Department of Surgery, Dartmouth-Hitchcock Medical Center

THE FIFTH ANNUAL STARS (SURGICAL TRAINEES ADVANCING RESEARCH SYMPOSIUM)

Thursday Nite Dinner with the Stars Featuring Keynote Presentation and “Quickshots” from surgical trainees. April 13th, 2017 from 5:30 – 8PM. Auditorium F

5:30 Reception and Hors D’ oeuvres

Order Time Author Title

1 6:03 PM Ravinder Kang, MD, MS The Effect of Neoadjuvant Chemoradiotherapy on Bowel Function for Patients with Rectal Cancer

2 6:10 PM Annah Vollstedt, MD Regional Variation in Diagnostic Testing for Overactive Bladder in the Female Medicare Population

3 6:17PM Ian C. Bostock, MD, MS Local Anesthesia in Elective EVAR Reduces Resource Utilization Without Affecting Outcomes

4 6:24 PM Lael Reinstatler, MD, MPH

Association of pregnancy with stone formation among US women: A National Health and Nutrition Examination Survey analysis 2007-2012.

5 6:30 PM

Karl Bilimoria, MD, MS, Northwestern University's Feinberg School of Medicine

Informing Policy with Evidence: The FIRST Resident Duty Hour Trial

6 7:45 PM Andrew Lambour, MD Improving the Discharge Timing of Surgical Inpatients Through a QI Initiative

7 7:52 PM Marc A. Polacco, MD Utility of Lingual Tonsillectomy in the Unknown Primary

8 7:59 PM Alyssa M. Flores Decision-Making in Low Risk Patients with Severe Aortic Stenosis – Are Frailty Markers Useful?

Friday Morning Rising Stars Presentations Full length presentations from Surgical Trainees. April 14th, 2017 from 6:45 – 8:30AM. Auditorium G

6:30 Reception with Coffee Service

Order Time Author Title

1 6:45 AM Karissa Tauber Medicare-Linked Registries Can Accurately Capture True Clinical Events

2 7:00 AM John M. Fallon, MD The incidence and consequence of patient prosthesis mismatch after surgical aortic valve replacement

3 7:15 AM Maureen V. Hill, MD An Educational Intervention Decreases Opioid Prescribing after General Surgical Operations

4 7:30 AM Jesse A. Columbo, MD Long-Term Mortality After Carotid Revascularization: A Novel Instrumental Variable Method for Valid Inference

5 7:45 AM Christopher Funderburk, MD, MS

Innovations in the Surgery Care Pathway: Using Telemedicine for Clinical Efficiency and Patient Satisfaction

6 8:00 AM Kevin Koo, MD, MPH The Burden of Cystoscopic Bladder Cancer Surveillance: Anxiety, Discomfort, and Patient Preferences for Decision-Making

7 8:15 AM Ravinder Kang MD, MS Receipt of Sentinel Lymph Node Biopsy (SLNB) for Thin Melanoma is Associated With Distance Traveled for Care.

Co-Directors: Philip P. Goodney, MD MS; Kari Rosenkranz, MD Prizes: $500 / $250 / $100 / $100 (1st/2nd/3rd, Young Investigator)

Judges: Sandra L. Wong, MD, MS; Srinivas J. Ivatury, MD, MHA; Alexander Iribarne, MD MS; Florian Schroeck, MD, MS; Eric Henderson, MD, MS; Karl Bilimoria, MD,MS

Ravinder Kang, MD, MS

The Effect of Neoadjuvant Chemoradiotherapy on Bowel Function for Patients with Rectal Cancer

Ravinder Kang, M.D., M.S., Jesse Colombo, M.D., S. Joga Ivatury, MD, M.H.A

BACKGROUND: Chemoradiation is known to cause gastrointestinal and mucosal irritation during treatment. However, its subsequent effect on bowel function is less clear. Rectal cancer patients undergo neoadjuvant chemoradiation for locoregional control. These patients may have altered bowel function at baseline. This study aims to evaluate the effect of neoadjuvant chemoradiation on the change in bowel function from baseline.

METHODS: Patients who underwent long-course neoadjuvant chemoradiation with either locally advanced, or early stage rectal cancer with significant medical comorbidity, were included in this prospective study. The COREFO questionnaire, a validated survey designed to assess bowel function in five domains (incontinence, frequency, social impact, medication, and stool-related aspects) was administered at time of presentation, and after completion of neoadjuvant chemoradiotherapy. Values for domain and total score range from 0 to 100, with a higher score representing greater functional disturbance. Demographic, preoperative stage, and tumor data were collected. Changes in COREFO scores were evaluated with paired t-tests.

