Using Practice-generated Patient and Outcomes Data to Perform Pain Management Research

There are several tools and resources available to pain practitioners who want to use their patient data to conduct a clinical trial.

During a session at the 2013 American Academy of Pain Medicine annual meeting, held April 12-14 in Fort Lauderdale, FL, Ajay D. Wasan, MD, and Robert R. Edwards, PhD, of Brigham and Women’s Hospital in Boston, MA, and R. Norman Harden, MD, of Northwestern University in Chicago, IL, described how clinicians can “utilize their own treatment outcomes to provide opportunities for clinical research,” explored the process of initiating clinical research and the basics of analyzing data, and explained how to systematically assess, track, and analyze treatment outcomes in routine clinical practice.

Many practitioners are unfamiliar with using treatment outcomes from patients in their clinical practice as part of clinical research, but are interested in learning how to do so. The presenters explained how to create a system in practice to track outcomes, outlined several key clinical research considerations, and reviewed basic statistical concepts and principles.

They reminded audience members that good clinical care can result in good clinical research at the same time if physicians learn to pay attention to what works and what doesn’t in clinical practice, and then effectively translate that into their clinical research efforts.

During the session, Wasan discussed several clinical research tools that are easily accessible and explained how to design a program within a physician’s practice that can benefit both patients as well as clinical research efforts. Wasan said these tools are simple to use and could be incorporated into general practice. For example, the brief pain score (BPS) is a simple questionnaire that clinicians can incorporate into their practice, which can help them understand how much pain patients are experiencing and how pain interferes with activities of daily living. In addition, other tools are available to measure sleep and function. Wasan noted that collaborating with others that are already involved in research or that could provide capabilities that a particular practice doesn’t have can advance research efforts. He said that research is “hard but doable if physicians persevere,” especially if they get excited about the innovative process and learn how to collaborate.

Harden expanded upon several practical clinical research paradigms and provided an overview on successfully transitioning clinical practice efforts into clinical research. He discussed study methods and design, barriers, disease models, study design, outcomes, functional scales used, and the publication process. At each step, he provided insight into the types of study methods (case reports/case series, retrospective vs. prospective, randomized, controlled, inclusion/exclusion, and stats), barriers to participation (randomized/ethics, control, blinding impacts, economics, referral bias and outcomes to use), study designs (cross sectional, parallel, cross over, blinded, controlled and small study population), blinding methods (open label, single label, double label and blind analysis) and publication (rationale/intro, protocol/methods, data analysis/results, and conclusions).

Edwards concluded the session with a discussion on basic statistics. While statistics may appear intimidating to the clinical practitioner who has never performed clinical research, basic statistics should not be overwhelming and practitioners can always enlist the help of a bioinformatics or statistical specialist. Edwards discussed several basic data types (discrete, categorical or continuous data), the concept of standard error, confidence intervals, p-values, errors (type I and II), sample size calculation or power analysis, statistical tests (parametric tests, continuous tests and normal distribution), comparator of groups (parametic tests, student T-test and analysis of variance), correlation, regression analysis (linear, logistical and multiple linear regression) and confounding factors.