Clinical Research Oriented Workshop (CROW) Meeting: Aug 30, 2018

Present:   Levi Bonnell, Nancy Gell, Kairn Kelley, Ben Littenberg, Jen Oshita, Gail Rose, Connie van Eeghen

1.                   Warm Up: Ben has budget $$$ for books and education materials; see him if you have requests.

2.                   Rodger Kessler & Connie van Eeghen: PRECIS evaluation of IBHPC study: pragmatic vs. explanatory continuum

a.       Background: Rodger Kessler, Stephanie Brennhofer, and Connie van Eeghen are working on a manuscript to describe the PCORI Integrated Behavioral Health in Primary Care study from a research study management perspective: the inherent complexity of large pragmatic trials using IBHPC as a starting point, supplemented with results from a literature review.  They have come to CROW to conduct an exercise in re-evaluating IBHPC on the PRECIS continuum of pragmatic/explanatory trials.

b.       One key issue: in using this continuum, discussion focused on who the participants are (recipients of the intervention) and who the practitioners are (those who deliver the intervention).  IBHPC has two kinds of recipients: patients and practice members.  It has two kinds of practitioners: practice members and “the practice.”  The group used both perspectives in evaluating the study. 

c.       Patient as Participant

                                                   i.      Primary Trial Outcome: 9; it’s not completely objective

                                                 ii.      Participant compliance with “prescribed” intervention: 10, hands down

                                               iii.      Practitioner adherence to study protocol: 10, also easy

                                               iv.      Analysis of primary outcome: 10

d.       Provider as Participant

                                                   i.      Experimental intervention – practitioner expertise: 9, some selection for friends of Rodger

                                                 ii.      Comparison intervention – flexibility: 10

e.       Current Radar Charts:

 

f.        Final CROW session next week: continue to focus on provider as participant to complete final 6 domains.

John King, MD gets a shout out for sharing a great website to match proposed manuscript titles or abstracts to potential journals for submission:

jane.biosemantics.org

Check it out and share whether you found it valuable too.

Levi Bonnell to present poster at APHA meeting

Congratulations to Levi Bonnell, MPH (CTS Doctoral candidate) and his team on the acceptance of their work as a poster at the American Public Health Association meetings.

Bonnell L, Littenberg B. Wshah S, Rose G. Automated identification of unhealthy drinking using routinely collected data: A machine learning approach. American Public Health Association Annual Meeting, San Diego, CA, November 13, 2018.

Background: Unhealthy drinking is prevalent in the United States and can lead to serious health and social consequences, yet it is under-diagnosed and under-treated. Identifying unhealthy drinkers can be time-consuming for primary care providers. An automated tool for identification would allow attention to be focused on patients most likely to need care and therefore increase efficiency and effectiveness. 

Objectives: To build a clinical prediction tool for unhealthy drinking based solely on routinely collected demographic and laboratory data. 

Methods: We obtained demographic and laboratory data on 89,325 adults seen at the University of Vermont Medical Center from 2011-2017. Logistic regression, support vector machines (SVM), k-nearest neighbor, and random forests were each used to build clinical prediction models. The model with the largest area under the receiver operator curve (AUC) was selected. 

Results: SVM with polynomials of degree 3 produced the largest AUC. The most influential predictors were alkaline phosphatase, gender, glucose, and serum bicarbonate. The optimum operating point had sensitivity 31.1%, specificity 91.2%, positive predictive value 50.4%, and negative predictive value 82.1%. Application of the tool increased the prevalence of unhealthy drinking from 18.3% to 32.4%, while reducing the target population by 22%. 

Conclusion: An automated tool, using commonly available data, can identify a subset of patients who appear to warrant clinical attention for unhealthy drinking.