RESULTS: All patients who met inclusion criteria from December 2014 to July 2016 were prospectively enrolled (n=19). The mean age was 68 years (standard deviation (SD): 12) and 10 (53%) of the patients were men. Preoperative local staging by magnetic resonance or endoscopic ultrasound demonstrated that 3 of the patients were stage I, 5 stage II, 10 stage III and 1 stage IV. The mean tumor location from the anal verge was 5 cm (SD: 3 cm). The median time between baseline and post-neoadjuvant questionnaire completion was 116 (interquartile range (IQR): 98-183) days, and the median time between completion of neoadjuvant therapy and administration of the second COREFO survey was 45 (IQR: 31-139) days. There was no difference in mean total COREFO score from baseline to post-neoadjuvant (Baseline: 23 (18), Post-neoadjuvant: 24 (16), p=0.86). Similarly, no difference was found in any domain score (Table 1).

CONCLUSION: Patients with rectal cancer who undergo neoadjuvant chemoradiation should not expect a significant change in their bowel function immediately following therapy from baseline.

Ravinder Kang, MD, MS

Annah Vollstedt, MD

Regional Variation in Diagnostic Testing for Overactive Bladder in the Female Medicare Population

Annah Vollstedt, Rachel Moses, E. Ann Gormley

Introduction: Overactive bladder (OAB) remains a common urologic ailment with direct healthcare costs exceeding billions annually. The American Urology Association (AUA) released the non-neurogenic OAB guideline in 2012 to guide the diagnosis and management of this costly ailment. OAB is a clinical diagnosis requiring a careful history, physical exam, and urinalysis. Additional work-up including post-void residual, urodynamics, cytology, and cystoscopy are not necessary in the uncomplicated patient. The purpose of this study is to determine rates of potentially unnecessary diagnostic testing in patients carrying an OAB diagnosis before and after the OAB guidelines publication.

Methods: Using the Dartmouth Institute’s Atlas Rate Generator exploring Medicare claims data, we identified females with a diagnosis of OAB by ICD-9 codes within 306 hospital referral regions (HRR). Rates of diagnostic tests within HRR were compared to the national average adjusted by age and race.

Results: The national average rate for diagnostic procedures performed on patients with OAB was 41% (163,919/399,004) in 2011, and only slightly decreased to 38.2% (169,706/443,512) in 2014. Comparing HRRs to the national rate, use of diagnostic procedures demonstrated almost 8-fold variation even after controlling for age and race for both years (Figure 1, 2). In 2011 the lowest rate was identified in Minot, ND (0.260) and the highest in Fort Myers, FL (2.036). By 2014, following the widespread dissemination of the AUA guidelines, the lowest rate was identified in Rapid City, SD (0.304) and the highest again in Fort Myers, FL (2.37).

Conclusion: There is significant regional variation in the work up of OAB. Interestingly, the rates of diagnostic testing did not appear to change significantly after the publication of the OAB guidelines. Further research is needed to identify how much of this diagnostic testing is inappropriate and to explore the relationship of diagnostic testing to management outcomes.

Annah Vollstedt, MD

Figure 1

Annah Vollstedt, MD

Figure 2

Ian C. Bostock, MD, MS

Local Anesthesia in Elective EVAR Reduces Resource Utilization Without Affecting Outcomes

Ian C. Bostock1 MD MS, Devin S. Zarkowsky MD2, Caitlin W. Hicks MD MS2, David H. Stone MD1, Shant N. Vartanian MD2, Philip P. Goodney MD MS1

1. Division of Vascular and Endovascular Surgery, Dartmouth-Hitchcock Medical Center, 1 Medical Center Drive,Lebanon, NH 03766

2. Division of Vascular and Endovascular Surgery, University of California3. Department of Surgery, The Johns Hopkins Medical Institutes, 1800 Orleans Street. Baltimore, MD 21287

Introduction: General anesthesia, especially when associated with prolonged intubation, has been suggested to detrimentally affect post-operative outcomes after elective EVAR. We aim to compare outcomes for patients who were and were not intubated in a national dataset.

Methods: All patients undergoing elective EVAR in the Vascular Quality Initiative dataset (01/2003-6/2016) were analyzed according to anesthetic group. Pre-, intra- and post-operative variables demonstrating <1% missing values were compared with univariate Chi-square, t-test, ANOVA and Kruskal-Wallis test by ranks. Coarsened exact matching (CEM) was performed to refine patient-level variation on independent survival covariates. Kaplan-Meier survival estimates and Cox proportional hazard modeling described survival and independent covariates with survival.

