-- BenSaville - 06 Nov 2009

Dept. of Pediatrics Collaborations, Ben Saville, Ph.D., Meng Xu, M.S.

This page documents the activity of the collaboration plan between the Dept. of Biostatistics and the Dept. of Pediatrics. Since it's beginning in Nov 2009, we have assisted many investigators in the department of Pediatrics on a variety of projects.

Productivity Statistics

Count 11/09- 06/10Sorted ascending 07/10-06/11 07/11-06/12
# published manuscripts 0 4 9
# manuscript submissions 3 9 17
# of grants submitted (internal and external) 7 17 13
# unique investigators 38 43 54
# of unique projects 46 55 68
# of requests (including repeats to same project) 59 89 89

Table of Contents

Useful links

Requested (no meeting yet)

  • Cecilia Di Pentima - T32 for fellow needs review
  • Tom Doyle - cardiologist, Study design questions
  • Sarah Scott (med student), Geoffrey Fleming: attending in the PCCU (VICTR submission)
  • Natasha Halasa, manuscript support
  • Forent Elefteriou, Melissa Dawn, T32 grant submission
  • Dan Moore - grant submission
  • Dia Beachboard - mentor?
  • Yaa Kumah - T32 grant submission (wasn't able to attend with mentor)

Active (already had meeting, work to do)

  • John Williams - manuscript support
  • Mary Romano - Comparing two time points of data on the outcome of interest. Lost Kelly's old code need to start over.
  • Kannankeril and Shuplock:working on the CIs of the difference on the outcomes before and after matching
  • Arnold: Propensity matching for BiPAP project. Wait for complete the info for "attending"
  • Jackie Shuplock, cardiology fellow, Dr. Prince Kannankeril, manuscript support
  • Jon Soslow, Scott Simpson

On Hold (waiting on something, expect to hear back in the near future)

  • Nick Ettinger, resident, had manuscript rejected (retrospective study of ED visits and admissions)
  • Shilpa Balikai, Peds critical care fellow: research project (education based). Want to validate survey, comparing physicians feedback to fellows. May want to consider confirmatory factor analysis, Bland-Altman plots, ICC. Goal is a manuscript in June. I recommended that they get a survey validation expert and I can help implement their suggestions.
  • Paul Moore and Michael DeBaun: R-34 grant proposal
  • Amelia Keaton, peds resident working with Barron Patterson, influenza study, deadline in February, will get back to me for initial meeting
  • Barron Patterson, Peds - manuscript support
  • Lindsay Roofe (Don Arnold), first year PEM fellow, working with Don Arnold, Estrada, and Abramo (Data collection in progress)
    • Lindsay to label research questions with variable names/labels, Meng to write AP
  • Kate Carlson, study of improvement in discharge instructions, never heard back about data!

Completed Items, 2012-2013 fiscal year

  • Cary Fu (Assistant Professor, Divisions of Pediatric Neurology/Epilepsy), Kevin Hess (neurology), Jan 2013
  • Watson - a couple additional p-values for manuscript, manuscript review, Jan 2013
  • Kathryn Miller: R-01 submission for June, similar to K23 study, assessing rhinovirus episodes and subsequent asthma, Jan 2013
  • Agarwal manuscript/resubmission review, Dec 2012
  • Ashley Shoemaker, multi-center RCT, Dec 2012
  • Erin Powell, Whitney Browning, advice on REDCap tool, Dec 2012
  • Charles Phillips, James Gay, wants to compare patient outcomes before and after resident duty restrictions implemented on July 1, 2011; Dec 2013
  • Isaac Thomsen, correlations for manuscript, Nov 2013
  • Simmons, Craig Sheedy manuscript review, Oct 2013
  • Jon Soslow - K23 submission, Oct 2013
  • Sara Horst - manuscript review, Oct 2013
  • Michael O'Connor & Don Arnold, comparing intra-rater reliability of 2 asthma instruments, Sept 2013
  • Agarwal manuscript review, Sept 2013
  • Jill Simmons - manuscript support, Sept 2013
  • Laurie Silva, Peds infectious disease, postdoc, wants help with manuscript statistics, Sept 2013
  • Agarwal - additional analyses for manuscript, Aug 2013
  • Drew Watson, needs a few additional analyses for manuscript (Kelly to provide), Aug 2013
  • Geoffrey Fleming, Assistant Prof of Peds: Survey study to describe the effect of implementation of a faculty development curriculum on junior faculty development, Sept 2013
  • Brian Bridges, Assistant Professor and Michael Northrop (PICU fellow), attended clinic, ECMO study, July 2013
    • Want to compare various outcomes (e.g. heparin, ACT, bleeding, variability of ACT) before and after intervention on Oct 1, 2011
    • Abstract due this summer
  • Rosen - manuscript support, July 2013
  • Ashwin Krishna, Clinical fellow, July 2013
  • Tom Abramo, Katie Rizzone, industry grant proposal, looking at ability of NIRS to detect concusions, July 2013
    • Ongoing discussions about aims and methods
  • Jon Soslow, K-23, submission in June, needs review of stats section for grant (June 2013)
  • Don Arnold, (June 2013) Associate Prof: has data collected on ~700 children with asthma; wants to know if new measure PEP of asthma severity does a good job. Wants to correlate PEP with FEV, IOS, and air score, adjusting for age, gender, race.
  • Simmons, Jill, Assistant Prof: manuscript: retrospective review of vitamin D levels in children with type 1 diabetes, June 2013
  • John Williams: Manuscript resubmission, wants to compare titers and positive rates for MPV vs. RSV viruses in prospective cohort of children. (June 2013)
    • 2 models, a linear regression model and a logistic regression model
    • f(outcome) = spline(age) + group + spline(age)*group
    • Give p-value for difference in curves, describe graphically, and give tests and estimated differences (with CI) at 3 months and 9 months. Consider log transformation for titers, and exponentiate back to get ratio of geometric means
  • Drew Bremmer, manuscript resubmission, June 2013
  • Romano/Baker - manuscript writing (stats section), June 2013
  • Jon Soslow, re-run analysis for manuscript after changes to data set (May 2013)
  • Neal Maynord, Assistant Prof: manuscript support for previous analysis (amino acids and hormone study, feeding tolerance), needs write-up plus additional figures (May 2013)
  • Drew Watson, Peds resident: manuscript support for previous analysis, predictors of aerobic capacity (VO2max) in children with sickle cell disease compared to matched controls, adjusting for fat free mass, gender, Tanner stage, or hemoglobin levels (May 2013)
    • Needs a couple additional analyses, e.g. univariate comparisons between groups of FFM, Tanner stage, hemoglobin; and between time periods within a group over time, additional tables for interpretation of previous models
  • Michael Rosen, analysis, abstract and manuscript (Kelly); gastrointestinal disorder (April 2013)
    • Retrospective chart review, wants to look at predictors of drug failure or drug escalation
    • Provided basic descriptive results for Sept. abstract submission, perhaps Cox models when rest of data are collected from other universities
  • Yuv, Kalra, 3rd year fellow in division of pediatric critical care (with Brian Donahue and Prince Kannankeril), data analysis (April 2013)
    • Have retrospective study of infants with complex heart disease. Goal is to identify whether thrombosis in this patient population is at increased risk of death or death/transplant.
  • Steve Bondi, Fellow in critical care medicine: matching cases & controls for study. Longitudinal study on diabetic patients with and without intervention, PAS abstract (Kelly): subgroup analysis, splines for the linear mixed model (April 2013)
    • Also has new outcome to analyze: appointment compliance
  • Sara Williams, division of ID but gen peds appt. in July, William Russell is mentor: K23 application, developing an interactive intervention/vaccine-hesitancy (April 2012)
  • Hemant Agarwal, Assistant Prof: - wants to include additional variables in the regression model and proportional odds model. (April 2013)
    • Also needs help with manuscript review
  • Lindsay Roofe, first year PEM fellow, working with Don Arnold, Estrada, and Abramo (April 2013)
    • Data collection in progress (40%) done, but have abstract submission due April 12th
  • Van Driest, manuscript revisions, March 2013
  • Jon Soslow, needs updates to figures, and 2 additional analyses with new data (including a 3rd group) for abstract submission, March 2013
  • Isaac Thomsen, Buddy Creech, need graphs for manuscript submission (Kelly), March 2013
  • Manolis Volanakis, Assistant Prof of Pediatrics, Oncology, K08 grant application comparing Luekemic cancer response between groups of mice with and without gene mutations (March 2013)
  • Mary Romano, Division of adolescent medicine, observational study comparing pre vs. post screening outcomes (Ben write up AP, give to Steve), still need to collect 6 months of data (Feb 2013)
  • Jessica Mouledoux, retrospective review of children with heart transplant (Kelly), requests additional analyses (e.g. Poisson regression for incidence rates), Feb 2013
  • Soslow - write-up for previous analysis in layman's terms (Feb 2013)
  • Nick Ettinger - issue with a graph (Feb 2013)
  • Michael Northrop - small grant, want estimate (Feb 2013)
  • Don Arnold, grant submission, 2 arm randomized clinical trial, Feb 5 submission
    • Primary outcome: Adherence rate at 12 months. 4 puffs of inhalers per day, control rate = 0.20, tmt rate: >40 or 45%
  • Lindsay Roofe, first year PEM fellow, advice on RedCap structure
  • Yaa Kumah - R code question about Wenli's analysis, Jan 2013
  • Sara Van Driest, Buddy Creech: Manuscript review, Jan 2013
  • Abby Williams, review stats comments for grant submissions/reviews, Dec 2012
  • Buddy Creech, Issac Thomsen: Comparing titers on 8 outcomes between kids with staff infections and healthy controls, manuscript (Steve), Dec 2012
  • Cecila Di Pentima, proposal for VICTR grant, Dec 2012
  • Larry Markham, Doug Sawyer, Jon Soslow, Joe Kornegay: U-54 grant submission, help with stat sections and power calculations, Nov 2012
  • Mark Reiderer, additional data and questions, abstract submission, Nov 2012
  • Grant Schulert, manuscript revisions (Kelly), he is pulling data together for requested additions, manuscript accepted! Nov 2012
  • Stacy Killen, Pediatric cardiology (Kelly), looking at predictors of pre-natal and post-natal heart arythmia, (Kelly) Nov 2012
  • Mark Reiderer, Assistant Professor (Kelly), concussion study (sports related), Nov 2012
    • Estimate incidence, stratify by gender
    • wants to know if change in proportion of adequate instructions after 2010 (CDC guidelines announced)
  • Kathryn Carlson, Asst Prof of Peds & Marissa Blanco; QI to provide language appropriate discharge instructions, need help with IRB submission, Nov 2012
    • Want to translate discharge instructions to Spanish, conduct phone surveys, gave them power calculations during meeting
  • Jessica Mouledoux, retrospective review of children with heart transplant (Kelly), needs help with writing manuscript edits, Oct
  • Michael O'Connor, Don Arnold, implement analysis comparing variability of treatments with asthma severity (Steve), Oct 2012
  • Killen/Maynord, grant for Gerber and Thrasher foundations, due in Dec, sent papers to read that we can discuss again, Sept 2012
  • Alaina Davis, Sarah McFadden (work with Baron Patterson) for PAS abstract: determine if screening of children with special health care needs is accurate and feasible, needs help setting up database (Steve) Aug 2012
  • Tom Abramo, reproduce analysis for NIRS/DKA data, including additional patients (Steve) Oct 2012
  • Rebecca Kasl, Tom Abramo: impact of obesity on ATV injuries (Steve), given to Mario Oct 2012 (emphasis student project)
  • Jon Soslow & Larry Markham, comparing DMD patients vs. controls on T1 times of the heart (Oct 2012)
  • Bill Heerman, grant proposal with Shari Barkin, needs 1-2 paragraphs on 10K grant submission for looking at association of maternal weight gain/BMI and child's growth trajectory (Oct 2012)
  • Abramo, another NIRS study, went to clinic and will go through BCC, Sept 2012
  • Watson, extensive additional analyses (replied we could not accomodate them); study is comparing oxygen consumed by Sickle cell anemia patients vs. controls, predictors of oxygen consumption, Sept 2012
  • Yaa Kumah-Crystal, fellow, new dataset/project, looking at Diabetes related cell phone use (Steve) Sept 2012
  • Jennifer Slayton - Not Peds faculty, (??Steve??), want to meet in August for data project with pharmacy and hem/onc MDs, never heard back
  • Sara Van Driest, Assistant Professor (Kelly), wants to replicate previous analysis on replication data set, Sept 2012
  • Shoemaker, reply to reviewers, Sept 2012
  • Allison Paroskie, retrospective data collection (grant/VICTR?) (Kelly), Sept 2012
  • Robert Sidonio, grant study design (Kelly), Sept 2012
  • Hemant Agarwal, revisit old analysis and potential manuscript (Steve/Yan), Sept 2012
  • Drew Bremer, longitudinal analysis comparing 2 groups of monkeys re diabetes (Steve), Sept 2012
  • Drew Watson, Ped resident, compare oxygen consumed by Sickle cell anemia patients vs. controls, predictors of oxygen consumption (Kelly), Aug 2012
  • Jeff Lomenick, Assistant Professor, Peds, additional analyses (Steve), Aug 2012
  • Amanda Berger, pharmacy resident, Haydar Frangoul (Steve), Aug 2012
  • Don Arnold - grant for STTR (Small Business Technology transfer grant), Aug 2012
  • Jessica Mouledoux, Clinical Fellow, cardiology. Wants additional analyses for manuscript (Kelly), July 2012
  • Jim Gay, manuscript resubmission, July 2012
  • Robert Sidonio, discuss aims of sickle cell protocol, design advice, July 2012
  • Tracy McGregor, VICTR submission question, July 2012
  • Ettinger, manuscript/ analysis revision (Wenli) July 2012
  • Drew Watson, manuscript revision; July 2012
  • Kathryn Miller, Jodie Jackson: Manuscript review & graphs (Kelly) July 2012
  • Jon Soslow, revisions to grant application (Kelly) July 2012
  • Andrew Smith, Assistant Prof. of Peds, cardiology (manuscript advice, Kelly)
  • Grant Schulert, additional analyses on RVP analysis (Kelly) (July 2012)
  • Kathryn Miller - manuscript submission (July 2012)

