81.06 Factors Affecting Likelihood of Obtaining Thoracoscopic Versus Open Lobectomy: A NSQIP Analysis

Posted on January 26, 2017

A. Osasona2, H. Mehta2, J. Goodwin3, K. Brown2, I. Okereke1  1University Of Texas Medical Branch,Cardiothoracic,Galveston, TX, USA 2University Of Texas Medical Branch,Surgery,Galveston, TX, USA 3University Of Texas Medical Branch,Internal Medicine,Galveston, TX, USA

Introduction:
Nationwide, video assisted thoracoscopic surgery (VATS) is being utilized with increasing frequency for pulmonary resection.  VATS lobectomy is associated with less morbidity than open lobectomy.  There appears to be a disparate rate of utilization of VATS across different patient populations, however.  Our goal was to determine which factors affect a patient’s likelihood of undergoing VATS versus open lobectomy.

Methods:
The National Surgical Quality Improvement Program (NSQIP) database was queried from 2005 through 2014 to identify all patients undergoing pulmonary lobectomy.  Multivariable logistic regression was conducted to identify patient demographics, clinical characteristics and surgeon specialty associated with an increased likelihood of receiving a VATS lobectomy versus an open approach.

Results:
From 2005 through 2014, 11,977 patients in the NSQIP database underwent lobectomy (open, 6,391; VATS, 5,586).  The mean age was 64.4 years and females comprised 53.4% of cases. The use of VATS lobectomy increased significantly over time, increasing from 17.5% in 2005—2008 to 53.1% in 2012—2014 (p<0.001).  On multivariable analysis, male gender (odds ratio [OR], 1.20; 95% confidence interval [CI], 1.12-1.30), African-American race (OR, 1.22; 95% CI, 1.04-1.43), age under 75 years (OR, 1.21; 95% CI, 1.11-1.32), worse frailty index (OR, 1.56; 95% CI, 1.06-2.28) and performance of the lobectomy by someone other than a non-cardiac thoracic surgeon (OR, 1.37; 95% CI, 1.25-1.51) were associated with an increased likelihood of receiving an open lobectomy.

Conclusion:
The frequency of VATS lobectomy has increased substantially over the last 10 years, and now accounts for over half of the lobectomies performed at NSQIP hospitals. Demographics, clinical characteristics, and specialty of surgeon influences a patient’s likelihood of receiving VATS versus open lobectomy. Identifying such disparities in surgical approach may help to direct the focus of interventions at the individual, hospital or organizational level to increase the use of VATS lobectomy.
 

81.05 Evaluation Of Navigational Bronchoscopic(ENB) biopsy of Lung lesions performed by a surgical service

Posted on January 26, 2017

T. D. Witek1, A. Pennathur1, J. D. Luketich1, M. Scaife1, D. Azar1, M. J. Schuchert1, W. E. Gooding2, O. Awais1  1University Of Pittsburgh Medical Center,Department Of Cardiothoracic Surgery,Pittsburgh, PA, USA 2University Of Pittsburgh Cancer Institute Biostatistics Facility,Pittsburgh, PA, USA

Introduction:

With increasing utilization of CT scans for lung cancer screening, and for surveillance of other cancers, thoracic surgeons are being referred patients with lung lesions for biopsies. While CT-guided biopsies have been used for biopsies of peripheral lesions, electromagnetic navigational bronchoscopy (ENB) guided lung biopsy is a relatively new technique for bronchoscopic biopsies of peripheral lesions.  Our objective was to evaluate the diagnostic yields and safety of electromagnetic navigational bronchoscopy (ENB) guided lung biopsy.

Methods:

We conducted a retrospective review of patients who underwent an ENB for diagnostic purposes, performed by a thoracic surgical service. We collected data on patient characteristics, lesion characteristics, procedure outcomes, and pathology information. General anesthesia and rapid on-site examination (ROSE) of cytopathology was used during all ENBs. All patients that were diagnosed with a malignancy from ENB findings were considered true positives. Lesions that did not reveal malignancy from ENB sampling were considered true negatives if sequential surgical or CT guided biopsy also revealed benign tissue or if serial imaging revealed stability or improvement; otherwise they were classified as false negatives.

Results:

A total of 121 lesions in 111 patients (men 46, women 65) underwent ENB guided bronchoscopic sampling of pulmonary lesions. The median size of the lesion was 27mm (Range 9-115 mm, IQR = 17 – 37 mm). Ninety-four (78%) of the lesions were malignant. Eighty-five (70%) of the 121 lesions had an accurate diagnosis. Accuracy increased with increased lesion size (odds ratio = 2.9, 95% CI = 1.3 – 6.3). The presence of “bronchus sign” was associated with 3 fold increase in odds of  accurate diagnosis (odds ratio 3.1, 95% CI= 1.1 – 8.2). Among patients with a “bronchus sign”, the predicted accuracy of the biopsies for 1 cm lesions was 66% (95% CI 40% – 85%; the predicted accuracy for 2 cm lesions was 77% (95% CI = 58% – 89%) for 3 cm lesions 85% (95% CI = 71% – 93%).  There was no procedure-related mortality. There were four (3%) instances of pneumothoraxes, requiring pigtail pleural catheters.

Conclusion:

Thoracic surgeons can perform ENB safely, with minimal morbidity and with good diagnostic yields. Accuracy increases with the presence of "bronchus sign" and increasing lesion size. Lesions larger than 2 cm have a higher likelihood of an accurate diagnosis.
 

81.04 Lung Resection for Non-Small Cell Lung Cancer: Does Perioperative Fluids Affect Outcomes?

Posted on January 26, 2017

M. Asai1, A. X. Samayoa1, C. Hodge1, Y. Shan1, H. Pak1, T. Vu1  1Abington Memorial Hospital,General Surgery,Abington, PA, USA

Introduction:
Intraoperative intravenous (IV) fluid infusion has been controversial in thoracic surgery. There has been some studies showing correlation of intravenous fluid infusion and acute lung injury in the perioperative phase. Although this is the case, there is limited evidence suggesting volume of perioperative IV fluids and any correlation with outcomes in patients undergoing thoracic surgery. In our study, we investigated outcome measures associated with patients undergoing lung resection for non-small cell lung cancer (NSCLC) and the correlation with perioperative IV fluids.

Methods:
A retrospective review of consecutive patients undergoing open lobectomy for primary NSCLC from January 2010 to June 2016. Exclusion criteria were patients who had previous lung resection(s), blood loss more than 500cc during surgery and patients receiving intraoperative blood transfusions. Patient were divided into those receiving intraoperative IV fluid greater or less than 7 cc/kg/hr, and another group receiving greater or less than 10 cc/kg/hr. Perioperative (intraoperative + 24 hour postop) IV fluid were also recorded. We compared this against patient’s demographics, intraoperative parameters and complication rate within 30 days of surgery. Complications were categorized as pulmonary, cardiac, renal complications as well as hemorrhage, death and reoperation.

