Biostatistics applications in surgery, anesthesiology, and emergency and critical care medicine Clinic Notes (2019)

---

2019 December 18

Rachel Boardman, Pharmacy

• Retrospective review to compare the use of dual hyperosmotic therapy (hypertonic saline + mannitol) against monotherapy in early management of TBI patients.
• Retrospective data pulled from EPIC within two years (December 2017-December 2019), N=195 total patients admitted into Vanderbilt emergency department with TBI. Each patient received either hypertonic saline, mannitol, or both. Patients' demographics and trauma related clinical variables at ED presentation were collected.
• Plan to perform propensity matching on the fluid received.
• The primary outcome will be GCS at hospital discharge. About 15% patients died prior hospital discharge and may assign worst score to include those patients in the analysis.
• Li Wang was present at the clinics. A VICTR voucher for biostatistical assistance is appropriate based on the scope of the study.

2019 December 11

Emily Long, Plastic Surgery

• We performed a preliminary study last year comparing cytokines in peritumoral skin of immune-suppressed vs immune-competent patients with squamous cell carcinoma. Our study was underpowered, and we identified trends, but did not reach statistical significance. We would like to determine how many samples we would need to adequately power a larger study going forward, and take steps to apply for a VICTR grant.
• VICTR voucher/Mentor confirmed

2019 December 4

Laura Shashy, Pediatrics- Neonatology

• Fellowship project looking at comparison of Hospital Anxiety and Depression scores in a group of NICU parents who journal vs those who do not journal. Previously attended clinic in September (Thursday).
• Would like to continue discussion about data analysis.
• VICTR voucher/Mentor confirmed
• Total 100 parents of NICU babies/children. Baseline anxiety and depression score were available and post scores were also measured. The goal is to compare post scores between intervention and control groups. For some kids, both parents data were included.
• Suggest look at outcome distribution first to decide whether use linear model or proportional odds model. Since some are from same family, consider mixed effects model.
• Expect about 30% missing outcom, table 1 should be baseline characteristics overall and among those with and without outcome missing. Can perform imputation as sensitivity analysis.

Sneha Patel, Pharmacy

• Mechanical ventilation is a common strategy to help maintain oxygen saturation and excrete carbon dioxide throughout the lungs of critically ill patients. Spontaneous awakening trials (SATs) and spontaneous breathing trials (SBTs) are utilized daily to facilitate weaning from mechanical ventilation when appropriate. Many patients who have passed the SAT/SBT trial are not extubated. I will be evaluating which steps limit patients from passing a combined SAT and SBT and achieving extubation. We are interested to see if patients who pass an SAT safety screen, SAT, SBT safety screen, and SBT but are not extubated will (1) be more likely to have received heavy sedation on the day prior to the SAT/SBT and (2) be more likely to have a neurological indication for mechanical ventilation. I am requesting support for data analysis discussion and assistance.
• There will be three main analysis.
• Analysis of daily failure of SAT in relation to sedations
• Analysis of daily failure of SBT in relation to sedation among those who pass SAT or do not take SAT
• Analysis of extubation among those who pass both SAT and SBT
• All analysis will be using mixed model with auto correlation structure
• VICTR voucher/Mentor confirmed.

2019 November 13

Mark Xu, School of Medicine

• Examining the clinical data of pediatric rhabdomyosarcoma patients, specifically what contributes to changes in vital status.
• Mentor confirmed

2019 October 23

Christin Giordano, Medicine/Nephrology

• We are hoping to look at pregnancy and fetal related outcomes in women with chronic kidney disease.
• VICTR voucher/Mentor confirmed

2019 October 9

John Power, Internal Medicine

• Redcap clinical database, analyzing with STATA but questions on how to perform multivariate analysis with high percentage of missing data

2019 October 2

Alan Makhoul, Plastic Surgery

• We are working on a case series of 20 patients who will have activty measured for 3-4 weeks before breast reduction surgery and for 4-6 weeks following surgery. We want to know how long it takes for post-operative physical activity to return to the pre-operative baseline.

  Questions for Biostatistician:

  1. Comments: Enrollement has begun (~3 subjects). 20 SUBJECTS IS LIKELY for feasibality. Could drop early observations for run in. Reccomend spagetti plots to show data. Analysis will bemostly descriptive. Could include table with row for each subject with age, diagnosis, time to baseline active, etc. Discussed missing

Megan Wright, Emergency Medicine/SOM

• Research question: Among adult patients transferred to the VUMC ED from outside hospital EDs in 2018, what patient and hospital factors are associated with potentially avoidable transfers?
• VICTR voucher, mentor confirmed
• Advised looking at CDP for chart abstraction. Advised of timeline for application and analysis, deadline is mid December for SAEM.