Completed Items, 2011-2012 fiscal year

  • Deborah Jones, Nephrology, referred to BCC
  • Steve Bondi, fellow in critical care medicine (just needs short meeting)
  • Sarah VanDriest (Kelly), June 2012
  • Abdalah Dalabih, manuscript revision, June 2012
  • Agarwal, manuscript review & submission, June 2012
  • Schulert graphs for manuscript (Kelly), June 2012
  • Dan Moore, study design/ power calculation for study of blood biomarkers and Type I diabetes, June 2012
  • Yaa Kumah-Crystal, fellow, wants to add additional covariates/patients to analysis (Wenli), June 2012
  • Katherine Kudyba, update and re-run analysis (Wenli), June 2012
  • Jon Soslow, cardiology, manuscript/analysis (Wenli), June 2012
  • Ashley Shoemaker - manuscript review, May 2012
  • Hemant Agarwal, review manuscript rejection (bleeding), May 2012
  • Courtney Horton (Candace McNaughton), Peds EM, May 2012 (Kelly), will need help writing paper later
  • Hemant Agarwal: needs help with manuscript, May 2012
  • Don Arnold, R01 application, RCT comparing asthma medication adherence program to usual care, May 2012
  • Grant Schulert, May 2012
    • Review manuscript, write stat methods and results section (Kelly)
  • Abdalah Dalabih, manuscript review, May 2012
  • Jodie Jackson and Kathryn Miller, May 2012
    • additional analysis question for manuscript
  • Neal Maynord, has data from VICTR application - Kelly, May 2012
    • Study looking at associations with hormones (PYY and Grehlin) with time to feeding tolerance for patients who are feeding intolerant at baseline
    • 50% missing data for PYY
    • Already have analysis plan in place from VICTR application. Modifications: add GEE to logistic model, correlation of Ghrelin vs. amino acids, other items sent after meeting by e-mail
  • Ashley Shoemaker, needs graph for manuscript (Wenli), May 2012
  • English Flack (to meet with Ben and Kelly), Clinical Fellow, Pediatric Cardiology (April 2012)
    • Small cohort of congenital heart surgery patients to investigate predictors of conduction recovery following postoperative complete atrioventricular block (CAVB)
    • Predictors: intermittent conductivity, age, JET (13 yes), sex, weight, bypass time, type of surgery
    • outcome: time to recovery
    • If not recovered at 9 days, patients get pacemaker, but they continue to follow patients for recovery (i.e. heart activity) even after receiving the pacemaker
    • Siddharth Vinod Jain, child nerology chief resident, (Engelhardt, neoatology; Barnes in NICU) assigned to Wenli
  • Daniel Moore, Instructor, dept of Peds (meeting with Wenli & Ben) Apr 2012
  • Sara Van Driest & Tracy McGregor: Vancomycin SNP data analysis - Kelly Apr 2012
  • Jessica Mouledoux, Clinical Fellow, cardiology (assigned to Wenli) Apr 2012
  • Don Arnold (Wenli) Apr 2012
    • Has binary outcome (need hospitilization), about 70 events in 700 patients, wants to look at predictive ability of about 20 variables
    • He has already fit models, so it's difficult to choose a post-hoc model with validity at this point
    • We suggested Bayesian model averaging. We can implement this easily, but are CIs available? Ben to look into this
    • Goal - manuscript, potentially an abstract in Mid march
  • Robert Sidonio, Assistant Prof, hematology/oncology; Mar 2012
    • VICTR Grant application
    • He will work on Aims (making safety a separate aim) and send to me
    • Will discuss methods section / sample size after aims are solidified
  • Mary Romano, Division of Adolescent Medicine & Behavioral Science, resident project (assigned to Wenli), Apr 2012
  • Courtney Horton, PEM fellow, project with Dr. Arnold (assigned to Kelly Lu) Mar 2012
  • Rebekah Brown, CF Center Director, needs help with manuscript revision (assigned to Kelly) Mar 2012
  • Abramo, Sarah Szlam: Emergency medicine (assigned to Wenli) Mar 2012
    • Study of DKA patients with suspected cerebral edema (kids who are combative)
    • Wants to describe Near-infrared (NIRS) over time, both before and after 3% saline solution is added
    • Particularly interested in delta at 5 minutes, 15 minutes, 30 minutes, 45 minutes
    • Other questions: Does baseline NIRS or baseline Ph levels affect the delta in NIRS?
    • Start with graphs, spaghetti plots, then basic analyses with just the deltas (e.g. 15 minutes), then longitudinal modelling
    • Abramo needs to restructure data sets
  • Kyle Brothers & Keith Meador, has grant needing sample size justification (to meet with Ben/Kelly), March 2012
    • Needs some tables of sample sizes for associations between social factor from survey and disease, and potentially the social factor moderating the effect of a gene on a disease
  • Allison Paroskie, application (assigned to Kelly)
    • Needs Methods write-up and sample size justification for Aim 1 of proposal
    • Chi-square tests comparing proportion with and without labs for those with a + or - diagnosis versus undetermined
    • Grant due March 1st
  • Sara Van Driest - opioid-dosing project in children after cardiac surgery, has initial data collected and needs stat advice, March 2012
    • Outcome: opioid exposure after surgery (probably right skewed)
    • Wants to know whether patients with down syndrome (n=44) require more opioids than matched controls (n=78).
    • Observational study, confounders include age, gender, cardiac diagnosis (5 levels), creatinine, pain score, other meds
    • I suggested she consider using propensity scores to adjust for the many covariates (11+ covariates).
  • Jon Soslow, cardiology, grant with Larry Markham, (clinical trial), March 2012
    • Needs stats section and sample size justification, will meet again in 3 weeks
    • Goal: compare 2 treatments for duchenne muscular dystrophy (DMD)
    • Stratify by walking vs. not, age
    • Compare delta in strain in a 12 month clinical trial
  • Steve Bondi, fellow in critical care medicine: longitudinal study on diabetic patients with and without intervention (Wenli), March 2012
    • Follow-up meeting to clarify research questions and data structure
    • Hypothesis: Intervention doesn't affect patients' A1C values and other outcome measures
    • Suggestions 1: to find control patients to adjust for the potential confunders
    • Suggestions 2: Fit a linear mixed effects model on A1C values regressed on a set of confounders
    • After receiving the data, Wenli will provide descriptive stats and then we'll move on to figure out detailed analysis plan.
  • Marty Blakely. basic statistical analysis (assigned to Kelly), Feb 2012
  • Revi Mathew, grant submission with Dr. Phillips, assigned to Wenli, Feb 2012
  • Jim Gay, comparison of sample of chart reviews (n=200) vs. all readmissions for manuscript of readmission preventability (Wenli) Feb 2012
  • Jefferson Lomenick, Assistant Prof, endocrinology, retrospective chart review (to meet with Ben/Kelly)
    • Retrospective study, agreed to sample 60 patients to compare generic drug vs. brand name drug
  • Geoffrey Fleming, Critical care, Feb 2012
    • Has grant deadline in 1 week, recommended that he find stat support elsewhere
  • Barron Patterson, General Pediatrics (and Natasha Halasa), - Kelly Feb 2012
    • Has vaccine data for patients in clinics
    • Wants to determine if there is an association between race (black, white, latino/hispanic, Other) and vaccine status (give, refused, deferred, missed/undocumented)
    • For abstract submission (eventual manuscript) due end of Feb.
  • Oscar Gomez, grant application, Feb 2012
  • Jonathan Soslow, pediatric cardiologist
    • Wants to determine association between progression of skeletal muscle disease and cardiac muscle disease
    • Internal grant, due Feb 15, to VPSD, Vandy Clinical and Translational Research Scholars Program
    • Muscular Distrophy Society, DMD grant is a possibiilty (2 yr grant, deadline in March)
    • Needs help re-writing Aims, matching analysis plan to Aims, and sample size justification. Grant is due Feb 15
    • Aims 1 - 3: correlate LVEF with walk time, biomarkers, and other vars. Need to account for starting points; perhaps use random effects.
  • Don Arnold protocol (Jan 2012)
  • Michael Rosen, design and power calculation for IRB submission (Kelly), also read paper to see if we can determine method for creating score (Jan 2012)
  • Jim Gay, compare 3 groups with respect to hospitalization data outcome variables (LOS, etc.), Jan 2012
  • Don Arnold, research letter to JAMA, Jan 2011
  • Jackson and Miller: several tweaks to analysis (graphs, etc.) Dec 2011
  • Nick Ettinger - additional requests for data analysis (Wenli) Nov 2011
    • Fit a logistic regression model on the binary outcome--admission Y/N with individual acuity level and interaction term
    • See how the regression model fitting results look like to decide whether need to remove average Acuity from the current model
  • Kudyba and Halasa (Wenli) - tweaks to analysis Nov 2011
  • Jim Gay: Manuscript support for "How preventable are (unplanned) readmissions?" (Wenli) Nov 2011
    • Have random sample of 200 readmissions, each classified by 3 panelists on a scale of 1-5 on preventability (1=Not preventable, 5 = Preventable)
    • There are 2 subcategories of rating=4
    • Goal 1: Need assessment of inter-rater reliability
    • Goal 2: Estimate rate of preventable re-admissions
    • Goal 3: Assess risk factors of preventability
    • We will focus on goals 1 and 2 now; will revisit 3 later
  • Dr. Puzanovova
    • Requested additional analyses from IBS vs. healthy patient analysis
    • Conduct comparisons between IBS and healthy patient groups in those outcomes of interest at baseline.
    • Explore the correlation between physiological factors( Heart rate) and psychological factors (severity ?)
  • Andrew Bremer: R-24 grant proposal (4 monkeys, needs to just focus on descriptive statistics and graph trajectories)
  • Don Arnold - tweaks to the cPRAM analysis (Nov 2011)
  • Tom Abramo & Mark Meredith (Nov 2011)
    • Evaluation of protocol change in management of diabetic pediatric patients on diabetic ketoacidosis (DKA)
  • Ashley Shoemaker, repeated measures model: Wenli to give output and graph (Nov 2011)
  • Thomas Morgan, manuscript review; Oct 2011
  • Grant Schulert & Paul Haine: additional revisions to analysis (Kelly); Oct 2011
    • We're waiting on him to add expected LOS as a covariate in the first model (model ratio of observed/expected?)
    • Change reference group on graphs to Gen.Peds=Yes
  • Kathryn Miller, Jodie Jackson (Kelly) Oct 2011
    • Follow-up meeting to clarify research questions
    • Are there differences in demographics and other variables between HRV (rhinovirus) + and HRV -?
    • Among HRV, are there differences in these variables between types A, B, or C?
    • Does proportion or incidence of HRV vary by year and/or season? (both overall and by HRV type)
    • Secondary questions: Is there an association between type of virus and diagnosis (e.g. asthma)
