31.03 Neoadjuvant Therapy Response in Esophageal Cancer Predicts Survival vs. Up-Front Esophagectomy

Posted on January 18, 2018

G. S. Chevrollier1, D. Giugliano1, F. Palazzo1, E. L. Rosato1, N. R. Evans1, A. C. Berger1  1Thomas Jefferson University,Department Of Surgery,Philadelphia, PA, USA

Introduction:
Despite advances in care, survival for patients with locally advanced esophageal cancer (EC) remains poor. It is known that patients who undergo neoadjuvant chemoradiation (nCRT) and have poor or no response (non-responders) have worse survival compared to those who respond. We sought to evaluate survival of patients who underwent nCRT compared to similarly-staged patients who underwent esophagectomy without nCRT (primary esophagectomy).

Methods:
Using our IRB-approved institutional esophagectomy database, we retrospectively identified all patients who underwent open or minimally invasive esophagectomy at our institution between January 1994 and June 2015. We separated patients into two groups: those who underwent nCRT (n=235) and those who underwent up-front esophagectomy with pathologic stage II or greater (primary esophagectomy, n=53). The neoadjuvant group was further separated into patients who were downstaged (responders, n=133) and those whose pathologic stage was the same or higher than their pre-nCRT clinical stage (non-responders, n=102). Overall and 5-year survival were compared between these three groups using Kaplan Meier survival curves and log-rank statistics, with significance set at p<0.05.

Results:
We identified 288 patients who met our inclusion criteria with an average age of 62 years. 53% of patients were male and 61% underwent open esophagectomy. The majority had adenocarcinoma (82%, n=236). Serious complication rates (grade 3 or higher according to the modified Clavien scale) were 28.4%, 26.3%, and 24.5% for non-responders, responders, and primary esophagectomy, respectively (p=NS).  The primary esophagectomy and non-responder groups had equal numbers of stage II (49% vs. 53%) and stage III (42% vs. 45%) cancers. Median survival was 36.2 months in the downstaged group (95% CI 27.2-42.6 months), 19.3 months in the non-responder group (95% CI 15-23.3 months), and 27.1 months in the primary esophagectomy group (95%CI 21.6-54.7 months) (p= 0.029). Five-year survival was 42% in the downstaged group, 25.8% in the non-responder group (HR 1.5), and 32.1% in the primary esophagectomy group (HR 1.2) (p=0.029).

Conclusions:
Patients with EC who fail to respond to nCRT have decreased survival compared to those who respond and those who undergo up-front esophagectomy. Neoadjuvant therapy in non-responders may delay definitive therapy in the form of esophagectomy, and may also expose patients to unnecessary morbidity and increased costs associated with nCRT. Further research is needed to identify potential non-responders with advanced resectable EC in order to provide more individually tailored treatment and avoid potentially harmful neoadjuvant therapy and delayed time to esophagectomy.

30.09 The Economics of Private Practice Versus Academia in Surgery: an Analysis of Sub-Specialization.

Posted on January 18, 2018

M. Baimas-George1, B. Fleischer1, J. R. Korndorffer1, D. Slakey1, C. DuCoin1  1Tulane University School Of Medicine,Surgery,New Orleans, LA, USA

Introduction:  In the surgical field, residents often make career decisions regarding future practice without adequate knowledge or exposure to the realities of professional life, particularly private practice. Currently there is a paucity of comparable data regarding the economic differences between practice models.  This study seeks to illuminate the financial disparities of surgical sub-specialties between academic and private surgical practice.

Methods:  Data was collected from the Association of American Medical College (AAMC) and the Medical Group Management Association’s (MGMA) 2015 reports of average annual salaries. Salaries were analyzed for eight comparative surgical sub-specializations, and regional data was combined for a national average. Fixed time of practice was set at 30 years. Assumptions for the calculation of lifetime revenue in academia included 5 years as assistant professor, 10 years as associate professor, and 15 years as full professor. The formula utilized is as follows: (average yearly salary) x [years of practice (30 yrs – fellowship/research yrs)] + ($50,000 x yrs of fellowship/research) = total adjusted lifetime revenue.

Results: As a full professor, academic surgeons in all sub-specialties make significantly less than their private practice counterparts. The largest discrepancy is in vascular and cardiothoracic surgery, with full professors earning 16% and 14% less than private practitioners respectively. Plastic surgery and general surgery are the only two disciplines that have similar lifetime revenues to private practitioners, earning only 2% and 6% less than their counterparts’ lifetime revenue respectively.  Surgical oncology is the only sub-specialty that regardless of practice model (academic vs private) or academic status (assistant, associate, or professor) grossed less lifetime revenue than general surgery.

Conclusion: Academic surgeons in all surgical sub-specialties examined earn less lifetime revenue compared to those in private practice.  This difference in earnings decreases but remains substantial as an academic surgeon advances from assistant to associate to full professor.  With limited exposure to the diversity of possible professional arenas, residents must be aware of this considerable discrepancy. 
 

30.10 Unnecessary Use of Plain Abdominal Radiographs in Patients of Acute Abdomen

Posted on January 18, 2018

D. Soares1, K. M. Pal1  1Aga Khan University Medical College,Surgery,Karachi, Sindh, Pakistan

Introduction:
Acute abdomen accounts for 5-10% of visits to the ER. An early and accurate diagnosis is essential in the management of these patients. Usually the first radiological investigation performed is an abdominal X-ray. However in most cases an abdominal X-ray is unable to reach a diagnosis and the patient then has to undergo further investigations. In our study, we wished to establish in how many patients presenting to the ER with acute abdominal pain was an abdominal X-ray done unnecessarily and did not lead to a final diagnosis. 

