72.08 Resource Utilization and Predictors of Readmission in Medically-Managed Diverticulitis

Posted on January 31, 2019

A. L. Mardock1, Y. Sanaiha1, S. E. Rudasill1, H. Khoury1, H. Xing1, P. Benharash1  1David Geffen School Of Medicine, University Of California At Los Angeles,Cardiothoracic Surgery,Los Angeles, CA, USA

Introduction: Acute diverticulitis is increasingly treated with medical management, reserving resection for complicated or recurrent cases. However, previous reports have shown that nearly one third of medically managed patients with diverticulitis recur within a year of discharge from initial presentation. The present study utilized a national database to assess the incidence and resource utilization of ninety-day readmission following medical management of diverticulitis in the United States.

Methods: This retrospective review of the 2010-2015 National Readmissions Database included all adults admitted for acute colonic diverticulitis and discharged without surgical intervention. Chi-squared univariate analysis was performed to identify differences in demographics and comorbidities between patients with and without 90-day readmission. Causes of readmission were identified using diagnosis related group codes.  Multivariable regression analysis was performed to assess independent predictors of readmission.

Results: Of the estimated 746,053 patients, 3,906 (0.5%) expired at index hospitalization, while 176,390 (23.6%) were readmitted within ninety days of discharge. Among those readmitted, 53.8% had a primary diagnosis at readmission related to small or large bowel procedure, digestive disorder, or gastrointestinal hemorrhage, all commonly associated with recurrent diverticulitis. Readmitted patients were older (62.8 vs. 60.5 years, p<0.01) and more likely to be female (59.4 vs. 58.5%, p<0.01) and insured by Medicare (49.6 vs. 42.4%. p<0.01). Readmitted patients also had higher adjusted costs ($8,488 vs. 6,667, p<0.01) and longer lengths of stay (4.9 vs. 3.7 days, p<0.01) at their index hospitalization. During the first readmission, 2,938 (1.7%) patients expired, and the mean length of stay was 5.9 days (95% CI 5.85-5.97), leading to a mean cost of $13,341 (95% CI $13,150-13,531). Independent predictors of 90-day readmission included intestinal or colovesical fistula, abscess, and concurrent malignancy, among others (see Table).

Conclusion: Readmissions following medical management of colonic diverticulitis represent a significant cause of resource utilization and are associated with increased mortality. Many such readmissions are related to recurrent gastrointestinal complications and procedures. It merits discussion as to whether surgical treatment at index hospitalization could conserve resources by preventing unnecessary readmission. A stronger understanding of factors that predispose these patients to readmission could guide decisions regarding medical versus surgical treatment of acute diverticulitis.

72.07 Surgical Therapy for Small Bowel Obstruction Decreases Readmissions and Increases Cost?

Posted on January 31, 2019

M. B. Richardson1, R. J. Reif1, S. Haruna1, H. Jensen1, S. Karim1, W. C. Beck1, J. R. Taylor1, K. W. Sexton1  1University of Arkansas for Medical Sciences,Little Rock, AR, USA

Introduction:  Small bowel obstruction (SBO) is common in patients hospitalized for acute abdominal pain. However, data on long-term follow-up of patients is lacking and no superior management strategy has been identified. We hypothesized that surgical management would decrease readmissions compared to medical management in the treatment of SBO.?

Methods:  This was a retrospective study of the 2010 – 2014 National Readmissions Database. Patients diagnosed with SBO were categorized into two groups: patients that were operatively treated (surgical), and patients managed conservatively (medical). We compared the in-hospital outcomes and readmission rates between the two groups (α=0.05). ?

Results: Within the study period, 778,599 patients diagnosed with SBO were identified. A total of 68,400 (8.8%) patients were treated surgically, compared to 710,199 (91.2%) patients in the medical group. Overall mortality (7.7% vs 4.4%, p<0.01) and length of stay (15.7 vs 7.3 days, p<0.01) were higher in surgically treated patients. However, while 83,007 (11.7%) of the patients treated medically were readmitted, only 4,795 (7.0%) of the patients treated surgically necessitated readmission to the hospital. Cost of care was higher for surgically treated patients both during initial hospital stay ($155,293 vs $67,918, p<0.01) and at readmission ($269,105 vs $123,334, p<0.01).?

Conclusion: Surgical treatment of SBO was associated with higher in-hospital mortality and longer length of stay. Patients who were treated medically for SBO had significantly higher readmission rates. Despite a higher rate of readmission, conservative treatment was associated with lower cost of care both at initial hospital admission and readmission. Non-operative management of SBO is a viable and cost-effective treatment strategy.?

 

 

 

 

 

72.06 The Longer It Takes The Longer They Stay: Outcomes Following Laparoscopic Ventral Hernia Repair

Posted on January 31, 2019

A. Y. Lee1, G. Sugiyama1, M. G. Sfakianos1, J. M. Nicastro1, G. F. Coppa1, P. Chung2,3  1Zucker School of Medicine, Hofstra Northwell,Department Of Surgery,Manhasset, NY, USA 2State University of New York Downstate Medical Center,Department Of Surgery,Brooklyn, NY, USA 3Coney Island Hospital,Department Of Surgery,Brooklyn, NY, USA

Introduction:
Laparoscopic ventral hernia repair is widely employed and associated with decreased postoperative pain and reduced length of stay (LOS). We sought to investigate factors that might influence LOS following laparoscopic ventral hernia repair, such as operative time.

Methods:
We used the 2010-2015 ACS NSQIP Participant Use Files (PUF) and identified cases with laparoscopic repair (CPT code 49652) for ventral hernia without mention of obstruction or gangrene (ICD9 codes 553.2, 553.20, 553.21, 553.29). We included only adult (≥18 years) patients that underwent elective surgery, admitted from home, performed by a General Surgeon. We excluded cases with small bowel resection (CPT code 44120), disseminated cancer, wound class III or IV, and missing data. Operative times <5 minutes and LOS >30 days were also excluded. Operative time was divided into quartiles (1st: <61 minutes; 2nd: 61-90 minutes; 3rd: 90-135 minutes; 4th: ≥135 minutes). Age was also divided into quartiles (1st: <47 years; 2nd: 47-58 years; 3rd: 58-67 years; 4th: >67 years). Risk variables included age, sex, race, morbid obesity status, history of diabetes, chronic obstructive pulmonary disease (COPD), congestive heart failure (CHF), dyspnea, bleeding disorder, ascites, hypertension, renal failure, dialysis dependent, functional status, smoking status, steroid use, weight loss, preoperative transfusion within 72 hours of surgery, ASA class, and operative time. Primary outcome was LOS. Multivariable analysis was performed using negative binomial regression adjusting for all risk variables.