      • Is there seasonality of viruses (overall and by type) or trends over the years?
    • Discussed using neg. binomial regression to model HRV +/- = Day + age + offset(time at risk in study) + (?repeated measures)
  • Hemant Agarwal: manuscript review (handover) (Oct 2011)
  • Tom Abramo & Abby Williams (Oct 2011)
    • Design of a randomized trial to evaluate 2 treatment arms (BiPap) vs. control (standard care)
    • Outcomes: PAS over a 4-8 hour period, but may need to change this to PRAM
    • Either look at delta in PRAM, or time to PRAM less than some threshold (i.e. normal threshold value)
    • Delta (i.e. linear model) may be problematic because not everyone is monitored for the same time (discharged after stabilization, etc.)
    • Need to get data from Abramo/Williams for the chosen outcome to determine best choice
  • Fyza Shaikh (Jim Crowe, Director Vaccine Center)
  • Laurie Cutting, Sheryl Rimrodt (Kelly), grant proposal
  • Hemant Agarwal: manuscript review (risk factors of bleeding)
  • Natasha Halasa, Katie Kudyba (Wenli) (Oct 2011)
    • Q1: Want to compare demographics and clinical presentations for those with rule-out sepsis (check coding of this variable - sepsis "positive" = rule-out sepsis) vs. those without rule-out sepsis for all patients < 6 months
    • Need to worry about Type I error rate. Use FDR?
    • Q2: Of those with rule out sepsis, compare virus positive vs. virus negative for the same list of variables
    • Q3: Of those with rule out sepsis and virus positive, how do RSV infections differ across months?
    • Q4: Of those with rule out sepsis, compare <28 days vs. >28 for virus positive vs. virus negative. No p-values for this table
  • Jessica Mouledoux, Peds fellow (Oct 2011)
  • Paula Williams, Dr. Killen (attended clinic with Wenli/Kelly on Aug 22), (Kelly)
    • Retrospective cardiac study of fetal arythmia for both in-utero babies and young children. Mothers or children are referred to them by primary doctor for a variety of reasons by; echo is done to check for Trachy Arytymia
    • Want to compare patients with pre-natal diagnosis vs. those with post-natal diagnosis
    • Is the timing of the presentation (positive echo) related to risk factors
    • Are there risk factors for severity of course (outcomes = # drugs,
    • Is the mechanism of arythmia (6 or so types) predict clinical course. Treatment is varied depending on mechanism
    • This study has way too many limitations to address these questions. First, the timing of presentation is not an accurate measure of when a patient develops arythmia, so it's not clear what "timing of presentation" is actually measuring. Second, risk factors are not assessed uniformly across all patients, e.g. pre-natal patients will have different kinds of information that post-natal patients, and pre-natal info is not available on post-natal patients
    • Controls would be all patients sent to them for echo who were negative for arythmia. They don't have these data at the time but will look into it.
    • Paula and Dr. Killen will revisit their research questions and focus on smaller and more focused research questions that their data can address. Will return to discuss questions later.
  • Kathryn Miller, Jodie Jackson: R-03 submission
    • Needs stat section
  • Nick Ettinger (clinic), - Wenli to write and implement AP, Sept 2011
    • Has data on 7,000 admissions. Q: Do admission times (or timing of bed requests) cluster around times of shift change?
    • Neg Bin regression may be suitable, with log (#bed requests/# At risk or waiting) = time (24 indicator variables for hour of day) + day + month + acuity (1-2 vs. 3-5) + time*acuity
    • Can use linear contrast to test 2,3,10,11 vs. other times. Time 2 is defined as 1:30-2:30
  • Sabina Gesell, project led by Yanna
    • RCT of 20 controls and 20 on treatment evaluating weight gain and loss during and post pregnancy
    • Intervention is a healthy lifestyle program
  • Oscar Gomez, Yesenia Romero-Herazo: IRB submission for NIH proposal (VICTR?), Sept 2011
    • Will help beef up stats section with more details and appropriate statistics
  • Drew Watson, follow-up analysis, wants to look at Reason for PTL (pre-term labor), Sept 2011
    • This variable was ignored in the first analysis because it has over 100 missing values (out of 750).
    • We will investigate multiple imputation
    • If not feasible, we can refit the primary model with this variable included (with the focus only on the PTL variable), but it will ignore 100 or so observations
  • Lou Muglia, help with comparing genotypes for a left-skewed gestational outcome variable, Sept 2011
    • One complication is that the sampling was not random; it was designed as a case control study (cases pre-term, controls full term)
  • Don Arnold, manuscript support, Sept 2011
    • Wants to look at agreement between 3 measures: PRAM, m-PRAM, c-PRAM (Bland Altman?, on different scales so standardize first?)
    • Wants to assess correlation/predictive ability of 3 measures on 2 outcomes: FEV and Rios
    • Will add me to REDCap, need to remove duplicates (take 1st record of each patient) and those with pneumonia
  • Don Arnold, write up stats section for revised R01, Aug 2011
  • Ted Brenkert, Manuscript editing and submission, Aug 2011
  • Natalia Jimenez, Epi student working with Buddy Creech
    • Has dissertation topic analyzing data from Vanderbilt athletes study
    • Aim 2 may involve Marginal Structural Models due to time varying predictors and outcomes (both binary)
    • Wants to include biostatistician on committee; I suggest that she talk to Jonathan Shildcrout to see if his research would apply to her data; I will talk with Bryan and Jonathan afterwards
    • She may benefit from 2 biostatisticians on committee. How will she learn MSM?
  • Josh Warolin, Peds GI fellow (Dr. Sari Acra), Aug 2011
    • Has 2 methods of determining reflux:
      • X-ray: initial 30 second observation classified as Y/N, with location if Y (Dist,Mid,Prox). If N, then monitor patient another 5 minutes and obtain another Y/N with location measurement
      • Probe Data: 6 sensors @ 3 general locations (Dist, Mid, Prox) - 2 in each. Get # of acidic and non-acidic episodes within 24 hours. Not everyone is observed for the full 24 hours. Also they collect 3 continuous measures: reflux index, Bolis index, Symptom index, all which are skewed right percentages.
    • Question: Want to show that the X-ray is not predictive of the Probe outcome data.
    • Strategy: With the 60 current patients, we'll look at confidence intervals for IDR (# episodes) using negative binomial regression (log time as offset) and difference in means on log-scale (3 percentage outcomes) (t-test). We can use this to see if more precision is needed. Can also consider using Bayes factors to find evidence for the null hypothesis.
    • Josh to restructure data, but also needs to refine question a bit (acidic vs. non acidic, Y/N predictor vs. severity predictor, Final X-ray results vs. standard (initial) result.
  • Thanh Nguyen, Anesthesiology, secondary appt. in Peds.: Help with study design and statistical analysis (proposal) for emergence delirium project
    • Needs sample size justification for Q: Does experimental combo drug decrease the proportion of those with delirium?
    • P.0 = 0.40, effect size 10 % (e.g. decrease from 40% to 30%)
    • We also discussed using delerium scale as a continuous variable, he needs to get mean and standard deviation for power calc., and effect size
    • Primary appt is in anesthesiology; should go to Jonathan for help first
  • Niru Madduri, Developmental Pediatrician, working on a Hobbs Discovery grant proposal (30K for 1 year), short grant (1.5 weeks) for grant submission
  • Hemant Agarwal, ECMO study
    • Wants to compare those with lesions vs. those without lesions, and whether timing of lesion diagnosis is associated with poor outcomes
    • Serious confounding issues because testing of lesions and timing of lesions is confounded with the outcome, as not all patients are tested for lesions
    • Testing of lesions is a clinical decision based on many factors, one of which is to test if patients are not able to be weaned from ECMO (due to poor vitals when attempting weaning)
    • Hemant has been advised of the confounding, and that he either needs to 1) modify the research question, or 2) start a new study
  • Grant Schulert, Hospital medicine (Kelly), July11
    • Has respiratory test data (+/-) given to patients within 24 hours of admission. Would like to know whether virus status is associated with outcomes: LOS, during of antibiotics, total # of lab tests, total $ of lab tests.
    • Potentially repeated measures
    • Outcome variables probably skewed right
    • Potentially different results for different types of patients, e.g. Cardiac, lung, renal, GI, hematology, other (interactions with test result)
    • N around 1,000, probably T-test or Wilcoxon Rank sum test (linear models if considering interactions), or other methods if repeated measures are a problem
    • Goal: Manuscript prep
  • Tracy McGregor (and Sara Van Driest), Assistant Professor of Pediatrics, Division of Medical Genetics: Pharmacogenetics in pediatric populations (Kelly), July11
    • Look at Vancomycin data / Matzke predictions
  • Lynn Walker: Additional analyses for stressor data, July11
    • First change coding for gender and CAP group, remove wordcount variable, then rerun analysis
    • Then create new document: Remove 4 time periods not of interest. Instead of controlling for baseline, model the baseline as part of the response. Create 5 linear (orthogonal contrasts) for each of the 3 outcomes.
  • Hemant Agarwal (Wenli), July11
    • Additional analyses on bleeding data
      • Wants to include many more predictors of bleeding. We'll need to use partial correlations/Spearman correlations in deciding which predictors to use (bootstrap validation?)
      • Also wants to include additional predictors in prop. odds models. Should this use propensity scores?
  • Ashley Shoemaker, VICTR submission power calculation, July11
  • Kathryn Miller, sample size calculations for VICTR submission, July11
  • Natalia Jimenez & Buddy Creech - manuscript review, July11
  • Kathryn Miller (Division of Allergy, Immunology and Pulmonary Medicine): Manuscript help with Jodie Jackson (researcher) and Kathryn's student, David.
    • Prospective study following patients who get swabs when sick, swabs diagnosed with virus (y/n) and type of virus if present
    • Want to compare clinical characteristics of patients with HRV A, B, and C. B only has 5 events, e.g. does the probability of asthma depend on whether you have HRV A vs. C? Should we adjust for potential confounders (age, race, etc.)
    • Want to look at trends over time. Problem is that the total number tested is unknown, so we don't know the prevalence of HRV virus
    • Jodie will meet with Kathryn to go over list of questions. Discussed authorship as well.

