94.17 INCIDENCE, PREVALENCE AND MORTALITY OF HPB MALIGNANCIES: 1990 TO 2016. GBD 2017 Study.

Posted on January 31, 2019

A. Weaver1, D. Denning1, J. Sanabria1,2,3,4  1Marshall University School of Medicine,Department Of Surgery,Huntington, WV, USA 2Marshall Institute for Interdisciplinary Research (MIIR),Department Of Surgery,Huntington, WV, USA 3Case western Reserve University,Nutrition And Preventive Medicine,Cleveland, OH, USA 4Wasnhington University,Institute For HUman Metrics And Evaluation (IHME),Seattle, WA, USA

Introduction: Cancer related disease is the second Global cause of mortality. As population ages. malignancies may become the most prevalent cause of morbidity and health expenditure. The purpose of the present study is to determine incidence, prevalence and mortality trends of malignant neoplasms from the liver, gallbladder (GB) and pancreas.  We also report on disability-adjusted life-years (DALYs), and Years Lost to Disability (YLD).

Methods:  The incidence, mortality, DALYs, and YLD by age, gender, year, and geography were found using datasets from the Global Burden of Disease (GBD) group, where the epidemiological data obtained were modelled in DisMod-MR 2.1, a Bayesian meta-regression tool which pools data-points from different sources and adjusts for known sources of variability. GBD data was extracted from 284 country-year from 1990 to 2016.

Results: Age-standardized global incidence rates (ASIR) per 100,000 for primary liver and pancreatic cancers rose from 12.49 to 14.55 and 6.11 to 6.37, respectively. Whereas the ASIR for GB decreased from 3.71 to 2.8. Age-standardized global mortality rates (ASMR) decreased for GB from 3.24 to 2.47, but stayed relatively constant for primary liver and pancreatic cancers at 12.54 vs12.13 and 6.19 vs 6.20 respectively. Age-standardized global DALY decreased for all three cancers; liver (331.71 vs 295.2), GB (63.19 vs 47.82), and pancreas (122.7 vs 119.49). Age-standardized global YLD rose slightly for liver cancer from 2.69 to 3.29, stayed the same for pancreatic cancer (1.24 vs 1.30), and felt for GB cancer (0.82 vs 0.63). While the highest ASIR of liver cancer in 2016 was observed in East Asia (34.1), High-Income Asia Pacific (22.94) and Western Sub-Saharan Africa (17.93), for GB cancer was noted in High-Income Asia Pacific (7.95), Southern Latin America (7.66) and Andean America (3.73), and for pancreatic cancer was High-Income Asia Pacific (11.8), High-Income North America (10.6), and Western Europe (9.93).

Conclusion: Global surveillance of HPB malignancies, and in particular changes in subpopulations, may provide with links to potential etiologies as well as stimulate further research. This may also help countries with rising incidences to formulate more effective strategies for prevention, screening, and treatment.
 

94.16 Clinicopathologic Features And Outcomes Of Early-onset Pancreatic Adenocarcinoma In the U.S.

Posted on January 31, 2019

J. E. Ordonez1, C. A. Hester1, H. Zhu1, M. Augustine1, M. R. Porembka1, S. C. Wang1, A. C. Yopp1, J. C. Mansour1, H. J. Zeh1, P. M. Polanco1,2  1University Of Texas Southwestern Medical Center,Division Of Surgical Oncology,Dallas, TX, USA 2Department of Veterans Affairs – North Texas Health Care System,Surgical Oncology,Dallas, TX, USA

Introduction: There is limited research regarding Pancreatic Ductal Adenocarcinoma (PDAC) diagnosed in Early-Onset (EO) patients.  EO is defined as cancer diagnosis before the age of 50. We aimed to characterize the clinicopathologic factors and treatment outcomes associated with EO patients compared to late onset (LO) patients and determine the impact of EO PDAC on survival.

Methods: The National Cancer Database was queried to identify EO and LO PDAC patients diagnosed from 2004 to 2013. Demographics, tumor characteristics, treatment regimens, and overall survival (OS) were compared between groups.Chi-square and Fisher’s exact tests were used to compare cohorts. Kaplan-Meier with log-rank univariate analysis was used to analyze survival. Cox proportional hazard regression was used to create a multivariable model.

Results:

207,062 patients were included in the study: 12,137 (5.9%) EO and 194,925 (94.1%) LO patients. EO patients were more likely to present with later stage of disease (Stage III/IV disease, 62.1% vs 55.2%, p <0.001) and larger tumor size (tumor ≥ 4cm, 36.7% vs 32.9 %, p<0.001) compared to those with LO PDAC.  LO patients presented with higher Charlson/Deyo comorbidity score (CDCC ≥1, 33.4% vs. 19.1%, p <0.001), and increased proportion of tumors located in the pancreatic head (52.4% vs 50.8%, p=0.002), compared to EO patients.  EO patients received more curative intent surgical resection, including surgery alone or with neoadjuvant/adjuvant therapies (24.0% vs 19.2%, p <0.001) and palliative chemoradiation (13.7% vs 10.2% p <0.001) compared to LO patients.  LO patients had a significantly higher proportion of patients who received no treatment (40.5% vs 23.3% p <0.001). EO PDAC was associated with improved OS among all PDAC patients (9.0 vs 6.2 months, p<0.001) and surgically resected patients (26.6 vs 24.3 months, p<0.001). EO PDAC was also independently associated with improved OS after adjusting for other significant clinicopathological factors.

 

Conclusion: In conclusion, EO PDAC comprised 5.9% of all PDAC cases.  EO patients presented with features characteristic of more advanced disease, including larger tumors and later stage.  EO patients also received higher rates of curative and palliative therapies.  Within curatively resected patients, EO patients received higher rates of neoadjuvant and adjuvant therapy.  EO PDAC was independently associated with improved OS compared to LO PDAC among all patients and also amongst curatively resected stage matched patients.
 

94.15 Impact of Preoperative MELD Score on Outcome After Minimally Invasive Robotic Hepatectomy

Posted on January 31, 2019

I. Sucandy1, J. Spence1, S. Schlosser1, T. J. Bourdeau1, S. B. Ross1, A. S. Rosemurgy1  1Florida Hospital Tampa,Surgery,Tampa, FLORIDA, USA

Introduction:  Minimally invasive hepatectomy for benign and malignant liver lesions has gained popularity over the past decade due to superior perioperative outcomes when compared to conventional ‘open’ hepatectomy. Patients undergoing hepatectomy present with varying degrees of liver dysfunction, which may affect their postoperative outcomes. The Model for End Stage Liver Disease (MELD) score is objective, is reproducible, and has been shown to predict mortality related to cirrhosis. This study was undertaken to investigate the impact of preoperative MELD scores on outcomes after minimally invasive robotic hepatectomy. 