2019 September 25

Michael Smith, Trauma and Surgical Critical Care

• We are planning to study our cohort of patients with traumatic brain injury around their patient-centered outcomes. This would probably be a multi-phase study for which we would ultimately seek external funding, so I would like to speak with the statistician team prior to starting anything.

2019 September 4

Jason Cook, Cardiology

• BNP is elevated in atrial fibrillation, however precise mechanism has not been determined. Our study aims to evaluate structural and functional assessments of the left atrial appendage to determine which factors have the largest impact on changes in BNP.
  
  • Mentor confirmed/VICTR voucher.
  • VICTR statistician has met with the investigators and discussed the study. A $5000 VICTR voucher is appropriate for the scope of the analysis.

2019 August 28

Wade Brown, APCCM

• Retrospective analysis of complication by level of exereince in MICU intubation
  
  • Mentor confirmed/VICTR voucher.

Melanie Whitmore, Pharmacy

• Retrospective assessment of use of chemical restraints for agitation and the impact on length of stay. Questions on data to collect and which statistical tests would match data collected
  
  • Mentor confirmed.

2019 August 14

Marie Kuzemchak, Cardiac Surgery

• Our project is looking at warm ischemic time in heart transplant patients and early postoperative graft dysfunction.
  
  • Mentor confirmed.

2019 August 7

Austin Adair, Pediatric Cardiac Critical Care

• Review of previous project with biostatisticians to change direction of project given sample size.Attended clinic in June.
• VICTR voucher.
• Mentor confirmed.

Taylor Coston, Internal Medicine

• I aim to answer the question, “Does the choice of balanced crystalloids versus saline for fluid resuscitation affect outcomes among critically ill adults with cirrhosis?” There is currently little evidence to guide the choice of resuscitation crystalloid in these patients. I plan to perform a subgroup analysis of SMART (a pragmatic, cluster-randomized, multiple-crossover trial comparing balanced crystalloids with saline for fluid resuscitation among adults admitted to 5 ICUs at Vanderbilt). I have identified a cohort of 530 patients with cirrhosis using a previously validated ICD-9 algorithm converted to ICD-10. The primary outcome was the proportion of patients that experienced a major adverse kidney event within 30 days - a composite of death, new receipt of renal-replacement therapy, or persistent renal dysfunction. In addition to the primary analysis I plan to perform a sensitivity analysis comparing saline vs LR only, given the decreased ability of cirrhotic patients to metabolize lactate. If data from the BASE trial is available, it would be interesting to compare LR vs plasmalyte in this population as well.
• VICTR Voucher/ Mentor confirmed.
• Project was discussed at the clinic. The study primary outcome was clearly defined. This is a substudy of SMART and the anallysis will be similar. The proposed work fits under the scope of VICTR biostatistics voucher.

2019 July 31

Alexandria David, Pharmacy

• To evaluate if adequate sedation can be achieved using a sedation protocol that focuses on utilizing opioid and dexmedetomidine infusions and minimizing benzodiazepine infusions in mechanically ventilated pediatric patients in a critical care setting.

  Endpoints:

  • Difference in total cumulative dose and duration of benzodiazepine infusions between control and study group
  • Comparison of the incidence and duration of delirium between control and study group
  • Comparison of total cumulative dose and duration of opioid infusion between control and study group
  • Comparison of utilization of adjunct sedation (ketamine, chloral, etc.) between control and study group
  • Comparison of length of mechanical ventilation, hospital length of stay, and ICU length of stay between control and study group.

We anticipate sample size to be small and would also like to know how to handle missing data.

• VICTR voucher.
• Mentor confirmed.

2019 July 24

Joydeep Baidya, Department of Anesthesiology

• Intrathecal baclofen pumps/catheters are placed in patients with cerebral palsy to treat spasticity. Following treatment, certain patients experience cerebrospinal fluid (CSF) leaks. While most recover as a result of conservative treatment, others require an epidural blood patch (EBP). We are trying to determine factors that predispose patients to have a CSF leak, and subsequently require an EBP.
  Question: There are a few factors that we have determined to be statistically significantly different between the two groups. We would like to conduct linear regression on the data and are looking for advice/suggestions on how to do so.
• Mentor confirmed.