Completed Items, 2010-2011 fiscal year

  • Revi Mathews, re-do analysis on different scale, address comments of reviewers, June 11
  • Maya Neeley, General Peds, survey research project, June11
    • Wants to assess whether new "tool" is an effective method of teaching interns/residents
    • The tool is an evaluation and feedback from the attending
    • Also has a survey of how comfortable the attending and intern are with the tool, using Likert scale
    • My advice: The only way to determine if the tool is useful is to have a randomized design, with 2 cohorts (one with tool, one without).
    • Use tool at beginning and end of year 1 for both cohorts, but use the tools every 2 weeks for the intervention group
    • Have the same set of reviewers rate the first and last endpoints, but keep them blinded to treatment assignment
    • Can have a different set of reviewers (unblinded) implement the tool (i.e. give bi-weekly evaluations/feedback)
  • Hemant Agarwal, manuscript review
  • Sheryl Rimrodt & Lauri Cutting, Associate Professor, Peabody (Education)
    • Want to design a clinical trial for a grant submission (NIH-R01) for children with Neuro Fibromutosis (NF); these children have severe learning disabilities
    • Have previously shown a benefit of treatment (tutoring) in NF/IRD patients vs. no treatment in IRD patients
    • Believe that adding drug lovostatin before tutoring will enhance effect of tutoring
    • 2*2 factorial design with drug (Y/N) * tutoring (Y/N)
    • Need power calculation; deadline July 5th
  • Jim Gay, manuscript
  • Martina Puzanovova, Lynn Walker (Wenli)
    • Want to look at associations between Chronic Abdominal Pain (CAP) patients, recovered CAP patients, and healthy controls, and interactions with gender and time (categorical)
    • 8 or so continuous outcomes, fit linear models adjusting for 4 confounders (use rest. cubic splines with 3 knots each)
    • Competing for renewal on July 5, would like to have some results by then in case it can help their renewal
  • Drew Watson & Bill Walsh, NICU, Case Control study for risk factors of intraventricular hemorrhage (IVH) (Kelly)
    • Goal: Determine why there are differences in IVH rates between inborns and outborns
    • 30 or so candidate predictors
    • 62/108 IVH in group 1, 120/290 IVH in group 2 (182/398 total IVH)
    • Subset analysis among outborns: Determine if transport variables are predictive of IVH
    • Originally they wanted to do matched pairs, but I proposed a logistic regression approach with propensity scores
  • Natasha Halasa: Review stats section, give estimated PhD/MS time for grant
    • Phase I clinical trial evaluating safety of high dose vs. standard (low) dose
    • Outcome: titers, or measures of antibodies in blood
  • Kathryn Miller, BOC March of Dimes grant proposal, needs stat section (due July 6)
    • piggyback grant on K-23, which is a cohort of pre-mature infants in Argentina
    • Aim 1: Prevalence of all viruses
    • Aim 2: Among those with any virus, does severity of illness predict outcomes such as wheezing and asthma?
    • Aim 3: IFN-lambda will mediate outcomes (increase ARI causes increase IFN which causes asthma)
  • Ashley Shoemaker, help with how to best capture data
    • Issue of noise when patients don't adhere to protocol
    • We decided best to identify intervals of stable values pre and post, can model repeated measures with GEE (exchangeable)
    • y = pre/post + Group + pre/post*Group (no time)
  • David Bichell, Ben Deschner, pediatric CT surgery (Wenli)
    • Want to compare new method of perfusion when reconstructing the aortic arch using retrospective longitudinal data
    • Has several continuous longitudinal outcomes that he would like to show improvement on: lactate, etc.
    • There are 2 studies: the Norwoods (about 40 in each group), and the aortic arch reconstruction (30 and 10 patients)
    • Want to adjust for confounders at baseline: weight, gestational age, pre-maturity, organ function, baseline outcome measures (e.g. baseline lactate)
    • I asked him to restructure data to fit on one .csv spreadsheet with column lables, long format
    • Goal is manuscript, but have abstract deadlines in April or May that would be helpful for Ben
  • Tom Abramo, Emergency Medicine: manuscript
  • Claudia Florez, observational study in newborns with RDS (Kelly)
    • Study of newborns with respiratory distress syndrome (and more severe pulmonary hypertension)
    • Poster presentation on May 21, needs stats around May 14 so there's time to print
      • Correlate plasma concentrations over time (0, 24h, 48h) with several outcomes that measure disease severity (e.g. days to discharge, death, NICU, etc., 2)
      • Correlate 6 genotypes (each with 3-4 levels) with the same outcomes
      • Update graphs that Wenli generated
    • Need list of predictors/outcomes from Claudia, and descriptive stats from Kelly after importing from REDCap
  • Matt Locklair, PEM fellow (Wenli)
    • Study of about 100 asthmatic patients, collect data on asthma severity and stroke volume, cardiac output, and cardiac index
    • Longitudinal (2-5 obs per patient), 2 measures of outcomes (SV, CO, CI) collected 5 minutes before and 5 minutes after each asthma score
    • Q1: estimate ICC; would like to show small intra-patient variability (within-patient) relative to inter-patient (between). Show graphs; shows machine gives plausible results
    • Q2: Is there a correlation between Asthma score and SV, CO, and CI? We'll already done this with a smaller subset of patients (Wenli analysis), repeat here
  • Natalia Jimenez (Buddy Creech), Pediatric infectious disease: manuscript writing
  • Tracy McGregor, Assistant Professor of Pediatrics, help with K23 submission (May 15)
    • Needs letter of support, biosketch (?), feedback on proposal
  • Ted Brenkert, fellow, pediatric emergency medicine. Manuscript of retrospective chart review: analysis of hypertonic saline (administered via saline) (Kelly - May 10)
    • Is Dose associated with change in sodium (has baseline and post measure), adjust for confounders
    • Additional questions about dose, LOS, indication, timing of dose (i.e. how fast administered), etc.
  • Don Arnold, Emergency Medicine (May 15 deadline)
    • R01 proposal: write stats section for prediction model
  • Kimberly Rogers (Arie Nettles), Associate Professor of Clinical Pediatrics (Developmental medicine): survey analysis (Kelly - May 5)
  • Don Anold R01: Sample size justification / Budget
  • Sabina Gesell - Research Assistant professor, Peds. Needs time estimate and budget (PhD/MS team)
  • Robert Sidonio, Pediatric hematologist (DeBaun), K12 submission
  • Patrick Drayna: Fellow, Emergency Medicine, manuscript review
  • Jim Gay, General Peds - manuscript #2
  • Revi Mathews - manuscript review
  • Drayna - manuscript review
  • Steven Bondi - grant submission (T-32)
  • Sara Horst, logistic regression model with restricted cubic splines
  • Drew Bremer, Manuscript review
  • Mathew Revi, manuscript review
  • Patrick Drayna, Don Arnold: help with extra analyses for manuscript submission (check power calc, time as a class variable, dose effect, etc)
  • Allison Paroskie, Michael DeBaun: Grant proposal
  • Kyle Brothers, Keith Meador: Grant proposal
  • Tracy McGregor, Division of medical genetics
  • Revi Mathews
    • Adrenal Insuff. Manuscript review
  • Don Arnold, responders vs. non-responders, logistic regression
    • Has a second outcome variable, those who "need hospitalization" (LOS > 24 hours or LOS<24 hours but have relapse) versus those who "Do not need hospitalization" (stay <24 hours and no relapse)
    • Has about 15 predictor variables, in addition to age and gender, (Total sample size = 145? or 700?)
    • He'll supply data and priority list for predictors, we'll fit logistic regression models to look at significant predictors
    • First look at descriptive stats, correlations between predictors, then we'll sit down to talk about interactions, splines, etc....
    • 2nd part
      • Has already done logistic regression on proportionate change (>10% increase = responder)
      • I suggested modeling change as a continuous variable, i.e. Change(Y2-Y1) = Y1 + predictors + Y1*predictors
      • We discussed using >10 unit increase (raw FEV), but decided this was not clinically relevant
  • Carolina Loria
    • Manuscript review
  • Andrew Bremer - K award
    • repeated measures ANOVA on several outcomes (or a linear mixed model with random intercept comparing various time points)
    • Probably need log transformation on most outcomes
    • Has 2 groups, 1) non-diabetics (n=25) and diabetics (n=4). Because different patterns are expected over time for diabetics, want to focus mainly on non-diabetics for all analyses. Dr. Bremer did formal comparisions originally on the small diabetic group, but due to sample size we should just give descriptive stats here * Paper currently looks at AUC (using ANOVA), which I think is probably just a linear transformation of the raw values. We should verify this for 1-2 variables and check the distributions of the AUC - perhaps this is more normal? If normal, we could base our analysis on AUC as well. If not, probably best to use log of the raw values and model change.
    • Provide estimates and confidence intervals for differences in time points (and p-values). May need to exponeniate these differences at the end (back transform) so that they are more clinically relevant (if log transformed the exponeniated estimate of the difference is the ratio of the geometric means). Don't adjust for multiple comparisons for now.
  • Kyle Brothers, Gen Peds
    • K-23 grant submission, needs power calculation and methods section writeup
    • Has two continuous outcomes: 1) How important is weight loss for patient (Q for patients), and 2) How likely is patient to follow advice of doctor (Q for provider)
    • 1 predictor: How important is genetics in determining obesity
    • Each patient or provider answers outcome for 4 patient scenarios: risk (normal/high) * Drug response (No vs. Yes)