Methods:
This was a cross-sectional study conducted at the Department of Surgery at Aga Khan University Hospital over a 6 month period from April to October 2016. Patients aged 16 to 60 years of any gender, who presented to the ER with non-traumatic abdominal pain, lasting more than 2 hours and less than 5 days in duration, and which measured more than 5 on the VAS were included in the study. The patients who presented with acute abdomen and undergoing an abdominal X-ray were followed. The principal investigator then reviewed how helpful the X-ray was in the diagnosis, and calculated the proportion of X-rays that were done unnecessarily. Data was analysed using SPSS version 19. 

Results:

A total of 110 patients were included in the study.

The initial diagnosis was intestinal obstruction in 47.3% (n=52), followed by acute pancreatitis in 15.5% (n=17), peritonitis in 9.1% (n=10), constipation in 8.2% (n=9), acute cholecysitis 5.5% (n=6) and acute appendicitis in 4.5% (n=5). 

The x-ray findings included a non-specific bowel gas pattern in 50% (n= 55). Significant findings included dilated small bowel loops in 23.6% (n=26) and fecal loading in 19.1% (n=21); air fluid levels, calcific opacity in the right lumbar region, dilated large bowel loops in 1.8% respectively; and diffuse haziness in the abdomen and a foreign body in 1 patient respectively.

The most common final diagnoses were intestinal obstruction (27.3%), acute pancreatitis (14.5%) and constipation (10%). 

The proportion of unnecessary X-rays was found to be 69.1% (n=76) with only 30.9% (n=34) actually leading to a final diagnosis.

We stratified different variables on the basis of the necessity of the x-ray. The location of pain (p = 0.007), the x-ray findings (p = 0.000) and the final diagnosis (p = 0.000) was found to be significantly associated with the unnecessary use of x-rays. Abdominal x-ray was found to have some usefulness in intestinal obstruction, ureteric caluclus, foreign body and constipation in geriatric patients. It was also found that the visual analog scale had a significant association with the use of unnecessary x-rays. On further analysis, it was found that patients with a VAS of 8 and above were more likely to have an unnecessary x-ray as opposed to patients with a VAS of 6-7. 

Conclusion:

The abdominal X-ray for acute abdomen was done unnecessarily in 69% of the patients presenting with acute abdomen.

30.08 Functional Status vs. Frailty in GI Surgery: Are They Comparable in Predicting Short Term Outcomes?

Posted on January 18, 2018

S. Y. Chen1, M. Stem1, S. L. Gearhart1, B. Safar1, S. H. Fang1, J. E. Efron1  1Johns Hopkins University School Of Medicine,Department Of Surgery,Baltimore, MD, USA

Introduction:
Functional dependence and frailty are important factors in assessing preoperative risk. No studies to date have compared functional dependence with frailty as predictors of surgical outcomes. We sought to compare the impact of functional dependence and frailty on early outcomes after gastrointestinal (GI) surgery, including readmission.

Methods:
Patients who underwent GI surgery were identified using the ACS-NSQIP database (2012-2015). Functional dependence is defined by NSQIP as “partial or total assistance with performing activities of daily living (bathing, feeding, dressing, toileting, and mobility) in the 30 days prior to surgery.” The 5-item modified frailty index (mFI) consists of: history of severe chronic obstructive pulmonary disease, congestive heart failure, functional status, hypertension requiring medication, and diabetes. Propensity score matching analysis was used to separately match dependent and independent patients, and patients with mFI<3 and mFI≥3 on baseline characteristics. Multivariable logistic regression analysis was utilized. Postoperative outcomes and reasons for readmission were compared. A subgroup analysis of colectomy patients was also performed.

Results:
Of 765,082 patients, 1.71% were dependent, and 1.49% had mFI≥3. Similar outcomes were observed in matched cohorts for dependent and mFI≥3 patients: readmission (15.61% dependent; 5.75% mFI≥3), overall morbidity (37.91%; 34.81%), serious morbidity (19.06%; 17.06%), mortality (6.73%; 5.43%), and reoperation (7.01%; 6.48%). Dependent and mFI≥3 patients had similar and increased odds of outcomes on adjusted multivariable logistic analysis (TABLE) and shared three of the top five indicators for readmission: complication of surgical procedure (11.46% dependent; 11.23% mFI≥3), intestinal obstruction (10.70%; 7.65%), and organ space surgical site infection (7.93%; 8.65%). Comparable outcomes and reasons for readmission were also obtained for dependent and mFI≥ 3 colectomy patients: overall morbidity (51.14% dependent; 49.03% mFI≥ 3), serious morbidity (25.12%; 23.11%), mortality (8.83%; 8.08%), reoperation (8.60%; 7.98%), and readmission (17.79%; 17.75%) Colectomy patients shared four of the top five reasons for readmission: 1) intestinal obstruction without hernia (13.06% dependent; 9.06% mFI≥ 3 ), 2)  complications of surgical procedure (9.44%; 10.40%), 3) organ/space SSI (8.06%; 9.40%), and 4) respiratory complications (6.94%; 8.39%).

Conclusion:
Functional dependence and frailty are comparable in predicting outcomes including readmission after GI surgery. Functional dependence should be considered an acceptable and practical alternative for preoperative risk stratification in a clinical setting.
 