Results:
We found 1,791 patients with mean age 56.2 years. Majority were women (n=1,082, 60.4%), and White (n=1,534, 85.7%). Median LOS was 1.0 (SD 2.35) days. Negative binomial regression showed that ASA class IV vs I (IRR 2.90, p<0.0001), ASA class III vs I (IRR 2.92, p<0.0001), class II vs I (IRR 1.98, p=0.00013), partially dependent vs independent functional status (IRR 2.15, p=0.024), history of COPD (IRR 1.60, p=0.00047), insulin dependent diabetes (IRR 1.43, p=0.0037), and 4th vs 1st quartile of age (IRR 1.39, p=0.0011) were associated with increased LOS. Additionally, Black vs White race (IRR 1.32, p=0.0015), 4th vs 1st quartile (IRR 4.35, p<0.0001), 3rd vs 1st quartile (IRR 2.79, p<0.0001), 2nd vs 1st quartile (IRR 1.91, p<0.0001) operative times were highly associated with increased LOS. Dialysis dependent (IRR 0.59, p=0.040), male vs female sex (IRR 0.84, p=0.0035), and morbid obesity (IRR 0.87, p=0.036) were associated with decreased LOS.

Conclusion:
In this large observational study using a national clinical database, operative time in patients undergoing elective laparoscopic ventral hernia repair is independently associated with increased LOS in patients. Additionally, Black vs White race was also found to be independently associated with increased LOS. Prospective studies are warranted to determine ways to decrease disparities in care.
 

72.05 Minimizing Opioid Prescribing in Surgery (MOPiS) Initiative: An Analysis of Implementation Barriers

Posted on January 31, 2019

J. M. Coughlin1,2, M. Shallcross1, W. Schäfer1, R. Khorfan1, J. Stulberg1,3,4, K. Y. Bilimoria1,3,4, J. K. Johnson1,3  1Feinberg School Of Medicine – Northwestern University,Surgical Outcomes And Quality Improvement Center, Department Of Surgery And Center For Healthcare Studies,Chicago, IL, USA 2Rush University Medical Center,Department Of Surgery,Chicago, IL, USA 3Feinberg School Of Medicine – Northwestern University,Chicago, IL, USA 4Northwestern Memorial Hospital,Chicago, IL, USA

Introduction: The United States is amidst an opioid epidemic. To reduce our reliance on opioids for pain management, our institution developed the Minimizing Opioid Prescribing in Surgery (MOPiS) initiative at five different hospitals. MOPiS is a multi-component intervention including (1) preoperative expectation setting on appropriate pain management (2) screening patients for addictive behavior and substance abuse risk, (3) optimizing perioperative non-opioid pain management, and (4) minimizing opioid prescribing at discharge. We conducted a pre-intervention formative evaluation to identify barriers to implementation, as well as solutions to address them.

Methods: Prior to implementation of the MOPiS initiative, we conducted semi-structured interviews at each hospital (n=5) with 23 key stakeholders (surgeons, nurses, pharmacists and administrators). The four components of the intervention were explained and interviewees were asked about perceived barriers and solutions for implementation. Interviews were audio recorded and transcribed. Responses were analyzed to identify common themes using the Theoretical Domains Framework (TDF), an integrative framework that applies theoretical approaches to interventions aimed at behavior change.

Results: Common themes of potential implementation barriers were identified and classified under five TDF domains: knowledge, social/professional role and identity, environmental context and resources, belief about consequences, and behavioral regulation (Table 1). Clinician knowledge of appropriate opioid stewardship is often lacking, and patients do not consistently receive comprehensive pain management education (knowledge). Providers of all levels cited time as the greatest barrier (environmental and context resources) to providing patient education. Pharmacists are not routinely incorporated into the treatment team, although they consistently express an interest in educating and advising (social/professional role and identity). Patients have varying expectations of post-operative pain management and physicians fear patient dissatisfaction if these expectations are not met (belief about consequences). Physicians who over-prescribe are not being held accountable for their prescribing practices (behavioral regulation).

Conclusions: Implementation barriers have the potential to derail any improvement initiative, and successful implementation of an opioid reduction initiative will require specific strategies to overcome barriers. In our MOPiS initiative, utilizing a pre-intervention formative evaluation process enabled the design of strategies to facilitate implementation of the components of the initiative.

72.04 A Nationwide View on Recent Trends in Partial Cholecystectomy

Posted on January 31, 2019

A. F. Sabour2, K. Matsushima1, E. Alicuben1, M. Schellenberg1, K. Inaba1, D. Demetriades1  1LAC+USC Medical Center,Acute Trauma Surgery,Los Angeles, CALIFORNIA, USA 2Keck School of Medicine of USC,Medicine,Los Angeles, CALIFORNIA, USA

Introduction:
Partial cholecystectomy provides a viable alternative to the proverbial “difficult” gallbladder. To date, few studies have observed the establishment of partial cholecystectomy as an increasingly common surgical practice. The purpose of this study was to assess nationwide trends of partial cholecystectomy through evaluation of operative variables, patient- and institution-level characteristics, and yearly rates in procedure preference.

Methods:
Data was obtained from the Nationwide Inpatient Sample (NIS) between 2003 to 2014. Patients with an ICD-9-CM diagnostic code for acute cholecystitis were categorized based on procedures for either open total, laparoscopic total, open partial, or laparoscopic partial cholecystectomy. Any patients younger than 18 years of age or with a preoperative stay longer than 1 week were excluded. Logistic regression analysis was performed to evaluate significant patient- and institution-level characteristics associated with the use of partial cholecystectomy.

Results:
A total of 298,009 patients were evaluated over 12 years. During our study period, the rate of partial cholecystectomy sharply increased from 0.22% of all cholecystectomy procedures to 0.80% (p<0.001). Conversion from a general laparoscopic to open partial cholecystectomy increased significantly from 0% in 2003 to 4.76% in 2014 (p=0.049). Partial cholecystectomies were performed at significantly higher rates in males (OR: 1.96, p<0.001), Asian Americans (OR: 2.15, p=0.037), and patients with alcohol abuse (OR: 2.23, p=0.024). Teaching hospitals (OR: 2.28, p<0.001) and those in rural areas (OR: 2.27, p=0.001) were also found with significantly higher numbers of partial cholecystectomies.

Conclusion:
Growing trends in the use of partial cholecystectomy suggest evolving surgical practices for acute cholecystitis. Current data suggests that gender, ethnicity, and hospital characteristics may already play a deciding role in procedure preference. Future studies are warranted to determine the indications for partial cholecystectomy and its outcome benefit.
 

