Data Studio Workshop

DATE: Wednesday, 1 February 2023
TIME: 1:30–3:00 PM

INVESTIGATORS:
Thomas Johnstone (1)
Gordon Lee (1)
Afaff Shakir (2)
(1) Department of Surgery, Stanford University School of Medicine
(2) Department of Surgery, University of Chicago Medicine

LOCATION: Conference Room X399, Medical School Office Building, 1265 Welch Road, Stanford, CA

WEBPAGE: https://dbds.stanford.edu/data-studio/

ABSTRACT
The Data Studio Workshop brings together a biomedical investigator with a group of experts for an in-depth session to solicit advice about statistical and study design issues that arise while planning or conducting a research project. This week, the investigator(s) will discuss the following project with the group.

INTRODUCTION: Breast cancer is the most frequently diagnosed cancer in American women and the third leading cause of death. The mainstay of breast cancer treatment centers on resection of neoplastic tissue: 36% of women with early-stage breast cancer and 60% of women with late-stage breast cancer undergo mastectomy. Depending on age, socioeconomic status, location, and other patient factors, 20% to 40% of these women will choose to undergo breast reconstruction with a plastic surgeon. Previous published works have evaluated transversus abdominis plane (TAP) blocks and found an improvement in post-operative narcotic usage in a variety of abdominal surgeries including abdominoplasty, laparotomy, and cesarean section. In autologous breast reconstruction, the results have not been as robust. A study by Hivelin et al. evaluated bilateral ultrasound-guided TAP blocks in patients undergoing unilateral deep inferior epigastric perforator (DIEP) flap reconstruction and found patients reported lower pain scores post-operatively and demonstrated lower narcotic usage in the first 24 hours after surgery, but not in subsequent time periods out to 48 hours after surgery.
STUDY DESIGN: We have conducted a clinical trial to evaluate prospectively the effect of TAP blocks in breast reconstruction patients. The design of this interventional clinical trial is classified as efficacy with double-blind masking and two-group parallel assignment via randomized allocation to either bupivacaine TAP block or sham saline for the purpose of treatment. Female patients scheduled to undergo breast reconstruction with abdominal free flap were enrolled. This includes patients undergoing deep inferior epigastric perforator (DIEP) flap or free muscle-sparing transverse rectus abdominis muscle (MS-TRAM) flap reconstruction. An a priori power analysis based on data from previous publications indicated that a target total patient population of 128 patients was required to detect a difference of 10 mg in narcotic usage with 80% power. Due to COVID-19, the trial concluded prematurely with 109 patients enrolled and thus fell short of its recruitment goal.
DATASET: Beginning on the day following each patient’s breast reconstruction, the postoperative narcotic consumption in oral morphine equivalents was measured and recorded every hour for the next two to three postoperative days. The length of narcotic consumption recording for each patient was determined by when they were discharged from the hospital. Therefore, the dataset consists of the patient’s demographic, reconstructive, and postoperative narcotic use information.
STATISTICAL MODELS: All analyses were conducted in R. A threshold of 0.05 was used to assess statistical significance. Mixed Analysis of Variance (MANOVA) was used to assess differences between the treatment and sham groups. Postoperative narcotic use was the continuous dependent variable measured in oral morphine equivalents. The between-subjects independent variable was treatment with bupivacaine TAP block or sham saline. The within-subjects independent variable was time. There were no outliers in either the between or within-subjects independent variables. Levene’s test was used to assess homogeneity of variances. Mauchly’s test was used to assess sphericity. The dependent variable was non-normally distributed on studentized residuals and Shapiro-Wilk test. Therefore, a Yeo-Johnson power transformation was applied to the dependent variable to satisfy the statistical assumptions of MANOVA to determine the primary outcome. Subsequent analysis removed the Yeo-Johnson transformation and considered untransformed postoperative narcotic usage. MANOVA with the Yeo-Johnson transformed outcome met all assumptions for statistical validity, while MANOVA with the untransformed outcome violated statistical assumptions. When a statistically significant difference in postoperative narcotic usage between treatment and sham groups is mentioned, it is in reference to the difference in Yeo-Johnson transformed narcotic usage, which is a statistically valid conclusion. By contrast, subsequent, untransformed analysis is presented to describe practically meaningful results that are friendly to the clinician. Subsequent subanalysis to investigate two-way interactions between variables was conducted with ANOVA. Differences in means were assessed by paired and unpaired two-sided t-tests. In subanalysis, Bonferroni correction was applied based on the context in which multiple testing occurred.

STATISTICAL ISSUES
(1) Are the assumptions of MANOVA adequately satisfied for the analysis presented?
(2) Is the Yeo-Johnson transformed outcome valid? Currently, we report untransformed results for the sake of interpretability. Is this appropriate?
(3) How should we handle the fact that different patients had different lengths of postoperative hospital stays, and thus different lengths of postoperative narcotic use monitoring?
(4) What issues or study limitations arise from the premature termination of this clinical trial?

ZOOM MEETING INFORMATION
Join from PC, Mac, Linux, iOS or Android: https://stanford.zoom.us/j/97196061848?pwd=ajY3MmJOUU9oYitMdFZXL3NQYmFEZz09
Password: 571460

Or iPhone one-tap (US Toll): +18333021536,,97196061848# or +16507249799,,97196061848#

Or Telephone:
Dial: +1 650 724 9799 (US, Canada, Caribbean Toll) or +1 833 302 1536 (US, Canada, Caribbean Toll Free)

Meeting ID: 971 9606 1848
Password: 571460
International numbers available: https://stanford.zoom.us/u/aeA1opIz3O

Meeting ID: 971 9606 1848
Password: 571460
SIP: 97196061848@zoomcrc.com
Password: 571460