Epstein, Irwin. (2010). Clinical data-mining: Integrating practice and research. London. Oxford University Press. Pages: 228. Cost: US$24.96
Reviewed by Clare O’Callaghan Phd RMT, Music Therapist at Peter MacCallum Cancer Centre & Caritas Christi Hospice, St Vincent’s Health/Department of Medicine & Faculty of the VCA and Music, The University of Melbourne, Australia
In multidisciplinary health care, the “practice research divide” is a phrase used to signify researchers’ frustration that practitioners may be reluctant to use research findings to inform their work, and practitioners’ frustration with difficulties using some research findings to inform practice. For example, intricate descriptions of elements involved in music therapy relationships are essential before one can consider research findings’ relevance to their own work contexts, however, these may not be detailed. While scepticism between social work practitioners and researchers described in Prof Epstein’s, “Clinical data-mining: Integrating practice and research”, may not be representative of music therapist researchers and practitioners, the book, nonetheless, offers exemplarily guidance for how therapists and researchers can work together, or therapists can work as researchers, to develop research knowledge highly relevant to practice. Importantly, the book illustrates how this can be done with data already existing in our workplaces. Through clinical data-mining (CDM), music therapists can do low or no cost research which can make work-life more satisfying, interesting, accountable, and professional.
I first came across Prof Epstein’s work when I attended a lecture he gave on CDM for allied health professionals in Melbourne, Australia, some years ago. In the book, Prof Epstein describes a 40 year history of teaching traditional research methods to social workers, wherein he published some seminal texts for his field. Concerned about the aforementioned “practice research divide” ten years ago, Prof Epstein focussed on seeking ways to engage practitioners in research. Certainly his description of CDM studies immediately engages this practitioner-researcher: “CDM studies are conducted by practitioners, are driven by practice-questions, and make use of practitioner-generated information, their primary purpose is to inform clinical, programmatic, or administrative decision making” (p. 12).
In the past decade, Prof Epstein has conducted many international CDM workshops for health carers, including in Australia and New Zealand. After distinguishing between practice-base research and research-based practice in the book, CDM is defined:
Clinical data-mining is a practice-based, retrospective research strategy whereby practitioner-researchers, alone or with the assistance of a research consultant, systematically retrieve, codify, analyse and interpret available qualitative and/or quantitative data from their own and other agency records in order to reflect on the practice, program and policy implications of their findings. (p. 71)
CDM as a method is located within Prof Epstein’s advocacy that health care work needs to be “evidence informed”. Evidence informed practice (EIP) is methodologically, structurally, and culturally pluralistic: “it values all ways of knowing and seeking useful knowledge – everything from RCTs to qualitative case studies … it values practice wisdom as well as research-based knowledge (and) positions practitioners as knowledge producers alongside academic producers” (p. 34).
Methodologically, three types of CDM studies are described. Those that begin with: (a) available quantitative data (e.g., work statistics) which is converted into a quantitative database for statistical analysis; (b) narrative data (e.gs., case notes, diaries) which is converted to a quantitative database and analysed statistically; and (c) qualitative data used for subsequent qualitative analysis. These may also be combined.
Prof Epstein then describes research studies, mostly postgraduate dissertations, to illustrate the CDM types. (Prof Epstein includes a study from my own Phd, which is described as a qualitative CDM.) This book made me realise that much of my music therapy research is CDM, including: qualitative research on practitioners’ wisdom through analysis of their journals or stories, and song lyric analysis. Content analyses on lyric analyses would be examples of narrative data being converted to quantitative. Also in my experience, CDM studies do not necessarily need tangible “records”: qualitative CDM studies can be conducted on music therapists’ practice wisdom through the transformation of clinical memories and reflections into text for analysis. Other music therapists’ research may also be conceptualised as the CDM of departmental and hospital records. Two apparent examples include Horne-Thompson, Daveson, and Hogan’s (2007) analysis of music therapy referral trends from palliative care team members, and Hilliard’s (2004) post-hoc analysis of music therapy services for nursing home residents receiving hospice care. Now I am wondering what other records in my own and other music therapists’ practices could be harnessed for research to enhance service provision. Potential univariate (e.g., service requests and presenting problems), bivariate (e.g., relationship between interventions and outcomes), and multivariate (e.gs., relationship between interventions, outcomes, and diagnostic or demographic factors) analyses suggested for social workers (pp. 127-8) are certainly also applicable for music therapists.
Prof Epstein is aware that, like all research methods, CDM has some “drawbacks”, such as “missing data” and discrepancies amongst workers about how work is described and records maintained (in CDM studies on departmental records), which affects the reliability and validity of practitioner-generated information. But, Prof Epstein questions, if practitioners’ records concerning clients’ attributes and outcomes and practitioner interventions are regarded as unreliable and invalid, “what justification is there for practitioners to work with and make intervention decisions about patients or clients in the first place?” (p. 66)
CDM is advantageous because of cost savings in using existing data within practices compared with creating data for prospective analysis. Perhaps there should be more endeavours to channel research grants into practice-based research, to maximise capacity from the health care/research dollar. I was, however, disheartened to read Prof Epstein’s frustrations with securing funding for CDM research. While music therapists may encounter similar frustrations, the good news is that: hospital ethics review committees are approving CDM studies; CDM studies are getting published in peer reviewed journals; and A/Prof Lynette Joubert from The University of Melbourne (who has done extensive work with Prof Epstein) was recently appointed as Honorary Senior Research Fellow in the Social Work Department at Peter MacCallum Cancer Centre (Melbourne, Australia) to develop research practices, including CDM. Perhaps music therapists could work toward the funding of comparable researcher-practitioner partnerships. I was fortunately involved in such partnerships in recent post-doctoral research. Shared research on music therapists’ practice wisdom can be a valuable learning and team building process:
This reflexive group supervision process changed the way some of the therapists worked with (palliative care) clients. In the final group …, some spoke of coming to regard music therapy sessions more as single sessions, because many patients could die at any time. They also felt more comfortable about using unpaid time to help patients to complete legacies when they felt it was important …. The therapists learned from hearing about each other’s practice styles. … people felt respected and not judged …. Three colleagues from one organization thought that the process had helped them as a team and felt more confident that the other therapists could work effectively with their own clients, if necessary. (O’Callaghan, Petering, Thomas, & Crappsley, 2009, p. 202)
This is a highly engaging book with interesting CDM research descriptions spanning widely varied clinical populations (e.gs., paediatric diabetes, adolescent mental health, domestic violence, liver transplantation, geriatrics, palliative care) across the lifespan. Some music therapists may question: “Why do we need such a book if we are doing some of this kind of research already? Do we need more jargon?” I found that the book’s research framework extended my thinking about how to justify the merits of my research, as well as furthered my thoughts about research possibilities. I believe that many music therapists – practitioners, researchers, and practitioner-researchers – could also find something useful and inspiring in this text. I recommend that you read it and imagine the rich data that awaits harvest for CDM studies in your own clinical contexts.
Hilliard, R. E. (2004). A post-hoc analysis of music therapy services for residents in nursing homes receiving hospice care. Journal of Music Therapy 41(4), 266-281.
Horne-Thompson, A., Daveson, B., & Hogan, B. (2007). A project investigating music therapy referral trends within palliative care: An Australian perspective. Journal of Music Therapy, 44(2), 139-155.
O’Callaghan, C., Petering, H., Thomas, A., & Crappsley, R. (2009). Dealing with palliative care patients’ incomplete music therapy legacies: Reflexive group supervision research. Journal of Palliative Care 25(3), 197-205.