Results: A total of 22,173 patients underwent elective EVAR during the study period. Of these 921 (4.2%) received local anesthesia, 964 (4.4%) regional anesthesia and 20,288 (91.5%) GETA. These groups differed significantly on 12 of 22 preoperative variables; a 1:1 CEM based on all of these variables created three statistically similar groups. Within the matched cohort, patients treated with local anesthesia demonstrated shorter procedure times, required less contrast and crystalloid, suffered less blood loss and had fewer endoleaks on completion angiogram (Fig 1, all P<0.05). Post-operatively, these patients required shorter hospital and ICU stays and required fewer transfusions (All P<0.05). They suffered complications and were discharged to home at a frequency similar to regional anesthetic and GETA patients (Fig 1, all P>0.05). Figure 2 is Kaplan-Meier survival analysis showed statistically similar 90-day survival (local 98% 95%CI 96-99 vs. regional 98% 95-99 vs. GETA 98% 96-99 Log-rank p=0.40). A Cox proportional hazard model in the unmatched cohort demonstrated a protective survival effect from regional and GETA in comparison to local anesthetic that became non-significant on CEM (unmatched Harrell’s C = 0.7, matched Harrell’s C = 0.7).

Conclusions: Local anesthesia for elective EVAR is associated with shorter procedure times, hospital stays and ICU stays with no difference in post-operative complications or survival when patients are matched by pre-operative characteristics. Continuing efforts to evolve EVAR towards less invasive surgical and anesthetic approaches are likely to lead towards lower healthcare costs without detriment to patient outcomes.

Ian C. Bostock, MD, MS

Figure 1. Operative and outcome variables associated with anesthetic technique.

Figure 2. Kaplan-Meier survival estimate.

Lael Reinstatler, MD, MPH

Association of pregnancy with stone formation among US women: A National Health and Nutrition Examination Survey analysis 2007-2012.

Authors: Lael Reinstatler, Sari Khaleel (affiliated w/ Geisel School of Medicine), Vernon M Pais Jr

BACKGROUND: Lithogenic urinary changes occur during pregnancy. Such changes may increase stone proclivity in working and child rearing aged women thereafter. However, such an association has not been previously identified.

METHODS: We analyzed nationally representative data from the 2007-2012 National Health and Nutrition Examination Survey to assess for an association between pregnancy and nephrolithiasis.

RESULTS: The weighted national prevalence of nephrolithiasis among women ≤50 was 6.4% (95% Confidence Interval (CI) 5.4%-7.6%). The prevalence of nephrolithiasis was significantly higher among women who had been pregnant compared with those who had never been pregnant (7.5% vs 3.2%, p=0.0004). On univariate regression, those who had been pregnant had over twice the odds of having had kidney stones (OR 2.44, 95% CI 1.50-3.98). An increased likelihood of nephrolithiasis among those with history of pregnancy persisted on multivariable logistic regression adjusting for age, ethnicity, obesity, history of diabetes, gout, hormone use, water intake and high sodium diet (OR 2.13, 95% CI 1.31-3.45). Finally, the adjusted prevalence of nephrolithiasis increased significantly with increasing number of pregnancies, from 5.2% in those with 0 reported pregnancies to 12.4% in those with 3 or more pregnancies (p=0.001).

CONCLUSION: Nephrolithiasis is strongly associated with prior pregnancies. Among women of reproductive age, the odds of stones are greater than doubled in those who had been pregnant compared with those never pregnant. Nephrolithiasis prevalence also increases with increasing number of pregnancies. Future investigation and identification of modifiable risk factors among pregnant patients may allow reduction in burden of stone disease in women.

Lael Reinstatler, MD, MPH

Figure. Adjusted* prevalence of nephrolithiasis among US women <50 years of age from NHANES 2007-2012 with increasing number of reported pregnancies.

*Adjusted for age, ethnicity, obesity, diabetes, gout, hormone use, water intake, and sodium intake>2300mg/day.

Andrew Lambour, MD

Title: Improving the Discharge Timing of Surgical Inpatients Through a Quality Improvement Initiative

Authors A. Lambour, MD1; C. Kerrigan, MD, MHCDS2; N. Batulis, MHA, LSSBB, CPHQ3; P. Goodney, MD4, and S. Wong,MD, MS, FACS1

1 Department of General Surgery, Dartmouth-Hitchcock Medical Center, Lebanon, NH2 Department of Plastic Surgery, Dartmouth-Hitchcock Medical Center, Lebanon, NH3 The Value Institute Learning Center at Dartmouth-Hitchcock, Dartmouth-Hitchcock Medical CenterLebanon, NH4 Department of Vascular Surgery, Dartmouth-Hitchcock Medical Center, Lebanon, NH

Background Timely discharge is a financial imperative for a sustainable health care system. At our tertiary care institution, we found that surgical discharges were clustered in the early afternoon - creating a barrier to timely discharge and a bottleneck to patient throughput.