Methods:  With IRB approval, patients undergoing robotic hepatectomy were followed prospectively. Demographic data, MELD score, and outcomes, such as operative time, estimated blood loss, complications, and length of stay, were collected. Regression analysis was used to compare preoperative MELD scores with outcome variables; significance was accepted with 95% probability. For illustrative purposes, data are presented as median (mean ± SD).

Results: 75 patients underwent robotic hepatectomy. Patients age was 64 (62 ± 14.2) years and BMI was 28 (29 ± 7.0) kg/m2; 56% were women. 55 (73%) of the hepatectomies were undertaken for malignancy (36% hepatocellular carcinoma, 24% colorectal metastasis, 9% intrahepatic cholangiocarcinoma, 7% gallbladder cancer, and 6% metastatic neuroendocrine tumor). On regression analysis, MELD score did not correlate with operative time [227 (262 ± 107.5) minutes, p=0.518] or estimated blood loss [125 (266 ± 324.4) mL, p=0.583]. MELD score did correlate with length of stay [3 (5 ± 4.7) days, p=0.002]. 7 patients (9%) experienced complications (i.e., acute respiratory failure, bile leak requiring ERCP, intra-abdominal fluid collection, pleural effusion, surgical site infection, enterocutaneous fistula, and atrial fibrillation); their MELD score was 7 (8 ± 2.5). 68 patients (91%) did not experience any complications; their MELD score was 7 (8 ± 2.8) and was not different than patients experiencing complications (p=0.803). 1 (1%) patient died as a consequence of an enterocutaneous fistula.

Conclusion: Robotic hepatectomy can be undertaken with few complications, a short length of stay, and low mortality. Preoperative MELD score does not predict operative blood loss, operative duration, the occurrence of postoperative complications, or postoperative mortality; preoperative MELD score does predict hospital length of stay following robotic hepatectomy. For patients being considered for hepatectomy, an elevated MELD score should not deter surgeons from utilizing minimally invasive robotic hepatectomy. 

 

94.14 Changes in Maximum Axial Diameter of Pancreatic Cystic Lesions Poorly Estimate Volumetric Changes.

Posted on January 31, 2019

V. Rendell1, A. Awe1,2, M. Lubner2, E. Winslow1  1University Of Wisconsin,Department Of Surgery,Madison, WI, USA 2University Of Wisconsin,Department Of Radiology,Madison, WI, USA

Background: Determining surveillance and treatment plans for indeterminate pancreatic cystic lesions (PCs) is challenging due to an overall lack of understanding of their natural history. Clinical guidelines use unidimensional size thresholds and growth criteria, but the correlation between cyst size and malignant potential has been variable. It is unclear if unidimensional size changes accurately capture true volumetric changes of PCs, which may better predict clinical risk. We aimed to determine if changes in maximum axial diameter (MAD) of PCs accurately reflect changes in volume.

Methods: We performed an imaging search at our institution from 2012-2013 for contrast-enhanced CT or MR scans demonstrating a PC>1cm. We selected the first and most recent scan for patients with two studies >1 year apart. Pseudocysts and solid tumors were excluded. Imaging analysis of PCs was performed using a novel analytics software, HealthMyne, to measure MAD, volume, sphericity, and surface area. To compare changes in MAD to changes in volume, the estimated change in volume over time based on MAD was calculated as follows: eΔvol =  4π(MADtime 2/2)3/3 – 4π(MADtime 1/2)3/3. The difference between measured (actual) change in volume, (Δvola = voltime2 – voltime1) and estimated change in volume was used to calculate the absolute value of the percent difference in volume change: %diffabs = ? (eΔvol – Δvola)/Δvola? x 100% . The %diffabs by high vs low PC sphericity and surface area to volume ratio (SA:V) were compared by chi-square analysis. 

Results: 153 patients had two scans, an average of 4.1 years apart (SD 2.7 years). PCs had an average MAD of 2.1cm (SD 1.0cm) on the first scan, and 2.5cm (SD 1.3cm) on the last scan. In total, 9 PCs (6%) decreased in size over time. The average change in MAD per year was 1.0mm/yr (SD 2.5mm/yr) or 5.7%/year (SD 13%/yr). The PCs had an average volume of 5.3cm3 (SD 10.1cm3) on the first scan and 9.2 cm3 (SD 16.7cm3) on the last scan. The average change in volume per year was 0.88 cm3/yr (SD 2.5 cm3/yr) or 4%/yr (SD 87%/yr). For 60 (39%), the actual volume change was larger than the estimated volume change. Sixty-four PCs (42%) had a >100% absolute difference between estimated and measured volumes changes. PCs with high sphericity had significantly better concordance between estimated and actual volumes, but no significant difference was found by SA:V (Table 1).

Conclusions: For many PCs, estimating volume changes using MAD has poor concordance with actual changes in cyst volume. When following PCs over time, volumetric changes may provide better assessments of changing cyst size. The value of volumetric measurements for risk stratification of PCs is in need of further study.

 

94.13 CT and MRI for the Pre-Operative Detection of Lymph Node Metastases in Gallbladder Cancer

Posted on January 31, 2019

E. A. De Savornin Lohman1, T. De Bitter2, C. Van Laarhoven1, R. De Haas3, P. De Reuver1  1Radboudumc,Surgery,Nijmegen, Netherlands 2Radboudumc,Pathology,Nijmegen, Netherlands 3University of Groningen Medical Centre,Radiology,Groningen, Netherlands

Introduction:
Lymph node metastases (LNM)  are an ominous prognostic factor in gallbladder cancer (GBC) and, when present, should preclude surgery. However, unclarity remains regarding the optimal imaging modality for pre-operative detection of LNM and international guidelines vary in their recommendations. The purpose of this study was to systematically review the diagnostic accuracy of computed tomography (CT) versus magnetic resonance imaging (MRI) in the detection of LNM of GBC.

Methods: A literature search of studies published until November 2017 concerning the diagnostic accuracy of CT or MRI regarding the detection of LNM in GBC was performed. Data extraction and risk of bias assessment was performed independently by two reviewers. The sensitivity of CT and MRI in the detection of LNM was reviewed. Additionally, estimated summary sensitivity, specificity and diagnostic accuracy of MRI were calculated in a patient based meta-analysis.