Results:
142 patients (69 male and 73 female) with the mean age of 66.5 years were identified following inclusion criteria were met. There was no significant difference in the hospital or ICU stay between any of the intraoperative or perioperative fluid groups.
Patients receiving greater than 10cc/kg/hr intraoperative IV fluid has higher rate of total complication, especially postoperative hemorrhage and reoperation rate. Patients receiving less than 7 cc/kg/hr intraoperative IV fluid has higher risk of developing acute renal injury. Patient who had greater than 1.5 cc/kg/hr perioperative fluid has significantly higher duration of chest tube in-situ, pulmonary complications and total complication rates.

Conclusion:
In our study, giving high amount of intraoperative and perioperative IV fluid has adverse effects on postoperative complications. Conversely, giving less (<7cc/kg/hr) intraoperative IV fluid has higher risk of renal dysfunction. We propose the optimal amount of intraoperative and perioperative IV fluid given should be in between these two extremes. To highlight these points further, a larger prospective randomized study should be performed.

81.03 Efficacy of Preoperative MRSA Screening and Decolonization in a Thoracic Surgical Oncology Clinic

Posted on January 26, 2017

G. M. Fitzpatrick6, H. Frum1, R. Quilitz2,5, R. L. Sandin7, D. Ruge7, J. L. Greene2,5, J. R. Garrett1, C. C. Moodie1, E. M. Toloza1,3,4  1Moffitt Cancer Center,Thoracic Oncology,Tampa, FL, USA 2Moffitt Cancer Center,Infectious Diseases,Tampa, FL, USA 3University Of South Florida Morsani College Of Medicine,Surgery,Tampa, FL, USA 4University Of South Florida Morsani College Of Medicine,Oncologic Sciences,Tampa, FL, USA 5University Of South Florida Morsani College Of Medicine,Medicine,Tampa, FL, USA 6University Of South Florida,Morsani College Of Medicine,Tampa, FL, USA 7Moffitt Cancer Center,Pathology,Tampa, FL, USA

Objective:   To study the efficacy of preoperative (preop) screening for methicillin-resistant Staphylococcus aureus (MRSA) and of decolonization of MRSA-positive (MRSA-POS) patients (pts) by a trimodality antibiotic regimen aimed at reducing postoperative (postop) MRSA infections.

Methods:   At preop clinic evaluation, pts scheduled for thoracic surgery were screened for MRSA via polymerase chain reaction (PCR) of nasal swab specimens.  Preop MRSA-POS pts were given oral doxycycline 100 mg twice daily, mupirocin 2% ointment to both nares, and chlorhexidine showers for 5 days for MRSA-decolonization preop, with doxycycline continued 7 more days postop.  Pts who were admitted to intensive care had a second MRSA screen immediately postop.  We retrospective analyzed 1106 consecutive thoracic surgery pts to determine prevalence of preop MRSA colonization, rates of POS-to-negative(NEG) and NEG-to-POS conversion between preop and immediate postop MRSA screens, and rates of postop MRSA and non-MRSA infections in MRSA-POS and in MRSA-NEG pts.  MRSA screens within 60 days prior to surgery and infections within 30 days after surgery were included.  Fisher Exact Probability Test was used, with significance at p<0.05.

Results:  Of 960 pts who had preop MRSA screen, 4.0% (38/960) were MRSA-POS.  Of these MRSA-POS pts, 15.8% (6/38) developed postop infections, compared to 18.8% (173/922) of MRSA-NEG pts (p=0.68).  The most common postop infection in each group was an uncomplicated urinary tract infection (UTI).  Excluding UTIs, 7.9% (3/38) of MRSA-POS pts and 11.8% (109/922) of MRSA-NEG pts had postop infections, respectively (p=0.61).  The only postop MRSA infection in this study was a case of MRSA pneumonia diagnosed on postop day 16 in a pt who was MRSA-NEG preop.  Of 530 pts who had both preop and postop MRSA screens, 95.1% (511/530) were MRSA-NEG preop, of which 1.4% (7/511) converted to being MRSA-POS on immediate postop screen.  Of the remaining 19/530 pts (3.6%) who were MRSA-POS preop, 68.4% (13/19) converted to being MRSA-NEG on immediate postop screen.  Of 15 MRSA-POS pts receiving the complete decolonization antibiotic regimen, 73.3% (11/15) converted to being MRSA-NEG on immediate postop screen, while of the 4 MRSA-POS pts who did not complete the decolonization antibiotic regimen, 50% (2/4) converted to being MRSA-NEG on immediate postop screen (p=0.56).

Conclusions:  Prevalence of MRSA colonization in thoracic surgery pts who were screened by PCR of preop nasal swab specimens is similar to those previously published.  Rates of conversion from preop MRSA-POS to immediate postop MRSA-NEG indicate overall efficacy of our trimodality decolonization regimen.  With adequate decolonization via trimodality treatment, risk of postop MRSA infections or even for non-MRSA infections is not increased in pts screened preop as MRSA-POS versus those screened as MRSA-NEG.

81.02 Aortic Z-Scores As Indicators for Aortic Repair

Posted on January 26, 2017

M. R. Helder1, H. V. Schaff1, A. Pochettino1, H. Connolly1  1Mayo Clinic,Cardiovascular Surgery And Cardiovascular Diseases,Rochester, MN, USA

Introduction: Practice guidelines recommend repair for aortic root aneurysms in adult patients with Marfan syndrome based on absolute diameter despite the recognized variation based on age, sex, and body surface area. Echocardiographic Z-scores have been used in multiple pediatric patient studies outlining a “normal z-score” range.  However, the aortic Z-score at which adult patients should be offered an operation has not been defined.

Methods: After IRB approval, we retrospectively examined the preoperative echocardiography data of all patients with Marfan syndrome who underwent elective aortic root repair from February 11, 2005 to September 3, 2010 based on either aortic root diameter threshold or increase in aneurysm diameter over time.  Aortic root diameters were measured at sinus level in all but two patients where the mid-ascending aorta was the largest diameter.  Age, sex, and body surface area were gathered from the electronic medical record and used to calculate a Z-score based using the following equation: Z = (measure diameter – predicted diameter)/0.261 cm.  Predicted diameter (cm) = 2.423 + (age X 0.009) + (BSA X 0.461) − (sex X 0.267), where male sex = 1, female sex = 2. 

Results: Fifty-eight adult patients, average age of 36±13 years (71% male), underwent aortic root repair.  Median aortic root diameter was 52 mm, range of 32 – 71 mm.  Average aortic root diameter was 54±5 mm.  This corresponded to a median aortic Z-score of 6 (range, -1.6 to 14).  Average aortic Z-scores were 6±3.  Mean ejection fraction, preoperatively, was 58±7%.  Eleven patients had an aortic root diameter < 45 mm; 4 of these patients (36%) had corresponding Z-scores greater than 4.

