Adapting Psychosocial Intervention Research to Urban Primary Care Environments: A Case Example

INTRODUCTION

During the past 3 decades, primary care practice-based research has made notable progress in establishing its feasibility and potential to address questions important to primary care and the communities we serve.^1–,⁵ Practice-based research can answer questions of greater relevance to primary care practice, test effectiveness of treatments in undifferentiated patient populations,³ and engage clinicians in the generation of new knowledge that can be readily assimilated into practice.⁵ Despite gains made, researchers note that challenges persist in successful implementation of studies in primary care settings.^6–,⁸ Moving psychosocial intervention research into urban community-based primary care settings that are not research oriented remains highly challenging. Disappointing results of well-designed intervention research may be due to practical difficulties of implementing studies in unpredictable community-based health centers rather than to lack of intervention efficacy.^9–,¹¹ Barriers, expected and unexpected, to implementing intervention research in community-based settings will remain^1,2,12 and will likely be exacerbated in the unique setting of safety net sites.¹³

In this article, we describe logistical hurdles encountered in testing an intervention to reduce depression in pregnant, low-income urban minority women in community health centers. We describe the varied elements that affected our research: the persons who receive interventions, the persons who provide interventions, and the persons who collectively create service delivery systems—contexts subject to the effects of unanticipated events. We offer several criteria for adapting intervention research to real-world circumstances.

METHODS

Participants were 187 African American and His-panic women with low-risk pregnancies (mean age = 25 years; mean education = 12 years). Women were screened for depressive symptoms in the third trimester, and those with elevated depressive symptoms were randomly assigned to an intervention group (n = 57) or group receiving treatment as usual (n = 43) offered by their health center. A third group of women who were not depressed (n = 87) was used for comparison. Data were collected again at 2 weeks’ and 3 months’ postpartum.^14,15 Depression was measured with the Beck Depression Inventory, second edition (BDI-II ), in which a score of 14 is considered the lower end of depressive symptoms.¹⁶ Total functional social support (ie, actual received support) was measured with the Norbeck Social Support Questionnaire, an instrument for use with pregnant women.¹⁷ Life stressors were measured with the Life Events Questionnaire,¹⁸ and functional abilities were measured with the Medical Outcome Study Short Form-36 (SF-36).¹⁹

The intervention consisted of 3 components: an 8-session cognitive behavioral treatment specifically developed for depression prevention in primary care,²⁰ 4 psychoeducation sessions using videotaped and written materials on infant development and maternal sensitivity, and ongoing social support building from the therapist. Therapists were graduate students in social work (all women of an ethnic minority completing clinical internships) who were trained and supervised in both the intervention and cultural sensitivity. Intervention sessions were offered in the health center or at home at least twice a month.

The settings were 3 health centers in the south Bronx, NY, all of which had prenatal teams of physicians, nurses, and social workers. The 2 larger, federally funded community health centers had formal psychosocial service teams. The smallest site was a family practice setting with a part-time social worker. The sites had little experience with research and none with intervention studies.

RESULTS

We describe the challenges and lessons learned in implementing and measuring study effects. Thirty-eight percent of the sample was lost to attrition. Of this 38%, 42% and 44% of those in the intervention and treatment-as-usual groups, respectively, did not complete the study (Table 1⇓). Only 32% of the women who were not depressed dropped out of the study. No significant differences were apparent in changes in total functional social support or reduction of depression in the 2 treatment groups.

Methodological and logistical challenges arose that affected outcomes, some from the realities of the research contexts and some through errors of commission and omission.

CONCLUSION

We concur with others that the adaptations and findings of field trials may actually be stronger than those of laboratory studies because of the more demanding, fl uid research environments in which they are conducted.²⁵ Thus, practice-based research has greater relevance and applicability to the primary care patients we serve.^2,3 We affirm the reality that intervention research is difficult to implement faithfully in urban primary care settings, especially those without a tradition or infrastructure for research. Our project represented an academic-community partnership, but we had insufficient involvement of the community practices to facilitate ownership at the site level or optimal problem solving, especially beyond its initial months. While we were attentive to cultural sensitivity, we lacked the true collaboration with participants necessary to define common goals and outcomes.^6,22

Our key message is that psychosocial intervention research in community-based primary care will require fl exibility in adjusting the research design before and during implementation.¹⁰ The concerns of multiple stakeholders, nonresearch settings, and real-life issues in patients’ lives can be reconciled by research designs that result in cycles of outcome assessments and procedure refinements.²⁶

Observing several criteria may help adapt research design to stakeholder and service system characteristics:

1. Assess what the target population can and will accept of an intervention. Using participatory research approaches, patients and knowledgeable staff can answer such questions as, What are the key issues in patients’ lives that are essential to target with the intervention? What kinds of interventions and in what dosages will patients accept? What kinds of outcomes do patients want and in what domains? Which outcomes will be most compelling to patients? Answers to these questions are contingent on understanding service and community contexts early on in a project. Operationally, this understanding may be accomplished by partnering with community representatives and conducting key stakeholder focus groups and interviews.

2. Aim toward interventions that clinicians in comparable settings can deliver, given the typical demands of practice in community settings. This criterion would pose such questions as, What will clinicians in community settings like this one need to know? What intervention and how much of the intervention can clinicians reasonably be expected to deliver? What will they embrace and what will they reject? Clinicians’ acceptance of empirically-based interventions in these circumstances is more likely to occur when combinations of treatments with small doses are available and where training requirements are not excessive. Despite the demands of the primary care setting, training on-site personnel to deliver the intervention might raise their ownership of the study and help identify potential problems and realistic solutions.

3. Assess the service setting’s capacity to accommodate intervention research. How does the intervention researcher integrate externally and internally induced changes in the service system into the research protocol? Countless predictable and unpredictable changes in participating practices can disrupt implementation and data collection. In every dimension of research, adaptations might be required that deviate from the original plans; these adaptations can be creatively pursued while maintaining the fundamental research design. This criterion obviates the need for ongoing discussions with providers and administrators in preparing for impending changes, particularly in nonresearch environments. Looking ahead at looming policy or fiscal realities that will bring changes to service systems and asking how the centers managed similar upheavals in the past can prepare the research project staff to reduce interruptions and adapt in advance.

4. Collect emerging data systematically that are unanticipated or not easily quantifiable but which may be important later. Typically, this information is not originally considered as part of the data to be collected. Yet the events that occur can have powerful effects on research implementation. The question then becomes, How can emerging, even anecdotal information, be collected and reviewed systematically to make appropriate design adaptations while maintaining scientific integrity? We recommend encouraging regular input of anecdotes and hunches from all members of the team, recording and examining these systematically for indications of trends and troubles. Our experience shows that unexpected findings or implementation problems can lead to creative and useful research adaptations.