Results: Nine studies including 292 patients were included for narrative synthesis and 5 studies including 158 patients were selected for meta-analysis. Sensitivity of CT ranged from 0.25 to 0.93. Estimated summary diagnostic accuracy parameters of MRI were as follows: sensitivity 0.75 (95% CI 0.6 – 0.85), specificity 0.83 (95% CI 0.74 – 0.90), LR+ 4.52 ( 95% CI 2.55 – 6.48) and LR- 0.3 (95% CI 0.15 – 0.45). Small (<10mm) LNM were most frequently undetected on pre-operative imaging. Due to a lack of data, no subgroup analysis comparing the diagnostic accuracy of CT versus MRI could be performed.

Conclusion: The value of current imaging strategies for the pre-operative assessment of nodal status in GBC remains unclear. Sensitivity of both MRI and CT appear limited, especially regarding the detection of small LNM. Additional research is warranted in order to establish uniformity in international guidelines,  improve pre-operative nodal staging and prevent futile surgery.

 

94.12 Practice Variation in Gallbladder Surgery; a Longitudinal Population Based Study

Posted on January 31, 2019

C. S. Latenstein1, S. Z. Wennmacker1, F. Atsma2, M. Noordenbos2, S. Groenewoud2, P. R. De Reuver1  1Radboudumc,Surgery,Nijmegen, NIJMEGEN, Netherlands 2Radboudumc,Scientific Institute For Quality Of Healthcare,Nijmegen, NIJMEGEN, Netherlands

Introduction:
Approximately, 700,000 cholecystectomies are performed in the US every year. Based on Medicare data the Darthmouth Atlas of Healthcare shows that the chance a patient will have surgical treatment for gallstones can vary across hospitals, with a factor score ranging from 2 to 4 fold difference. Nationwide longitudinal data about cholecystectomy rates in all patients presenting at the surgical outpatient clinic are lacking. We aimed to determine the longitudinal practice variation for cholecystectomy in the Netherlands.

Methods:
A population based analysis was performed for gallstones patients in all Dutch general hospitals from 2013 up to 2015. Operation rates were determined for each hospital by dividing the absolute number of operated patients by the total  patients who consulted a surgeon for gallstones. Differences in operation rates between academic, teaching and non-teaching hospitals were evaluated. Operation rates were adjusted for differences in  age, sex and Social Economic Status (SES) to summarize a factor score for practice variation. This score was calculated by dividing the adjusted operation rate of the 95% percentile by the adjusted operation rate of the 5% percentile to asses a trend in practice variation over time.

Results:
A total of 89,316 patients (33.1% male, mean age 54.1 years) with gallstones were included from all hospitals during the study period. The average nationwide operation rate for cholecystectomy was 76.1%. The difference in operation rate ranged between centers ranged from 44.7% to 97.7%. The average operation rate in eight academic hospitals was 60.7%, 75.7% in 44 teaching hospital and 79.6% in 31 non teaching hospitals. The summarized annual factor score for 75 general (teaching and non-teaching) hospitals was 1.3 in all three years.

Conclusion:
Unadjusted operation rates for cholecystectomy vary among hospitals, but the factor score illustrates a relatively low practice variation and comparable adjusted operation rates in the Netherlands. Compliance to evidence based guidelines and uniformity in decision making could potentially contribute to a low practice variation in gallbladder removal compared to other countries.
 

94.11 Epidural Analgesia Improves Early Pain Control But Impedes Timely Discharge After Pancreatic Surgery

Posted on January 31, 2019

S. S. Kim1, X. Niu1, I. A. Elliott1, J. P. Jiang1, A. M. Dann1, L. M. Damato1, H. Chung1, M. D. Girgis1, J. C. King1, O. J. Hines1, S. Rahman1, T. R. Donahue1  1David Geffen School Of Medicine, University Of California At Los Angeles,Los Angeles, CA, USA

Introduction:  Epidural analgesia (EA) is frequently used after pancreatic surgery for postoperative pain control and offers many benefits over intravenous narcotics. However, there is inconsistent evidence on the effect of EA on postoperative hospital length of stay (LOS), expeditious discharge, and pain control in patients undergoing pancreatic surgery.

Methods:  A retrospective review of American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) databases from 2014 to 2015 was conducted to determine the effect of EA on LOS after open pancreatic surgery with multiple logistic regression analysis. A single-institution pancreas surgery database was also reviewed to determine the impact of EA on patient-reported postoperative pain. 

Results: There were 6185 patients who underwent open pancreaticoduodenectomy (PD) and 1913 patients who underwent open distal pancreatectomy (DP) in the 2014-2015 ACS-NSQIP databases. The frequencies at which patients were administered EA were 23.2% for PD and 19.4% for DP. EA was associated with a statistically significant longer median LOS for both PD and DP (p= 0.007, 0.001 respectively). There was a significantly greater proportion of patients with EA discharged after 5 to 8 days for PD and 5 to 9 days for DP. A detailed comparative analysis was conducted for patients discharged before vs. after 7 days for PDs and 6 days for DPs, as they were the most frequent LOS for these procedures and generally represent an efficient and complication-free postoperative course. There were significantly more patients who received EAs in the longer LOS groups for both procedures (PD: 24.5% vs. 18.3%, p< 0.001, DP: 21.3% vs. 15.5%, p= 0.001). On multivariable analysis controlling for pancreatectomy-relevant complications, EA remained significant for the longer groups for both PD and DP (PD: OR 1.465, p< 0.001, DP: OR 1.417, p= 0.004). On review of single-institution pancreas surgery database, patient-reported pain scores were significantly lower in the EA groups (PD: n= 20, DP: n= 24) than intravenous narcotics groups (PD: n= 20, DP: n= 18) on the day of surgery only (PD: p= 0.008, DP: p= 0.031).

Conclusion: Epidural analgesia impeded early discharge after surgery for patients undergoing open PD and DP in the 2014-2015 ACS-NSQIP databases. EA led to improved pain control only on the day of surgery in our institutional database. Based on these results, we suggest an earlier transition from EA to intravenous and oral opioids so as to not delay an early discharge, provided pain is well-controlled and the patient appears to be on course for an uncomplicated recovery.

94.10 Impact of Routine Drain Placement and Drainage Duration on Outcomes After Pancreaticoduodenectomy

Posted on January 31, 2019

P. Addison1,3, P. C. Nauka1, K. Fatakhova1,3, L. Amodu1,3, N. Kohn2, H. L. Rodriguez Rilo1  2Feinstein Institute for Medical Research,Department Of Biostatistics,Manhasset, NY, USA 3Zucker School of Medicine at Hofstra/Northwell,Department Of Surgery,Hempstead, NY, USA 1Zucker School of Medicine at Hofstra/Northwell,Pancreas Disease Center,Manhasset, NY, USA

Introduction:  The decisions to routinely place a drain after pancreaticoduodenectomy and how long to leave the drain remain controversial due to conflicting evidence and significant variations in clinical practice. This study aims to address those questions by utilizing a large national database and a rigorous analytical model.