    • Methods: Linear mixed model or GEE restricted cubic splines? Y = risk + drug + genetics + genetics*risk + drug*risk
    • Power analysis: Simple correlation (Precision??), n=200
  • Geoffrey Fleming, retrospective study, Excel data, has grant money (referred to Jeffrey Blume)
  • Andrew Bremer, R24 grant proposal (Jan deadline)
    • 4 year grant proposal of insulin resistance in monkeys, with 0-1 year collecting baseline data, years 1-2 receive drugs, 3 therapies and 1 control, maybe 60-80 monkeys
    • Aim 1: Prove High fructose diet in year 1 will increase insulin resistance, cholesterol, fat, and prevalence of diabetes
    • Aim 2: Compare each treatment vs. control for change in above outcomes after receiving treatment
    • Has previous data on 25 monkeys on high fat diet at baseline, 6 months, and 12 months (receiving no therapy?), use in power analysis
  • Jill Simmons, Endo, sample size estimation for multi-institutional consortium
    • Has a 45,000 patient database with Type I diabetes, main goal is to quantify incidence of Celiac disease by age (in children and adults)
    • Patients can have multiple records over the 5 years
    • Need power calculation for number of patient blood samples needed
    • I've asked her for 3-4 reasonable sample sizes (n), and estimates of incidence in 5 year intervals
    • Will base power calculation on precision of incidence estimate
  • Patrick Drayna (fellow), Don Arnold PAS abstract/manuscript requiring repeated measures.
    • check power calculation (delta = 2, sigma = 2.7, n=25 for paired t-test), Arnold didn't input a correlation for pre vs. post
    • Has intraoccular pressure measured for both left and right eyes at 0,1,3,5,15,30 minutes, don't see any clinically meaningful differences (based on average of left, right)
    • Need to restructure data so that it's tall, not wide
    • First give histograms of Isop by time to check normality assumptions
    • If normality satisfied, fit linear model with spline for time: y = time + dose, use Robcov for repeated measures
  • Revi Mathews and Andrew Bremer; Premature Pubarche
    • FDA or Hochberg correction, for Wilcoxon and Spearman analyses, separately
  • Yaa Kumah-Crystal, has data on A1C, wants to compare pre vs. post values, n = 30, about 3-6 measurements per patient (half pre/ half post)
  • Revi Mathew: review manuscript (GHST)
  • Sara Horst (and Lynne Walker), logistic regression with splines: compare R results to Stata to make sure she's looking at correct p-values (No deadline)
  • Matthew Locklair, PEM fellow, questions regarding Ausculon study, waiting on data (Nov 17 abstract due)
  • Jim Gay, confidence intervals for manuscript resubmission
  • Arie Nettles, Kimberly Rogers, Vickie Thompson, Developmental medicine, OIHE survey
  • Hemant Agarwal, figures for poster (accepted abstracts)
  • Revi Mathews and Andrew Bremer; Premature Pubarche (Abstract due Nov 15)
  • Kathryn Miller, grant power calculation
  • Claudia Florez, abstract submission (Nov 17 deadline)
  • Don Arnold, Abstract submissions re. albuterol pilot data in preparation for R01 grant (Nov 17th)
  • James Gay, manuscript revisions (No deadlines)
    • Provide suggestions for table
  • Natasha Halasa, basic descriptive stats and demographic tables by group status (about 4 distinct binary outcomes) (Nov 9 deadline)
  • Buddy Creech: graphs of raw proportions across time for different outcome classifications, demographic tables, p-values for Spearman correlations (Nov 4)
  • Julie Taylor, grant revisions, biosketch (Nov 12 deadline)
  • James Gay, manuscript revisions
    • Provide statistical tests for select variables. For dichotomous variables, provide chi-square p-values and confidence intervals for proportions
    • For ordinal variables, provide quartiles and Wilcoxon rank sum test p-value
  • Larry Markham, Assistant Professor, Pediatric cardiology,
    • Finish grant proposal input on study design and power calculations
  • Don Arnold:
    • VICTR proposal: review analysis plan and sample size justification
  • Arie Nettles, Associate Professor of Clinical Pediatrics, questions regarding survey called OIHE
  • Clarence Buddy Creech, Assistant Professor, Pediatric Infectious Diseases and Vanderbilt Vaccine Research Program
    • 630 pregnant women measured at enrollment, birth, discharge, 2 months post and 4 months post
    • 630 (ish) children measured at birth, discharge, 2 months post and 4 months post
    • Q1: Does maternal colonization at enrollment or birth predict child colonization at birth? (vertical transmission)
    • Q2: Does maternal colonization at enrollment or birth predict child colonization post birth? (horizontal transmission)?
    • Proposed analysis: GEE logistic regression: logit(y) = maternal.colonization + time + time*maternal.col interaction
      • 2 centers (VUMC and Memphis), Memphis did not collect many data at 2 and 4 months
      • Include 2-4 confounders, for twins use 1st born, watch out for missing data
  • Puzanovova, additional graphs
  • Neal Maynord, check REDCap database construction and outcomes
  • Shari Barkin, manuscript revision
    • use BMI z-scores instead of percentile, and use restricted cubic splines
  • Mathew Revi, additional requests regarding previous analysis
  • Hemant Agarwal, review abstracts
  • Scott Borinstein, Assistant Professor, Hematology/Oncology, FIND Someone for micro-array analysis!
  • Martina Puzanovova, requests additional analyses involving time*predictor interactions
  • Sara Horst, gastroenterology fellow; prelim data for thesis, Sept 23
  • Ashley Shoemaker, needs analysis plan for VICTR grant, Sept 24
  • Hemant Agarwal
    • has 3 studies
    • 1) Compare pre vs. post survey questions and compare post survey data vs. handoff data
    • 2) Assess risk factors for bleeding and determine whether bleeding is predictive of morbidity and other outcomes
    • 3) Determine whether patients with longer bypass times have worse outcomes, adjusting for severity of surgery (1 year and 3.5 year data sets)
  • Thomas Abramo, MD, VCH Emergency Dept.
    • Get copy of Baxter letter
    • 1st manuscript: Would like to describe (either with stats or figures) the within-subject variances of pre-tapped values
    • 2nd manuscript
      • Would like to compare the variances (and means) of pre tapped vs. post tapped, where the post tapped values might be moderated by proximal/distal shunt
      • October deadline for R01 submission
      • Needs sample size calculation for comparing the mean and standard deviations for the pre, during, and post observations; say for 150 patients
      • Also wants to show that the measurements are reliable within post-tapped values (we've already looked at pre-shunt values)
  • Jenny Slayton
    • Advice on Redcap database
    • put her in touch with Teresa and biostat IT?
  • Kathryn Miller, Pulmonary Division (waiting on better preliminary data for power analysis)
    • Needs power analysis (and write-up) for VICTR application
    • Compare a change in delta-delta score for asthma vs. non-asthmatic kids (from time 0 to time 8)
    • delta-delta score for given time period is calculated as the difference between infected and non-infected deltas from a single person (where deltas are difference between IL29 and B-actin)
    • Wants to recruit 100 patients with Asthma and 100 patients without asthma, expect 60% to actually grow cells, so 120 patients total
    • Has data for asthma patients (with different levels of severity), but only the deltas because they don't have infected and non-infected from the same person
    • Primary endpoint is change from 0 to 8 hours, but also have data for 24 and 48 hours, and also have another 8-9 cytokine outcomes.
  • Revi Mathew, Associate Professor Endocrinology: questions on adrenal insufficiency project
    • Patients have cortisol test, if cortisol <20 then get ATCH test; if ATCH cortisol <18 then get Metyrapone test
    • Give counts for each of the different tests (a tree pyramid)
    • Does initial cortisol test predict whether patients are treated with steroids? (sensitivity/specificity, although # with steroids is small)
  • Jenny Slayton, graph for Dr. Gitlan's address
  • Don Arnold & Hollye Gallion: "Our Kids" requesting biostat support from Terri Scott
  • Jim Gay, General Peds, administrative data, has large data set of hospital admissions and re-admissions (Feb 23, 2010)
    • Would like to compare characteristics of re-admission population (within 15 days) to non-re-admission
    • Goal: Manuscript, authorship already discussed
  • Lou Muglia (list of names)
  • Abdallah Dalabih, discuss data analysis for writing the manuscript
    • Would like graph for analysis
  • Dupree Hatch, PGY-3 Resident (Neonatology): Tues, July 27
  • Natasha Halasa, review grant material
  • Bill Walsh - produce graph / review manuscript
  • Abdallah Dalabih, discuss data analysis for writing the manuscript
    • Omit 2 subjects from entire analysis; re-run 1st model (2.1.1) adjusting for age and race (waiting on race variable)
  • Natasha Halasa: Review grant application
  • Carolina Loria; resident wants to add stats to her study
  • Review Erin Albers manuscript
  • Abramo/Granger: edits for ETC02 manuscript; re-run analysis on new data
  • Ashley Shoemaker, questions from VICTR pre-review
  • Yaa Kumah, pediatric resident, endocrine department
    • study design questions: comparing simple insulin regimen to complicated insulin regimen (pre-post), n=30-40.
    • May collect data on controls who were never switched to simple regimen (about 50)
    • Find date of switch, look 1 year before and 1 year after, try to collect about 3 measurements of A1c (outcome) for pre and 3 for post
  • Sarah Rippel, Gastroenterology Fellow
    • study design questions: Does the severity of EE (Esophilitus) correlate with geography and/or pollen count/grass for those areas? (among those with EE)
    • Has data for zip codes: recommended collecting data at county or census tract level, and getting raw # of eosinophils rather than categorized bins