72.03 Recurrence Rate and Healthcare Expenditures following Component Separation for Abdominal Wall Hernias

Posted on January 31, 2019

J. R. Montgomery2,3, J. Henderson1, J. B. Dimick2,3, D. A. Telem2,3  1University Of Michigan,Statistics,Ann Arbor, MI, USA 2University Of Michigan,General Surgery,Ann Arbor, MI, USA 3University Of Michigan,Center For Healthcare Outcomes & Policy,Ann Arbor, MI, USA

Introduction: Performance of component separation (CS) as part of an open hernia repair has doubled over the past 6 years. The impact of CS as part of open hernia repair (OHR) on hernia recurrence and healthcare expenditures when compared to OHR alone remains unknown. We aim to characterize the comparative risk of hernia recurrence after CS and associated difference in overall healthcare expenditures.

Methods: Using Truven Marketscan Data, patients aged 18 to 65 who underwent an abdominal wall hernia repair from 2009 to 2015 were identified. Recurrence was defined as any hernia repair within two years of index operation. Multivariate logistic regression models and time-to-event analyses were used to determine factors associated with hernia recurrence. Healthcare expenditures were defined by total inpatient payments and both price-standardized and inflation-adjusted.

Results: A total of 22,157 patients underwent hernia repair during the study period. For those patients who completed two years of follow-up (n=7,547), CS (n=1,074) was associated with a decreased incidence of hernia recurrence when compared to OHR (n=6,473) alone for both recurrent (7.5% vs 11.8%, p=0.028) and nonrecurrent (5.2% vs 9.6%, p<0.001) hernias. In a two-factor logistic regression, CS was protective for hernia recurrence (OR 0.55, 95%CI 0.42-0.72, p<0.001), whereas recurrent hernias were more likely to recur again (OR 1.29, 95%CI 1.07-1.56, p=0.008). In a proportional hazard models using all index repairs regardless of follow-up and adjusting for baseline recurrent hernia, CS remained associated with decreased chance of subsequent hernia recurrence (HR 0.51, 95%CI 0.42-0.61, p<0.001) [Figure 1]. Compared to OHR, CS was associated with an increased initial cost of $8,406 ($24,484 vs $16,078). However, given its lower postoperative recurrence rate, it is subsequently associated with a decreased cost of $131 per year, per patient.

Conclusion: Our analysis reveals that CS as part of an OHR is associated with decreased recurrence rates at 2-years as compared to OHR alone. Although initially more expensive, its decreased postoperative recurrence rate is associated with a subsequent savings of $131 per year, per patient. Utilization of CS for patients with highest risk of postoperative hernia recurrence may result in decreased overall healthcare expenditures.

 

72.02 Watchful Waiting for Ventral Hernias: A Large Single-Institution Descriptive Analysis

Posted on January 31, 2019

E. R. Dadashzadeh1,2, D. Van Der Windt1, R. Handzel1,2, J. Moses1, P. Bou-Samra1, V. P. Anto1, M. Hossain1, A. Tsung1, M. R. Rosengart1  1University Of Pittsburgh,Department Of Surgery,Pittsburgh, PA, USA 2University Of Pittsburgh,Department Of Biomedical Informatics,Pittsburgh, PA, USA

Introduction:

Ventral hernias remain a frequent complication of abdominal surgery. To avoid incarceration, operative repair is the current accepted treatment. While the morbidity and mortality of ventral hernia elective repair (ER) are well-documented, the same cannot be said for watchful waiting (WW) as we lack knowledge of the natural history of untreated ventral hernias, including the true incidence of incarceration events. To date, no prospective randomized trials comparing ER to WW have been completed, and the largest retrospective cohort analysis was a 2016 European single-center study consisting of 569 patients. The purpose of this study was to leverage our institution’s large patient population by performing a descriptive analysis of the ventral hernia experience in Western Pennsylvania.

Methods:

This retrospective cohort study was conducted by utilizing linked quality improvement health administrative databases from the University of Pittsburgh Medical Center spanning from January 1, 2010 to December 31, 2017. ICD-9 and ICD-10 coding was used to identify patients with a diagnosis of ventral hernia. Additionally, surgical CPT coding and admission data were used to classify patients into one of the following 3 groups: Elective Repair (ER), Watchful Waiting (WW), and Failure of Watchful Waiting (FWW). Manual audits of randomly selected patients were performed to confirm accurate classification. All-cause mortality was determined using our inpatient database linked with the Social Security Death Index.

Results:

After excluding patients under the age of 18 and those who presented with an incarceration event as their initial encounter within our system, 24240 unique patients were identified with a diagnosis of ventral hernia. 4447 patients underwent ER, 79% of them within 3 months from their initial ventral hernia diagnosis. 19793 patients underwent WW, of whom 264 suffered an acute incarceration event (FWW). The incidence of incarceration events in our population was 3.7 per 1000 patient-years. The all-cause mortality for the FWW cohort was significantly higher than both the ER and WW cohorts (12.1% vs 3.0% and 3.8%, P<0.0001). The median follow-up duration for the entire study was 51 months.

Conclusion:

While watchful waiting appears to be a safe strategy for the majority of patients presenting with a ventral hernia, the mortality associated with its failure is fourfold that of those who underwent elective repair. Instead of awaiting the results of randomized prospective trials in the hope of a single, collective approach for all patients presenting with a ventral hernia, future studies can evaluate leveraging both clinical and imaging data to identify and select those patients with the highest risk of incarceration for elective repair.

72.01 Cost-effectiveness of Index Treatment Strategies for Gallstone Pancreatitis

Posted on January 31, 2019

S. W. Knight1, S. Scaife2, J. D. Mellinger1, S. Ganai1,3  1Southern Illinois University School Of Medicine,Surgery,Springfield, IL, USA 2Southern Illinois University School Of Medicine,Center For Clinical Research,Springfield, IL, USA 3Southern Illinois University School Of Medicine,Population Science And Policy,Springfield, IL, USA

Introduction: Guidelines for management of gallstone pancreatitis (GSP) recommend cholecystectomy at the index admission to limit risk of readmission from recurrent pancreatitis, cholangitis or acute cholecystitis. It is uncertain whether endoscopic retrograde cholangiopancreatography (ERCP) with or without sphincterotomy or stenting is sufficient management for this patient population. We hypothesized that index cholecystectomy (chole) for GSP would be the more cost-effective strategy using a time horizon of a single quality-adjusted life-year.

Methods:  A retrospective cohort analysis was performed using the National Readmission Database (NRD) from 2013-2015 across 1st-3rd quarters to obtain full 90-day readmission data for all analyzed patients. Inclusion criteria identified adults admitted with pancreatitis and various cholelithiasis ICD-9 codes. Exposures were procedures (ERCP, cholecystectomy, none, or both) coded during the index hospitalization. Outcomes included cost, likelihood of 90-day readmission, and risk of death. A societal perspective framework for cost-effectiveness was used based on likelihood of readmission, median cost of each admission, mortality with intervention or readmission, and negative utility of a readmission (NUR), factoring the perceived impact of a 90-day readmission in respect to quality of life (QOL)..