Methods: The American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) 2015-2016 Pancreatectomy Participant Use Data Files were utilized to identify patients who had undergone pancreaticoduodenectomy (n = 7583). Univariate and multivariate logistic regression analyses were performed to control for potential confounders and various preoperative risk factors.

Results:Of 7583 patients, drains were placed in 6666 (87.9%). Drain placement decreased the risk of developing serious morbidity (relative risk [RR] 0.73, 95% confidence interval [CI] 0.65-0.82), overall morbidity (RR 0.79, 95% CI 0.72-0.87), and organ space surgical site infection (RR 0.72, 95% CI 0.61-0.85); drain placement did not change the risk of developing a clinically relevant postoperative pancreatic fistula (RR 0.96, 95% CI 0.78-1.19). However, the duration of drain use was independently associated with serious morbidity (hazard ratio [HR] 3.06, 95% CI 2.65-3.53), overall morbidity (HR 2.48, 95% CI 2.20-2.80), and organ space surgical site infection (HR 1.47, 95% CI 1.25-1.74).

Conclusion: Routine drain placement following pancreaticoduodenectomy may decrease postoperative complications, including serious morbidity, overall morbidity, and organ space surgical site infections; however, increased drain duration was associated with serious morbidity, overall morbidity and organ space surgical site infection. These results support the routine placement and early removal of intraoperative surgical drains in pancreaticoduodenectomy.

 

94.09 Total Neoadjuvant Therapy in Pancreatic Cancer: Experience with Complete Pathologic Responses

Posted on January 31, 2019

E. E. McGillivray1, M. Hill1, E. O’Halloran1, S. Murthy1, C. Mayemura1, K. Liang1, N. Goel1, J. Farma1, A. Karachristos1, J. Hoffman1, S. S. Reddy1  1Fox Chase Cancer Center,Department Of Surgical Oncology,Philadelphia, PA, USA

Introduction: The benefits of neoadjuvant therapy in the treatment of pancreatic adenocarcinoma are well known, and with its routine use, pathologic complete response (pCR) has been observed in post-resection specimens. We report our experience with patients who underwent total neoadjuvant therapy (TNT) and had a pCR.

Methods:  We conducted a retrospective chart review of all patients who received TNT and underwent a pancreatoduodenectomy, distal or total pancreatectomy at our tertiary care institution from 2002-2018 (n=80). Total neoadjuvant therapy was defined as chemotherapy alone (Gemcitabine or Fluoropyrimidine (5FU) based regimens), followed by radiosensitizing chemotherapy with radiation. Clinical data and pathological information were reviewed, and pCR was defined as those specimens with no remaining tumor cells identified. Preoperative serum CA19-9 levels were analyzed using unpaired T-Tests. Disease free survival (DFS) and overall survival (OS) were examined and compared using the Cox proportional hazard model.

Results: Eight of the 80 patients (10%) were noted to have a pCR. The average age of diagnosis was 61 (46-83), and 50% were male. Clinical stage was stage II (n=5, 62.5%), followed by stage III (n=2, 25%) and stage I (n=1, 12.5%). The median initial CA19-9 was 429.5 U/mL (normal: 0-37), and the median post neoadjuvant CA19-9 was 25.6 U/mL. Overall, 75% of pCR patients experienced a greater than 50% reduction in their CA19-9 values compared to 60% in the non-pCR group (p=0.436). Of the 8 patients, 63% received Gemcitabine based TNT, and 37.5% received 5FU based TNT. Postoperative adjuvant chemotherapy was given to 1 (12.5%) patient 4.6 months after surgery. The 2 year OS rate in the pCR group was 100% with a median OS of 42.2 months, while the 2 year OS for patients that did not have a pCR (n=72) was 61% with a median OS of 31.5 months (p=0.013). Median DFS was 33.6 months in pCR patients compared to 24.9 months in the non-pCR group (p=0.0032). Recurrence was documented in 3 (37.5%) patients with pCR with a median time to recurrence of 26.7 months, compared to recurrence in 41 (56.9%) patients in the non-pCR group with a median time to recurrence of 21.3 months.

Conclusion: Although uncommon, patients who undergo TNT and have a pCR have improved outcomes over those who do not. More investigation is needed into determining which pre-operative therapies and patient factors contribute to a pCR.

94.08 “Percent Hepatectomy” vs. “Partial Hepatectomy” to Risk-Stratify Parenchymal-Sparing Liver Resection

Posted on January 31, 2019

H. A. Lillemoe1, R. K. Marcus1, B. J. Kim1, N. Narula1, C. H. Davis1, T. A. Aloia1  1The University of Texas MD Anderson Cancer Center,Department Of Surgical Oncology,Houston, TX, USA

Introduction: As the majority of liver resections have become parenchymal-sparing, revised terminology is proposed to risk-stratify the broad range of procedures often referred to as "partial" or "minor" hepatectomies.

Methods: Consecutive patients from 2015-2017 classified by Brisbane classification as minor hepatectomy were analyzed. Extent of hepatectomy was further defined by the operating surgeon based on the percent of total liver parenchyma removed. Postoperative outcomes were compared based on whether a patient was classified as < or ≥ 30% hepatectomy. Recovery was defined using patient-reported outcomes (PROs) and surgeon-determined readiness to Return to Intended Oncologic Therapy (RIOT). Categorical variables were analyzed using chi-square or Fisher’s exact test and continuous variables were analyzed using the Mann-Whitney U test.

Results: Percent hepatectomy was determined for 81 patients. The median age was 55 years (IQR 46, 63) and 52% of patients were female. The median percent hepatectomy was 15 (IQR 7.5, 20). Procedures included: partial right hepatectomy (21), partial left hepatectomy (20), caudate resection (1), left lateral bisegmentectomy (4), central hepatectomy (5), and multiple partial hepatectomies (30). Table 1 reflects the percent hepatectomy by procedure title. The majority of operations (47%) were performed for resection of colorectal liver metastases. On analysis of partial hepatectomy of < or ≥ 30%, baseline demographics including age, sex, BMI, and ECOG performance status were similar. More procedures in the ≥ 30% group were performed via open surgical approach, with a substantially longer median operative time (264 minutes vs. 195 minutes, p<0.001) and higher median estimated blood loss (200 cc vs. 100 cc, p<0.001). The ≥ 30% hepatectomy group also had a longer length of stay (5 vs. 4 days, p=0.013) and a higher incidence of surgical complications (44% vs. 18%. p=0.033). The ≥ 30% hepatectomy group trended toward longer time to RIOT (median 28 days) compared to the < 30% group (median 21 days, p=0.081). PRO analysis determined that after ≥ 30% hepatectomy, patients were twice as likely to report severe life interference by symptoms (63% vs. 33%, p=0.057).