Methods This quality improvement project utilized a multidisciplinary, Rapid Process Improvement Workshop for root cause analysis and intervention generation. Four interventions were implemented using a Plan-Do-Study-Act approach: interdisciplinary discharge rounds, preoperative discharge-planning questionnaire, job aids, and new EMR discharge navigator. The primary outcome was average discharge time of day on two surgical floors (2W and 4W). Secondary outcomes include geometric mean length of stay (GMLOS), 30-day all cause readmissions, and patient satisfaction scores (by unit and sub-specialty service).

Results Baseline average discharge time on 2W and 4W was 14:48 and 14:52; at 90-days post-intervention, mean discharge time was 14:11 (p=0.43) and 14:07 (p=0.01), respectively. There were no significant changes in percent GMLOS or 30-day readmissions: GMLOS 114% vs. 117% (2W) and 120% vs. 119% (4W), and readmissions 12.7% vs. 8.1% and 12.9% vs. 13.1%, respectively. Patient satisfaction with speed of discharge was not significantly different (49% vs. 62% (2W) and 59% vs. 52% (4W); however a difference within Vascular and Urology was identified 48% vs. 63% and 53% vs. 72%, respectively.

Conclusion We demonstrate the successful implementation of an intervention to improve discharge timing for surgical patients. A significant change in discharge timing for a subset of surgical patients was observed. Ongoing work includes continuing to hardwire the interventions, monitor outcomes for further improvement, and exploring reasons associated with this improvement, with the intent to extrapolate these findings to other patient groups.

Marc A. Polacco, MD

Utility of Lingual Tonsillectomy in the Unknown Primary

Polacco MA, Sudoko C, Gosselin BJ, Paydarfar JA

Objectives: Localization of the primary site in patients presenting with metastatic head and neck carcinoma of unknown primary (HNCUP) is critical in order to achieve accuracy of diagnosis and minimize treatment morbidity. This can be challenging with standard direct laryngoscopy and palatine tonsillectomy. Previous studies have shown promising diagnostic utility by adding lingual tonsillectomy (LT) using either transoral robotic surgery (TORS) or transoral laser microsurgery (TLM), with localization rates up to 80%. In this study we report our experience with LT in the diagnostic workup of HNCUP.

Methods: This was a retrospective chart review of all patients with HNCUP presenting to Dartmouth-Hitchcock Medical Center between 2011 and 2016 who underwent LT in addition to standard staging laryngoscopy/palatine tonsillectomy as part of their diagnostic workup. HNCUP was defined as a lack of findings on physical exam, flexible laryngoscopy, and lack of primary localization on PET/CT.

Results: Eleven patients met inclusion criteria. All patients underwent preoperative PET/CT and contrast-enhanced neck CT scan. LT was performed either during the standard staging laryngoscopy/palatine tonsillectomy (5/11) or as a secondary procedure (6/11). LT was performed using TORS (2/11), TLM (7/11), or trans-oral microsurgery with cautery (2/11). The addition of LT resulted in a detection rate of 27% (3/11). Breakdown by technique was TORS (0/2, 0%), TLM (2/7, 29%), and trans-oral resection with cautery (1/2, 50%).

Conclusion: Localization of the primary from HNCUP can lead to significant therapeutic benefits. Although previous studies have shown detection rates of 70-80% following LT, we find a lower detection rate which appears independent of surgical technique.

Alyssa M. Flores

Decision-Making in Low Risk Patients with Severe Aortic Stenosis – Are Frailty Markers Useful?

Alyssa M. Flores, BS1; Dave-Gregory Chackery, MD1; Alexander Iribarne, MD, MS1,2 1Section of Cardiac Surgery, Dartmouth-Hitchcock Medical Center, West Lebanon, NH. 2 The Dartmouth Institute for Health Policy and Clinical Practice, West Lebanon, NH.

Introduction: In treating severe aortic stenosis, appropriate risk assessment is essential to guiding patient selection between surgical or transcatheter aortic valve replacement. Frailty status has been shown to impact prognosis of high and intermediate-risk patients, however it is unclear if frailty has a significant role in risk stratification in lower risk populations. In this analysis we assess the utility of frailty markers on predicting in-hospital outcomes among patients undergoing surgical aortic valve replacement (SAVR) with a low Society of Thoracic Surgeons predicted risk of mortality (STS-PROM).

Methods: Data on 6 frailty markers (Katz ADL, IADL, grip strength, KCCQ, 5-meter walk test, and albumin) were prospectively collected on 86 patients with severe aortic stenosis considered low risk for SAVR or SAVR + coronary bypass grafting based on their STS-PROM. One point was assigned to each frailty marker if considered positive to develop a frailty score. Patients were then followed prospectively for the primary endpoints of: ICU and hospital length of stay (LOS), post-operative complications, and mortality.