Results: In total, 396,978 index admissions with a diagnosis of GSP were studied. Readmissions were noted in 32.8% of patients receiving no procedure (n=336393), 12.6% receiving chole (n=54917), 13.5% of patients receiving chole and ERCP (n=2260), and 38.6% of patients receiving ERCP (n=3408).  Median cost for those who were not readmitted was $21k for no  procedure, $55k for chole, $59k for chole/ERCP, and $41k for ERCP.  Median cost for those who were readmitted was $72k for no procedure, $83k for chole, $149k for chole/ERCP, and $108k for ERCP. Readmission mortality was 1.45% for no procedure, nil for chole and chole/ERCP, and 0.85% for ERCP.  Figure 1 models cost-benefit ratio using the above data controlling to the cost of a no procedure strategy across varying NUR.

Conclusion: Our cost-effectiveness analysis supports the hypothesis that cholecystectomy performed at the index admission for gallstone pancreatitis is superior to ERCP alone. ERCP is more favorable than ERCP/chole only when the expected surgical mortality is high or the NUR is less than 0.2 (<10 weeks of impaired QOL with readmission), primarily because of higher readmission rates and index+readmission mortality risk. ERCP alone prior to discharge is not an optimal strategy for an average-risk patient.

71.10 Overtreatment: A Qualitative Analysis of Surgeons, Endocrinologists, and Patients with Thyroid Cancer

Posted on January 31, 2019

C. B. Jensen1, M. C. Saucke1, J. L. Jennings1, H. J. Khokhar1, C. I. Voils1, S. C. Pitt1  1University Of Wisconsin,Endocrine Surgery Division – Surgery Department,Madison, WI, USA

Introduction: Overtreatment is a significant problem in the United States, particularly in patients with low-risk thyroid cancer. In order to reduce the harms of unnecessary care, it is essential to understand stakeholders’ attitudes and beliefs about overtreatment. 

Methods: We conducted 34 semi-structured interviews with surgeons, endocrinologists, and patients with low-risk thyroid cancer. Interviews probed about decision-making for thyroid cancer, including less extensive and non-surgical management options. We used content analysis to identify themes related to overtreatment and created concept diagrams to map observed relationships between these themes.

Results: Surgeons and endocrinologists discussed overtreatment of low-risk thyroid cancer as resulting directly from overdiagnosis. They believed the process commonly starts with incidental discovery of a thyroid nodule on imaging and viewed biopsy as a habitual action driving overdiagnosis. Providers ascribed the reflexive biopsy to lack of adherence to or knowledge of guidelines, radiology recommendations, and the desire of patients and physicians to minimize diagnostic uncertainty. Providers described the subsequent diagnosis as an event that lets “the cat out of the bag” or “opens Pandora’s box.” Providers acknowledged that the resulting cancer diagnosis provokes a strong instinctive and culturally rooted need to proceed with surgery. As a consequence, most providers believed it is easier to prevent overdiagnosis than overtreatment. They suggested overdiagnosis can be addressed with provider-focused educational interventions, resetting patients’ expectations, and engaging the media. In contrast, patients did not discuss overdiagnosis or overtreatment. Some patients described the linear process from an incidental finding to surgery. Their statements confirmed the “need to know” was a major motivation for biopsying their nodule. Most patients felt that once they had a cancer diagnosis, surgery was a foregone conclusion. Patients admitted their knowledge about thyroid nodules and cancer was low, leaving room for education about the need for biopsy and less extensive treatments. They expressed significant trust in their surgeon and willingness to consider less extensive management options if recommended.

Conclusion: Surgeons’ and endocrinologists’ attitudes and beliefs about overtreatment focus on the automaticity of overdiagnosis. Both patients and providers acknowledge the established pathway that propels patients from incidental discovery of a thyroid nodule to surgery. Research is needed to determine if this seemingly inevitable progression can be interrupted with educational and behavioral interventions. 

71.08 Impact of a Patient Navigator on a Free Surgery Program for the Uninsured

Posted on January 31, 2019

J. Heimroth1, S. Edwards1, J. Matzke3, S. Walling2, E. SUTTON1  1University of Louisville School of Medicine,Hiram C. Polk, Jr, MD Department Of Surgery,LOUISVILLE, KY, USA 2University of Kentucky School of Medicine,Department Of Surgery,Lexington, KY, USA 3Medical College of Wisconsin,Milwaukee, WI, USA

Introduction: Surgery on Sunday Louisville, Inc. (SOSL) is a nonprofit organization whose mission is to provide free endoscopy and outpatient surgery to people who are uninsured or underinsured.  Two years ago, SOSL began to collect patient outcomes such as wait time from referral to consultation and wait time from referral to surgery.  Seeing lengthy wait times, SOSL hired a patient navigator to attempt to improve these initial two outcomes.  Our hypothesis was that a patient navigator would improve wait times by aiding patient throughout in our referral system.

Methods: Patient outcome data such as date of referral, days from referral to consultation, days from referral to surgical procedure, and dates of follow up were prospectively collected for 278 patients referred to SOSL from February 1, 2016 to June 1, 2018.  The patient navigator, who spoke the primary language of the patient majority (Spanish), was hired on March 1, 2017.  Data were retrospectively analyzed for the 13 months prior to her hiring (Group A) and compared to the 15 months after her hiring (Group B).  Patients were excluded if they did not show up for their clinic appointment and thus had no values for inclusion.  Mean days from referral to consultation and referral to surgical procedure were calculated for each group and compared using a student’s t test for normally distributed data.

Results: Patients referred before the hiring of a patient navigator (Group A, n=99)) experienced a mean wait time of 49 ± 52 days until consultation and 126 ± 98 days from referral to surgical procedure.  Patients referred after the hiring of a patient navigator (Group B, n=160) experienced a mean wait time of 49 ± 51 days until consultation and 149 ± 57 days from referral to surgical procedure.  Thus, there was no significant difference between days to consultation (p=1.0, 95% CI -12.94 to 12.94) between Groups A and B.  There was however a significant difference in days to surgery (p=0.02, 95% CI -41.96 to -4.04) in favor of Group A.

Conclusion: Hiring a patient navigator maintained the wait time from referral to consultation, yet a navigator was not able to impact the wait time from referral to surgery in a model that attempts to delivery surgery free of charge to the patients.  The impact of increased referral volume cannot be determined from this study.  These results suggest another solution to lengthening surgical wait times should be sought.