Completed Items, 2009-2010 fiscal year

  • Rachel Biber: Review manuscript (June10)
  • Tom Abramo, needs edits for NIRS first manuscript manuscript
  • Daniel Jensen, Fellow, Pediatric Endocrincology (June 3, 2010)
    • Sample size calculations for IRB submission
    • Has 3 groups: Wants to compare Type I diabetes to other two groups: Type II diabetes, controls
    • Has 10-20 continuous variables
    • Phase 2: If they find associations, look at significant variables in pre-diabetic patients. Is variable predictive of progressing to Diabetes?
    • Needs paragraph of statistical analysis plan and power calculations
    • No data on variables; report things in terms of standard deviations. Recommend ANOVA with pairwise comparisons. Power with t-tests.
    • No deadlines
  • Jim Gay (June 3, 2010)
    • Review first manuscript
  • Mary Romano, Division of Adolescent Medicine & Behavioral Science (June 1, 2010)
    • Needs descriptive statistics on RedCap survey
    • Provide means, medians, stdv, histograms of continuous variables
  • Hemant Agarwal (June 3, 2010)
    • Review of manuscript for submission
  • Martina Puzanovova, Division of Adolescent Medicine and Behavioral Science (June 1, 2010)
    • Case/control study comparing IBS (20) vs. healthy (22) individuals and boys vs. girls (expect interactions between gender and case/control)
    • Has 7 periods (baseline, stressor, recovery, 3 stressor periods, and recovery), will lump two recovery times together
    • Around 12 outcomes: 3 blood pressure measurements, 3 skin conductivity vars, 3 heart rate vars, 3 Spector analysis vars; all normally distributed except spector, which are log-normal
    • Baseline: She has already done t-tests and Wilcoxon sum rank tests at baseline
    • Recommend linear mixed model with 3 time points as outcomes (2, 3&7, 4-6), predictors = Baseline + time + IBS + gender + IBS*Gender, with random intercept for subject
    • She will provide more specific data (FFT=1, collapse over 3&7, 4-6); will see if I can show her how to fit model in STATA (compare R and STATA results)
    • Manuscript production (authorship agreed); usually referees want to see ANOVA (2-way ANOVA?)
  • Bill Walsh, Chief of Nurseries (June 2, 2010)
    • Need help with manuscript
    • See e-mail for details; would like write-up of 12*2 table analysis and 45/91 proportion testing vs. p=.25
    • Provide graph with percents for 12 groups
    • Deadline end of June, Ok with June 18
  • Rachel Biber (May 26)
    • Paragraph for manuscript submission
  • Natasha Halasa (May 21, 2010)
    • Wants additional power calculations (n=1300), with actual numbers instead of standard deviations; also needs grant support (10% faculty, 15% MS?)
  • Sarah Horst (May 18)
    • Additional questions about analysis
  • Kathryn Miller (& John), Vaccine Clinic data in microsoft access format (5/11), don't know how to get data out.
    • Recommended to contact Janey Want to put Access into REDCap database
  • Abdallah Dalabih, Pediatric critical care fellow
    • Has data on severe Asthma patients. Outcome is time to leaving ICU with no censoring. Wants to determine if Aminophylline (a drug) influences the time in the ICU. Initially thought that the drug would decrease time in the ICU but has found the opposite (using Wilcoxon test). Is this due to the drug or to the possibility that the drug is given to the sickest patients? (doctor makes the decision on the drug)
    • Attempted to match the drug group with the no drug group
    • The drug has a dose (0-15), 0-5 is non-therapeutic, 5-15 is therapeutic, >15 is toxic
    • Q2 Within the drug group, does the dose correlate with time in ICU? This is likely to have the same problem as analysis above
    • Q3 Does time to mild status correlate with time to leaving ICU? This should correlate because total time = time to mild + time from mild to leaving ICU
    • Recommend using Cox regression: Time to leave ICU *event = drug + Asthma score (either continuous or categorical)
  • Hemant Agarwal, more follow-up questions for 2nd manuscript; recommended that he attend clinic for a 2nd opinion on the stats
    • He wants to find what % of death is attributable to different systems. Problem: only has 20 deaths; asking more of data than possible
  • Erin Albers, Cardiology Fellow, needs help with manuscript (May 5 2010)
    • Albers: run repeated measures ANOVA on her data
  • Natasha Halasa, sample size calculations for grant (May 5 2010)
    • Wants power calculation for 2 continuous measures (correlation/linear regression), Vitamin D and RSV severity, among patients with RSV
    • Sample size around 1,000, but could do more if needed.
    • We will focus on precision, or how narrow the confidence interval will be for different correlations and different sample sizes
    • Try rho = 0.5, 0.65, 0.85, with n=750,1000,1500
    • Deadline: May 7th
  • Dejan Milatovic, Ph.D., Research Assistant Professor Toxicology, Grant Application
    • Has 2 studies
    • 1first study: compare 5-6 endpoints between three groups of people (3 locations, one with no exposure to pesticides, 1 with low exposure, and 1 with high exposure)
      • Do ANOVA to compare 3 groups, all pairwise comparisons, and determine correlation of endpoints with eachother (not everything just the main ones)
    • 2nd study: Rat study
      • compare same outcomes (+1 more - brain activity) among three doses (none, low, high) and 3 lengths of exposure (1 week, 3 weeks, 8 weeks), and all possible combinations
      • Either two-way ANOVA or linear model
    • Needs analysis plan for both aims/studies as well as power calculations. Deadline: Tuesday, April 27th.
  • Joyce Granger, PEM fellow, (deadline: April 16th, presentation Apr 22nd)
    • Has data from an asthma/ETCO2 study. Wants to show that new instrument has high reliability
    • Also has outcome data (Did they go home, therapy, ICU, etc.)
  • Eric Austin, junior faculty, Pediatrics (mentor: Jim Loyd)
    • Needs help with Program Project grant submission (May)
  • Carolina Loria, Fellow, Hematology-Oncology. Has finished chart review and needs data analysis for MSCI project (Completed Apr 16)
  • Sara Rippel, biostats for oral presentation/graphics (Apr 9)
  • Shari Barkin, additional analyses for Rachel's abstract (completed Apr 9)
    • Add category to Wenli's table
    • Wants to assess correlation between each of the 3 variables, adjusting for overweight status or BMI percentile
    • Interested in using absolute BMI instead of percentile, but also adjust for age and gender.
  • Thomas Abramo, MD, VCH Emergency Dept. (Apr. 2)
    • Needs help analyzing inter and intra variability
  • Miranda Reines, 4th year med student working for Brent Graham, Division of Pediatric Rheumatology (March 30, 2010)
    • Has pilot data, wants to determine if increased anxiety in parents predicts the perceived symptoms of the mother
    • predictor: Yes to any of the anxiety questions. Outcome: Total number of symptoms checked
  • Hemant Agarwal, edits for 1st manuscript submission (March 26)
  • Brad Strohler, Rapid response team database (scheduled Thurs, March 25th)
    • Talk with Teresa
    • Talk with Zhouwen
    • Give Brad details about Redcap meetings (Terri)
  • Palmer Johnston, neonatology fellow, needs help with presentation to department (March 26)
    • Has two groups (pocket aid only vs. pocket aid with guidance), several outcome variables
  • Sarah Horst, Gasteroenterology Fellow, Vanderbilt University (March 25th)
    • Questions regarding logistic regression analysis for MPH project
    • Has 350 kids with stomach Stomach pain, follows them several years later as adults
    • Wants to find predictors of stomach pain as adults
  • Hemant Agarwal (completed March 17)
    • Wants to completely rework 2nd manuscript analysis
    • Has around 20 Y/N predictors, denoting whether a child had a certain type of event
    • Will produce 2*2 tables and Fisher's ch-square test for testing whether p1=p2, proportion of deaths in group 1 (event) equal proportion of deaths in group 2 (No event)
    • Will also produce quartile summaries and Wilcoxon Sum rank test comparing LOH, LOI, LOPI for the two groups (event vs. non event) for each predictor
    • Goal is to rank the predictors by strength of association; problematic because of low death count (21); failing to reject Ho does not prove there is no association!
  • Mark Riederer, Division of Pediatric Hospital Medicine (Fri, March 12), needs analysis for abstract due March 19th (Complete March 18th)
    • Wants to compare outcomes of a resident team (Team A) vs. a non-resident team (Team D)
    • Part 1: Has 7 questions on Likert scale, wants to test difference in means between Team A and team B
    • Part 2: Has 2008 Team A data, and 2009 Team A & D data. Wants to compare 08 A vs. 08 A, and 09 A vs. 09 D.
      • Needs a series of t-tests and chi-square test of proportions
  • Ian Kane, resident in Peds, study design questions (March 16)
  • Joyce Granger, PEM fellow, descriptive stats for ATV project (March 16)
    • Has 8 years of data regarding ATV, wants to determine if there are trends over time for several outcomes:
      • ISS< GCS, Helmet vs. no, Race, # events, $ hospital charges, severity of injury, etc...
  • Shari Barkin, replicate analysis using SASified data (March 15)
