Relevant Data Components
We anticipate that the advancement of precision mental health will require greater use of data sources not yet fully tapped by current approaches to mental health symptom assessment, such as educational- or employment-related functioning, cognitive and neurological testing, and other bio-social indicators (relevant to #1, 6, and 7 above). There is no conceptual reason why these data elements cannot be increasingly integrated into feedback systems, particularly as systems move to be largely digital and cloud-based with rapid real-time reporting possible (Lyon et al. 2016). However, since many of the measures developed in this area are laboratory-derived, a translational process might be necessary to make them feasible in the real world. For example, Bickman and colleagues developed a battery of measures that are designed for use in real-world settings where time is short (Bickman and Athay 2012).
Moreover, precision mental health provides an opportunity for the field to move beyond traditional self-report data. Almost all the data currently collected are based on clients’ or others’ completion of questionnaires. Although such an approach provides critical information about clients’ perceptions of their own difficulties, this mono-method dependency is problematic. While we are aware that we still have much work to do to integrate and understand self-report data (De Los Reyes 2011), we are missing new and rapidly emerging sources of information. For instance, Torous and Baker (2016), as well as many others, have noted that the new technologies based on smartphones and wearable sensors offer access to data and events that are not possible with electronic or paper-based questionnaires completed in the office or clinic. Although there are numerous complex issues that need to be resolved with the use of these new technologies (e.g., privacy, security, validity), there is significant potential to transform what we know about mental health and mental health services.
Among the data sources captured in the list presented above, information about the intervention itself (#4 above) is particularly underdeveloped. Physical medicine is going through a major cultural shift from the practice of medicine as an art to medicine that is evidence based and follows guidelines and standards. However, this has not been a simple journey, and some of the problems encountered may be remedied by an emphasis on precision medicine (Greenhalgh et al. 2014). Although many evidence-based treatments exist in mental health, research indicates that these are not yet part of the mainstream clinical culture (Becker et al. 2013). Moreover, there is currently little incentive for providers to use these treatments and monitor their fidelity. Thus, most care is described using the imprecise—and typically heterogeneous—term “treatment as usual”. Many of the feedback studies to date have introduced feedback practices into that “treatment as usual” context, which may not be optimal. This lack of precision in describing treatment is a handicap for feedback systems, because is it unclear not only what data to relay, but also what actions the clinician should take based on the feedback. The use of frameworks to identify intervention components (e.g., Chorpita et al. 2005; Michie et al. 2014) should continue to be advanced, but they are not yet widely embedded in practice, as will be noted in many of the contributions to this special issue.
Building Precision Mental Health Databases
The mental health field lacks high-quality, large databases that include linked data from #1 to #7 above. Databases currently available to form the basis for precision medicine are likely to be drawn from three sources: clinical trials, routine care, and cohort studies. While we could find no systematic data on the sizes of clinical trials, ClinicalTrials.gov, as of December 2015, lists 192,475 trials, 7366 (3.8 %) of which deal with some aspect of mental health. Most of these include some elements of #1–7, but not all. Furthermore, many will be limited in the populations covered. Cohort studies including those developed by groups of volunteers are a potentially useful source of data (Precision Medicine Initiative Working Group, 2015), but the mental health aspects of such databases are typically limited. For the foreseeable future, routine care is likely to be the key source of data for pursuing precision mental health. However, these datasets are likely to be highly flawed and incomplete, suffering from the challenges common to administrative datasets (e.g., missingness, inadequate specification) and exacerbated by the fact that, in mental health, we will have to depend on typical community-based treatment. Significant sources of data for health care are hospital data systems and laboratory test results. Hospitals and laboratories have a long history of collecting and maintaining relatively high-quality data, but outpatient mental health services often do not share this tradition. Furthermore, most existing data systems are not designed to “talk” to each other. This interoperability problem exists in physical medicine, but there are financial incentives for providers to develop such systems (e.g., Blumenthal and Tavenner 2010). Moreover, there are large investments being made by governments to create solutions.
Presently, ROM and MFS are in the forefront of developing technologies suitable for mental health to obtain the needed data. However, given the lack of similar incentives and financial resources, and the lack of standardized and widespread measurement, progress will be slow. The quality of mental health data from routinely collected data sources is therefore likely to remain a problem for some time to come. Many of the papers in this special issue deal with the problems inherent in collecting such data in the real world.
Facilitating Ease of Data Capture and Use in Mental Health
One of the major challenges this field faces concerns the implementation of data capture and use in the context of under-resourced and overstretched services. In many cases, new measures must be developed because the existing measures were developed for research projects without severe time restrictions for data collection. The resources available in research settings stand in contrast to the conditions of service delivery in the real world, where assessment is often seen as “stealing” time from treatment. Furthermore, the focus on monitoring makes more relevant individualized (i.e., idiographic) assessment approaches that are typically used for intra-individual comparisons (i.e., comparing individuals with themselves over time), rather than comparing individuals with established norms from a larger population (Haynes et al. 2009; Weisz et al. 2011). Many of the articles in this issue address the issue of implementation and draw on implementation science for suggested ways forward.
MFS and ROM Support Precision Mental Health
The current special issue contains two companion sections that showcase projects designed to support the elements of precision mental health listed above. They address some of the challenges previously identified via different technical (i.e., training, consultation, learning collaborative) and technological (i.e., digital measurement feedback systems and electronic health records) strategies. The special issue arose because of a range of work going on across the United States, United Kingdom, and elsewhere (e.g., the Netherlands) where researchers and practitioners were experiencing common challenges and concerns. Originally designed as two separate contributions, the commonalities between the groups became clear and therefore they were brought together in one issue while treating each section with its own introduction and overview. For specific information about the individual article author contributions, the reader is referred to the individual special section introductory papers. Specifically, Edbrooke-Childs, Wolpert and Deighton (this issue) have prepared a section focused on the use of patient-reported outcome measures (PROMs), which includes consideration of training and support necessary to allow for implementation. Lyon and Lewis (this issue) oversee a section that focuses on the development and implementation of digital MFS technologies explicitly designed to support ROM practice.
Papers in both sections stress that implementation and long-term sustainment of using patient-reported outcomes and other data to inform practice can be fraught with challenges, such as varying levels of organizational buy-in, long timelines, and mounting costs. Nevertheless, they also demonstrate the potential payoffs of successfully installing these innovations. Furthermore, the papers make it clear that the implementation of feedback technologies involves many of the same issues as those involved in the implementation of other evidence-based practice changes in behavioral health. Thus, they require good design and packaging to make them accessible and useable for practitioners, and to facilitate their uptake and long-term use. This may be accomplished by explicitly incorporating stakeholders and stakeholder perspectives into structured processes for the development, selection, and implementation of new innovations. Consistent with the broader implementation literature (Beidas and Kendall 2010), effective training and consultation procedures are necessary regardless of the type of innovation being implemented. Furthermore, both sections make clear the value of qualitative, quantitative, and mixed-methods approaches to (1) evaluate clinician and service recipient views toward the technological and practice changes that characterize implementation of feedback approaches, (2) tailor the practices or technologies to meet their needs, and (3) determine their effectiveness in promoting positive service outcomes. With this special issue, we hope to advance the science and practice of precision mental health by considering the capture, feedback, and use of data in community service settings, as well as the processes and strategies through which these innovations are developed, implemented, and evaluated.