 

71.07 Do Surgeons and Patients Value Shared Decision Making in Surgery?

Posted on January 31, 2019

E. M. Carlisle1, L. A. Shinkunas2, C. J. Klipowicz3, L. C. Kaldjian2  1University Of Iowa,Division Of Pediatric Surgery/Department Of Surgery,Iowa City, IA, USA 2University Of Iowa,Program In Bioethics And Humanities,Iowa City, IA, USA 3University Of Iowa,Department Of Anthropology,Iowa City, IA, USA

Introduction: Shared decision making (SDM) is presumed to be the preferred approach to patient counseling.  However, few data exist regarding whether patients prefer SDM over a surgeon-guided approach during complex surgical decision making. Even less data exist regarding surgeon preferences toward SDM. In this systematic review we identified studies that address patient and surgeon preferences toward SDM in surgery.

Methods: We conducted a PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) based systematic review of articles published in Medline, EMBASE, and Cochrane databases that evaluated surgeon/patient preferences toward SDM in adult surgery. Two investigators independently reviewed all identified articles.  Articles were included if they specifically investigated preferences of surgeons/patients toward SDM in adult surgery.

Results:The search yielded 20359 articles. 4988 duplicate articles were removed yielding 15371 articles for title/abstract review. 86 articles addressed preferences toward SDM in surgery.  Complete analysis of the articles is currently underway. Preliminary results from the first 35 articles demonstrate the following: 13 subspecialties were represented: Surgical Oncology (34%), General Surgery (20%), Cardiac Surgery (11%), Orthopedic Surgery (11%), Plastic Surgery (11%), Gynecology (9%), Urology (9%), Vascular Surgery (9%), Neurosurgery (6%), Thoracic Surgery (6%), Transplantation (6%), Colorectal (3%), and Otolaryngology (3%).  43% of articles discussed decision making for patients with cancer. Of those, 80% focused on breast cancer. 60% were from non-US institutions, and 71% focused on outpatient decision making. 83% concentrated on patient preferences, and 20% focused on surgeon preferences.  51% discussed decisions between operative or nonoperative management, 34% discussed decisions among different surgical procedures, and 11% discussed decisions regarding the timing of surgery. No articles addressed decision making for emergency surgery.  In the 7 articles addressing surgeon preferences, most surgeons favored SDM. For patients, 46% favored SDM, 23% surgeon-guided decision making, and 14% independent decision making.

Conclusion:Despite recommendations that SDM is the best approach to clinical counseling, our systematic review identified very few articles that evaluate patient and surgeon preferences toward SDM in surgery. The majority of articles focused on non-emergent, outpatient decision making related to oncologic diagnoses. Additionally, most studies were preformed outside of the US where different health care systems or attitudes toward physicians may impact decision making preferences. Further research is needed to understand whether patients and surgeons may prefer a more surgeon-guided approach to decision making in emergent, inpatient decision making. Additionally, more work is needed to assess decision making preferences for patients and surgeons focused on non-oncologic diagnoses.

 

71.06 Prevalence of Thymic Parathyroids in Primary Hyperparathyroidism during Radioguided Parathyroidectomy

Posted on January 31, 2019

S. Dream1, B. Lindeman1, H. Chen1  1University Of Alabama at Birmingham,Birmingham, Alabama, USA

Introduction:

Radioguided surgery has been an effective tool for identifying hyperfunctioning parathyroid glands, including both adenomas and hyperplastic glands during routine parathyroidectomy for hyperparathyroidism.  The purpose of this study was to examine the role of radioguided surgery for the identification of intrathymic adenomatous and hyperplastic parathyroid glands.

Methods:
Between March 2001 to February 2018, 2291 patients underwent parathyroidectomy by one surgeon for primary hyperparathyroidism.  Of these patients 158 (7%) were identified to have an ectopic intrathymic parathyroid gland. All patients underwent radioguided parathyroidectomy with preoperative injection of 10 mci of TC-99m sestamibi.  Ex vivo radionuclide counts were used to confirm parathyroid excision with specimen radioactivity of >20% of the background level.

Results:
The mean age was 56 ±1 years with 74% of the patients being female.  Preoperatively, 122 patients underwent sestamibi scan with the scan correctly identifying the the affected gland 61% of the time.  Mean preoperative calcium was 10.7± 0.1 mg/dL and the mean preoperative parathyroid hormone(PTH) was 112 ± 6 pg/mL. Mean background radionuclide count was 208 +/-7, mean ex vivo radionuclide count was 127 ± 9, with ex vivo counts of removed glands were >20% in all patients.  Thymectomy was performed in 140 of the patients. Mean postoperative calcium was 9.3 ± 0.1 mg/dL and the mean postoperative PTH was 46 ± 3 pg/mL.

All ectopic parathyroid glands were successfully identified using gamma probe.  Ex vivo counts found to be significantly higher in patients with adenomas. Patients with parathyroid adenomas also were older in age and had higher preoperative calcium levels (see table).  While 10% of patients with primary hyperparathyroidism have hyperplasia, 42% of patients with thymic parathyroids had hyperplasia.

Conclusion:
Radioguided parathyroidectomy is useful in detecting ectopic parathyroid glands in the thymus.  Patients with hyperplasia disproportionately have thymic parathyroid glands.
 

71.05 Safety of Thyroid Surgery in The Elderly: A Propensity-Score Matched Cohort Study.

Posted on January 31, 2019

V. Papoian1, F. P. Marji1, J. E. Rosen1, N. M. Carroll1, E. A. Felger1  1Georgetown University Medical Center,Surgery,Washington, DC, USA

Introduction:
Thyroid surgery is becoming more common in the elderly as the proportion of the population that is elderly continues to grow. Unfortunately, there are limited studies evaluating the complication rates of thyroidectomy in elderly patients. We aim to evaluate the relative risk of morbidity from thyroidectomy in patients greater than 75 years of age.

Methods:
Medical records were used to identify all patients older than 75 years undergoing thyroidectomy between 2001 and 2018 in a multi-hospital network. A matched control group was selected with use of a propensity score based on gender, ethnicity, type of surgery, insurance status and comorbidities. The Charlson Comorbidity Index was used to quantify comorbidities. Total complications included both thyroid surgery specific complications, including recurrent laryngeal nerve injury or dysfunction, dysphagia, symptomatic hypocalcemia, hematoma, and wound complications, in addition to systemic complications. Analysis was performed with the use of chi-square analysis and two sample t-tests. A subgroup analysis was performed for patients older than 80 years of age.