    • Create 2 new tables for WC and BIA percentile, (by normal/overweight, girls/boys, age(3,4,5)), give mean +- sd
  • Sheryl Rimrodt, Assistant Professor, Developmental Medicine, letter of support for K08 application
  • David Guis, Associate Professor, Department of Radiation Oncology, sample size calculations
  • Dedrick Moulton, Pediatric Gastroenterology, Hepatology, grant submission
    • Follow-up meeting to solidify power analysis in application
  • Dedrick Moulton, Pediatric Gastroenterology, Hepatology, grant submission (Feb 26, 2010)
    • Study AIM 1: Compare the number of eosinophils in cases (EE) vs. the number of eosinophils in 1) Normals and 2) GERD. Expectation: Normals<EE
    • Study AIM 2: Among the EE cases, correlate the number of eosiniphils in nasal secretions to number of eosiniphils in 2 locations of Esophogus
    • Needs paragraph with sample size calculations (2-sample t-test or ANOVA and correlation)
  • Hemant Agarwal, follow-up questions regarding analysis (completed Feb 23,2010)
    • Produce figure for Table 4
  • Ashley Shoemaker, Endocrinology fellow, helped with sample size calculations and write-up for research proposal
  • Carolina Loria, Fellow, Hematology-Oncology
    • Review main goals of the oncology study (manuscript); prep for Studio next week
    • Has persons with and without virus (but all sympomatic), would like to test association between virus (y/n) and the following outcomes: type of infection (lower,upper),oxygen+vent (y/n), oxygen(#cont), as well as predictors of oxygen, oxygen+vent,type of infection(neutropenia, lymphopenia, transplant(y/n),steroids(y/n)
    • Main issue is how to deal with repeat infections on the same person (GEE repeated measures logistic regression?)
  • Erin Albers, has data now, did not find correlation between cardiac bypass time and change in isoprostanes
    • Study results not as expected, but not much we can do about it (negative result)
  • Sara Rippel, Pediatric GI fellow: clinical research project (Feb 16,2010)
    • Questions regarding subset analysis of Lynn Walker's grant
    • Study 1 (manuscript): 290 patients with GI at enrollment, would like to develop a scale for severity of GI disorder (scale is based on pictures)
      • Was told by Chris Slaughter to have 3 pathologists evaluate 10 slides each; assess inter-rater reliability to determine total # of cases needed
      • I also suggested repeating study a week later to assess intra-subject reliability
    • Study 2 (manuscript & conference): comparing GI disorders and headaches using chi-square tests (confirmed stats are Ok)
  • Julie Taylor, Assistant Professor General Peds; K01 grant (biostatistical collaborator), power calculations, stats plan
  • Erin Albers, Clinical Fellow, Cardiology: Questions regarding statistical analysis plan for MSCI project (Tues, Jan 12)
  • Maria Gillam-Krakauer, neonatology fellow (Jan 22, 9am)
    • Sample size calculations
    • Needs paragraph to present to group next week
  • Lou Muglia, Needs help with sample size calculation and allocation ratio (Jan 18,1pm)
  • Neal Maynord, Fellow in Critical care, questions regarding VICTR request (Fri, Jan 15, 9am)
    • Wants to predict feeding tolerance/intolerance using 2 hormone levels
    • Data collected at baseline (outcome and predictor), followed every 24 hours for 2-5 days
    • I will provide paragraph giving analysis plan and sample size justification
  • Claudia Florez, neonatology, observational study in newborns with RDS (Jan 19th, 11am); helped with RedCap database needs (don't convert numeric to categorical!)
  • Hemant Agarwal, follow-up questions
    • Look at correlation of # of cardiac events and survival (additional data set); this ideally will not be as strong as total # of events
    • Use logistic regression to look at relative impact of severe, moderate, mild events
    • Clean up previous analysis with clean data; include p-values for number of events;
  • Bill Walsh, questions regarding statistical plan for grant proposal (Tues, Jan 12)
  • Erin Albers, Clinical Fellow, Cardiology: Questions regarding statistical analysis plan for MSCI project (Tues, Jan 12)
  • Deb Friedman, needs my biosketch to include on T32 training grant. Need to update CV/biosketch!
  • Tracy McGregor (meeting Dec 15), questions on data collection for RedCap
  • Ashley Shoemaker, fellow. Needs help with sample size calculations
    • Sample size calculation comparing 2 groups (unequal sample size) for thermogenic effect. Thermogenic effect = (post-pre metabolic rate)/food intake
  • Hemant Agarwal, ICU, manuscript help for 2nd study
    • He has created a scale for cardiac ICU patients, based on mild, moderate, or severe organ failures. Would like to determine whether scale is predictive of death, and how it correlates with 2 standard scales. He can also break thing down by organ system.
    • We will revisit the analysis in January
    • Has a 3rd study just starting; would like feedback on design and data collection
  • Carolina Loria, CTSA studio request for respiratory study
    • Have 80 samples from cancer patients, want to determine prevalence of certain viruses. Some patients have 2 samples
    • Approach: create summary variables for each outcome: Did patient ever have Virus? (n=50-60) Then calculate proportion and confidence interval
    • Can test via binomial one sample test whether proportion differs from population values
    • Can us logistical regression for determining markers of severe vs. less severe infections
    • Studio will be in January, data ready in Spring, CTSA graduation deadline in May (papers/publications to result)
  • Pinki Prasad: Help with sample size calculations and wording of analysis plan for grant proposal
    • Has 2nd grant proposal for Hodgkins lymphoma; needs help understanding/and or re-writing statistical analysis plan (Meeting Dec 2nd)
    • Discussed this with Tatsuki: Pinki and all other Peds Cancer investigators should work through the Cancer Biostats group
  • Sara Horst, 24 Nov 09, gastroenterology, Stat related questions for Abstract submission (Fellow)
  • Candace McNaughton, 24 Nov 09: sample size calculation for EM study
  • Hement Agarwal, Thurs, Nov 6, 2009
    • Has data for mult-disciplinary, mult-specialty educational program. He has completed questionairre based on 11 questions with Likert scale. He would like to demonstrate that the program was succesful, and compare the different types of respondents with respect to their view of the programs success.
      • 2nd part of the study is based on clinical outcomes, time in ICU and time on ventilator. Subjects can go off ventilator and come back on, and some patients die. Hement has done a t-test for these times (omitting individuals who died), but found the data are skewed. A Wilcoxon test would be more appropriate. However, we may want to use survival analysis to incorporate the subjects who died.
    • Has requested help for completing statistical analyses for manuscript preparation.
    • Completed!
  • Bill Walsh , Thurs, Nov 6, 2009
    • Requests manuscript help in calculating positive predictive value regarding diagnostic screening tool for congenital heart failure in newborns. Study has unique design in that it wasn't a research study. Hence there are some questions on what population or what individuals to use for the relevant tables. I gave him a formula for confidence intervals for a proportion, and agreed with his interpretation of the tables. * B.C. Paria, Tues. Nov 3, 2009
    • Requests sample size calculations for grant proposal. Has several experiments (10?), each of which has a different outcome and a different study design. We discussed one of the study designs, in which there are two types of tissues sampled from each animal in the study. Within each tissue, 4 cultures are grown, of which 3 are given various doses of a treatment intervention and 1 is left as the control. The outcome is the color density.
    • Questions,
      • Does tissue A have higher values than tissue B?
      • Are higher doses associated with greater color density?
    • A paired t-test would be appropriate for a simple sample size calculation but we need an estimate of the range of the outcome and the effect Dr. Paria would like to detect
    • To Do: supply formula, wait for effect sizes and stdev estimates.
  • Shari Barkin, Tues. Nov 3, 2009
    • Has longitudinal research study "Salud con la Familia", evaluating effect of program intervention for childhood obesity in Hispanic population. Has baseline data and would like some simple statistical analyses for 3 abstract submissions
      • For children, give descriptive stats for BMI percentile, waist circumference, and BIA (fat mass). Determine if the 3 measures are correlated.
      • Give descriptive stats for 8 SNP's (0,1,2 variables) and genetic risk score. Determine if there is an association between adult BMI and genetic risk score, and between child BMI percentile and genetic risk score. Also compare genetic risk score for adults BMI >= 27 vs. else, children BMIp >= 85% vs. else, and children >= 95% vs. else
      • Compare adults vs. children for BMI, waist circumference, and BIA. Do adult values predict child values?
  • Revi Mathew, review manuscript (waiting for response - never heard back)

