Teaching Statistics from the Operating Table Minimally Invasive

Teaching Statistics from the Operating Table: Minimally Invasive and Maximally Educational http: //www. amstat. org/publications/jse/contents_2015. html Amy Nowacki, Ph. D Associate Professor of Medicine nowacka@ccf. org

Every class I gave my students a clean little formatted dataset … Inadvertently reinforcing the mistaken equivalence of statistics with data analysis failing to convey the larger investigative cycle. INVESTIGATIVE CYCLE

Guidelines for Assessment and Instruction in Statistics Education (GAISE) • College Report 2012 “Activities should include design, data collection and analysis so students can see the whole process at work. ” • Pre. K – 12 Report 2005 “Students should be taught to appreciate that it is difficult and time-consuming to formulate problems and to get quality data to address the right questions. Most students may not realize this until they go through this experience themselves. Optimally, this encounter should be in a supportive learning environment (such as the classroom) and not on the job where high stakes are often encountered. ”

Develop a new activity for the first session of Intro Biostats: Objective: Prepare students for study start-up - Identify components of a clearly defined research question Compare and contrast types of studies (experimental vs. observational) Define independent and dependent variables Classify types of variables (numeric, categorical, binary, etc. ) Build a database

• Advantages gained by incorporating fun activities and active learning into the classroom (Lesser and Pearl 2008) • One such example is the use of games (Lesser 2007) • Consider desirable characteristics of classroom activities as outlined in the ASA’s GAISE guidelines - activity should mimic a real-world situation activity should involve the class in some of the decisions about how to conduct the activity should include design, data collection and analysis so students can see the whole process at work activity should be fun! GOAL: develop a classroom activity in adherence with those guidelines which utilizes the popular Hasbro board game Operation

STUDENT PREPARATION Complete three brief reading assignments: 1. Introduction to the following statistical topics: flow of a research study, population vs. sample, observational vs. experimental study designs, randomization, independent vs. dependent variables, covariates, and qualitative vs. quantitative variables 2. Two journal articles addressing the controversial topic of what characteristics make for a successful surgeon • reviews the available data on the impact of left-handedness on surgical training (Tchantchaleishvili and Myers Journal of Surgical Education 2010) • identifies factors (hand dominance, gender and experience with computer games) that influence surgeons’ performance (Grantcharov et al. Surgical Endoscopy 2003)

GROUP EXPERIMENT DESCRIPTION Ask each group to discuss the assigned readings regarding what characteristics make a successful surgeon. Have each group generate a hypothesis that they would like to test motivated by the articles. • provides an opportunity to discuss the features of good testable research hypotheses (PICO method)

Characteristics of a Successful Surgeon (CHAOSS) Study Nowacki et al. (2015) Primary Aim: Investigate whether handedness or gender is associated with being a successful surgeon.

GROUP EXPERIMENT DESCRIPTION Together as a class, design the study both defining terms and identifying the: • population - surgeons • sample – surgeons in training (class name) on (class date) • study designs and their features - prospective cohort study • Independent variables • Dependent variable - successful surgery If during the surgery the surgeon avoids touching the edge of the opening with the surgical tool (success); otherwise a buzzer sounds and the patient’s nose lights up red (failure).

For each of the variables to be collected in this study, classify by circling all taxonomy that applies. • HAND qualitative quantitative nominal numeric categorical binary • TIME

CHAOSS Study You envision your data looking something like this: Surgeon Hand Gender Ailment Experience Outcome Time Attempt 1 right male writer’s cramp yes success 4. 1 yes 2 right female funny bone no failure 3. 8 no 3 left male wish bone yes success 6. 7 no 4 right male brain freeze no success 5. 9 yes 5 left female bread basket yes failure 10. 5 no 6 right male failure 2. 5 no 7 left female Adam’s apple success 4. 2 yes writer’s cramp no yes

Set-up 1. Select one group members’ computer for data collection 2. Select one group members’ computer/smartphone to use as a random number generator to determine ailment 3. - computer: http: //www. randomnumbergenerator. com/ - Smartphone: Random Number Generator + App (free) Select one group members’ computer/smartphone to use as a timer - computer: http: //www. online-stopwatch. com/full-screen-stopwatch/ - Smartphone: stopwatch feature of clock

Now let’s collect some data… Study Protocol 1. Select the operating surgeon 2. Record baseline data (SURGEON, HAND, GENDER and EXPERIENCE) 3. Identify and record the ailment (use ailment code sheet) 4. Time the surgery (seconds) and declare surgery outcome (success/failure) 5. Record TIME and OUTCOME data 6. Repeat with new surgeon (do as many surgeries as time allows) 7. When instructed, save JMP data file and email to instructor (nowacka@ccf. org)

Teaching Applications Using the Data from the Experiment • • exploration of descriptive statistics estimation of a mean or proportion confidence intervals t-test / Wilcoxon rank-sum test one-way ANOVA chi-square Fisher’s exact Be sure to revisit this dataset minimally to answer the primary research question (a chi-square test to compare the success rates and a t-test/rank-sum to compare the length of surgeries among the genders). Additional Issues: rare exposures, confounding, data management

Implementation Issues Number of board games • first year: 1 board game for 32 students and 50 minutes to collect data n = 64 surgeries, I recorded the data myself • second year: 4 board games for 32 students and 30 minutes to collect data n = 36 surgeries, students recorded data • now: 8 board games for 32 students and 30 minutes to collect data n = 147 surgeries, students recorded data Rough estimate ≈ 2 minutes per surgery Cost • • range from $5 to $20 depending on the version of the game slowly build inventory utilize e. Bay and Christmas sales Batteries!

THANKYOU QUESTIONS? ? ? Cleveland Clinic Statistical Education Dataset Repository http: //www. lerner. ccf. org/qhs/datasets/