30.07 M&M Combined with Critique Algorithm-Based Database Reliably Evaluates Quality of Surgical Care

Posted on January 18, 2018

A. C. Antonacci1, S. Dechario1, J. Nicastro1, G. Coppa1, C. Antonacci2, M. Jarrett1  1North Shore University And Long Island Jewish Medical Center,Surgery,Manhasset, NY, USA 2Tulane University School Of Medicine,New Orleans, LA, USA

Introduction:

Collection and critique of actuarial complication data following surgery has been a historically difficult endeavor. Weekly Morbidity and Mortality conference (MMC) review combined with a standardized critique algorithm as part of a relational database can provide valuable cumulative data useful for evaluation of surgical quality.

 

Methods:

From January 2014 to July 2017,  62,377 general surgery operative procedures were performed at two major university based medical centers within our health system. We collected weekly Morbidity/Mortality reports from a total of 741 cases comprising 1714 adverse events (2.75% complication rate) and 194 deaths (0.31% mortality rate).  Approximately 250 cases were presented in detail at MMC. However, all cases were analyzed for adverse event incidence, Clavien-Dindo risk profile, error assessment (i.e., diagnostic, judgment, technical, communication and system), management and high-risk surgery.  Management evaluation was  determined by a small group of senior surgeons not involved with individual cases. Reports were reviewed at the department and provider level, and used to guide quality improvement processes.    

Results: 

The overall mortality rate for the study group was 0.31%. Yet,  the mortality rate for patients sustaining an adverse event was 25.9% (194/741), or 11.3% (194/1714) of adverse events.  Patients without mortality sustained an average of 1.7 complications per case and patients who expired sustained an average of 2.84 complications per case. There were no statistically significant differences in the management of survivors vs. non-surviviors.  Returns to the operating room (RTOR), death, intrabdominal abcess, return to interventional suite (RTIS), and hemorrhage requiring transfusion were the most common adverse events reported overall.  Technical (60%), judgment (20.1%), system (13.1%) and diagnostic (6%) errors occurred with equal frequency between both campuses. Denominator adjusted complication and mortality rates in high-risk surgical procedures  ranged from 6.5% to 23.5%, and as high as 2.8%, respectively. Over eighty-five percent (85%)  of  reported cases had Clavien Dindo scores between Grade IIIa and Grade V, confirming that  post-operative RTIS, RTOR, ICU care for systemic disease and death were important features of the complication profile.

 

Conclusion:

 Denominator adjusted morbidity and mortality rates are elevated well beyond reported overall rates. The number of complications following surgery are statistically associated with mortality, and  patients who sustain a complication  have an eleven percent (11%) risk of death. This methodology has implications not only for focused quality improvement, but for teaching a logical approach to self-assessment in the context of residency training. This project describes the feasibility of combining MMC  with a standardized critique algorithm-based database to provide accurate risk-adjusted data useful for comprehensive assessment of  surgical quality.

 

 

30.04 Geriatric Syndromes Predict the Timing of Early Postoperative Do-Not-Resuscitate (DNR) Orders

Posted on January 18, 2018

M. A. Hornor1,2, R. A. Rosenthal1,3,5, T. N. Robinson1,4,6  1American College Of Surgeons,Chicago, IL, USA 2Ohio State University Wexner Medical Center,Department Of Surgery,Columbus, OH, USA 3Yale University School Of Medicine,Department Of Surgery,New Haven, CT, USA 4University Of Colorado Anschutz Medical Campus,Department Of Surgery,Aurora, CO, USA 5Veterans Affairs Connecticut Health Care System,New Haven, CONNECTICUT, USA 6Veterans Affairs Eastern Colorado Health Care System,Aurora, COLORADO, USA

Introduction:  The timing of and risk factors for new DNR orders following surgery in older adults are not well defined. The goal of this study was to investigate the timing of new postoperative DNR orders and to determine if identifiable geriatric syndromes are associated with early postoperative DNR.

Methods:  We performed a retrospective cohort study using data from the American College of Surgeons’ (ACS) National Surgical Quality Improvement Project (NSQIP) Geriatric Surgery Pilot Project that collects an additional 20 geriatric and palliative care-specific variables at 26 hospitals.  Patients aged ≥ 65 who underwent an inpatient operation were included. The timing of postoperative DNR orders was determined and univariate and multivariate analyses were performed to examine the association between patient factors and early postoperative DNR orders, defined as a new DNR order placed on postoperative day 0-2.

Results: Of the 29,864 patients included in the study, 717 (2.4%) patients had a DNR order placed postoperatively, 329 (1.1%) of which were classified as early. Over half of the patients with early postoperative DNR’s underwent emergency surgery (58.1%). In the adjusted multivariate model, preoperatively identifiable geriatric syndromes were significantly associated with early postoperative DNR [Table 1].  

Conclusion: Early postoperative DNR orders are highly associated with preoperative geriatric syndromes and emergency operation status. The consideration of geriatric syndromes such as cognitive and functional status in shared decision making conversations prior to surgery may better inform advance care planning and surgical decision making. 

 

30.05 Influence of English Proficiency on Patient Provider Communication and Shared Decision Making

Posted on January 18, 2018

A. Z. Paredes1, J. Idrees1, E. W. Beal1, Q. Chen1, E. Cerier1, V. Okunrintemi1, G. Olsen1, S. Sun1, T. M. Pawlik1  1Ohio State University,General Surgery,Columbus, OH, USA

Introduction: The proportion of Hispanic and Asian persons in the United States is expected to increase over the next 50 years. In turn, the number of patients who speak a language other than English will also continue to increase. The effect of English proficiency on health care outcomes has been poorly studied, yet may be important. Therefore, we sought to define the impact of English proficiency on self-reported patient provider communication and shared decision-making.