Results: All patients underwent successful SAVR with a tissue valve. The mean age was 73.2 ± 8.4 years and the mean STS-PROM was 2.7 ± 1.5 %. The distribution of frailty points was: 0 points = 47.7% (n=41), 1 point = 34.9% (n=30), ≥2 points = 17.4% (n=15). The median ICU LOS was 1.7 [1.0-3.1] days, median hospital LOS was 6 [5-7] days, and there were no in-hospital deaths. When controlling for STS-PROM, frailty score was associated with ICU (p=0.01) and hospital LOS (p=0.01) [Figure 1]. However, there was no correlation between risk of post-operative complications and frailty score (p=0.079).

Conclusion: Among patients with a low STS-PROM who undergo SAVR, frailty markers may predict length of ICU and hospital stay but not risk of post-operative morbidity. With the expansion of TAVR into lower risk populations, additional markers may be needed to optimally risk stratify such patients.

Karissa Tauber

Title: Medicare-Linked Registries Can Accurately Capture True Clinical Events

Authors: Karissa Tauber, Jesse Columbo, Kathleen Leinweber, Regis Hila, Andy Hoel, David Stone, Philip Goodney

Introduction: Disease registries are often linked to Medicare claims to improve follow-up. The accuracy of captured events in linked data is not known. Our objective was to compare events found within a Medicare linked disease registry to a gold standard of medical chart review.

Methods: We reviewed the charts of all patients undergoing endovascular aortic aneurysm repair (EVAR) at our institution from 2003 to 2013 (n=729). We then compared the incidence of reintervention between three data sources: chart review, the Vascular Quality Initiative (VQI) disease registry, and VQI registry data linked to Medicare claims (VQI-Medicare). We excluded patients if VQI follow-up was unavailable (n=68, 9%), or if Medicare data were unavailable (n=114, 16%). Our primary endpoints were the agreement between the three data sources, and the Kaplan-Meier rate of reintervention at 1, 2, and 3 years after EVAR. We contacted patients by telephone interview when data sources were discordant.

Results: All patients in the final cohort (547/547, 100%) had follow-up available from all three data sources. There was strong patient level agreement between chart review and VQI-Medicare data (kappa=0.93). Compared to chart review, VQI-Medicare was 92% sensitive and 96% specific at three-year follow-up. In the first year after EVAR, the Kaplan-Meier estimated VQI reintervention rate was lowest (2.7%) while chart review and VQI-Medicare linked rates were higher (6.0% and 6.4% respectively; Figure 2). Reintervention rates from VQI-Medicare data remained similar to rates found on chart review over three-years of follow-up (log rank p=0.59).

Conclusions: Chart review after EVAR demonstrated a 16% three-year reintervention rate, and 92% of these events were accurately captured using VQI data linked to Medicare claims. Registries linked to Medicare can accurately capture true clinical events without the need for labor intensive chart review.

John M. Fallon, MD

The Incidence and Consequence of Patient Prosthesis Mismatch After Surgical Aortic Valve Replacement

John M Fallon MD, Joseph Desimone MD, Matthew Brennan MD, Dylan Thibeau MS, Anthony Discipio MD, Sean O'Brien PhD, Jeffrey Jacobs MD, David Malenka MD.

Objective: There remains concern about the consequences of patient prosthesis mismatch (PPM) after aortic valve replacement (AVR). The goal of this study was to determine the relationship of PPM to long-term survival and to assess if growing concern about PPM has resulted in a decreased incidence over time.

Methods: Using the STS Adult Cardiac Surgery Database, we identified 59,779 patients ≥ 65 years old who underwent isolated surgical AVR between 2004-2014. The degree of PPM was calculated using literature derived effective orifice areas for commonly used valves. Outcomes to ten years were stratified by degree of PPM.

Results: The distribution of PPM was 35% none (n=21053), 54% moderate (n=32243), and 11% severe (n=6483). Compared to those with none, patients with moderate or severe PPM had significantly increased risk of readmission for heart failure (HR 1.15 95% CI 1.09-1.21, HR 1.37 95% CI 1.26-1.48) and redo AVR (1.41 95% CI 1.13-1.77, HR 2.68 95% CI 2.01-3.56). Survival was significantly worse for any degree of PPM (moderate to none HR 1.08 95% CI 1.05-1.12, severe to none HR 1.32 95% CI 1.25-1.39) with ten year adjusted survival rates of 46%, 43%, and 35% for none, moderate, and severe (p<0.001). The incidence of severe PPM decreased by 55% over the study period from 13.8% in 2004 to 6.2% in 2014.