Conclusion: This study describes aortic of Z-scores of patients that underwent aortic root repair based on indications dictated by current practice guidelines.  Z-scores of some patients indicated a larger relative aneurysm than would have been clear by aneurysm diameter alone.  Aortic Z-scores should be studied further and incorporated into the clinical decision making of whether or not to offer a patient aortic repair.  Based on this descriptive data, a z-score > 3 could be used as an operative indication for aortic repair in patients with Marfan syndrome.

 

81.01 Factors That Alter the Relationship Between Peak Postoperative CKMB and Troponin T After CABG

Posted on January 26, 2017

K. M. Mehta1, J. Pruszynski1, M. Peltz1, L. C. Huffman1, P. Bajona1, M. A. Wait1, R. Correa1, W. Ring1, M. Jessen1  1UT Southwestern Medical Center,Department Of Cardiovascular And Thoracic Surgery,Dallas, TX, USA

Introduction: Peak postoperative creatine kinase MB fraction (CKMB) and Troponin T (TnT) levels have been measured after cardiac surgery to assess perioperative myocardial damage, evaluate myocardial protective strategies and predict adverse events.  However, the relationship between peak levels of both enzymes has not been fully established in this setting.  We compared peak levels of CKMB and TnT in patients after CABG to test the hypothesis that patient and operative characteristics influence the correlation between the values of these biomarkers.

Methods: Data were prospectively collected from 171 consecutive patients undergoing on-pump CABG at a single institution between July 1, 2014 and Dec 31, 2015. Peak values were selected from all serum levels of CKMB and TnT collected during the hospital stay following surgery.  Clinical variables were collected based on definitions in the STS Adult Cardiac Surgery Database version 2.181.  Linear regression models were used to statistically compare the slope of the linear relationship between peak postoperative CKMB and TnT for the patient cohort. Models were created to compare the slopes by pre-defined clinical variables including (1) gender, (2) age (< or >70), (3) race, (4) tobacco use, presence or absence of (3) hypertension, (4) dyslipidemia, (5) diabetes, (6) renal dysfunction (GFR<60), (7) MI within 21 days, (8) EF (< or > 40%), preoperative use of (9) ACE-inhibitors, (10) beta-blockers, and (11) anticoagulants; and operative variables including (1) cross clamp time (< or > 70 min), (2) CPB time (< or > 100 min), and (3) whether or not intra-operative blood products were received. A lower slope implies less change in CKMB compared to the change in TnT.

Results: Overall, the correlation between peak postoperative CKMB and TnT was robust in patients undergoing CABG (m = 19.6, r= 0.783).  However, the slope of the relationship was significantly lower in males, patients > 70 years, diabetics, non-smokers, patients with renal dysfunction, patients with lower EF, patients receiving anticoagulants, and patients undergoing CABG following a recent MI.  The slope was significantly greater in patients with longer clamp times and who were receiving beta blockers and ACE-inhibitors (Table I). In all other models, the slope of the relationship was similar. 

Conclusion: The relationship between CKMB and TnT following CABG appears to be influenced by patient and operative characteristics. These data do not assess which enzyme more accurately reflects myocardial injury, but does suggest conclusions about myocardial damage may be affected by the biomarker selected in the presence of certain variables. Further study to assess the association between these biomarkers and patient outcomes is warranted.

78.16 Incidence and Survival of Anal Cancer in the Young Male Population. Can We Overcome the Odds?

Posted on January 26, 2017

M. M. Bojko1, R. Kucejko1, J. L. Poggio1  1Drexel University College Of Medicine,Colorectal Surgery/Surgery,Philadelphia, Pa, USA 2Drexel University College Of Medicine,Philadelphia, Pa, USA

Introduction:

Recent studies have identified an increased incidence of squamous cell carcinoma of the anus (SCCA) in the young male population. This may be attributed to changing sexual practices leading to increased susceptibility to infection with Human Papillomavirus. Prior studies have identified black race as contributing to higher incidence, but they were limited by region and did not specifically address age or socioeconomic status. We aim to identify populations at risk so screening programs and community outreach may be appropriately directed to best affect the rising public health burden.

Methods:

The National Cancer Institute’s Surveillance of Epidemiology and End Results database was queried for data on patients with SCCA for the years of 2000-2013. Cases were selected by ICD-0-3 site codes corresponding to the anus, and only cases identified histologically as SCCA were included. SEER*Stat version 8.3.2 and SPSS version 24.0.0.0 were used to calculate incidence and survival using cox-regression analysis. Age, sex, race, and socioeconomic status were evaluated for their effect. Significance was calculated at the 95% confidence interval, and all P-values were 0.05 or below. 

Results:

16,908 cases were identified with SCCA. Overall, the incidence rate of SCCA was twice as high in black men compared to white men aged <35 years between 2000-2013 (p < 0.001). Cox regression analysis of male patients <49 identified black race to have a hazard ratio of 1.55 (CI 1.33 to 1.81) when compared to white race. It also identified a hazard ratio of 0.90 (CI 0.86 to 0.94) associated with each $10,000 increase in median family income. Among this sample, black men had an incidence rate of 2.7 and 2.8 times that for white men for regional and distant disease, respectively. Figure 1 illustrates the lower survival among black men compared to other races. 

Conclusion:

In the male population less than 49 years old, the incidence of SCCA is significantly higher in black men compared to white men, and their disease is diagnosed at later stages. Higher median family income is an independent predictor of survival, and when adjusted for stage and median family income, black men have a significantly lower survival compared to white men. Treatment parameters are established for SCCA but there is no consensus on screening parameters. New screening programs have been developed in the recent years, and data suggest they have a positive impact on incidence and survival.  Given these results and our own data, we encourage all physicians to identify high-risk individuals and proceed with screening for anal cancer. Furthermore, work is needed to develop health outreach programs to target populations at higher risk, such as the low-income, black male population.

 

77.06 Fast Track Surgery For Abdominal Surgery In Rwanda: A Randomized, Controlled Trial

Posted on January 26, 2017

L. NDAYIZEYE2, J. Rickard1,2, A. K. KiSwezi2, L. NDAYIZEYE2  1University Of Minnesota,Minneapolis, MN, USA 2University Of Rwanda,Kigali, KIGALI, Rwanda

Introduction:
Fast Track Surgery (FTS) was started in the 1990’s initially for colorectal surgery, but later became applicable to other aspects of surgery. Core elements of FTS include optimal pain control with early initiation of mobilization and feeding. FTS in high-resource settings has been shown to be beneficial with decreased hospital stay, reduced complications and lower costs. There is limited data on FTS in low-resource settings. The aim of this study was to evaluate the impact of FTS among patient undergoing abdominal surgery at a Rwandan referral hospital

Methods:
Adult patients undergoing elective, abdominal surgery were randomized to FTS versus conventional management. Interventions for FTS included early feeding, early mobilization, pain control and discharge planning. Primary outcome was length of hospital stay and secondary outcome was complication rate

Results:
Sixty two patients were enrolled in the study with 31 randomized to FTS and 31 patients undergoing conventional management. 30 (97%) FTS patients received enteral nutrition within 24 hours postoperatively compared with 15 (48%) patients with conventional management (p<0.0001). 30 (97%) FTS mobilized within 24 hours compared with 7 (23%) patients with conventional management (p<0.0001). Pain was adequately controlled with non-opiate medications in 24 (77%) FTS patients. In the conventional management group, 26 (84%) patients received opiates. The mean hospital stay for FTS patients was 2.1 days and the mean hospital stay for patients with conventional management was 5.3 days (p<0.001). There was no difference in complication rates between the groups.