Methods: The 2013-2014 Medical Expenditure Panel Survey database was utilized to identify respondents who spoke a language other than English and who had self-rated their proficiency in English. Patient provider communication (PPC) and Shared Decision Making (SDM) were characterized into three categories using a composite score that ranged from 4 to 12 (score 4-7: “poor," 8-11: “average,” and 12 “optimal”). The relationship between PPC, SDM and English proficiency was analyzed using regression analysis.

Results: 13,880 respondents spoke a language other than English and self-rated their English proficiency. Most respondents were white (n=10,281, 75%), age 18-39 years (n=6,677, 48%), male (n=7,275, 52%), middle income (n=4,125, 30%), born outside of the United States (n=9,125, 65%), and currently lived in the Western region of the United States (n=5,812, 42%). English proficiency was rated as “very well” (n=7,221, 52%), “well” (n=2,378, 17%), “not well” (n=2,820, 20%) or “not at all” (n=1,463, 10%). Among individuals who self-reported English proficiency as “not at all,” 81% had the medical interview conducted completely in the patient’s native language with or without the use of translator (“well” 38% vs. “not well” 72%  p=<0.001). On multivariable analysis, compared with “very well,” patients who self-reported English proficiency as “well” (OR 1.21, CI 1.033–1.42) or “not well” (OR 1.21, CI 1.04–1.43) were more likely to report "poor" PPC (both p<0.02). Similarly, SDM was more commonly self reported as “poor” among patients who reported English proficiency as “not well” (OR 1.31, CI 1.04–1.65, p=0.02). Compared with patients with “very well” English proficiency, individuals who reported “not at all” English proficiency had comparable PPC (OR 1.0, CI 0.82–1.23) and SDM (OR 0.96, CI 0.72–1.28) scores (p>0.05, both). Of note, the majority of patients who reported “poor” PPC had self-reported their proficiency as “well” and therefore had their interview conducted in English (n=413, 72%).

Conclusion: Decreased English proficiency was associated with worse self-reported PPC and SDM. Among patients for whom English was a second language, PPC was “poor” even among patients who reported English proficiency as “well” when the interview was conducted in English. Attention to the patient language needs is critical to patient satisfaction.

30.06 MELD Underestimates Morbidity and Mortality in Cirrhotic Patients for General Surgical Procedures.

Posted on January 18, 2018

M. Fleming1, F. Liu2,4, Y. Zhang2,3, K. Pei1  1Yale School Of Medicine,Department Of Surgery,New Haven, CT, USA 2Yale School Of Medicine,Section Of Surgical Outcomes And Epidemiology, Department Of Surgery,New Haven, CT, USA 3Yale School Of Public Health,Department Of Environmental Health Sciences,New Haven, CT, USA 4Beijing 302 Hospital,Beijing, BEIJING SHI, China

Introduction:
Ascites and the Model for End-Stage Liver Disease (MELD) score independently predict surgical morbidity and mortality. However, MELD, unlike other scoring systems for chronic liver disease such as the Child’s-Turcotte-Pugh, does not include the presence of ascites. Recently, MELD score has been shown to underestimate morbidity and mortality for cirrhotic patients undergoing colectomy for diverticulitis. We sought to ascertain whether this previously reported underprediction was generalizable to cirrhotic patients with ascites across a multitude of general surgery procedures.

Methods:
We performed an analysis of the American College of Surgeons National Surgical Quality Improvement Program (NSQIP) database from 2005 through 2014 to calculate risk adjusted morbidity and mortality of cirrhotic patients with and without ascites undergoing the following common surgical procedures including both open and laparoscopic approaches: inguinal hernia repair, adhesiolysis for small bowel obstruction, cholecystectomy for acute cholecystitis, and ventral hernia repair. Stratification was performed by MELD score and presence of ascites. Patients with and without ascites were compared within each MELD stratum (low, moderate, and high) utilizing low MELD and no ascites as a reference group.

Results:
Overall there were 30,391 procedures included of which 19,311 (63.54%) were open and 11,080 (36.46%) were laparoscopic. Compared to the low MELD strata with no ascites, each group had increased risk for complications (all p < 0.0001) and within each MELD stratum the presence of ascites portends increased risk for complications (low MELD with ascites adjusted OR 3.22 CI 2.00-5.18, moderate MELD no ascites adjusted OR 1.72 CI 1.55-1.90, moderate MELD with ascites adjusted OR 3.70 CI 2.64-5.19, high MELD without ascites adjusted OR 2.93 CI 2.53-3.39, high MELD with ascites adjusted OR 6.38 CI 4.39-9.26). The same findings hold true when evaluating mortality (all p < 0.0001, low MELD with ascites adjusted OR 9.40 CI 3.53-25.01, moderate MELD without ascites adjusted OR 3.22 CI 2.36-4.40, moderate MELD with ascites adjusted OR 15.24 CI 8.17-28.45, high MELD without ascites adjusted OR 7.01 CI 4.90-10.05, high MELD with ascites adjusted OR 28.56 CI 15.43-52.88).  These trends hold true for all 4 general surgical procedures when adjusted morbidity and mortality were analyzed by procedure.

Conclusion:
Ascites increases the risk of perioperative morbidity and mortality across a myriad general surgery procedures in chronic liver disease patients when stratified by MELD score. These findings suggest that ascites plays a critical physiologic and predictive role for surgical patients that is not incorporated into MELD. Further studies should attempt to prospectively validate a novel clinical score inclusive of ascites that may predict outcomes with better accuracy.
 