Conclusions: Any degree of PPM significantly decreased long-term survival and increased re-admission rates for both heart failure and re-operation for AVR. Temporal trends show a significant decrease in the incidence of PPM over the past decade.

Figure 1) Distribution of the incidence of PPM over time

Maureen V. Hill, MD

An Educational Intervention Decreases Opioid Prescribing after General Surgical Operations

Maureen V. Hill MD1, Ryland S. Stucke MD1, Michelle L. McMahon BS2, Julia L. Beeman BS1, Richard J. Barth Jr. MD1

1Department of Surgery, Dartmouth Hitchcock Medical Center, Lebanon, NH 2Geisel School of Medicine at Dartmouth, Hanover, NH

Objective: We recently analyzed opioid prescription and use for 5 common outpatient operations at our institution: partial mastectomy (PM), partial mastectomy with sentinel lymph node biopsy (PM SLNB), laparoscopic cholecystectomy (LC), laparoscopic inguinal hernia repair (LIH) and open inguinal hernia repair (IH). We found that opioids were over-prescribed. We formulated guidelines for opioid prescribing that would half the number of pills prescribed and also satisfy 80% of patients’ opioid requirements. Our objective was to determine whether an educational intervention was sufficient to decrease opioid prescribing after these general surgical operations.

Methods: We discussed our findings and opioid prescribing guidelines with surgeons at our institution. We recommended that surgeons encourage patients to use a non-steroidal anti-inflammatory drug (NSAID) and acetaminophen before using opioids. We then evaluated opioid prescriptions and use in 246 subsequent patients undergoing these same operations.

Results: The mean number of opioid pills prescribed for each operation markedly decreased: PM 19.8 vs 5.1; PM SLNB 23.7 vs 9.6; LC 35.2 vs 19.4; LIH 33.8 vs 19.3 and IH 33.2 vs 18.3; all p < 0.0003. The total number of pills prescribed decreased by 53% when compared to the number that would have been prescribed prior to the educational intervention. Only 1 patient (0.4%) required a refill opioid prescription. Eighty five percent of patients used either a NSAID or acetaminophen.

Conclusions: By defining post-operative opioid requirements through patient surveys and disseminating operation specific guidelines for opioid prescribing to surgeons we were able to decrease the number of opioids initially prescribed by more than half. Decreased initial opioid prescriptions did not result in increased opioid refill prescriptions.

In review: Annals of Surgery

Effect of education intervention on total number of opioid pills prescribed

Operation Patients in post-education group

Pre-education mean # opioids prescribed

# opioids would have been prescribed

# opioids actually prescribed

% Decrease

PM 58 19.8 1148 295 74.3 PM SLNB 62 23.7 1469 598 59.3 LC 58 35.2 2042 1129 44.7 LIH 27 33.8 913 520 43.0 IH 18 33.2 598 330 44.8 Total 224 6170 2872 53.3

Jesse A. Columbo, MD

Title: Long-Term Mortality After Carotid Revascularization: A Novel Instrumental Variable Method for Valid Inference Authors: Jesse A. Columbo; Pablo Martinez-Camblor; Ravinder Kang; Todd A. MacKenzie; Douglas O. Staiger; A. James O’Malley; Philip P. Goodney Introduction: Studies comparing carotid endarterectomy (CEA) vs stenting (CAS) are conflicted. Randomized trials show no difference in long-term mortality, while observational studies of “real-world” practice show CEA to be superior. We employed a novel instrumental variable method to allow valid inference for long-term mortality after CEA vs CAS in a large disease registry. Methods: We included all patients who underwent CEA or CAS within the Vascular Quality Initiative (VQI) disease registry from 2003-2015 (CEA n=28,712, CAS n=8,117; male: 60.5%; mean age: 69.9 years, ±9.6 years; total person-years of follow-up: 132,016). We created crude and adjusted Cox-regression models after the index procedure. We then determined the proportion of CEA out of the total carotid procedures (CEA+CAS) performed at each hospital. We used this proportion as an instrumental variable and applied our new instrumental variable to the Cox-regression model. Our primary outcome of interest was all-cause mortality. Results: All-cause mortality was 13.7% (3,955/28,712) for CEA and 9.9% (807/8,117) for CAS during follow-up. The crude hazard ratio (HR) for CEA vs CAS was 0.72 (95% confidence interval (CI): 0.67-0.78), indicating that patients who underwent CEA were 28% less likely to die under follow-up. The HR adjusted for age, sex, race, and co-morbidities was 0.75 (CI: 0.69-0.81), indicating a persistent protective effect of CEA after adjustment for confounding variables. However, this protective effect on mortality for CEA was no longer significant after instrumental variable corrected adjustment, with a HR of 0.95 (CI: 0.81-1.11; Figure). Conclusions: Estimated mortality for CEA vs CAS is dependent on the method of analysis used. Our novel instrumental variable method demonstrated no significant difference in mortality, a finding consistent with randomized trials. Researchers should consider instrumental variable analysis when attempting to draw valid inference from observational data.