Conclusion: Within operations classified as "partial" (or "minor") hepatectomy, there are a spectrum of clinical outcomes that can be predicted by classifying cases by the percent of liver resected. Specifically, a cutoff of 30% liver resection may be a clinically relevant risk-stratifier of patients undergoing a partial hepatectomy.

94.07 Hypothermia Prevention in Hepatopancreatobiliary Surgery: A Multidisciplinary Perioperative Protocol

Posted on January 31, 2019

R. A. Sorber2, T. C. Crawford2, E. Haut2,3, C. L. Wolfgang2,4, C. Atallah2  2The Johns Hopkins University School Of Medicine,Department Of Surgery,Baltimore, MD, USA 3The Johns Hopkins University School Of Medicine,Johns Hopkins Surgery Center For Outcomes Research (JSCOR),Baltimore, MD, USA 4The Johns Hopkins University School Of Medicine,Divison Of Hepatobiliary Surgery And Surgical Oncology,Baltimore, MD, USA

Introduction:

Perioperative hypothermia, defined as a patient core temperature below 36oC, has been examined for a number of surgical procedures and associated with numerous adverse outcomes among patients such as surgical site infection, cardiac arrhythmia, increased blood loss and prolonged length of stay. This work describes the implementation and efficacy of a multidisciplinary perioperative protocol to prevent hypothermia among a group of elective hepatopancreatobiliary (HPB) surgery patients at a large academic hospital.

Methods:

This work is a case-control study of 200 elective HPB surgery patients at the Johns Hopkins Hospital from 2016 to 2017, both before and after the implementation of a perioperative warming protocol. The protocol involved pre-op and operating room nurses, OR staff, anesthesia and surgery providers and consisted of applying warming blankets, assuring temperature control in the operating rooms, use of intraoperative underbody warmers and warming blankets and use of warmed IV fluids. Data was collected via retrospective chart review of the intraoperative anesthesia record and analysis was performed both with and without propensity matching to reduce standardized percentage bias.

Results:

Mean core temperature at incision time for HPB patients significantly increased in the intervention period (36.1oC ±0.4) versus the pre-intervention period (35.6oC±0.5, p<0.001). The percentage of patients achieving normothermia significantly increased at incision and in the first 2 hours of the operation (71% and 77% versus 19% and 58%, p<0.01). All of these results remained significant with propensity matching. There was no significant difference elicited in this small study for in-hospital mortality, surgical site infection, intraoperative blood loss or median hospital stay. 

Conclusion:

Implementation of a multidisciplinary perioperative warming protocol significantly reduced hypothermia among elective HPB surgery patients at incision time and in the first two hours of the operation.

94.06 Post-Operative Day 3 Discharge After Pancreaticoduodenectomy

Posted on January 31, 2019

K. A. Baugh1, G. Van Buren1, H. Tran Cao1, E. Silberfein1, C. Hsu1, C. Chai1, N. Massarweh1, W. Fisher1  1Baylor College Of Medicine,Michael E DeBakey Department Of Surgery,Houston, TX, USA

Introduction:  We hypothesized that the implementation of an Enhanced Recovery After Surgery (ERAS) pathway targeting discharge on post-operative day (POD) 3 following pancreaticoduodenectomy (PD) would reduce length of stay (LOS) without affecting post-operative morbidity or mortality.

Methods:  The ERAS pathway was implemented by one surgeon for 15 months starting in November 2016 (ERAS group).  Outcomes in this group were compared to patients operated on by other surgeons at the same institution not following the pathway during the same time period (contemporary group) and to a retrospective cohort (retrospective group) of patients from the same single surgeon, between October 2014 to November 2015. A prospectively maintained database with documentation of all complications within 60 days of surgery was retrospectively reviewed. Analysis was performed using chi-squared or Fisher exact tests for categorical variables and Student’s T-tests or Mann Whitney tests for continuous variables.  

Results:  117 patients were included in the study, 41 were in the ERAS group, 41 in the retrospective group and 35 in the contemporary group.  There were no significant differences in patient demographics, co-morbidities, or fistula risk scores between groups. Median LOS was reduced by 40% in the ERAS group compared to the retrospective group (3 vs 5 days, P<0.001) and by 50% compared to the contemporary group (3 vs 6 days, P<0.001). In the ERAS group, 24% of patients experienced a significant complication (Accordion ≥grade 2) vs 37% in the retrospective group (P=0.23) and 29% in the contemporary group (0.68). There was no difference in clinically relevant post-operative pancreatic fistulas between groups (ERAS 7% vs retrospective group 10% vs contemporary group 3%; P=1, 0.62). Delayed gastric emptying occurred less in the ERAS group than the retrospective groups (4.9% vs 19.5%, P=0.048) but not when compared to the contemporary group (4.9% vs 8.6%, P= 0.66). There were no other differences in post-operative complications, rates of re-operation, or 60-day mortality.

Conclusion: Discharge on POD 3 following PD can be done safely and without impacting morbidity or mortality.

 

94.05 Neoadjuvant Chemoradiation Effect on Positive Lymph Node status in Surgically Resected Pancreatic Cancer

Posted on January 31, 2019

K. B. Golisch1, E. Price1, T. S. Riall1, A. K. Arrington1  1University Of Arizona,Surgery,Tucson, AZ, USA

Introduction: Pancreatic cancer continues to be a leading cause of cancer mortality. Even local or resectable tumors have poor survival outcomes and treatment options are suboptimal. The current NCCN standard of care for pancreatic cancer is surgery first in resectable disease. Clinical trials are ongoing investigating the benefits and role of neoadjuvant chemotherapy and prior literature has shown favorable outcomes. Neoadjuvant chemoradiation (NACXRT) is becoming standard consideration in the setting of borderline resectable cancers, to improve R0 resection rates. We hypothesize NACXRT may effect lymph node (LN) positive status and down-staging of the patient at the time of surgery.

Methods:  This is a NCDB retrospective study of resected pancreatic adenocarcinoma from 2008-2015. Three treatment groups were investigated: surgery first (n= 21,723), neoadjuvant chemotherapy (NAC) (n= 1905), and NACXRT (n= 1806). Charlson-Deyo index, location of tumor, pathologic stage, clinical stage, positive LNs, total LNs were compared between treatment groups. 

