Present: Abby Crocker, Amanda Kennedy, Rodger Kessler, Ben Littenberg, Connie van Eeghen
1. Check In:
a. Ben: Complex Systems’ journal club this morning reviewed recent Science article this month on Google Books, and meta-studies coming from 5 million books digitized and analyses of the words that appear in them, e.g.
i. Frequency of scientists’ names
ii. Evolution of verbs
iii. Lifeline of frequency of specific years and specific phrases
iv. Lifeline of historical epidemics
v. Identification of words not found in dictionaries: e.g. “aridification”
2. Abby – Analytical Plan
a. Teaching component of CTS program – on its way
b. Research question/idea: develop a predictive model to determine the presence of NAS and test it (“What predicts the occurrence of NAS?”); longitudinal health outcomes of children with NAS at perinatal, 1 year, and 5 year marks (“What are the outcomes over time?”). Develops expertise in area of high-risk children, a professional goal.
i. Will help with skill building
ii. Can be used to complete the Masters’ requirement (quasi-defense will be on March 11)
iii. Robust data base available, especially compared to what has been published so far
c. Research plan for question 1: What predicts the occurrence of NAS; gold standard is NAS score at delivery hourly for 24 hours, every four hours from 24 – 48 hours. (No studies exist re: when NAS manifests itself – another good question to study.)
i. Subjects: High risk babies delivered at FAHC from 2005-2011; n=350 (babies with NAS). 2005 is the start of formal documentation of NAS. 550 were identified as potentially having NAS, and 200 proved negative. “High risk” determined by opiate use by mother for any reason.
ii. Purpose of this study (why do it?): For those identified at risk for NAS, can NAS status be confirmed at time of delivery. Benefits may include reduced costs, improved family bonding, intervention started earlier, etc. Possible practical goals:
1. Precisely and accurately predict NAS (develop a point scoring system for providers involved in delivering high risk babies) NOT YET
2. Confirm a hypothesis
3. Understand/describe/explain the relationships among predictors THIS ONE
iii. Dependent variables: from registry, problem lists, medication lists, and other sources; could include delivery characteristics (ultrasound, medications prescribed to mother, social variables, height, weight, APGAR score...). Could also include the reason the mother used opiates (high risk criteria), maternal age, prematurity, delivery data, maternal meds, SES, diagnosis, prenatal care, prior use of health care resources (# nights in the hospital in the past year, SA admissions), PCP, presence of significant other and/or kids, employment, tobacco, ETOH, maternal labs/vitals/physical exam, glucose tolerance. Would be helpful to consider race relative to metabolism of opiates.
iv. Retrospective now; prospective when validating model in the future
v. Develop predictive model and compare prediction to actual results in retrospective analysis. The predictive model needs to be simple enough to be used at the time of delivery.
vi. Questions to explore:
1. When does NAS occur?
2. What explains NAS?
3. Can we predict NAS (is a useful way, whether at delivery or before discharge, depending on the outcome of question 1)?
4. Does the prediction rule help?
vii. Logistics:
1. Access to data (yes)
2. Clean data?
3. Process and time for chart review
viii. FINER checklist of 0 - 4
1. Feasibility: 4
2. Interesting: 4
3. Novel: there is a predictive model for NAS from Boston, smaller number of variables, included some things that might be useless; about the same sample size. Found that breast feeding intention or action was relevant: 3 (Consider journal clubbing that article with this group)
4. Ethical: could be used in a harmful way? Confidentiality? 3
5. Relevance: Could improve care, provide new interventions, new knowledge, 120 kids/year at $xx/day cost savings, and affect public health policy: 4
3. Next Workshop Meeting(s): Thursday, 2:30 p.m. – 4:00 p.m., at Given Courtyard Level 4
a. February 3: Connie – QI through Office System Analysis - Case 1 analysis
b. February 10: Rodger - discuss a research design for evaluating mental health contribution in BP PCMH practices; Craig Jones and Beth Tanzman (new member of his team) will be guests
c. February 17: Rodger & Connie’s draft paper for PPRP (No Amanda, Ben)
d. February 24: (no Amanda, Kairn, Connie, Ben – schools are on winter break)
e. March 3: (no Rodger)
f. March 10: (no Rodger)
g. March 17: John Senders - engineer and psychologist with deep expertise in human error and an international authority on medical errors; Ben as host
h. March 24: Rodger: NCQA survey data (no Amanda)
i. Future agenda to consider:
i. Rodger: Mixed methods article; article on Behavior’s Influence on Medical Conditions (unpublished)
ii. Future: Review of different types of journal articles (lit review, case study, original article, letter to editor…), when each is appropriate, tips on planning/writing (Abby)
Recorder: Connie van Eeghen