Jesse A. Columbo, MD

Figure 1: Baseline Characteristics of the Crude and Propensity-Matched Cohort

Christopher Funderburk, MD, MS

Innovations in the Surgery Care Pathway: Using Telemedicine for Clinical Efficiency and Patient Satisfaction Christopher Funderburk, M.D., M.S., Nicole Batulis, M.S., Alison Evans, APRN, Sheila Aubin, R.N., Nickolay Markov, M.D., Tatyana Kalinchuk, P.A., and John Nigriny, M.D. INTRODUCTION Surgical nurses receive numerous calls from patients in the post-operative period. Baseline data collection in the Section of Plastic Surgery revealed an average of 50 follow-up calls per week. Prior to initiating this project, patients were frequently offered default recommendations to return to clinic to be seen by a provider. These visits were often unnecessary and a burden to patients—many of whom travelled long distances and missed work opportunities. Offering telemedicine may allow us to properly triage patients and avoid unneeded visits. Our goals with this project were to open access for patients, decrease clinic burden, and, ultimately, to improve patient satisfaction. METHODS Seventy-five plastic surgery patients were surveyed pre-operatively and after their post-operative telehealth visits from July 2015 to August 2016. The preoperative survey sought to determine travel distance, comfort with technology, access to the Internet and video-enabled devices, and the patient’s interest in telehealth. The post-operative survey focused on patient satisfaction with the experience. For this project, we collaborated with the Visiting Nurse Associations of Vermont and New Hampshire to facilitate twenty-minute in-home telehealth visits. RESULTS

Our pre-operative survey revealed that a majority (73%) of patients preferred an in-person follow-up visit at the clinic. However, our post-operative survey distributed after the telehealth encounter found that a great majority (97%) of patients were either satisfied or very satisfied with the telehealth experience. Ninety-six percent (96%) of patients said that their questions were answered appropriately, and ninety-seven percent (97%) of patients stated that they would use telehealth again in the future.

CONCLUSIONS

Telehealth encounters enable real-time clinical decision-making by providing patients and visiting nurses access to DHMC providers while decreasing patient transportation needs and wait times. While initially hesitant to opt for a telehealth encounter in lieu of a traditional clinic visit, the great majority of patients voiced satisfaction with the telehealth experience.

Kevin Koo, MD, MPH

The Burden of Cystoscopic Bladder Cancer Surveillance: Anxiety, Discomfort, and Patient Preferences for Decision-Making Kevin Koo, MD, MPH, MPhil, Lisa Zubkoff, PhD, Brenda E. Sirovich, MD, MS, Philip P. Goodney, MD, MS, Douglas J. Robertson, MD, MPH, John D. Seigne, MB, and Florian R. Schroeck, MD, MS

Introduction Periodic cystoscopic surveillance involves a tradeoff for patients with non-muscle-invasive bladder cancer (NMIBC), who must balance their discomfort and anxiety related to cystoscopy against the risk for cancer recurrence. Evidence-based guidelines for NMIBC specifically recommend shared decision-making for these patients, although evidence on the topic is scarce. We examined discomfort, anxiety, and preferences for decision-making in patients undergoing surveillance cystoscopy for NMIBC. Methods Veterans with a prior diagnosis of NMIBC were invited to complete validated survey instruments assessing procedural discomfort, worry, and satisfaction, and to participate in semi-structured focus groups about their experience and desire to be involved in surveillance decision-making. Focus group transcripts were analyzed qualitatively, using (1) systematic iterative coding, (2) triangulation involving multiple perspectives from urologists and an implementation scientist, and (3) searching and accounting for disconfirming evidence. Results Twelve patients participated in three focus groups. Median number of lifetime cystoscopy procedures was 6.5 (interquartile range 4−10). Based on survey responses, two-thirds of participants (64%) experienced some degree of procedural discomfort or worry, and all participants reported improvement in at least two dimensions of overall well-being following cystoscopy. Qualitative analysis of the focus groups indicated that participants experience pre-procedural anxiety and worry about their disease (Table). While many participants did not perceive themselves as having a defined role in decision-making surrounding their surveillance care, their preferences to be involved in decision-making varied widely, ranging from acceptance of the physician’s recommendation, to uncertainty, to dissatisfaction with not being involved more in determining the intensity of surveillance care. Conclusions Many bladder cancer patients experience discomfort, anxiety, and worry related to surveillance cystoscopy and potential disease progression. While some patients are content to defer surveillance decisions to their physicians, others prefer to be more involved. Future work should focus on defining patient-centered approaches to surveillance decision-making and developing effective decision support tools.