Results: A total number of patients 25,434 were included in the study. Positive lymph node status is significantly lower in NACXRT compared to surgery first (χ2= 838, p<.0001) and compared to NAC (χ2 = 216, p<.0001).  Positive LN status is also significantly lower in NAC compared to surgery alone (χ2 = 66, p<.0001). Figure 1 demonstrates the difference in percentage of positive LN among groups. Of cases with both pathologic and clinic stage available, a significant higher percentage of cases were downstaged with NACXRT (38.2%) compared to the other groups (14.6% NAC and 2.5% surgery respectively) while fewer cases were upstaged (34.7% NACXRT, 46.2% NAC, 54.5% surgery).  The number of LN examined was similar for each treatment group. Mean(SD): surgery first= 16.9(9.6), NAC= 20.0(10.5), NACXRT= 16.2(9.3).

Conclusion: Overall, patients who received neoadjuvant therapy had lower positive LN status at time of surgical resection and a higher rate of downstaging. This effect was most significant in the NACXRT group. Currently, NACXRT is primarily used to improve R0 resection rates in borderline resectable pancreatic cancers. However, these results suggest that resectable pancreatic cancer patients should be considered for NACXRT prior to surgery. Though many clinical trials are looking at NAC in resectable pancreatic cancer, more evaluation of NACXRT in pancreatic cancer is certainly warranted.
 

94.04 Characteristics of Gallbladder Polyps on MRI and Dedicated US; a Prospective, Exploratory Study

Posted on January 31, 2019

S. Z. Wennmacker1, E. De Savornin Lohman1, P. De Reuver1, J. Drenth3, J. Hermans2, C. Van Laarhoven1  1Radboudumc,Surgery,Nijmegen, Netherlands 2Radboudumc,Radiology,Nijmegen, Netherlands 3Radboudumc,Gastroenterology And Hepatology,Nijmegen, Netherlands

Introduction:  Preoperative evaluation of gallbladder (GB) polyps is commonly done by transabdominal ultrasound (TAUS). However, reported sensitivities of TAUS to detect GB polyps vary greatly and diagnostic accuracy for differentiating neoplastic polyps remains suboptimal. Therefore, additional preoperative imaging modalities should be considered. We aimed to identify imaging characteristics of (different types) of GB polyps on MRI, to verify previously identified TAUS characteristics of different GB polyp types, and to assess diagnostic accuracy of both modalities.

Methods:  We performed a prospective, exploratory study including patients ≥18 years of age with GB polyp(s) identified on previous ultrasound, who were considered to undergo cholecystectomy. Preoperatively, patients underwent a dedicated, standardized TAUS examination (evaluating presence, size, number, shape, surface, echogenicity, internal structure and subtype of gallbladder polyps), and a high b-value diffusion weighted (DWI-)MRI. Characteristics of polyp subtypes on MRI (presence, size, number, shape, internal pattern, signal intensity on T1, T2 and DWI, contrast enhancement on T1, and ADC-value on DWI) were described. Histopathology (PA) after cholecystectomy was used as reference standard. Polyp characteristics on TAUS were compared to characteristics from literature. Polyp size on TAUS and MRI were compared to PA using McNemars’ test, and sensitivity and specificity for presence and neoplastic nature of polyps were calculated.

Results: In total 27 patients were included of whom 20 (74%) had a polyp on PA (14 cholesterol polyps, 3 adenomyomatosis, and 3 adenomas). On MRI, all hyper intense polyps on T1 weighted image were cholesterol polyps, and none of the hyper intense polyps on DWI were cholesterol polyps. Two out of three adenomas showed a specific intensity pattern on DWI and ADC. However, five polyps were too small to establish polyp type on MRI, and small size and artifacts prohibited optimal analysis of several polyps. Polyp characteristics on dedicated TAUS from this study and literature are shown in Table 1. Sensitivity and specificity for presence of GB polyps were 95% and 57% for both MRI and TAUS, and for neoplastic nature these were 67% and 76% for TAUS, and 67% and 83% for MRI, respectively. Polyp size on TAUS, but not MRI, significantly differed from PA (p=0.03 and p=0.32, respectively).

Conclusion: Diagnostic accuracy of dedicated TAUS and MRI is similar. In larger polyps MRI might of additional value regarding polyp size evaluation, and differentiation of polyps based on signal intensity on T1 weighted images and DWI, and ADC patterns. MRI characteristics illustrated in this study should be confirmed in a large prospective cohort study.
 

94.03 Jejunostomy Tube Placement at the Time of Pancreaticoduodenectomy

Posted on January 31, 2019

Y. Song1, A. D. Tieniber1, C. M. Vollmer1, M. K. Lee1, R. E. Roses1, D. L. Fraker1, R. R. Kelz1, G. C. Karakousis1  1Hospital Of The University Of Pennsylvania,Philadelphia, PA, USA

Introduction: Pancreaticoduodenectomy (PD) has been associated with significant morbidity, including development of postoperative pancreatic fistula (POPF) and delayed gastric emptying (DGE). Jejunostomy tube (JT) placement has been used at time of surgery to mitigate the impact of these complications. We studied a national sample to determine whether the decision to place JT is predicated upon the risk of POPF and/or DGE and to evaluate the influence of this approach on postoperative outcomes after PD.

Methods: Patients undergoing PD were identified using the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) and Procedure Targeted Pancreatectomy (2014-2015) databases. JT placement status at the time of PD and development of POPF and DGE were determined. Multivariate logistic regression analyses were used to identify pre- and intraoperative factors predicting JT placement, POPF, and DGE, as well as postoperative outcomes associated with JT placement. Statistical analyses were performed using R version 3.5.1.

Results: Of 3,600 patients who underwent PD, 322 (8.9%) underwent concurrent JT placement. The factors most strongly associated with JT placement included preoperative radiation (Odds Ratio (OR) 4.45, P<0.001), preoperative albumin < 3.5 g/dl (OR 2.43, P<0.001), and white race (OR 1.54, P=0.008). Notably, these were not associated with development of either POPF or DGE. Factors associated with POPF included soft pancreas texture (OR 2.65, P<0.001), pancreas duct size < 3mm (OR 2.15, P<0.001), and male sex (OR 1.41, P<0.001). Factors associated with DGE included preoperative admission (OR 1.56, P=0.004), preoperative bilirubin < 2mg/dl (OR 1.51, P=0.002), and male sex (OR 1.47, P<0.001). Rates of postoperative complications, including development of DGE (25.5% vs. 14.9%, P<0.001), return to the operating room (9.3% vs. 4.7%, P<0.001), and other medical complications (23.3% vs. 16.8%, P=0.004) were higher in the JT than non-JT group. In multivariate analyses, JT placement was associated with DGE (OR 1.92, P<0.001), return to the operating room (OR 1.92, P=0.004), other medical complications (OR 1.41, P=0.022), and increased length of stay (P<0.001), but not with mortality (P=0.14). Among 314 patients with low preoperative albumin (< 3g/dl), JT placement does not increase overall morbidity (P=0.88) or mortality (P=1.00). Additionally, among 967 patients who developed POPF or DGE, the development of other complications (P=0.36) and mortality (0.92) did not differ by JT status.