30.02 Gender Disparities in Retention and Promotion of Academic Surgeons: A Prospective National Cohort

Posted on January 18, 2018

N. Z. Wong1, J. S. Abelson1, M. Symer1, H. L. Yeo1,2  1Weill Cornell Medicine,Surgery,New York, NY, USA 2Weill Cornell Medicine,Healthcare Policy And Research,New York, NY, USA

Introduction: Women comprise 38.3% of general surgery residents in the U.S., but only 9.8% of full professors in academic general surgery. Previous studies have identified factors contributing to the underrepresentation of women in academic surgery, but no study has quantified the rates of retention and promotion of early and mid-career female academic surgeons.  As a result, we used data from the American Association of Medical Colleges (AAMC) Faculty Roster to track a national cohort of academic surgeons over time to evaluate gender disparities in retention and promotion.

Methods: Data were extracted from the AAMC Faculty Roster for all first-time appointments of full-time assistant and associate professors of surgery starting their academic careers between January 1, 2003 and December 31, 2006; these faculty were individually followed over 10 years to determine if they stayed in full time academic practice (retained) or were promoted.  Cumulative counts of retained or promoted faculty at the end of the 10-year follow up period were compared using Fisher’s exact test. The impact of gender on retention and promotion during the study period was analyzed with survival analysis by log-rank test.

Results: The analysis included retention and promotion data for 3,966 early and mid-career (assistant and associate professors) academic surgeons. Over the 10-year follow up, there were no differences in retention rates between women and men for assistant professors (50% vs. 46%, p=0.10) or associate professors (39% vs. 35%, p=0.27). Survival analysis did not demonstrate a significant difference in retention rates by gender for either academic level (assistant/associate). However, when comparing rates of promotion, women both at the assistant (29% vs 34%, p=0.02) and associate (32% vs. 42% p=0.01) level were promoted at significantly lower rates compared to their male collogues. Furthermore, 10-year survival analysis demonstrated a significant difference in promotion rates in full-time academic surgery for both assistant and associate professors (log-rank p=0.03 and p=0.03, respectively).

Conclusion: This study is the first to quantify gender disparities in retention and promotion rates among U.S. academic surgeons using a comprehensive and prospective national database. Findings suggest that academic surgery retention rates are similarly low between women and men, while promotion rates are significantly lower for women faculty. These findings demonstrate that women surgeons are at increased likelihood of non-promotion in academia, likely contributing to decreased gender diversity at the full professor level. We should consider strategies to improve retention of junior faculty (both men and women) over time.  Additional research on the relationship between gender and promotion will be critical to effectively increasing and maintaining workforce diversity.
 

30.03 Medicare's HAC Reduction Program Disproportionately Affects Minority-Serving Hospitals

Posted on January 18, 2018

C. K. Zogg1,2, J. R. Thumma2, A. M. Ryan2, J. B. Dimick2  1Yale University School Of Medicine,New Haven, CT, USA 2University Of Michigan,Center For Healthcare Outcomes And Policy,Ann Arbor, MI, USA

Introduction: In FY2015, Medicare began reducing payments to hospitals with high adverse-event rates. Termed the Hospital Acquired Condition (HAC) Reduction Program, concern has been expressed that HAC penalties could adversely affect minority-serving hospitals, leading to reductions in resources and exasperation of disparities among hospitals with the greatest need. The objective of this study was to examine the extent to which a hospital’s percentage of minority patients associates with FY2017 a) overall/domain-specific HAC scores and b) HAC penalty receipt. Differences in socioeconomic status (SES) and hospital receipt of DSH payments (a marker of safety-net status) were also assessed.

Methods:  Older adult (≥65y) inpatients presenting for eight common surgical conditions were identified using 2013-2014 100% Medicare fee-for-service claims. Records were matched to risk-adjusted FY2017 HAC scores/penalties and hospital-level data from Medicare Hospital Impact files and the AHA Annual Survey Database. Differences were compared using multilevel logistic regression and calculation of absolute percentage-point change. Restricted analyses addressed the possibility that marginal changes among the most vulnerable (likely to be penalized) institutions could be driving the differences observed.

Results: As a hospital’s percentage of minority patients increased, climbing from 1.0 to 25.1%, average HAC scores also increased, rising from 5.8 to 6.3 (higher values indicate worse scores). Increases in penalties did not follow the same stepwise increase, instead exhibiting a marked jump within the highest decile of minority-serving extent (45.7 vs 36.7%; OR[95%CI]: 1.45[1.42-1.47])—absolute difference +8.9% (Figure). Similar patterns were seen for safety-net (1.44[1.42-1.47]) and low SES-serving (1.38[1.35-1.40]) hospitals. Restricted analyses accounting for the influence of teaching status and severity of patient case-mix both accentuated differences in penalties when limiting hospitals to those at highest risk (more residents-to-beds, more severe)—absolute differences +13.9% and +20.5%. Restriction to high operative volume, in contrast, reduced the penalty difference—absolute difference +6.6%.

Conclusion: Minority-serving hospitals are being disproportionately affected by the HAC Reduction Program. While scores followed a stepwise increase, disparities in penalty allocation were isolated to hospitals with the largest minority-serving extent—a finding which became more pronounced among hospitals with an already heightened risk of penalty receipt. As the program continues to develop, efforts are needed to identify and protect patients in vulnerable institutions in order to ensure that disparities do not increase.

 

30.01 Domestic responsibilities for physician mothers across specialties.

Posted on January 18, 2018

H. G. Lyu1, R. E. Scully1, J. S. Davids2, N. Melnitchouk1  1Brigham And Women’s Hospital,Surgery,Boston, MA, USA 2University Of Massachusetts Medical School,Surgery,Worcester, MA, USA

Introduction: Despite an increase in women in medicine, there are disproportionately few in academia and leadership positions. We hypothesized that early career physician mothers face greater burnout and career dissatisfaction due to unequal domestic responsibilities.