2019 June 26

Austin Adair, Pediatric Cardiac Critical Care

• All pediatric patients undergoing OHT during the designated study time who had either a cardiac catheterization, CT-angiogram of the chest, or cardiac MRI as a part of their pre-transplant evaluation will have their Nakata index and McGoon ’s ratio as measures of pulmonary artery size compared to right ventricular dysfunction, ventilator hours, graft failure, hospital mortality, ICU length of stay, hospital length of stay, maximum vasoactive infusion score in the first 24 hours after transplant. We need help with the data analysis plan.
• VICTR voucher.
• Mentor confirmed.

Lana Boursoulian, Pulmonary

• Data managemnt needs
• Mentor not confirmed.

2019 June 12

Ricardo Lugo, Cardiology

• Determining predictors of a “steam pop” event during radiofrequency ablation in the human ventricle. We have collected data of repeated observations within multiple patients. I would like to request assistance with using the R package for GEE analysis.
• This is a retrospective study of RFAs conducted in a single center. n~39 patients, ~1,000 ablations. ~17 patients had at least one event (~32 events in all).
• Could use sandwich estimator for simple linear model.
• Suggest using graphics highlighting individual data points where possible.
• Suggest VICTR voucher. This meeting can serve as initial clinic visit for voucher.
• Mentor confirmed.

2019 May 29

Daniel Sack, Epidemiology

• Association between completed primary care visits at Shade Tree clinic and visits to the VUMC emergency room, adjusted for zip, CCI, mental health diagnosis, and recent ED visits
• Mentor confirmed by phone.

2019 May 15

Caroline Eskind, Medicine/Infectious Disease

• Assessing lung transplant microbiome with 16s rrna sequencing. first need assistance with analyzing demographic data
• Needs statistical help with comparing demographics and clinical factors between groups., as well as analyzing sequencing data.
• Suggest apply VICTR Biostatistics voucher in amount of $5000.

Ben Fernandes, Pediatric Critical Care Medicine

• Please provide a short description of your project and the questions you’d like to address: Conducting a prospective pilot study of single ventricle patients in the CVICU. Checking 3 biomarkers prior to discharge after first surgery and readmission dates
• VICTR Biostatistics voucher, mentor confirmed.

2019 May 8

Janesh Lakhoo, Radiology

• Comparing calcifications on MR to CT. Need voucher for statistics tests. Planning on paired t-tests at the moment for comparison.
• For each patient, there will be one CT measure of calcification, and additional three MR measures (by three methods). We want to compare each of the MR measure to the CT measure. The primary outcome is the measure of calcification, which is ordinal. Consider use Bland-Altman method to measure agreement, which plots the difference vs. the average. We might also need to take level dependence agreement into account. Denim regression can be fit to measure the relationship between CT and MR.
• Suggest apply $5000 VICTR Biostatistics voucher for the help of statistical analysis and manuscript preparation..

2019 April 24

James Law, General Ophthalmology

• Univariate analysis of oculoplastics data
• Protocol with no expected funding support, mentor confirmed.
• Outcome include number of treatment (continuous) and type (categorical). Want to compare the outcomes by various factors like location (with more than 3 categories). Can use proportional odds test for number of treatment, and Chi-squared test for categorical outcomes. Suggest apply $5000 VICTR voucher for biostat support.

2019 April 10

Chelsea Isom, General Surgery

• I have a paper that was originally designed to detect a complication rate difference in a cohort. One of my reviewers is requiring that I now perform a multivariable model to look for factors that maybe associated with my outcome. I wanted help to perform a power calculation with my current dataset to see if I even have power to detect a meaningful difference in ORs with the data I have.
• Protocol with no expected funding support, mentor confirmed.

2019 March 06

Kristina Betters, Pediatric Critical Care Medicine

• We are doing a retrospective large database study looking at utilization of physical and occupational therapy in pediatric critical care patients. We have done some analysis on our own but want to verify with statistician. Specifically, we are looking at differences in patients and outcomes between centers that use more PT/OT services, and have run multivariate logisitic regressions for factors related to development of bed ulcers and discharge to inpatient rehab facility.
• Protocol with no expected funding support
• We have discussed the statistical approach for the analysis of the project. We suggest apply for $5000 VICTR voucher which would be sufficient for VICTR biostatistician to complete the analysis and help with the manuscript.

2019 February 27

Christine Helou, OBGYN

• The objective of this study is to investigate whether implementation of an enhanced recovery after surgery pathway facilitates reduced length of admission for patients undergoing minimally invasive gynecologic surgery. The study will use data retrospectively collected at our institution beginning with the implementation of ERAS in February of 2018. These patients will then be compared to controls matched by surgical procedure and surgeon who underwent surgery in the year prior to protocol implementation at our institution.

  Questions we would like to address relate to study design and controlling for confounding with variations in provider practice, compliance with the protocol, and multiple interventions implemented as part of this protocol

• Protocol with no expected funding support, mentor confirmed.