Conclusions: JT placement during PD is infrequent and not driven by clinical factors associated with POPF or DGE. Among patients with low preoperative nutritional parameters or who develop POPF or DGE, JT placement does not increase the immediate postoperative morbidity profile. If utilized during PD, efforts for more rational patient selection should be made.

94.02 Discharge Disposition Following HPB Surgery: Impact on Readmission Risk and Site of Readmission

Posted on January 31, 2019

A. Z. Paredes1, R. Mehta1, E. W. Beal1, F. Bagante2, A. Diaz1, K. Merath1, M. E. Dillhoff1, J. M. Cloyd1, T. M. Pawlik1  1Ohio State University,Department Of Surgery, Division Of Surgical Oncology,Columbus, OH, USA 2University of Verona,Surgery,Verona, Italy

Introduction: Hepatopancreatobiliary (HPB) surgeries are complex procedures that may require discharge to a facility other than home.  As surgical patients age, the number of patients requiring post-discharge care may increase. We sought to identify factors associated with discharge disposition, as well as assess readmission risk relative to discharge services.

Methods: The National Readmissions Database (NRD) was used to identify patients who underwent HPB surgery between 2010-2015. Patients were stratified based on discharge destination: home, SNF/intermediate care, home with home health care (HHC). Factors associated with discharge disposition were assessed and 30-day and 90-day readmission were compared.

Results: Among 47,955 patients who underwent HPB surgery, mean age was 58 years (IQR, 50-69), 53.4% were female and 48.9% had >3 comorbidities. Surgery consisted of hepatectomy (N=26,135, 54.5%), pancreatectomy (N=13,180, 27.5%) or biliary procedure (N=8,640, 18.0%).  At the time of discharge, the majority of patients were discharged to home (N=34,046, 71.0%), while a smaller proportion were discharged to SNF/intermediate care (N=3,385, 7.1%) or HHC (N=10,524, 21.9%).  On multivariable analysis, older age (OR 1.07, 95%CI 1.07-1.08), >3 comorbidities (OR 2.09, 95%CI 1.46-2.99), and a high preoperative mortality risk score (OR 3.28, 95%CI 2.32-4.62) were associated with an increased risk of discharge to SNF/intermediate care (all p<0.05). Similarly, a high preoperative mortality risk score (OR 1.43, 95%CI 1.23-1.67), and loss of function (OR 3.21, 95%CI 2.48-4.16) were associated with discharge to HHC.  While 13.7% and 15.8% of patients were readmitted following discharge to home or HCC, respectively, the incidence of 30-day readmission was 23.5% following discharge to SNF/intermediate care (p<0.001). In contrast, by 90 days the incidence of readmission was similar among patients discharged to SNF/intermediate care (34.2%) or HHC (30.5%), yet lower for patients who had been discharged home (19.1%) (p<0.001). Of note, patients discharged to SNF/intermediate care (23.1%) were at a higher odds of being readmitted to a non-index hospital (home, 13.7%, HHC, 15.8%)(p<0.001).

Conclusion: Following HPB surgery, 1 in 3 patients were either discharged with HHC or to SNF/intermediate care. Discharge with HHC or to SNF/intermediate care was associated with increased risk of both 30- and 90-day readmission, as well as readmission to a non-index hospital.

94.01 Safety of Combined Modifiers of BIRC5 Expression: Metformin-Simvastatin-Digoxin (C3)

Posted on January 31, 2019

S. F. Markowiak1, S. Liu1, J. Nemunaitis2, J. Austin3, R. Schimmoeller4, L. Hammerling4, F. C. Brunicardi1  1University Of Toledo Medical Center,Dept Of Surgery,Toledo, OH, USA 2University Of Toledo Medical Center,Dept Of Medicine,Toledo, OH, USA 3University Of Toledo Medical Center,College Of Pharmacy,Toledo, OH, USA 4Promedica Health Systems,Toledo Hospital,Toledo, OH, USA

Introduction:  We screened an FDA approved drug library using a “pro-cancer” target gene BIRC5 super-promoter and identified a novel combination of metformin, simvastatin and digoxin, which suppressed expression of BIRC5 in cancer. In vivo testing of the combination of metformin, simvastatin and digoxin (C3) showed marked response in patient-derived PDAC cell-line mouse-model.  A phase 1 protocol of the safety and response to C3 has been designed. Prior to trial initiation, safety assessment of 72 patients who received concurrent metformin, statin, and digoxin for non-cancer disorders was undertaken.

Methods: The ProMedica Health Systems database of 2.5 M patients was queried using Epic Systems® SlicerDicer for patients concurrently prescribed metformin, digoxin, and any statin.  A matched, 2:1 cohort not taking these medications was identified for comparison.  Each patient underwent extensive chart review for dosage, toxicity, adverse reactions, co-morbidities and cancer history. Safety data, including adverse reaction incidence, was compiled using Clinical Pharmacology© and the most recently published, FDA-approved package inserts. Independent samples t-test and X2 test were used.

Results: 72 consecutive patients (mean age 67-years, 44% female, 95% Caucasian) were observed for 18-months. 12.5% (n=9) expired during the observation period (n=7 cardiovascular, n=1 melanoma, n=1 unknown). The adverse drug reaction profile of the 72 patients was comparable to package insert criteria (p=0.999) and negative cohort (p=0.999) (see table 1). The average Modified Marshall Scoring System for Organ Failure of the 72 patients taking the three-drug combination was 0.7, revealing minimal toxicity.  All patients took metformin once or twice daily for diabetes at an average of 652mg per dose. Digoxin was dosed every-other-day or daily at an average of 175mcg/dose, 54% for atrial fibrillation and 46% for CHF.  For statins, 51.4% of patients took atorvastatin (n=37), 29.2% simvastatin (n=21), 16.7% pravastatin (n=12), and 2.8% lovastatin (n=2).  There was no difference in mortality among the statins (p=0.280). 13 patients (18.1%) were long-term survivors of cancer: prostate (n=3), breast (n=5), colon (n=4), and bladder (n=1). 