Results:
We identified 313 patients over the age of 75 years with a propensity score matched group of 313 patients. There was no difference between the percent female (73% vs 73%, p=0.92), race composition (p=0.91), insurance status (p=0.99), percent undergoing total thyroidectomy (84% vs 84%, p=0.91) and Charleston Index (2.6 vs 2.69, p=0.70) of the two groups. There was no statistically significant difference between post operative emergency room visits (7% vs 6%, p=0.61), readmissions (11.5% vs 8.6%, p=0.18), cardiovascular (1.3 vs 0.6%, p=0.42), pulmonary (3.2 vs 0.9%, p=0.07), or neurologic complications (1.0 vs 0.3%, p=0.34). No re-operations were noted in either group. Elder patients did have a longer length of stay (2.64 vs 1.29 days, p<0.01). The findings for the sub-analysis for patients over the age of 80 showed comparable findings to the entire cohort.

Conclusion:
Elderly patients did have a longer length of stay when compared to a matched younger population. Although, there was a trend with higher complication rates in the elderly, those differences did not reach statistical significance. The current results indicate that thyroidectomy in the elderly is as safe as it is in younger patients when accounting for comorbidities.
 

71.04 An Institutional Experience with Primary Hyperparathyroidism in the Elderly Over Two Decades

Posted on January 31, 2019

K. L. O’Sullivan1, T. W. Yen1, K. Doffek1, S. Wagner1, I. Mazotas1, D. B. Evans1, T. S. Wang1  1Medical College Of Wisconsin,Endocrine Surgery,Milwaukee, WI, USA

Introduction: Parathyroidectomy is the only curative treatment for primary hyperparathyroidism (pHPT) and is associated with low morbidity. An increasing number of elderly patients are undergoing elective surgery and are at greater risk for morbidity and mortality. The aim of this study was to examine the presentation and indications for surgery based on age for pHPT patients at a high-volume institution over 20 years.

Methods:  This is a retrospective review of all patients who underwent initial parathyroidectomy for sporadic pHPT from 1/1999-3/2018. Elderly patients were defined as ≥75 years. To study the progression of pHPT over time, the cohort were divided into 3 timeframes: 1999-2007, 2007-2012, and 2013-2018. Demographic and clinical data were collected.

Results: Of the 1900 patients, 1508 (79%) were female. The median age was 59.7 years (range, 18-94); 202 (11%) were ≥75 years. For the entire cohort, preoperative median serum calcium, ionized calcium, parathyroid hormone (PTH), and 24-hour urine calcium levels decreased over time, while 25-OH vitamin D levels and patient body mass index (BMI) increased (Table). There was no difference in 24-hr urine calcium levels (p=0.06). Over the 3 timeframes, the elderly had lower preoperative serum calcium (11 vs 10.7 vs 10.7;p=0.05) and PTH (150.4 vs 111.9 vs 107.9;p<0.001) levels, but higher 25-OH vitamin D (16 vs 28 vs 31;p<0.001) levels. Fewer patients had fragility fractures (27% vs 20% vs 14%;p=0.005) and more reported symptoms of gastroesophageal reflux (24% vs 41% vs 46%;p<0.001). When compared to patients <75 years, the elderly had similar preoperative serum calcium levels (10.8 vs 10.9;p=0.91), higher PTH (102 vs 121;p<0.001) and creatinine (0.8 vs 0.9;p<0.001) levels, lower 24-hr urine calcium (315 vs 196;p<0.001) levels, and lower BMI (29.4 vs. 27.4;p<0.001). The elderly were more likely to be taking vitamin D (52% vs 43%;p=0.01), have osteoporosis (58% vs 21%;p<0.001), and a history of fractures (20% vs 10%;p<0.001); younger patients had higher rates of nephrolithiasis (27% vs 16%;p=0.001). Postoperatively, by age groups, there was no difference in rates of recurrent laryngeal nerve injury (1.5% in both groups) or hypoparathyroidism (1.4% vs. 2.0%).

Conclusion: Over the 3 timeframes, elderly patients who underwent parathyroidectomy for sporadic pHPT had lower serum calcium and PTH levels, although the clinical significance of these findings is unclear. There was no difference in endocrine-specific complications between the age groups, suggesting that parathyroidectomy in the elderly is not associated with higher morbidity and that elderly patients with hypercalcemia should be evaluated for pHPT and considered for surgical referral.

 

71.03 Predictive Value of Surgeon Performed Ultrasound In the Diagnosis of NIFTP

Posted on January 31, 2019

M. S. Sussman1, M. B. Mulder1, O. Picado1, J. I. Lew1, J. C. Farra1  1University Of Miami,DeWitt Daughtry Department Of Surgery: Division Of Endocrine Surgery,Miami, FL, USA

Introduction:  The reclassification of noninvasive encapsulated follicular variant papillary thyroid carcinoma (FVPTC) to noninvasive follicular thyroid neoplasm with papillary-like nuclear features (NIFTP) has been shown to decrease the reported risk of malignancy (ROM) for all categories within the Bethesda System for Reporting Thyroid Cytology (BSRTC), with the greatest impact seen in atypia /follicular lesion of undetermined significance (AUS/FLUS) or Bethesda III thyroid nodules. There are currently no clinical factors to help predict malignancy vs. NIFTP in indeterminate thyroid nodules. This study evaluates the utility of gene expression classifier (GEC) testing and surgeon performed ultrasound (SUS) features as predictive factors for NIFTP in patients with thyroid nodules.

Methods:  A retrospective review of prospectively collected data of 847 patients who underwent thyroidectomy at a single institution from 2010 to 2016 was performed. Pathology slides with a diagnosis of FVPTC (n=146) were re-reviewed by endocrine pathologists for reclassification to NIFTP. Risk of malignancy (ROM) overall and within each BSRTC classification was determined before and after the reclassification of NIFTP. GEC testing and SUS characteristics were compared in FVPTC vs. NIFTP patients to evaluate for predictive value with significance defined as P<0.05.

Results: Of 146 patients who underwent thyroidectomy for FVPTC, 22% were reclassified as NIFTP (n=32). Of the NIFTP group, 35% (n=11) had AUS/FLUS thyroid nodules. GEC testing was performed in 25 patients, of which 22 had a suspicious result. Suspicious GEC results between FVPTC (12%) and NIFTP (12%) pathologies were identical. On multivariate regression, SUS characteristics of echogenicity and microcalcifications were independent predictors of NIFTP vs. FVPTC. Isoechogenicity was predictive of NIFTP, whereas hypoechogenicity was predictive of FVPTC (OR 3 95% CI 1.3 – 7, p<0.05). Additionally, microcalcifications was predictive of FVPTC compared to NIFTP (OR 4 95% CI .9-18, p<0.05).