Conclusion:
In a low-resource setting, implementing FTS management with early nutrition, mobilization and decreased opiate use reduces length of hospital stay without increasing complications.
 

77.05 Not all Readmissions are Created Equal – Index vs. Non-Index Readmissions After Major Cancer Surgery

Posted on January 26, 2017

S. Zafar1, A. A. Shah1, H. Channa2, L. L. Wilson4, N. Wasif3  1Howard University College Of Medicine,General Surgery,Washington, DC, USA 2Purdue University,Agricultural Economics,West Lafayette, IN, USA 3Mayo Clinic In Arizona,Surgical Oncology,Phoenix, AZ, USA 4Howard University College Of Medicine,Surgical Oncology,Washington, DC, USA

Introduction:
Hospital readmissions after major cancer surgery pose a major healthcare burden and are associated with increased costs and worse outcomes. Increasing regionalization of cancer surgery has the inadvertent potential to lead to fragmentation of care if readmissions occur at a different hospital from the index facility.  Using a national dataset we aim to quantify rates of readmission to non-index hospitals after major cancer surgery and to compare outcomes between index and non-index hospital readmissions.

Methods:
We used the National Readmissions Dataset (2013) as our data source. All adult patients undergoing a major cancer operation (defined as esophagectomies/gastrectomies, hepatico-biliary resections, pancreatectomies, colorectal resections, and cystectomies) within the first 9 months of the year were selected. Readmission characteristics including timing, cost, morbidity and mortality were analyzed. Discharge weights were used to calculate national estimates. Adjusting for clustering by facility, we used multivariate logistic regression to identify factors associated with non-index vs. index readmissions and also to identify differences in mortality, major complications and subsequent readmissions.  Generalized linear modeling was used to test for differences in length of stay (LOS) and hospital costs between the two groups.

Results:
A total of 57,362 patients with 86,362 hospital admissions were analyzed. Overall, the 90 day readmission rate was 23.1% and median time to readmission was 42 days (IQR 20-70 days). Weighted analysis revealed the total national cost for 90 day readmissions to be $682 million. Of the 17,020 readmissions, 22.0% were to a non-index hospital. Independent factors associated with 90 day readmission to a non-index hospital included younger age,  male gender, type of procedure, more comorbidities, Medicaid insurance, longer LOS, in-hospital complications, discharge to a nursing facility, and index surgery at a teaching hospital (p<0.05).  Following risk adjustment, patients readmitted to non-index hospitals had 32% higher odds of mortality (OR 1.32, 95% CI: 1.03-1.70) and 26% higher odds of having a major complication (OR 1.26, 95% CI: 1.10-1.43). Subsequent readmissions and hospital costs were not significantly different between the two groups.

Conclusion:
The 90 day readmission rate following major cancer surgery in the United States is 23.1%, of which a further 22% are to a non-index hospital. When compared to patients readmitted to the index hospital, readmission to a non-index hospital is associated with higher mortality and morbidity. Targeted interventions to reduce non-index readmissions may mitigate fragmentation of postoperative care and result in improved outcomes.
 

77.04 Trends in Post-Surgical Opioid Prescription by Specialty and the Impact on Rates of Overdose and Abuse

Posted on January 26, 2017

G. Brat1, D. Agniel2, A. Beam2, N. Palmer2, B. Yorkgitis1, M. Homer2, A. Salim1, I. Kohane2  1Brigham And Women’s Hospital,Trauma, Burns, And Critical Care,Boston, MA, USA 2Harvard School Of Medicine,Department Of Bioinformatics,Brookline, MA, USA

Introduction:
Age-adjusted opioid overdose rates now rank as the leading cause of unintentional injury-related death. Significant national, state, and hospital guidelines have been published to curb this epidemic. Despite these attempts, few documents are specifically focused on post-surgical patients. These patients receive post-discharge opioids between 50-80% their visits, nearly 4 times more often than their non-surgical counterparts. In this study, we hoped to explore the epidemiological change in prescribing patterns of surgical specialties and if such changes have translated into reduced rates of abuse.

Methods:
We used a de-identified administrative claims database of >70 million patients covered at a commercial managed healthcare company to identify individuals with health insurance coverage between 2008 and 2016 who underwent surgery. Our inclusion criteria were 1) 6 months of pre- and 1 year of post-surgical coverage, 2) 60 days of pre- and 90 days of post-surgical prescription drug coverage, and 3) no pre-surgical opioid or abuse exposure. We used previously described ICD-9 codes to identify pre- and post-surgical opioid abuse. A subset of 23 surgeries and their associated CPT codes, as defined by NSQIP, were used to categorize specialties. Post-discharge opioid use was collated from pharmacy data using total morphine equivalents (MME) prescribed divided by the total days prescribed, adjusted for overlap. Multivariable logistic regression was used to quantify the effect of year in changing surgical abuse rates independent of surgery.

Results:

Of the 1,000,309 narcotic naïve patients who met criteria, 1,913 (0.19%) developed an abuse or overdose code within 1 year. Figure A shows the trends in dose and duration by specialty and over time. From 2008 to 2014, median MME/day fell from 56.2 to 50. Specialties generally followed this same pattern–with vascular and urological surgical specialties making the greatest changes in dosing.  However, these efforts did not translate into changes in outcome. Despite significant reductions in rates of opioid prescription over time, abuse rates increased (Figure B) from 202.1 to 297.4 per 100K patient-years. An adjusted model showed that progressive years were an independent risk factor for overdose and abuse.

Conclusion:
Physicians struggle to appropriately dose post-discharge opioids while addressing the very real needs of post-operative acute pain. Our data suggests that surgical specialties have generally behaved in similar ways to reduce their prescribed doses. Despite these small changes, abuse rates continue to rise. Successful interventions in surgery must move beyond dose changes to succeed.