Conclusion: The combination of metformin, digoxin, and any statin (C3) taken concurrently over an 18-month period appears to confer no significant toxicity, as measured by Marshall Score and chart review for adverse drug reactions. Several unexpected long-term cancer survivors were identified in the cohort.  Based upon these safety and toxicity data for patients taking C3 for advanced metastatic disease, we conclude that it would be safe to proceed with a phase 1 prospective trial using C3 for BIRC5-expressing cancer therapy. 

 

93.20 Outcomes and Costs of Patients Admitted with Isolated Blunt Intracranial Hemorrhage

Posted on January 31, 2019

B. N. Cragun1, M. R. Noorbakhsh1, F. Hite Philp1, A. S. Philp1  1Allegheny General Hospital,Division Of Trauma Surgery,Pittsburgh, PA, USA

Introduction: Blunt traumatic intracranial hemorrhage (ICH) is associated with significant morbidity and mortality.  The management of these patients often includes ICU admission, neurosurgery consultation, and interval imaging.  While the outcomes for these patients are well established, less is known regarding the costs incurred.  We evaluated this patient population in order to identify outcomes and cost patterns.

Methods:  We retrospectively identified patients admitted to a single level 1 trauma center with isolated blunt ICH from February 2016 to August 2017.  We conducted chart reviews and obtained cost and payment information from the hospital finance department.

Results: We identified 556 patients in our 20 month study period, with a mortality of 12.6%.  ICH was associated with a mean total cost of index hospitalization of $13,820.  Mean cost per survivor was $15,811.  Decedents were significantly costlier than survivors ($18,100 vs $13,224, p=0.001).  Patients admitted to the ICU had a significantly higher mean total cost than patients admitted to the floor ($15,734 vs $4,692, p<0.001), and higher mean cost per day ($3,144 vs $2,192, p=0.02).  Traumatic ICH associated with oral anticoagulant use had a mean total cost of $16,638, compared with $13,969 in patients not on anticoagulants (p=0.05).  Mean total cost was significantly higher for patients that required neurosurgical intervention ($42,289 vs $10,821, p<0.001).  

Conclusion: Patients admitted to the hospital after traumatic ICH require costly care and have high mortality, leading to elevated cost per survivor.  Patients admitted to the ICU incur greater costs than patients admitted to the floor.  Traumatic ICH is particularly costly for patients who require neurosurgical intervention and for those taking oral anticoagulant medications.  Our findings provide information regarding the financial impact of this common traumatic injury.

 

93.19 Revisiting The Risk Of Intra-Abdominal Abscesses With Drain Use In Liver Trauma

Posted on January 31, 2019

A. A. Smith1, H. Mejia Morales1, R. Fabian1, J. Friedman1, C. Guidry1, P. McGrew1, R. Schroll1, C. McGinness1, J. Duchesne1  1Tulane University School Of Medicine,Surgery,New Orleans, LA, USA

Introduction:  The management of patients with liver trauma represents a wide range of clinical practices. Despite this evolution, there are very few studies that critically evaluate current practices of trauma surgeons for the management of liver trauma. Historical studies have recommended against the use of drains, but there has not been a modern investigation of this issue.  The objective of this study was to compare drain use following operative management of liver trauma. 

Methods:  A retrospective chart review of all adult patients presenting as trauma patients to a Level I trauma center from 2012-2018 was performed. The primary endpoint evaluated was post-operative intra-abdominal abscess (IAA). Univariate and multivariate analyses were performed. 

Results: 144 patients with operative management of liver traumawere included in the study. Penetrating trauma was the most common mechanism. Drains were utilized in 29.9% of post-operative patients. IAA were higher in the drain group (27.9% vs 5.9%, p=0.0006).  Drains were used in patients receiving more units of PRBCs (19.2+3.6 vs. 12.9+1.4, p=0.0492).  Drain use was found to be an independent risk factor for post-operative intra-abdominal abscess on multivariate analysis (OR 6.5, 95% CI 1.9-22.6, p=0.003).  

Conclusion: Results from this study support previous conclusions that drain use for the management of operative liver trauma is associated with an increased risk of IAA formation. Future studies should focus on the development of specific guidelines for the use of drains, especially with the evolving management of liver trauma. 

 

93.18 Increased Incidence of Pre-existing Mental Illness in Adult Trauma Patients

Posted on January 31, 2019

F. E. Scott1, D. Tatum1  1Our Lady of the Lake Regional Medical Center,Trauma Specialist Program,Baton Rouge, LA, USA

Introduction: Mental illness is a growing public health problem and an independent risk factor for unintentional injury. The CDC estimates that 26% of adults in the United States (US) will suffer mental illness this year and that half of all adults will develop at least one mental illness in their lifetime. Patients with psychiatric disorders have been shown to have higher risk of in-hospital mortality, more complicated clinical courses, and longer hospital stays compared to those without a pyschiatric disorder. Recent benchmarking data for US trauma centers revealed mental disorder incidence in trauma patients admitted to our institution to be higher than the mean reported incidence of all other reporting hospitals (14% vs 10%). We sought to identify those most at risk of unintentional injury in an adult, non-elderly population with history of mental illness.

Methods: All admissions from 2014 – 2016 in adults aged 18 – 64 years were identified in the trauma registry. Exclusions included eldery age (65+ years), dementia, prior TBI, history of substance or alcohol abuse alone, and intentional injury.

Results: Of 4270 patients included, 527 (12.4%) had a reported mental illness. Incidence increased from 11% in 2014 to 14% in 2016 (P = 0.02). Median age in the mental illness cohort was higher [44 (IQR 31 – 56) vs 36 (IQR 25 – 50); P<0.001], median hospital stay longer (5 (IQR 3 – 9) vs 4 (2 – 8); P<0.001], and Injury Severity Score (ISS) lower [9 (4 – 13) vs 9 (4 – 16), P = 0.013]. The percentage of females was significantly different between mentally ill and non-mentally ill (48% vs 24%, P<0.001). Race differed significantly (P<0.001), as the nonmentally ill group was 47% Caucasian/ 44% African American, while the mentally ill group was primarily Caucasian (73%). Mechanism of injury did not differ between the groups.

Conclusions: Incidence of patients presenting with unintentional injury in the setting of mental illness has increased significantly. Older, non-elderly adults, especially women, with mental illness are at significantly increased risk of injury by unintentional means. Longer LOS despite lower ISS may be due to difficulty in arranging appropriate discharge disposition or obtaining mental health resources, especially in Medicaid funded patients. Recognition of these risk factors should prompt targeting of injury prevention efforts and resources to the mentally ill.

 

« Previous Page
Next Page »