Conclusion: A significant proportion of AUS/FLUS thyroid nodules are NIFTP on final pathology. Although GEC testing has limited utility, SUS features such as isoechogenicity and the absence of microcalcifications may favor a diagnosis of NIFTP in such thyroid nodules. This may help guide and determine extent of thyroidectomy in these select cases. 

 

71.02 Surveillance Strategies for Benign Non-functional Adrenal Incidentaloma:A Cost Effectiveness Study.

Posted on January 31, 2019

K. O. Memeh1, M. A. Guerrero1, F. B. Maegawa1,2  1University Of Arizona,General Surgery,Tucson, AZ, USA 2Southern Arizona VA Health Care System,Surgery,Tucson, AZ, USA

Introduction:
There is significant global variation in both society guidelines and clinical practice regarding the surveillance of an initially diagnosed non-functional, radiologically-benign adrenal incidentaloma. Given the low likelihood of these lesions becoming functional or malignant, some clinician have questioned the utility of frequent long term surveillance ( FLTS) strategy compared to no surveillance(NS) strategy.  We sought to evaluate and compare the cost effectiveness of current guidelines in the United States ( FLTS) and Europe( NS). We hypothesized that the FLTS strategy would not be a cost effective approach to managing this group of patients.

Methods:
A Markov transition- state model was created comparing the FLTS and NS strategy for a 60 year old patient diagnosed with a non-functional, non-malignant adrenal incidentaloma after adequate initial work up. Cost estimates were obtained from published Healthcare Cost and Utilization Project and Medicare reimbursement databases. Utility and outcome probabilities were estimated from published literature. Sensitivity analysis was performed to determine the uncertainty of cost, outcome probability and utility estimates on the model.  A threshold of $ 100,000/ quality adjusted life year ( QALY) was used to determine cost effectiveness. 

Results:
The FLTS strategy produced an incremental cost of $12,521 with incremental effectiveness of 0.26 QALY giving an incremental cost- effectiveness ratio ( ICER) of  $181,773/QALY which exceeds the $100,000/QALY threshold for cost effectiveness. The FLTS strategy was not cost effective and this result was confirmed on multi-way sensitivity analysis using Monte Carlo simulation. 

Conclusion:

Frequent long term surveillance ( as described in the current US adrenal incidentaloma guideline) is not a cost effective strategy for the management of non-functional, radiologically-benign adrenal incidentaloma. 

 

71.01 Malignancy Rate of FDG-PET Avid Thyroid Nodules: Results of a US-based Single Institutional Cohort

Posted on January 31, 2019

A. G. Ramirez1, N. Nuradin1, U. Syed1, V. Grajales2, M. A. Zeiger1, J. B. Hanks1, P. W. Smith1  1University Of Virginia,Surgery,Charlottesville, VA, USA 2University Of Pittsburg,Urology,Pittsburgh, PA, USA

Introduction:
The incidence of 18F-fluorodeoxyglocose positron emission tomography (FDG-PET) avid thyroid incidentalomas is 1-2% with an associated 35% malignancy rate. Thus it is recommended that all FDG-PET avid thyroid lesions be evaluated with ultrasound (US) and fine needle biopsy. North American studies examining prevalence are mixed and difficult to interpret due to poor rates of clinical evaluation of these incidentalomas. Socio-demographic and clinical factors associated with surgical resection of incidentalomas and malignancy are also not well-defined. This study’s objective was to assess our single-institutional malignancy rate, and characterize factors associated with further evaluation of incidentalomas, and surgical resection.

Methods:
All patients undergoing FDG-PET from February 2000-March 2015 with focal thyroid uptake were identified. Those with a history of thyroid cancer or previously evaluated thyroid lesion were excluded. Patient characteristics, US and FDG-PET findings including maximum standardized uptake value (SUVmax) pathology results were reviewed. Descriptive statistics were performed using Student’s t-test and X2-squared test as appropriate. Factors were compared using parametric statistical methods and logistic regression to control for confounders.

Results:
Of 15,399 FDG-PET scans performed, 179 thyroid incidentalomas were identified (1.2%). 59/179(33%) underwent US and 49 (27.4%) had further histological evaluation. 13/49(26.5%) were resected with 10 (16.9%) confirmed malignancies.(Figure) Resection and malignancy were associated with higher SUVmax, p=0.0002 and p =0.0003, Bethesda classification 5-6, p=0.004 and p<0.0001, respectively. After adjustment for confounders, patients with a prior non-thyroidal cancer diagnosis regardless of stage (OR 0.19 p=0.004) were less likely to pursue evaluation of a thyroid incidentaloma. Patients who did not receive adjuvant therapy within 6 months of the FDG-PET (OR 3.9 p=0.012) and lesions with higher SUVmax (OR 1.13 p=0.004) were more likely to obtain US. Nodule size, TSH, and socio-demographic factors including age, race, sex, and insurance status were not significantly different for patients undergoing further evaluation of incidentaloma, receiving surgery, or malignancy.

Conclusion:
Our malignancy rate for imaged thyroid incidentalomas, evaluated and resected, was 16.9%, which is lower than previously reported (35%). These data are suggestive of epidemiological variation and differences in patient selection and preferences. Higher SUVmax and Bethesda classification were associated with malignancy. Despite this lower rate of malignancy, US and biopsy should be pursued when appropriate in the context of the patient’s overall clinical status.
 

70.10 Post-Operative Opioid Prescribing Practices: Do Pills Equal Satisfaction?

Posted on January 31, 2019

M. Flannery1, S. Stokes1, A. Jacobs1, T. Varghese1, R. Glasgow1, B. S. Brooke1, L. C. Huang1  1University Of Utah,General Surgery,Salt Lake City, UT, USA

Introduction:

In the era of patient-reported outcomes, patient satisfaction has become a key quality metric for grading providers and hospitals. A patient’s postoperative pain experience can affect these metrics. Providers may be tempted to prescribe excess pain medications to improve patient satisfaction scores. We hypothesized that satisfaction with pain control was not related to the quantity of opioids prescribed to the patient at discharge.

Methods:
We designed a prospective observational study to evaluate post-operative opioid prescribing and satisfaction among patients undergoing a broad spectrum of general, vascular, colorectal, and plastic surgery procedures at a single tertiary academic medical center. All patients received a survey to determine opioid use and satisfaction with pain control at their first post-operative follow-up visit. We extracted the quantity of opioids prescribed at discharge (normalized to hydrocodone 5 mg tablets) from the electronic medical records. Post-operative pain control satisfaction was compared with the quantity of opioids prescribed at discharge. We constructed hierarchical, mixed effects models using forward step-wise variable selection clustered by procedure to identify risk factors for patient dissatisfaction with pain control while adjusting for potential patient- and procedure-level confounders.