77.03 Impact of Pain after Trauma on Long Term Patient Reported Outcomes

Posted on January 26, 2017

J. P. Herrera-Escobar1, M. Apoj2, G. Kasotakis2,3, A. J. Rios-Diaz1, E. Lilley1, J. Appelson1, B. Gabbe4, K. Brasel5, E. Schneider1, H. Kaafarani6, G. Velmahos6, A. Salim1, A. H. Haider1  1Brigham And Women’s Hospital,Boston, MA, USA 2Boston University,Boston, MA, USA 3Boston Medical Center,Boston, MA, USA 4Monash University,Melbourne, VIC, Australia 5Oregon Health And Science University,Portland, OR, USA 6Massachusetts General Hospital,Boston, MA, USA

Introduction:  The Institute of Medicine has recently called for incorporating long-term patient reported outcomes (PROs) to improve trauma care quality and increase patient-centeredness. In order to understand the value of collecting post-discharge PROs, we sought to describe the burden of self-reported pain at 6 and 12 months after injury, and determine its association with important PROs such as: Post-Traumatic Stress Disorder (PTSD), return to work, and new need of assistance at home.

Methods: Trauma patients with an ISS ≥9 were identified retrospectively using the institutional trauma registry of two Level I trauma centers and contacted 6 or 12 months post-injury to participate in a telephone interview evaluating PROs measures: Trauma and Health Related Quality of Life (T-QoL and Short Form-12 [SF-12]), PTSD screening (Breslau), return to work, and residential status. Multivariable logistic regression models clustered by facility and adjusting for confounders (age, sex, ISS, and length of stay) were used to obtain the odds of positive PTSD screening, not returning to work, and new need of assistance at home, in trauma patients who reported to have pain on a daily basis compared to those who did not.

Results: We conducted 305 interviews: 141 at 6 months and 164 at 12 months after injury. 48/52% (6/12 months) reported pain on a daily basis, 24/19% took pain medications daily, 47/48% were limited by pain in the things they are able to do, and 36/37% reported that pain moderately, quite a bit or extremely interfered with their normal work. There were no differences in age, gender, ISS, mechanism of injury, and length of stay between patients with and without pain on a daily basis at 6 and 12 months (all p>0.05). Compared to patients without pain, patients with pain at 6 months were more likely to screen positive for PTSD (OR: 1.92[1.28-2.88]), required assistance at home without prior need (OR: 2.86[1.82-4.48]), and did not return to work (OR: 2.85[2.46-3.30]). Similarly, at 12 months, patients with pain had higher odds of positive PTSD screening (OR: 5.95[3.87-9.15]) and requiring assistance at home without prior need (OR: 5.53[1.48-20.71]). However, we did not find a statistically significant difference in return to work compared to patients without pain (OR: 2.35[0.79-7.03]) 

Conclusion: There is an enormous amount of self-reported pain after trauma that is not being captured by current trauma registries that are limited to outcomes at discharge. Pain after trauma is associated with poor PROs such as positive PTSD screening, delayed return to work, and new need of assistance at home. Inclusion of long-term PROs in trauma registries will enable quality improvement that is more inclusive of all aspects of recovery after an injury.

 

77.02 Explicating Hospital-Based Violence Intervention Program Risk Assessment Through Qualitative Analysis

Posted on January 26, 2017

E. J. Kramer1,2, J. Dodington1, T. Henderson2, R. Dicker2, C. Juillard2  1Yale University School Of Medicine,New Haven, CT, USA 2University Of California San Francisco,Surgery,San Francisco, CA, USA

Introduction:  Violent injury is the second most common cause of death among 15-24 year-olds in the US. Up to 45% of violently injured youth return to the emergency department with a second violent injury. Hospital-based violence intervention programs (HVIPs) have been shown to reduce injury recidivism through intensive case management to victims of violence at high risk for recidivism. To date, case manager gestalt has guided identification of victims at high risk for re-injury; no validated guidelines for risk-assessment strategies in the HVIP setting have been published. We aimed to use qualitative methods to investigate the key components of risk assessments employed by HVIP case managers in a mature HVIP with demonstrated effectiveness. We propose a risk assessment model based on this qualitative analysis, combined with literature review and review of current best practices of established HVIPs. 

Methods:  A qualitative approach was used due to the complexity and interconnectivity of inherently non-binary and socially-influenced risk factors. An established academic hospital-affiliated HVIP served as the nexus for this research. Thematic saturation was reached with 11 semi-structured interviews and 2 focus groups conducted with HVIP case managers and key informants identified through snowball-sampling. Interactions were audiotaped, transcribed, and analyzed by a four-member team using Nvivo and employing the constant comparison method. Risk factors identified were used to create a set of models presented in 2 follow-up HVIP case managers and leadership focus groups.

Results: Key themes emerged revolving around the imminent threat of violence, history of incarceration, dissociative behavior, weapons use, and pursuing “street”/gang lifestyle. In total, 141 potential risk factors for use in the risk assessment framework were identified. The most salient factors were incorporated into eight models that were presented to the HVIP case managers. A 29-item algorithmic structured professional judgment model was selected by focus group consensus. The model contains four categories of risk factors: high-risk (A), behavioral (B), severe conditional (C), and moderate conditional (D) factors. The presence of 1+ A factor indicates high-risk, while combinations of B/C/D factors can lead to an assessment of high-risk [Fig 1].

Conclusion: Qualitative methods identified four tiers of risk factors for violent re-injury that were incorporated into a proposed risk assessment instrument. Further research should assess the validity and generalizability of this instrument. A valid risk assessment instrument has the potential to identify high-risk individuals in diverse clinical settings, who may benefit from violence intervention strategies.

77.01 Low-Mortality Hospitals Provide Better End-Of-Life Care for Non-Survivors After Severe Brain Trauma

Posted on January 26, 2017

E. J. Lilley1,2, J. W. Scott1, A. Krasnova1, E. Schneider1, A. H. Haider1, Z. Cooper1  1Brigham And Women’s Hospital,The Center For Surgery And Public Health,Boston, MA, USA 2Robert Wood Johnson – Rutgers,Department Of Surgery,New Brunswick, NJ, USA

Introduction: Severe traumatic brain injury (TBI) in older patients is associated with poor prognosis, with rates of mortality or functional dependence exceeding 80%. For non-survivors, aggressive, rescue-directed care may include burdensome treatments at the end of life and impede appropriate provision of hospice, an established quality marker for end-of-life (EOL) care. It is not known whether hospitals with low in-hospital mortality also deliver high-quality EOL care to non-survivors. Therefore, we sought to compare EOL care of older non-survivors with severe TBI treated at low-mortality hospitals against all others.

Methods: Using 2001-2011 Medicare data, we identified patients > 65 years with severe TBI (head AIS > 3) who survived > 1 day in-hospital. Patients at low case-volume hospitals (< 5 eligible patients per year) were excluded. Two categories of non-survivors were analyzed: those who died in-hospital and those who died < 30-days post-discharge. The primary independent variable was in-hospital mortality ranking of the treating hospital: Hospitals were divided into quartiles based on in-hospital mortality and those in the lowest quartile were considered high-performing. EOL care outcomes included intensity of treatment (gastrostomy and tracheostomy) and hospice enrollment. Outcomes of patients treated at high-performing hospitals were compared with others using multivariate Cox regression (30-day post-discharge mortality) and multivariate logistic regression models (EOL care outcomes) adjusting for age, sex, race, and comorbidity.