Methods: Data from 2,360 U.S. physician mothers were gathered via an anonymous, IRB-approved online survey. Univariate analysis was performed using Chi-squared tests.

Results: The majority of respondents (97.7%) were married or partnered. Physician mothers reported having sole responsibility for the majority of domestic needs, compared to their significant other, including routine childcare plans  (56.8% vs 12.0%), back-up childcare plans (44.0% vs 26.7%), cooking (43.0% vs 25.6%), groceries (45.1% vs 24.7%), shopping for clothing (85.0% vs 2.9%), vacation planning (50.0% vs 15.6%), helping with homework (21.6% vs 3.6%), and laundry (46.6% vs 14.1%). By contrast, physician mothers reported that their significant others were more likely than them to have sole responsibility for home repairs (62.7% vs 12.0%), finances (45.7% vs 30.0%), and automobile maintenance (57.4% vs 11.3%). Compared to physician mothers whose significant other was a stay-at-home parent, those whose significant other was a physician were significantly more likely to be solely responsible for routine child care plans, (72.5%% vs 8.9%, p<0.001); this was even more pronounced for the subset with surgeon spouses, (80.8% vs 8.9%, p<0.001). Female physicians who report having an unsupportive significant other are more likely than those with supportive significant others to report burnout or desire to switch careers (8.3% vs 4.2%, p<0.001).

Conclusions: Female physicians continue to carry more domestic responsibility than their significant others, even when they are both physicians. Increased domestic responsibility correlated with having an unsupportive partner, as well as a desire to switch to a less demanding career or specialty.

29.10 Blunt Cerebrovascular Injury Incidence, Stroke-Rate, and Mortality with the Expanded Denver Criteria

Posted on January 18, 2018

A. Grigorian1, N. Kabutey1, S. Schubl1, M. Dolich1, V. Joe1, D. Elfenbein1, C. De Virgilio2, J. Nahmias1  1University Of California – Irvine,Division Of Trauma, Burns & Surgical Critical Care,Orange, CA, USA 2University Of California – Los Angeles,Harbor-UCLA Medicine Center, Department Of Surgery,Los Angeles, CA, USA

Introduction: Screening for blunt cerebrovascular injury (BCVI) (carotid artery injury [CAI] or vertebral artery injury [VAI]) requires computed tomographic angiography (CTA) of the neck. Conventional indications for screening may miss up to 20% of BCVI. Expanded indications for BCVI screening (expanded Denver criteria [eDC]) were created in 2012. The eDC includes additional signs, symptoms and risk factors to help capture that “missing” 20%. We hypothesized that the introduction of eDC would lead to an overall increase in the incidence of BCVI.  We also sought to identify risk factors for CAI and VAI, as well as whether an increased detection of BCVI would lead to a decrease in stroke rate and mortality.

Methods: The National Trauma Data Bank was queried for all blunt trauma admissions between 2007-2015. Two groups were stratified based on pre-eDC (2007-2011) or post-eDC era (2012-2015). The primary outcome was the incidence of BCVI. Secondary outcomes were stroke-rate and mortality. After a univariate logistic regression model identified significant covariates we performed a multiple logistic regression for analysis.

Results: Of the total 5,635,700 blunt trauma admissions there were 11,741 BCVIs (97.7% CAI, 2.9% VAI). The pre-eDC group was younger (median, 43.5 vs 47.1) with a higher injury severity score (mean, 10.8 vs 9.3). The post-eDC group had a higher prevalence of smokers (14.1% vs 6.1%), history of stroke (2.3% vs 1.9%) and hypertension (29.6% vs 21.3%) (all p<0.001). There were 5,085 BCVI in the pre-eDC group (0.20%), and 6,656 BCVI in the post-eDC group (0.23%) (p<0.001) translating to a 15% increase incidence of BCVI (OR 1.25, CI 1.20-1.30, p<0.001). The stroke-rate in the post-eDC was significantly higher (5.74% vs 13.49%) (OR 2.75, CI 2.40-3.15, p<0.001). There was no difference in mortality or number of patients with traumatic brain injury (p>0.05). The strongest predictors for BCVI were skull base fracture (OR 3.61, CI 3.46-3.77, p<0.001) and cervical spine fracture (OR 3.43, CI 3.29-3.57, p<0.001). The most significant independent traumatic risk factor for VAI was cervical spine fracture (OR 19.98, CI 15.85-25.19, p<0.001) while skull base fracture was the most significant for CAI (OR 3.62, CI 3.46-3.78, p<0.001). CAI was more likely to be associated with stroke than VAI in blunt trauma victims (OR 19.62, CI 18.25-21.10, p<0.001).

Conclusion: The incidence of BCVI following blunt trauma has significantly increased in the past few years.  This increase may be related to the adoption of expanded criteria for CTA of the neck in at risk patients.  Skull base fracture was the strongest traumatic risk factor for BCVI. CAI is more likely to be associated with stroke than VAI. Although detection may have increased, the stroke-rate doubled. Future research will be needed to investigate this further.

29.08 Effect of a Dedicated Pain Management Service on Trauma Patients with Rib Fractures

Posted on January 18, 2018

S. A. Bellister1, R. D. Betzold1, S. E. Nelson1, D. P. Stonko1, R. A. Guyer1, T. J. Hamilton1, J. P. Wanderer1, O. L. Gunter1, O. D. Guillamondegui1, B. M. Dennis1  1Vanderbilt University Medical Center,Division Of Trauma And Surgical Critical Care,Nashville, TN, USA

Introduction:   Rib fractures are a source of significant morbidity. Inadequate pain
control compromises respiratory function which can lead to respiratory complications and
adverse outcomes. A dedicated pain management service provides expertise in
multimodal pain management techniques which may mitigate these events. We sought to
assess the effect of a comprehensive pain service (CPS) on the outcomes of patients with
rib fractures.