Kevin Koo, MD, MPH

Table. Representative themes and example quotes from focus group discussions.

Themes Explanatory variables

Examples

Process factors

Perceived control over frequency

“From the time that they determine yes or no on the biopsy, then there will be an amount of time that the doctor will determine that he wants to have another cysto… It’s not in my hands; I’m just their patient. [Moderator: And are you comfortable with that?] Oh, yes. It’s because I believe in the doctor.”

Communication of results

“A lot of times [the result] comes back no problem… But in the meantime I do understand that you worry.”

Psychosocial factors

Pre-procedural anxiety

“When they [the urologist] called me they said that this is actually good cancer. It’s something we can take care of. So it’s always in your mind that maybe the next time, it could be bad cancer.”

Worry about disease “You always worry whether it’ll come back or not… So there’s a lot of anxiety associated with it. And is this going to be a good experience, or is somebody going to tell me it’s back? That’s what you worry about.”

Personnel Identity/training of proceduralist

“I feel that if the doctor isn’t going to do [the cystoscopy] and you’ve got a resident or an intern, you should have the option of whether you want them to do it or not. I want to be told. I didn’t know [the proceduralist] wasn’t a doctor…and I wasn’t asked.”

Preferences for decision-making

Acceptance of physician’s recommendations

“If the doctor says we need to see it every three months until we know everything is okay, and then we go to six months. And we follow through with that, and we’ve let [the doctor] make those decisions.”

Uncertainty “[The doctor] was saying maybe come back in two years. But I don’t know. I’ll think about it. I don’t know if it’s necessary, to tell you the truth.”

Dissatisfaction with not being involved more

“No, you’re [the patient] not involved. They [the doctors] tell you.”

Self-efficacy Empowerment “To take care of yourself, you know what I mean? As much as you can. There’s a good sense of satisfaction there, that you’re helping yourself, taking care of it, the best you can.”

Holistic effects “I’ve changed. I’ve cleaned up my act considerably, you know what I mean?… You can set your body up, I think, so that cancer won’t live in it. And I’ve done a lot of changes towards that goal.”

Ravinder Kang, MD, MS

Receipt of Sentinel Lymph Node Biopsy (SLNB) for Thin Melanoma is Associated With Distance Traveled for Care. Ravinder Kang MD, MS; Jesse A Columbo MD; Sandra L. Wong MD, MS, FACS Introduction: Controversy surrounding the use of sentinel lymph node biopsy (SLNB) for thin melanoma (<1 mm in depth) exists. Some surgeons will perform a SLNB in the setting of ulceration, high mitotic index, or young patient age. However other factors may also influence this decision. Our aim was to define the impact of distance traveled on the likelihood of receiving a SLNB for thin melanoma. Methods: We used the National Cancer Database to identify 130,425 patients diagnosed with thin melanoma (1998-2011). We created crude and adjusted regression models to assess the impact of distance traveled for care on the likelihood of receiving a SLNB. We defined distance as short (<12.5 miles), intermediate (12.5-49.9 miles) or long (>50 miles). Results: SLNB was used in 33.3% of thin melanomas. Compared to those who traveled a short distance, those traveling an intermediate distance and a long distance were 12% and 21% more likely to undergo SNLB, respectively (crude odds ratio (OR) 1.12; 95% confidence interval (CI): 1.09-1.15, and 1.21; CI: 1.17-1.26; p-trend<0.001). Adjusting for ulceration and mitotic index did not impact the findings; and further controlling for patient, and hospital characteristics results in an adjusted odds ratio of 1.18 (CI: 1.10-1.26) and 1.29 (CI: 1.25-1.44) for intermediate and long distance respectively). Subgroup analysis for very thin melanoma (<0.75 mm) noted a 27% higher chance of undergoing SLNB if a long distance was traveled for care. Conclusions: There is a stepwise increase in the likelihood of receiving a SLNB for thin melanoma with increasing distance traveled for care.

Ravinder Kang, MD, MS

Impact of Travel Distance on Likelihood of Receiving a Sentinel Lymph Node Biopsy for Thin Melanoma

*

Adjusted for mitotic index and ulceration. ** Adjusted for age, sex, race, comorbidity index, income, insurance status, mitotic index, ulceration, facility type and urban/rural setting.

Figure 1: Increased travel distance is associated with a higher likelihood of receiving a sentinel lymph node biopsy in the setting of thin melanoma (< 1.00mm).