Results:Among 35,057 older patients hospitalized with severe TBI, 57% died in-hospital or 30-days post-discharge. By definition, in-hospital mortality was lower at high-performing hospitals than at others (36% vs. 49%), yet 30-day post-discharge mortality was the reverse: 26% in high-performing hospitals vs. 19% at others (HR 1.41 [CI 1.32-1.50], p < 0.01). Compared with patients treated at other hospitals, in-hospital non-survivors at high-performing hospitals had similar rates of gastrostomy (3.9% vs. 4.1%, OR 0.94 [CI 0.77-1.15], p = 0.57), tracheostomy (6.9% vs. 6.4%, OR 1.08 [CI 0.92-1.26], p = 0.34), and no clinically significant difference in hospice enrollment (2.2% vs. 1.6%, OR 1.38 [CI 1.05-1.83], p = 0.02). In contrast, 30-day post-discharge non-survivors at high-performing hospitals had fewer high-intensity treatments and a greater proportion were enrolled in hospice (Figure). 

Conclusion:These findings suggest that high-performing hospitals are better able to identify patients with poor prognosis and provide appropriate EOL care.

 

74.10 The Surgeon Volume-Outcome Relationship – Not Yet Ready for Policy

Posted on January 26, 2017

J. Modrall1, R. M. Minter1, A. Minhajuddin4, J. Eslava-Schmalbach3, G. P. Joshi2, S. Patel1, E. B. Rosero2  1University Of Texas Southwestern Medical Center,Department Of Surgery,Dallas, TX, USA 2University Of Texas Southwestern Medical Center,Department Of Anesthesiology And Pain Management,Dallas, TX, USA 3Hospital Universitario Nacional De Colombia,Bogota, COLOMBIA, Colombia 4University Of Texas Southwestern Medical Center,Department Of Clinical Sciences,Dallas, TX, USA

Introduction: Studies have demonstrated improved clinical outcomes for high-volume surgeons performing index operations. As skills may transfer between operations, we hypothesized that there may be surrogate experience that yields similar outcomes for surgeons with a low-volume experience with a specific index operation, but who have high-volume experience with related procedures. Using esophagectomy as a sample index operation, we compared the outcomes of surgeons with a high volume of index operations to low-volume surgeons with the index operation who had significant experience with related surrogate procedures.

Methods: The Nationwide Inpatient Sample (2003-2009) was analyzed to identify patients undergoing open esophagectomy. Surgeons were stratified into low-, medium-, and high-volume groups based on annual volume of esophagectomy (operation-specific volume), corresponding to the <10th, 10-90th, and >90th percentiles of volume, respectively. Surrogate volume was defined as the aggregate annual volume per surgeon of several upper GI operations: excision of esophageal diverticulum, gastrectomy, gastroduodenectomy, and repair of diaphragmatic hernia. Three-level hierarchal generalized mixed models were used to test the association between mortality and operation-specific and surrogate case volume, adjusting for both patient and hospital characteristics.

Results: 26,795 esophagectomies were performed nationwide (2003-2009) with a crude in-hospital mortality rate of 5.2%. As expected, in-hospital mortality decreased with increasing volume performed annually: 10.1%, 5.5%, and 3.8% for low-, medium-, and high-volume surgeons, respectively (P<0.0001). A similar relationship was seen between surrogate surgeon volume and in-hospital mortality for esophagectomy, as low-, medium-, and high-volumes of surrogate operations were associated with mortality rates for esophagectomy of 9.8%, 5.7%, and 2.6% (P<0.0001). Among surgeons with a low-volume esophagectomy experience, increasing volume of surrogate operations improved the outcomes observed for esophagectomy: 12.2%, 9.5%, and 1.9% for low-, medium-, and high-surrogate-volume surgeons, respectively (P=0.026). Hierarchical multivariate analysis showed that both esophagectomy and surrogate volume were significant predictors of lower in-hospital mortality for esophagectomy (Table).

Conclusion: Both operation-specific volume and surrogate volume are significant predictors of in-hospital mortality for esophagectomy. These data suggest that the relationship between surgeon volume and outcomes may be more complex than previously recognized. Based on these observations, it would be premature to limit hospital privileges based on operation-specific surgeon volume criteria.

 

74.09 Why Do Patients Refuse VTE Prophylaxis? A Nursing-Focused Qualitative Evaluation

Posted on January 26, 2017

L. J. Kreutzer1, C. J. Minami1,2, L. Saadat1, K. Y. Bilimoria1,2, A. D. Yang1,2, J. K. Johnson1,2  1Northwestern University,Surgical Outcomes And Quality Improvement Center,Chicago, IL, USA 2Northwestern University,Center For Healthcare Studies In The Institute For Public Health And Medicine,Chicago, IL, USA

Introduction: Venous Thromboembolism (VTE) is a serious medical condition that results in preventable morbidity and mortality. Optimal VTE prophylaxis in hospitalized patients includes three components: ambulation, sequential compression devices, and chemoprophylaxis; however, patients can refuse one or more of these components, which can increase their VTE risk.  Previous studies have shown that nurses can influence patient compliance with VTE prophylaxis. Our objectives were to identify nursing-related drivers of patient refusal of VTE prophylaxis and to develop customized interventions to reduce refusal rates.

Methods: We conducted focus group interviews (n=14) with day and night shift nurses from 5 units (2 medical, 3 surgical) at 1 hospital to assess nurse understanding of VTE prophylaxis and perception of why patients refuse each prophylaxis component. Four units were selected for participation by their high patient refusal rate along with the unit that had the lowest rate. Focus groups were recorded and transcribed verbatim. Nurse perception of drivers of patient refusals of VTE prophylaxis were analyzed using the Theoretical Domain Framework (TDF), which is an integrative framework that applies theoretical approaches to interventions aimed at behavior change.

Results: The process for ordering and administering VTE prophylaxis allows identification of potential points of patient refusal (Figure 1). Focus group findings highlight that patient refusals are influenced by three main TDF domains: environmental context and resources, knowledge, and beliefs about consequences.  One key environmental context and resource barrier identified was the lack of patient education materials on VTE prophylaxis. Nurses did not have the resources required to supplement their explanation to patients about the significance of prophylaxis. Furthermore, the knowledge barrier was highlighted by the many nurses having the misconception that all three components of prophylaxis are not necessary. This overlapped with the barrier on beliefs about consequences because many nurses felt that chemoprophylaxis was not necessary in ambulating patients.