Jackson Cabo, Urology

• We are planning a large, prospective study assessing disparities in post-operative opiate practices. Outcomes will include keeping of leftover opiate medications beyond the post-operative period, as well as safe disposal practices. We seek to determine how these practices vary according to health literacy (as measured by BHLS scores) as well as according to the patients place of residence. In particular we are interested in assessing how these outcomes vary according to rural/urban status, as well as distance to tertiary care center. We will be assessing these outcomes prospectively in a large cohort of general surgery and urology patients. Particular data on opiate use and disposal will be assessed via telephone interviews, with prescription data verified using the Tennessee CSMD.

  (1) How should we structure our variables and analyses to best assess how disparities in literacy and geographic area may impact opiate use and disposal practices?

  (2) In order to specifically test for how these “disparities based factors” may impact opiate practices, what demographic variables should we be sure to control for. What sort of analysis would be optimal for asking this question?

• VICTR Voucher, mentor confirmed.

2019 February 20

Benedicto Fernandes, Pediatric Critical Care Medicine

• Retrospective chart review looking at the predictive value of BNP and hCRP in the prediction of readmission, heart failure, and failure to thrive in patients with congenital heart disease.
• VICTR Biostatistics voucher, mentor confirmed.

Marcus Tan, Surgical Oncology

• I have three tumor subtypes and I’m screening these tumor subtypes for 8-10 different markers to evaluate whether a combination of these markers can separate the tumor subtypes.

  For at least one of the markers, the expression was lost in the tumor subtypes in 60%, 10% and 0%.

  I need help with the power calculation (would 20 tumors of each subtype be sufficient?) and statistical analysis

• VICTR Biostatistics voucher

2019 February 13

Mack Goldberg, Ob/Gyn

• I am examining retrospectively estimated blood loss of certain gynecologic procedures that do and do not use of a drug called tranexamic acid. Looking for assistance on analyzing data for case control study
• Protocol with no expected funding support
• Outcomes: blood loss, significant blood loss, transfusion
• Subjects undergoing D&E.
• Txa, yes/no (standard dose)
• 17 cases/control.
• Wilcoxen test for primary outcome of blood loss
• Could use propensity score adjustment and use a linear model to predict blood loss using txa and propensity score
• Suggest dot plot plus boxplot to show all data points by group (txa vs. no txa)

2019 February 6

Lindsey Safley, Pharmacy

• Venous thromboembolism (VTE) is a disorder that includes deep vein thrombosis (DVT) and pulmonary embolism (PE). Historically, patients were admitted to the hospital and treated with intravenous unfractionated heparin (UFH) or subcutaneous low molecular weight heparins (LMWH) as a bridge to therapeutic anticoagulation with an oral vitamin k antagonist (VKA) determined by the international normalized ratio (INR). An outpatient management protocol was implemented at Vanderbilt University Medical Center for low-risk patients who present to the emergency department with VTE to be discharged on apixaban.The purpose of this study is to retrospectively evaluate the safety and efficacy of using apixaban in low- risk VTE patients in an outpatient setting

• Applying for VICTR grant and hoping to obtain feedback/results for middle of March. Mentor confirmed.

• Note: Tight timeline, mid-March for abstract, conference deadline, voucher will take at least two weeks.
• Process, need analysis plan
• Outcome: Return in 7/30 days
• Time to analysis? Do have date of return. This could be a more powerful approach.
• Data: May want to do an audit of data to confirm data. Can do a second reviewer, and measure concordance berween reviewers. ~125 chart reviews. Could review 100% of primary predictors, discrepencies adjudicated. For other predictors, could do a random audit. Randomly select charts for independent reviewers.
• Next steps, send protocol to Kim Hart and Chris Lindsell, Chris Lindsell will be responsible for analysis. Can list biostat personnel as KSP to use identified data.

Parisa Samimi, OB/GYN

• I would like assistance with a retrospective review examining the effect of urethral length on post-operative slings.

• Applying for VICTR grant. Mentor confirmed.

• Preoperative imaging measure uretha length, shorter length may lead to improved recovery/outcome.
• Outcome: Failure within one year-binary outcome (yes/no). Time to outcome is problematic, given episotic visits.
• Also: relationship of length to urinary retention and UTI? How to proceed? Be explicit upfront--pre-specify analysis. Flesh out other hypotheses--are the mechanisms the same or different? If not, can we answer the question? Pre-specification is key.
• Restricted cubic slines, to see shape of curve and allow shape of relationship to vary.
• Pr(y): logistic regression. Covariates, weight, height, (not BMI, it can be a poor summary), age?, race?, other?.
• Next steps: send protocol to Kim Hart and Chris Lindsell, draft manuscript!

