing the progress made toward goals
- Advocating as case managers on behalf of the patients’ mental health needs with medical, legal and social systems
Psychotherapy
Function: increase patients’ coping skills and understanding of their symptoms; alleviate symptoms and their impact
Components:
- Facilitating mind-body awareness; teaching and practicing relaxation skills
- Providing psychoeducation on the relationship between trauma/stress and symptoms, treatment options for mental health symptoms, use of medications, and the doctor-patient relationship in Western medical culture
- Developing and teaching compensatory strategies for taking medications accurately and following health plan instructions that accommodate impairments in memory/concentration and other mental health symptoms
- Applying evidence-based trauma-focused treatments to reduce symptoms of depression, anxiety, and posttraumatic stress
- Problem-solving with patients to decrease impact of symptoms and distress by changing coping behaviors and thought patterns
- Advocating as psychotherapists on behalf of patients’ mental health needs with medical, legal and social systems
_____________________________________________________________________________________
Control Group
Participants in the control group received care as usual, without CVT involvement beyond administration of outcome measures. Once randomized, CAU patients could be referred to a full range of behavioral health services by their primary care physician. Use of behavioral health services by patients in the CAU groups was monitored by primary care providers but not by the study.
Data Collection and Measurement
Demographic characteristics were collected prior to randomization. Pre-specified outcomes were mean change in depression, anxiety, PTSD, pain and social functioning scores over the year of enrollment. Outcomes were collected at baseline, 3, 6 and 12 months using instruments found to be reliable and valid with refugee populations. Presence and severity scores of symptoms associated with MDD and Generalized Anxiety were measured on a 4-point Likert scale using the Hopkins Symptom Checklist-25 (HSCL-25) [53]. Presence and severity of symptoms associated with PTSD were similarly measured using Part 3 (17 PTSD symptoms) of the Posttraumatic Diagnostic Scale (PDS) [54] adapted to assess DSM-V diagnostic criteria. Presence and severity of pain was measured using an internally developed 5-item Pain Scale with adequate internal consistency of α =.76. Social functioning in meeting basic needs, stabilization, employment, social support, adjustment, and community engagement was measured with a 37-item standardized instrument on a 7-point Likert scale validated with refugees [55]. Instruments were selected based on extensive research indicating high prevalence of depression, anxiety, PTSD and pain in refugee populations [3, 56, 57]. Torture, war, and resettlement also impact social functioning, including basic needs, legal status, social support and involvement, employment and education, and engagement with one’s geographic community. Measures were administered by a trained assessor, blinded to treatment condition, who followed scripted protocols and used a professional interpreter. Assessors had no contact with CVT providers to minimize breaches to blindness and bias. The only exception occurred when a participant expressed intent to harm self or others. In these instances, the protocol allowed for appropriate crisis response without breaching assessor blindness.
Sample size
Power analysis was originally conducted a priori using depression symptom scores as the outcome variable based on assumption of a 20% attrition rate [58]. Due to lower than expected attrition (10%), we re-calculated a sample size of at least 95 in each treatment group (190 participants total) to detect statistical significance at the alpha <.05 level with power of 80% or greater.
Analysis
Mean (SD) baseline characteristics of participants randomized to the intervention or control groups were analyzed using t-tests for continuous and chi square tests for categorical data (see Table 2). Standardized t-scores were created for all outcomes using the normed population distribution collected at CVT [59]. All dependent variables met the statistical assumptions of normality, independence, homoscedasticity and sphericity prior to inferential analysis. Treatment effects were examined through repeated measures analysis of variance. Comparisons between groups were pre-specified and all tests were two-sided. Pairwise comparisons were performed post hoc with Sidak adjustment for comparison of mean scores at each time point between groups. All analyses were conducted according to intention-to-treat methods [60]. An alpha cutoff of p ≤ .05 was used to assess statistical significance. Effect sizes were calculated using partial eta squared and interpreted as 0.010 - .059 = small, 0.060 -.139 = medium, > 0.14 = large [61]. Statistical analyses were conducted in Statistical Package for the Social Sciences (SPSS) 24 [62] and R version 3.4.4. [63]. As cases were allocated in a non-random way, weighing pragmatic concerns of case load, provider availability, and where possible gender matching, no analysis of provider-related differences was included in the analysis.