Results:
A total of 346 patients were contacted following surgery, and 289 patients completed the survey (response rate 84%). The distribution by specialty was 33% general, 22% vascular, 20% colorectal, 12% surgical oncology, 7% plastics, and 6% foregut. 83% of patients were satisfied with their pain control, 6% neutral, and 11% were dissatisfied. The median quantity of opioids prescribed normalized to hydrocodone 5mg tablets was 30 (IQR 15-40). After adjustment for time to follow-up, procedure, operative approach (e.g., open versus minimally invasive), and inpatient/outpatient stay, there was no statistical difference in the median number of opioids prescribed to dissatisfied patients compared to satisfied or neutral patients (41.7 ± 5.5 tabs vs. 33.1 ± 2.6 tabs, respectively; p=0.123). On univariate analysis, patient risk factors associated with dissatisfaction with pain control were past opioid use (p = 0.015), smoking history (p < 0.001), current alcohol use (p = 0.049), history of sexual abuse (p = 0.003), and history of attention deficit disorder (p = 0.015). After adjustment using multivariable regression models, we found a significant association between dissatisfaction and past opioid use (OR 3.33, 95% CI 1.12-9.94) and current alcohol use (OR 3.77, 95% CI 1.16-12.22).

Conclusion:
Prescribing more opioids in this study was not associated with greater patient satisfaction after surgery. Patients with past opioid or alcohol use are more likely to be dissatisfied with their pain control. Further research is needed on how to improve pain control and satisfaction in this challenging patient population.

70.09 Sunshine in my Pocket: Industry’s Payments to Surgeons

Posted on January 31, 2019

R. Ahmed1,2, D. Segev2, S. Bloom1, A. Massie2, S. Eubanks1  1Florida Hospital,General Surgery,Orlando, FL, USA 2Johns Hopkins University School Of Medicine,General Surgery,Baltimore, MD, USA

Introduction: Collaboration between industry and surgeons is essential in developing new approaches to treat surgical patients. The Physician Payment Sunshine Act (PPSA) was implemented to publicly disclose financial transactions between industry and physicians, thus informing patients of potential conflicts-of-interest. The objective of this study is to characterize industry payments to surgeons.

Methods: We used the most recent PPSA data (January 2017-December 2017) to assess industry payments made to physicians listed as surgeons or surgical specialists in the CMS Open Payments website.

Results: Surgeons (N=61,2014) received a total of $660,474,480 during the 2017 fiscal year. The median (IQR) was $80 ($22-269). Among surgeons 25% received <$100; 41% received $100-$999; 26% received $1K-$10K; 7% received >$10K-$100K; and 1.6% received >$100K. The top 3 payment categories were royalties or licensing fees ($393,097,177); Consulting fees ($98,089,885); and speaker fees ($53,849,118). A total of 1,004 companies made payment to surgeons ($7 to $69,256,633), of which 10 comprised 66.5% of all payments. The highest to lowest median(IQR) payments by surgical specialty were: orthopedic surgeons $146 (37-782); pediatric orthopedic surgeons $114 (34-477); thoracic surgeons $104 (27-338); transplant surgeons $102 (26-432); neurosurgeons $98 (27-330); hand surgeons $90 (22-209); plastic surgeons $87 (24-203); vascular surgeons $77 (21-239); surgical oncologists $63 (20-198); general surgeons $56 (19-187); colorectal surgeons $55 (19-157); and critical care surgeons $55 (19-147).

Conclusion: The 2017 PPSA data demonstrate that 60% of total amount were made for royalties and licensing fees tp 3.5% of surgeons; 66 % of surgeons received <$1,000. Orthopedic surgeons were the highest paid specialists. Awareness of the PPSA data is critical for surgeons, as it provides a means to prevent potential public misconceptions about industry payments within surgeons that may affect patient trust.

 

70.08 Accurately Predicting 30-day Unplanned Postoperative Readmission Using Eight Predictor Variables

Posted on January 31, 2019

A. B. Singh1, D. R. Gibula3, M. R. Bronsert1, W. G. Henderson1, K. E. Hammermeister1, N. O. Glebova4, R. A. Meguid1  1University Of Colorado Denver,Aurora, CO, USA 3University Of Utah,Salt Lake City, UT, USA 4Mid-Atlantic Permanente Medical Group,Vascular Department,Rockville, MD, USA

Introduction: Unplanned postoperative readmissions may indicate inferior healthcare quality which adversely impact patient recovery and quality of life. Patients desire to know their risk of unplanned readmissions and the surgeons need to know the risk to adequately counsel their patients.  The existing Surgical Risk Preoperative Assessment System (SURPAS) shared decision making tool is a parsimonious model including eight predictor variables: Current Procedural Terminology-related risk, operative complexity, age, functional health status, American Society of Anesthesiologists physical status classification, in- or outpatient status, surgeon specialty, and emergency or elective operation.  Developed from the American College of Surgeons’ National Surgical Quality Improvement Program (ACS NSQIP) dataset, SURPAS applies to >3000 operations in nine surgical specialties and predicts mortality, overall morbidity and eight clusters of common complications, and is incorporated into our health system’s electronic health record (EHR). We aim to develop an accurate preoperative prediction model for identifying the risk of unplanned postoperative readmission related to the primary procedure for integration into the EHR using all ACS NSQIP preoperative non-laboratory predictor variables and compare it to a model limited to the eight SURPAS predictor variables.

Methods: The full model was developed using logistic regression from all twenty-eight non-laboratory variables from the ACS NSQIP 2012-2016 dataset. It was compared to the model of the eight SURPAS predictor variables using the c-index as a measure of discrimination, the Hosmer-Lemeshow observed-to-expected plots testing calibration, and the Brier score, a  combined metric of discrimination and calibration

Results: Of 3,715,921 patients,149,648 (4.03%) experienced an unplanned readmission related to the initial operation.  The SURPAS model’s c-index, 0.727, was of 99.2% of that of the full model, 0.733, and Brier score of 0.0375 equal to the full model. Hosmer-Lemeshow analyses indicated similar calibration between the two models (see Figure).

Conclusions: The eight variable SURPAS model detects patients at risk for postoperative unplanned, related readmission as accurately as the full model developed from all available non-laboratory preoperative variables in the ACSNSQIP dataset.  Therefore, unplanned readmission can be integrated into the existing SURPAS tool providing accurate prediction of postoperative readmission without necessitating the collection of additional predictor variables.

We aim to develop an accurate preoperative prediction model for identifying the risk of unplanned postoperative readmission related to the primary procedure for integration into the EHR using all ACS NSQIP preoperative non-laboratory predictor variables and compare it to a model limited to the eight SURPAS predictor variables.

 

« Previous Page
Next Page »