Methods: A retrospective analysis on all adult patients (age ≥ 16) with more than 2 rib
fractures at a level 1 trauma center from September 2010 through December 2015 was
executed. 1:1 propensity matching was performed on the likelihood of receiving a CPS
consult. Demographic, injury data and medication use were examined. The primary
outcome was in-hospital mortality, secondary outcomes included pneumonia,
tracheostomy, 30-day ventilator-free days and 30-day ICU-free days. Mortality,
pneumonia and tracheostomy were analyzed using logistic regression, while 30-day
ventilator-free days and 30-day ICU-free days required proportional odds ordinal logistic
regression.

Results: 3,215 patients that met inclusion criteria, with a final matched cohort of 1,022
patients receiving CPS consults and 1,022 without consult. Demographics (mean age)
and injury (ISS and rib fractures) were similar in both groups. CPS consult was associated
with decreased mortality (OR 0.52, 95% CI 0.30-0.88). CPS consultation was associated
with decreased pneumonia (OR 0.58, 95% CI 0.37-0.89), tracheostomy (OR 0.54, 95%
CI 0.36-0.81), and 30-day ICU-free days (OR 0.68, 95% CI 0.58-0.80). There was
increase in 30-day ventilator-free days (OR 1.28, 95% CI 1.03-1.60) with CPS consult.

Conclusion: A comprehensive pain service consultation in rib fracture patients is
associated with a nearly 50% reduction in mortality, as well as reductions in pneumonia
and tracheostomy rates. There also is an increase in ventilator-free days in patients with
CPS consults.

 

29.09 Isolated Parafalcine Subdural Hematoma: A Clinically Insignificant Finding

Posted on January 18, 2018

B. N. Cragun1, M. R. Noorbakhsh1, F. Hite Philp1, M. F. Ditillo1, E. R. Suydam1, A. D. Murdock1  1Allegheny General Hospital,Pittsburgh, PA, USA

Introduction:
Isolated parafalcine subdural hematoma (SDH) represents a common cause of trauma admission.  Although no distinction is made with regard to location or type of bleed in the guidelines for management of SDH, parafalcine SDH may represent a distinct clinical entity with differing clinical behavior.  We hypothesize that isolated parafalcine bleeds, as compared to other SDH, were unlikely to require neurosurgical intervention and do not benefit from critical care monitoring. 

Methods:
Trauma registry data was used to identify patients presenting to a single level I trauma center with isolated intracranial hemorrhage (ICH) from February 2016 to April 2017.  Isolated ICH was defined as abbreviated injury score (AIS) of ≥3 for head and <3 for any other body location, and we further identified patients with isolated SDH.  Data reviewed included: neurosurgical interventions, radiographic worsening of the bleed, mortality, level of care, GCS on admission and discharge, disposition, and demographics.

Results:
We identified 164 isolated SDH, of which 45 had isolated parafalcine ICH.  Antiplatelet or anticoagulant use was equally prevalent in both groups (49% of parafalcine bleeds vs 54% in other SDH).  Average age was 68 ± 21.  Parafalcine SDH had a similar rate of radiographic progression (8.9% vs 11.8%), but none had neurological deterioration and none required neurosurgical intervention.  Mortality was significantly lower in parafalcine SDH as compared to other SDH (0% vs 10.1%, p=0.04).  Mortality, length of stay (LOS), neurological deterioration, radiographic worsening, and need for neurosurgical intervention are compared between the two groups as summarized in Table 1.

Conclusion:
Our data showed no mortality, no clinical worsening, and no neurosurgical intervention in parafalcine SDH, suggesting that parafalcine bleeds represent a benign entity as compared to other SDH.  While several patients with parafalcine bleeds had radiographic progression, this did not translate to neurological deterioration or need for intervention.  This data suggests that admission to an ICU, as well as interval imaging, are unlikely to be helpful in the setting of isolated parafalcine SDH.  Collectively, these data begin to build evidence for our consideration of a change in practice in the management of parafalcine SDH, as they appear to be a distinct clinical entity from other types of SDH in our trauma population. 

29.07 Trauma Recidivism and Mortality Following Violent Injuries in Young Adults

Posted on January 18, 2018

A. M. Kao1, K. A. Schlosser1, M. R. Arnold1, P. D. Colavita1, R. F. Sing2, T. Prasad1, A. E. Lincourt1, B. R. Davis1, B. T. Heniford1  1Carolinas Medical Center,Gastrointestinal And Minimally Invasive Surgery,Charlotte, NC, USA 2Carolinas Medical Center,Trauma/Critical Care,Charlotte, NC, USA

Introduction: Trauma recidivism accounts for a significant number of emergency department and trauma center admissions. Injuries associated with recurrent violent trauma result in increased treatment costs and are a significant public health burden due to higher rates of morbidity and mortality in this patient cohort.

Methods: A prospectively maintained registry of patients presenting to a Level 1 Trauma Center was queried for patients ages 18 to 25 years who sustained a gunshot wound (GSW), stab wound, or blunt assault between 2009-2015. Demographics, injury data, and discharge disposition were reviewed. Patients presenting with violent injuries were compared using Chi square, Fisher’s exact tests and Kruskal-Wallis test. Primary outcomes included mortality and trauma recidivism, identified by patients who presented with at least two unrelated violent traumas during the study period. Re-hospitalization for complications resulting from the initial injury was excluded. Out-of-hospital mortality was identified using the Social Security Death Database.