Conclusion: Nursing-related drivers of patient refusal of VTE prophylaxis include lack of knowledge and misperceptions of the consequences of missed elements of VTE prophylaxis. These factors appear to be modifiable targets for quality improvement, particularly by focusing on equipping nurses to address potential patient refusals and by engaging patients in their care. Interventions can be tailored to specifically address these vulnerabilities. Future initiatives will use similar methods to identify how physicians can influence patient refusals of VTE prophylaxis.

74.08 Familiarity and Communication Between Health Care Provider Dyads in the Operating Room

Posted on January 26, 2017

L. L. Frasier1, S. R. Pavuluri Quamme1, Y. Ma1, D. A. Wiegmann1,2, E. H. Dugoff3, G. Leverson1, C. C. Greenberg1,2  1University Of Wisconsin,Surgery,Madison, WI, USA 2University Of Wisconsin,Engineering,Madison, WI, USA 3University Of Wisconsin,Population Health,Madison, WI, USA

Background: Communication is implicated in many adverse events in the operating room (OR). Although familiar OR teams have fewer communication- and teamwork- related failures, working with unfamiliar providers is the norm rather than the exception. Providers in focus groups indicated that increased verbal communication could compensate for team unfamiliarity. We tested the resulting hypothesis that there is an inverse correlation between familiarity and communication frequency between provider pairs (dyads) in the OR.

 

Methods: Ten open operations were audio-video recorded over 13 months. We asked six providers (anesthesiology attending, in-room anesthetist, circulator, scrub, surgery attending, surgery resident) about the number of shared cases with each other and assigned familiarity scores to dyads. We identified communication events, coded for dyad(s) involved, calculated dyad communication rates and evaluated the relationship between dyad familiarity score and communication rate using Poisson regression.

 

Results: We generated 48 hours of footage (mean case length, 4.8 hours; range, 3.3 to 6.7 h). There were 2574 communication events (range, 137 to 510 per case). Seventy events (2.7%) could not be coded for content and/or dyad and were excluded. On average, 58.0 communication events occurred per hour (range, 29.4 – 76.1) resulting in a communication event every 62 seconds. Overall, the mean communication rate for provider dyads was 6.8 events / hour of shared room time.

Fifty-four of 59 participants returned questionnaires (91% response rate). Fifty-four percent of participants were male (25% nursing, 79% anesthesia, 60% surgery). For 51% of participants, the recorded operation represented the first shared case with another provider in a week; for 22% of participants, it represented the first shared case in a month. Evaluating total shared cases ever, 11% reported that this was the first ever shared case with another provider, while nearly a third reported sharing 40 or more cases.

In multivariable analysis, dyad familiarity score was not significantly associated with communication rate (p=0.69). The only significant predictor of communication rate was cross-disciplinary status (p<0.001). Compared to intra-disciplinary dyads, which averaged 10.0 communication events/ hour of shared room time, nursing-surgery, anesthesiology-surgery, and anesthesiology-nursing dyads have communication rates of 5.8, 5.2, and 2.8 events per hour of shared room time, respectively.

 

Conclusion: Despite provider perceptions of increased communication frequency when working with unfamiliar team members, we found no relationship between familiarity and verbal communication rates in the operating room.  Our results suggest the persistence of  traditional disciplinary silos in the OR. Given the prevalence of multi-disciplinary teams with limited familiarity, these findings represent a critical gap for improving communication and safety in the operative setting.

74.07 Surgical intensity and risk factors for prolonged opioid use following spine surgery

Posted on January 26, 2017

A. J. Schoenfeld1, W. Jiang1, M. A. Chaudhary1, R. Scully1, A. Haider1  1Brigham And Women’s Hospital,Center For Surgery And Public Health,Boston, MA, USA

Introduction:
There is a growing concern that the use of prescription opioids following elective surgical interventions may predispose patients to chronic opioid use and abuse. This concern is especially prevalent for patients undergoing elective orthopaedic procedures and spine surgery in particular.  The prevalence of sustained opioid use in this population, especially among narcotic naive populations, has not been explored previously.  In addition, risk factors for sustained opioid use and the role of surgical intensity in mediating such risks are not well understood.  We sought to estimate the proportion of patients using opioids up to 1-year following discharge for common spine surgical procedures and identify factors associated with sustained opioid use.

Methods:
This study utilized 2006-2014 claims data from TRICARE insurance maintained by the Department of Defense.  Adults (18-64 years) who received one of four common spine surgical procedures (discectomy, decompression, posterolateral fusion or interbody fusion) between 2007-2013 were identified. Patients with prior opioid use, as defined by use within the 6 months preceding the index procedure, were excluded. Posterolateral fusion and interbody fusion were considered procedures of higher intensity, discectomy and decompression were considered to be lower intensity. Covariates included demographic factors, pre-operative diagnoses, comorbidities, post-surgical complications and mental health disorders. Prescription opioid use was evaluated up to one-year following discharge. Risk-adjusted Cox Proportional-hazards models were used to evaluate predictors of sustained opioid use.

Results:
This study included 4,798 patients.  Eighty-one percent filled at least one opioid prescription following discharge. At 30 days following discharge, 9% continued opioid use, while 2% continued use at 3-months and 0.3% at 6-months. When adjusted for pre-operative diagnosis, less intense surgical procedures were associated with a higher likelihood of opioid discontinuation (HR 1.42, 95% CI 1.32, 1.53). Lower socioeconomic status (HR 0.79, 95% CI 0.66-0.95), depression (HR 0.80, 95% CI 0.71, 0.91) and anxiety (HR 0.68, 95% CI 0.49, 0.95) were significantly associated with a decreased likelihood of discontinuing opioid use. 

Conclusion:
By 6-months following discharge, nearly all patients had discontinued opioid use after spine surgery. As only 0.3% of patients continued opioid use more than 6-months following surgery, these results indicate that spine surgery among opioid naive patients is not a major driver of long-term prescription opioid use.  Socioeconomic status and pre-existing diagnoses of anxiety and depression may be factors associated with sustained opioid use after spine surgery.
 

74.06 New Persistent Opioid Use Following Surgery: A Chronic Postoperative Complication

Posted on January 26, 2017

J. S. Lee1, H. M. Hu1, D. C. Cron1, C. M. Brummett2, M. J. Englesbe1, J. F. Waljee1  1University Of Michigan,Department Of Surgery,Ann Arbor, MI, USA 2University Of Michigan,Department Of Anesthesiology,Ann Arbor, MI, USA

Introduction:

Opioid misuse and abuse is a national public health crisis with an estimated 44 deaths each day from prescription opioid overdose. Moreover, chronic opioid use confers a disproportionately high risk of morbidity, mortality, and increased healthcare expenditures. Although 40% of opioid prescriptions are written for postoperative pain relief, the risk of transitioning to chronic use is not well understood. In this study, we sought to define long-term trends in opioid use following surgery.