Inactive (Waiting on something, haven't heard from in a long time)

  • Sarah Szlam (and Rob Ferre), fellow, emergency medicine. study design question
  • Jill Obremskey, Assistant professor Peds, study design (Kelly) Dec 2011
    • Design of study to evaluate the implementation of a national guideline of managing/controlling asthma pediatric patients during visit to a pediatric emergency department fast track
    • Investigator is interested to see improvement of long-term control of asthma after guideline is implemented
    • Guideline will be implemented in the second week of January 2012
    • Will collect data before guideline implementation (now has 6 months data, Jan 2011-Jul 2011, around 700 obs)
    • Outcome:
      • number of visits within a year
      • time to the second visit
      • questionnaire score ( Asthma Control Test (ACT) )
    • After receiving the data, Kelly will provide descriptive stats, e.g.: number of visits / year, time to next visit, etc
    • Perhaps do power and sample size calculation for IRB submission
  • Bill Walsh & JoAnn Romano-Keeler, manuscript help (Kelly)
    • Have manuscript rejected, want to address reviewers comments
    • Study compares growth of premature infants between infants with high protein regimen and standard regimen
    • Delta = baseline + Gestage@birth + Gender + DaysSinceBirth@baseline + Tmt + birthweight + Time since baseline + interactions? (birthweight*Tmt and Time*Tmt)
    • Outcomes: Head circumference (weekly), length (weekly), weight (daily), LOS
    • ITT analysis only needed if data are collected after patients stop adhering to treatment.
  • Abramo - additional questions for Locklair's data set.
  • Jessica Lilley, Fellow Pediatric Endocrinology working with Sergio Fazio (adult endocrinology), needs help with VICTR submission (power analysis)
    • Had two groups to compare: Patients with FH obtained from father and patients with FH obtained from mother (FH = Familial hypercholesterolemia)
    • Wants to collect data prospectively using VICTR funds in her clinic, up to 15/week for 1-2 years
    • Outcomes to compare between groups: LDL, insulin, CRP, Age, HDL, TG, etc.
    • Likely to be skewed variables
    • She will extract a data set from a retrospective study (Framingham Data) with thousands of patients in order to obtain information about the distribution (mean, sd, normality) for power analysis
  • Sheila McMorrow, Emergency Medicine (Kelly), Aug 2011
    • Has trial comparing those with a positive test for PCR (Ehrichia, a tick virus), n=102, but no controls. Wants to determine risk factors for positive test.
    • She will research whether she has access to controls.
  • Erin Powell (intern), Whitney Browning,
    • Want to implement intervention of scheduled mock simulations for interns (1st year residents) to see if it leads to improved performance, interns randomized to intervention or control (n=26 total)
    • Test/Code to be assessed at beginning and at end of year
    • Performance to be assessed by 2-3 blinded judges, recorded on videotape
    • Discussed how to grade performance, suggested Likert scale or continuous slider bar with Likert labels to calibrate responses.
    • Next step: They will write up the items that they want to score them on (10-15 items) and bring back for review
  • Haydar Frangoul, Associate Professor, Division of Hem/Onc, Oct 2010 (no deadlines): Refer to Tatsuki (Cancer center)!!
    • Study of Ped. Leukemia, wants to describe subset of population who experience relapse and determine if certain variables are predictive of survival.
    • Goal: Manuscript. Follow general strategy of journal article.
  • Brad Stohler (Jenny Slayton, John Grantham), Aug 2010
    • Wants to assess the relationship between markers of hospital stress and the number/incidence of Rapid Response (RR) requests
    • Has data for 2009, has the number of RR for each month, as well as Case Mix Index (CMI, a measure of illness severity assigned at discharge), % capacity (number of rooms occupied/# of rooms available), and other secondary risk factors
    • Others: Staff/patient ratio or % overbudget on staff; % patient turnover (more turnover means more work for staff); these variables might not change much at the hospital level
    • Can measure % occupancy at 12pm and at 12am
    • Might be a non-linear relationship between CMI/Occupancy and RR rate.
  • Yaa Kumah-Crystal, Additional requests:
    • Determine if gender, age (numeric or dichotomous), weight, # of chronic visits are confounders for time. This involves interactions with time and predictors, but we may be limited on the # of predictors we can include due to sample size.
  • Michael Rosen, Gastroenterology, Hepatology, & Nutrition: has retrospective study of response rates to infliximab for pediatric patients hospitalized with severe colitis
    • AIM 1) Describe infliximab levels for patients (overall, non-responders, and responders if possible)
    • AIM 2) Determine predictors of non-response. Has about 5 potential predictor variables (continuous vars)
    • Sample size will be limited (40-50 total, about half responders), wants to know if they need to recruit another clinic
    • Response is determined by PUCAI severity index score, which could be analyszed as a continuous measure for increased power, but response vs. non-response is more clinically relevant
  • Connie Fauntleroy, IRB submission (Barkin referral)
  • Sara Rippel, manuscript preparation, Clinical outcomes of former pediatric patients with histologic reflux esophagitis
    • Has about 10 outcomes, mostly dichotomous, a couple continuous (depression scales), wants to compare 3 groups (GER, Normal Endoscopy, Controls)
    • For binary outcomes, use logistic regression with outcome(Y/N) = group + Age@Time0 + gender + TimeElapsed + .... (education, etc.)
    • Include pairwise comparisons for 3 groups, adjust for inflated type I error?
    • How to deal with outcomes with few events? (Dysphagia, etc.), limited number of predictors.

Topic revision: r465 - 31 Oct 2014, AdminUser
 

This site is powered by FoswikiCopyright © 2013-2022 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Vanderbilt Biostatistics Wiki? Send feedback