2019 January 30

Don Arnold, Pediatrics/Emergency Medicine

• NIH guidelines recommend %-predicted peak expiratory flow (%-PEF) or forced expiratory volume in 1-second (%-FEV1) measurement in children with acute asthma exacerbations to categorize severity (≥40%, mild-moderate; <40%, severe) and response to treatment. To our knowledge, the validity %-PEF to predict %-FEV1 as a criterion measure of lung function and response to treatment during childhood asthma exacerbations has not been examined. We sought to examine whether %-PEF predicts %-FEV1 in children during asthma exacerbations.
  Methods: We prospectively studied children aged 5–17 years with acute asthma exacerbations in a pediatric ED. Participants performed PEF and spirometry in accordance with American Thoracic Society (ATS) standards. I anticipate including data from those with spirometry meeting ATS quality criteria in multivariable regression models to examine associations of pretreatment %-PEF with %-FEV1 and proportionate change of %-PEF with proportionate change of %-FEV1 after 2 hours of treatment. Model covariates include age, gender, race and pretreatment severity measured using the validated, 0-16 point (16 most severe) Acute Asthma Intensity Research Score (AAIRS).

  1. Appropriate inferential test(s) to examine these associations, including whether methods to examine for nonlinear associations should be used.

These analyses will be used for:

1. 1. Manuscript
   2. Preliminary data for R03 or R21 application

• VICTR Biostatistics voucher, mentor confirmed
• Have worked with Chris Slaughter in the past, on other projects
• Have submitted to SAEM (abstract)
• Data from K23, data previously published
• Would like voucher
• Modeling ability of PF to predict FEV

Jeffrey Birnbaum, Pediatrics/Emergency Medicine

• I am having providers perform Acute Asthma Intensity Research Scores (AAIRS) on patients independently at 2 time points. I would like to look at inter-rater reliability of scores overall and by each subcomponent of score. I do not know if I will need VICTR funding for formal biostats assistance.
• May want voucher, mentor confirmed.
• Apply for voucher
• Kappa, ICC may be appropiate
• ~20 enrolled, would like sample size calculation, and assitance with analysis plan.
• IRB exempt, limited data collection

2019 January 23

Austin Adair, Pediatric Critical Care Medicine

• Discuss potential analysis strategies of data set of transport.
• VICTR Biostatistics voucher, mentor confirmed

Yuxi Zheng, Ophthalmology

• Outcomes for long-term followup of surgical correction of head positioning (ordinal variable) and strabismus (continuous variable) associated with infantile nystagmus syndrome.
• Question: I would like help interpreting the results of multiple regression.
• My mentor was present during first joint meeting with James Law, Cathy Jenkins, and Li Wang on 11/7.

2019 January 16

Jillian Hayes, Pharmacy

• I am currently working on a project involving the impact of feedback on the prescribing habits of physicians assistants and nurse practitioners in the outpatient Vanderbilt Health at Walgreens clinics. We are attempting to determine the best way to analyze the change in prescribing rates over time. We have a few ideas, but wanted confirmation on the best way to do this.
• Abstract, mentor confirmed

Joshua Bland, VUSM

• Chart review of mental/behavioral health patients who present to VCH ED over a 1 year period. Data are collected. Outcomes include diagnosis subgroup, disposition, and length of stay. We have performed linear and logistic regressions and would like feedback about how best to refine these analyses to our data.
• Abstract, mentor confirmed

2019 January 9

Austin Adair, Pediatric Critical Care Medicine

• Discuss with Li Wang analysis of data pertaining to obesity in single ventricle patients undergoing bidirectional Glenn.
• VICTR Biostatistics voucher, mentor confirmed

Alexander Hawkins, General Surgery

• The number one driver for readmission and overall health care utilization in patients with a new ileostomy is dehydration. The early stages of dehydration are difficult to assess. We want to see if the use of at home urine osmolarity testing via dipsticks would decrease health care utilization (the exact outcome measure is up for discussion, I was thinking either a composite count of ED visit, readmission, clinic visit versus a count of days of ED visits, readmission length of stay and clinic visit). I am attaching a brief research overview of the project.

  Biggest question revolves around the randomization. It will be far easier for staff to randomize by month (or even week) rather than by patient. I want to understand the ramifications of this.

  Would also like to discuss power calculations. Current rate of health care utilization is around 25%. We would look to halve that.

• VICTR Biostatistics voucher, independent investigator