Methods:

Using a national dataset of employer-based insurance claims, we identified U.S. adults age 18 – 64 years old undergoing major (bariatric surgery, colectomy, incisional hernia repair, reflux surgery, hysterectomy) and minor elective surgery (varicose vein removal, laparoscopic cholecystectomy/appendectomy, hemorrhoidectomy, transurethral prostate surgery, thyroidectomy, parathyroidectomy, carpal tunnel release) from January 2013 – December 2014 (n=57,760). Opioid prescriptions were obtained from pharmacy claims and converted to oral morphine equivalents (OME) for comparison. Patients were stratified into 5 groups based on preoperative opioid use: naive, intermittent, and chronic (high, medium, low) opioid users. Naive patients with new persistent opioid use 90 days after surgery were analyzed as a separate group. Primary outcomes were daily opioid dose and concurrent benzodiazepine use, given its strong association with opioid overdose fatalities. Outcomes were evaluated at 30-day intervals from 180 days before surgery to 180 days after surgery, and compared using a mixed model approach.

Results:

58.9% of patients were opioid naive with no persistent postoperative opioid use, 3.8% developed new persistent opioid use, 29.1% were intermittent users, and 8.2% were chronic users. All groups were alike in demographics and case mix, but had significantly different trends for the primary outcomes (p<0.0001). Chronic and intermittent opioid users returned to baseline levels of opioid use following surgery (Figure 1). In contrast, patients with new persistent opioid use continued taking high doses 6 months after surgery without decline, equivalent to 6 tablets/day of hydrocodone/acetaminophen 5/325, and at similar doses to chronic users. They also had escalating rates of concurrent benzodiazepine use, increasing from 2.3% to 5.2% six months after surgery.

Conclusion:

New persistent opioid use continues long after surgery with maintenance of high daily doses and rising rates of concurrent benzodiazepine use. Given the substantial morbidity and mortality associated with chronic opioid exposure, it is critical to identify patients at risk of developing postoperative opioid dependence and target interventions to mitigate persistent use.

74.05 Preoperative Chronic Opioid Use is Associated with Morbidity after Elective Surgery

Posted on January 26, 2017

D. C. Cron1, J. S. Lee1, M. Hu1, L. Zhong2, P. E. Hilliard3, M. J. Englesbe1, C. M. Brummett3, J. F. Waljee2  1University Of Michigan Medical School,Surgery,Ann Arbor, MI, USA 2University Of Michigan Medical School,Plastic Surgery,Ann Arbor, MI, USA 3University Of Michigan Medical School,Anesthesiology,Ann Arbor, MI, USA

Introduction:  Chronic opioid use is increasingly prevalent, poses challenges for perioperative management, and may increase risk of surgical complications. To better understand this growing high-risk surgical population, we studied the effect of preoperative chronic opioid use on postoperative complications.

Methods:  We used the Truven Health MarketScan Commercial Claims and Encounters Database, encompassing over 100 U.S. health plans, to identify adults who underwent elective surgery between 7/2009-12/2012 (N=338,011). We included minor surgeries (N=259,873; varicose vein ablation, hemorrhoidectomy, laparoscopic appendectomy, laparoscopic cholecystectomy, prostatectomy, thyroidectomy, parathyroidectomy, and carpal tunnel release) and major surgeries (N=78,138; colectomy, ventral hernia, bariatric, gastric reflux, and hysterectomy). The primary exposure, preoperative opioid use, was defined as ≥1 opioid prescription filled both within 30 days and also between 30-90 days preoperatively. The secondary exposure was preoperative 90-day mean morphine equivalents (MME). The outcome was 30-day major complications, and we used logistic regression to risk adjust for procedure type and patient clinical and demographic characteristics. 

Results: The prevalence of preoperative chronic opioid use was 5.5% overall, 5.3% among minor surgery, and 6.2% among major surgery. In opioid users compared to opioid-naïve, risk-adjusted complication rates (≥1 complication) were 1.4 times higher overall (5.1% vs. 3.7%; p<0.001), 1.3 times higher after minor surgery (3.3% vs. 2.5%; p<0.001), and 1.4 times higher after major surgery (10.9% vs. 7.9%; p<0.001). Opioid use in the overall group was significantly associated with surgical site infection (1.7% vs. 1.2%), pulmonary failure (1.1% vs. 0.7%), renal failure (0.8% vs. 0.7%), hemorrhage (1.2% vs. 0.8%), and GI bleeding (0.9% vs. 0.6%). The figure shows overall complication rates by preoperative MME; there was a statistically significant trend overall with increasing preoperative opioid dose associated with increased complications (p-value for trend <0.001). 

Conclusion: Chronic opioid use prior to elective surgery was associated with increased risk of postoperative complications in this population-based study. This higher risk surgical population may require more aggressive and tailored perioperative care to prevent and manage complications. Preoperative interventions attempting to wean opioids will be difficult but may improve surgical outcomes.

 

74.04 The Good the Bad and the Ugly: Patient Perceptions About Their Doctor in Two Words.

Posted on January 26, 2017

B. A. Singletary1,3, N. K. Patel3, J. M. Dodd2, H. Chen1, M. J. Heslin1,3  1University Of Alabama At Birmingham,Department Of Surgery,Birmingham, AL, USA 2Univesity Of Alabama At Birmingham,Department Of Quality And Safety,Birmingham, AL, USA 3University Of Alabama At Birmingham,Chief Of The Medical Staff Office,Birmingham, AL, USA

Introduction: Physicians are increasingly being measured by patient’s perceptions of care and experience. Current measurement tools many times fail to resonate with physicians and are complex in their interpretation. This study evaluated descriptive words about the patient doctor interaction, and validated this tool against current Centers for Medicare & Medicaid Services (CMS) standards.

Methods: Between 1/1/2015 and 4/30/2016, an addition to the current outpatient patient experience survey was made, asking individuals: “Please describe your physician in today’s visit in two words”.  Words were scored as positive and negative, based on sentiment analysis, and then analyzed categorically.  Correlation analyses were completed to validate this measure with the existing, yet more extensive Consumer Assessment of Healthcare Providers and Systems Clinician & Group Survey (CG-CAHPS). 

Results: 52,328 surveys included two word data, with responses analyzed on 594 physicians across 20 departments at a single institution.  An average of 28 surveys were completed per physician, with respondents were more likely to be female (65%), caucasian (81%), and showing a mean age of 58 (SD 15).  Overall, physicians received 96% positive word results.  297 physicians (50%) received all positive responses compared to 1 physician (0.2%) only receiving negative words.  Positive word responses showed highly significant correlation with CG-CAHPS measures of top-box scores, r=0.97 p=<0.0001, while negative word responses showed a significant negative correlation with institution percentile rank, r= -0.33 p<0.0001.

Conclusion: The two-word measurement tool shows a significant correlation with current CMS measures, and provides a more concise means of patient response collection. Thematic result rates and visual data provide leadership the opportunity to advocate for high performers and hold accountable those requiring remediation.  These results suggest this tool as a valid and efficient method for assessing a patient’s perceptions of their provider.

 

« Previous Page
Next Page »