Full Summary – HEALing Communities Study Design Workshop
Introductory remarks were offered by ADM Brett P. Giroir, M.D., Assistant Secretary for Health and Senior Advisor for Mental Health and Opioid Policy, U.S. Department of Health and Human Services. ADM Giroir highlighted the scope and urgency of the opioid crisis and the vision for the study, with an emphasis on the devastation of the opioid crisis and the ambitious solutions that are required to meet the goal of decreasing fatalities while accounting for challenges in geographic diversity. This collaborative effort between NIDA and the Substance Abuse and Mental Health Services Administration will support a study that leverages existing interventions, data sources, and comprehensive systems of care to create sustainable and potentially replicable innovations.
Meeting Goals and Structure
Over the course of the one-day meeting, six discussion sessions focused on challenges and options related to: (1) study design; (2) study outcomes; (3) the opioid prevention and treatment cascade; (4) health economics research; (5) implementation research; and (6) infrastructure, partnerships, and collaborations. A summary of major issues discussed in each session is provided below. To view the proceedings, see the archived video at: https://videocast.nih.gov/summary.asp?Live=27969&bhcp=1.
Session 1: Study Design
The first part of this session focused on site selection; the second addressed research design options.
Measurably reducing overdose fatalities is an important outcome for the study, which adds complexity as many factors outside the control of researchers contribute to overdose deaths. Accurately measuring overdose events and fatalities is challenging, and there is wide variability in the measurement and categorization of overdose events, the quality of data, and trajectories of opioid use across communities. In addition to using overdose rates, other ways to identify communities affected by the opioid crisis might include the development of composite measures incorporating information on rates of opioid use disorder (OUD), overdose fatalities and events, and rates of neonatal opioid withdrawal syndrome. Modeling could be used to identify the geographic areas most at risk based on methods—developed by investigators and the Centers for Disease Control and Prevention—that use data on community-level risk and protective factors. The study context varies across communities in terms of factors such as difficulty with transportation, drug supply (in particular, fentanyl availability), law enforcement, treatment capacity, workforce shortage, and economic conditions. Communities could be defined as counties, neighborhoods, or other justifiable geographic areas within states that have high opioid overdose mortality and opioid-related morbidity. Rural areas have unique challenges (e.g., lack of transportation) that will need to be taken into consideration.
Several study design options were presented along with their strengths and challenges, including:
- Group- or Cluster-Randomized Trial is well suited for evaluation of multilevel interventions and supports strong causal inferences. This study design does not require multiple waves of data collection; however, it requires a large number of sites.
- Stepped Wedge Design also supports strong causal inferences with all sites receiving the intervention. This study design requires a longer study duration with more frequent data collection.
- Regression Discontinuity Design uses a quantitative score to determine intervention assignment, and causal inferences would be supported. This study design requires a larger number of sites and has not yet been used for multilevel interventions.
- Time Series Design could be done across multiple sites and is well suited to study health and/or policy interventions. This study design requires repeated and reliable data measurements, including baseline data.
- Adaptive/Flexible Intervention, also called stepped care, allows for intervention adaptations to be made over time to reflect responses to the intervention. This study design requires specifying in advance the decision points when changes to the intervention may be made, the tailoring variables used to make those decisions, the intervention options, and the decision rules. Adaptive interventions are more difficult to employ in the context of a multilevel intervention.
Effect sizes could be estimated in multiple ways: pilot study, meta-analysis, modeling, or determination of the primary outcome and establishment of a meaningful change based on clinical or public health impact.
A careful balance will be required between standardizing the interventions and implementing interventions that are tailored to each locality, and it will be important to have a centralized data repository for rapid evaluation of findings. Existing resources, systems, and local knowledge should be incorporated into the study design, and confounding variables will need to be tracked carefully.
Session 2: Study Outcomes
The prevention and treatment cascade was presented as an orienting framework for outcomes. For OUD, this cascade includes diagnosis, linkage to care, medication-assisted treatment (MAT) initiation, ongoing MAT for at least six months, and sustained recovery. In the progression from diagnosis of OUD through sustained recovery, successful movement from one point on the treatment cascade to the next is dependent on successful engagement in care on the prior step (e.g., receiving MAT is dependent on receiving a diagnosis of OUD). Three levels of study outcomes were discussed: systematic structural outcomes (or outputs), organizational process outcomes (or outputs), and individual and public health outcomes. While it was agreed that measurable reduction of overdose fatalities is the ultimate goal, it is also important to measure more proximal outcomes (outputs) as well as outcomes related to prevention, drug use trajectory, and quality of life. Decreasing nonfatal overdose rates, as well as opioid misuse and addiction rates, while increasing retention in MAT treatment for more than 6 months and the provision of recovery support services are key secondary outcomes.
Structural outcomes will improve service capacity and coordination across community domains. Structural outcomes to consider include prescribing patterns, the availability of prevention and treatment services in communities, the number of specialized addiction treatment providers that offer MAT, the number of general medical providers offering MAT, the availability of naloxone, infrastructure to connect individuals who have experienced non-fatal overdoses to treatment, the availability of screening and brief intervention for opioid use disorder, and collaboration between sectors to address the unique challenges of individuals who are re-entering the community after incarceration.
Challenges to measuring these outcomes include variable data quality, lag time in data availability, and a lack of baseline data in some domains. Also, careful consideration is needed for analysis of how changes in contextual variables such as drug supply will affect the outcomes. Data quality, including that of data characterizing overdose fatalities, is essential and important to address in the study.
Session 3: Opioid Prevention and Treatment Cascade
Evidence-based prevention provided during early adolescence can make a meaningful contribution toward reducing addiction in the long term. The chronic relapsing nature of opioid addiction may require iterative progression through the treatment cascade. Crucial areas to address are the extent to which prevention and care is evidence-based and the availability of necessary preventive services, treatment services, and recovery supports, ranging from effective preventive interventions for early use and misuse of opioids to transportation to service provider locations to screening, assessment, and intervention for OUD.
Session 4: Health Economics Research
Ensuring that the HEALing Communities Study addresses health economics research questions will be vital to informing system design and providing evidence for policy development. Economic evaluation of the interventions should extend beyond cost assessment to answer questions about the economic impact and cost-effectiveness of increasing access to prevention and treatment services for OUD. Quality of life and qualitative assessments should also be incorporated into economic analyses.
Session 5: Implementation Research
Effective proven interventions may yield limited impact if the quality of implementation is inadequate or deteriorates over time. The intervention must be delivered with fidelity, reach the intended population, and be sustained over time; each of these steps can be negatively impacted by stigma. In opioid treatment, implementation of proven interventions will be important due to the inadequacy of many treatment practices, such as use of drug-free treatment instead of MAT, and inadequate counseling and recovery supports in conjunction with MAT. Capacity expansion is needed, so it will be necessary to address workforce shortages and improvement of care through both training and coaching. Moreover, interventions will need to be structured for maximum sustainability.
Session 6: Infrastructure, Partnerships, and Collaboration
Throughout the meeting, experts also emphasized the need to engage multiple systems within a community or state to ensure sustainability. In particular, engaging state governments is critical and could be facilitated by partnering with governmental officials able to influence these multiple systems. Advisory councils can support collaboration by coordinating input from the full range of stakeholders, including state and community representatives and people in recovery. It will be important to solicit input from the communities to identify both specific needs and existing programs and resources that could meet those needs. Other initiatives are likely ongoing in these areas of high need, and HEALing Communities should build on these.