Results:A total of 6,484 patients between 18-25 years presented to the Level 1 Trauma Center; 1,215 (18.7%) had sustained a blunt assault, GSW, or stab wound. Patients with violent injuries were 87.4% male, with mean age of 22.1±2.2 years; the distribution of injuries included 64.4% GSW, 21.1% stab, and 14.8% blunt assault. Compared to patients in the same age cohort who sustained non-violent injuries, patients with violent injuries had a greater risk of mortality (8.0% vs. 2.1%, p<0.0001).  Out-of-hospital mortality was 1.3% (vs 0.46% in non-violent, p<0.0005), with average time to death of 5.2±14.6 months from initial injury. The delayed mortality was significantly more likely in patients who initially presented after a GSW (89.7% vs 5.2% stab wound or 5.2% blunt assault, p<0.0001). Recidivism was 23.5% with mean time to second violent injury at 31.9±21.0 months; 15.0% had two unrelated trauma readmissions and 6.0% had 3 or more unrelated admissions. 90% of subsequent injuries occurred within 5 years, with 19% in the first 12 months. Initial injury in recidivists was GSW in 63.3%, compared to 22.3% with initial injury of stab wound and blunt assault 22.3% (p<0.001). 59.6% of patients who returned with an unrelated, second violent injury sustained a blunt assault, followed by GSW (26.6%) and stab wound (13.7%). There was no difference in age, length of stay, initial ED vitals, or injury severity score that correlated with trauma recidivism.

Conclusion:
In young trauma patients sustaining a violent injury, the burden of injury extends past discharge as patients have a significantly higher rate of mortality after discharge.  Nearly one-quarter of patients will represent due to violent trauma.  Improved medical, psychological, and social collaborative treatment of these high-risk patients is needed to interrupt the cycle of violent injury.

29.05 Alcohol Intoxication and Burn Injury Outcomes: A Propensity Score Analysis

Posted on January 18, 2018

L. T. Knowlin1,2, B. A. Cairns1, A. G. Charles1  1University Of North Carolina At Chapel Hill,Surgery,Chapel Hill, NORTH CAROLINA, USA 2Howard University Hospital,Surgery,Washington, DISTRICT OF COLUMBIA, USA

Introduction: Approximately half  of burn-injured patients have detectable blood alcohol levels at the time of hospital admission. Alcohol use has been hypothesized to exacerbate the immunosuppression process that occurs following burn injury leading to increase morbidity and mortality. We sought to examine the effects of alcohol intoxication on burn injury outcomes.

Methods: Patients ≥14 years old admitted to a large, tertiary care referral burn center between 2002 and 2012 were eligible for inclusion. The effect of alcohol intoxication on infection complications and in-hospital mortality was evaluated using standardized Cox proportional hazard regression. Models were standardized using inverse-probability of treatment weights to account for confounding by patient demographics and burn characteristics.

Results:A total of 1,719 patients were included in the study. Of these,19% (n = 329) had blood alcohol concentration (BAC) > 0 on admission and 13% (n =221) had a blood alcohol concentration above the legal limit (> 0.08). 31% (n=548) developed an infection complications such as pneumonia, wound infectin, urinary tract infection, or sepsis. There was no difference in the crude mortality rate for patients with any positive alcohol use (11% vs 8.6%, p> 0.05) and those with a BAC above the legal limit (12.6% vs 8.6%, p>0.05)  compared to sober patients. Intoxicated patients had a higher infection rate compared to sober patients (26% vs 15%, p<0.05).  A weighted Cox regression estimated increase hazard of progression to 30 day in-hospital mortality of 12% (HR = 1.12, 95% CI = 0.63-2.00) for positive alcohol use and 64% (HR = 1.64, 95% CI 0.84- 3.22) for BAC above legal limit compared to sober patients but was not significant.The adjusted 60-day cumulative risk for infection complications in alcohol intoxicated patients with inhalation injury was the highest at 35% compared to sober patients with inhalational injury, alcohol intoxicated without inhalational injury and sober without inhalational injury (35% vs. 29.1%, 27.2%, and 22.1%, respectively).(Figure 1)

Conclusion:Alcohol intoxication did not significantly increase risk of inpatient mortality compared to sober individuals following burn injury. However, alcohol intoxication had a significantly higher cumulative risk of infection complications. The immunosuppression response after burn injury is magnified in alcohol intoxicated individuals prompting the need for future therapeutic interventions to reduce poor outcomes among burn patients who drink alcohol.

 

29.06 Nationwide Analysis of Motor Vehicle Collision Readmission and Reinjury

Posted on January 18, 2018

J. Parreco1, T. L. Zakrison1, A. D. Badilla1, R. Rattan1  1University Of Miami,Miami, FL, USA

Introduction:

No nationally representative studies of motor vehicle collision reinjury have tracked readmission across different hospitals, having been previously limited to single hospitals. The purpose of this study was to perform the first nationwide analysis of risk factors for motor vehicle collision reinjury and readmission, including to different hospitals, among non-motorcycle and motorcycle collisions.

 

Methods:

The Nationwide Readmissions Database for 2013-2014 was queried for all admissions involving motor vehicle collisions. Survey weights were used to provide national estimates. Reinjury and all-cause readmission were identified. Multivariable logistic regression was used to identify risk factors.

 

Results:

In the 399,763 patients admitted during the study period index admission mortality rate was 2.8%, 1-year readmission rate was 10.1%, and 1-year reinjury rate was 0.85%. Motorcycles were involved in 17.7% of collisions and had a higher reinjury rate (0.99% vs 0.82%, p<0.01), and had higher median admission cost ($15,014 [$8,650-$29,559] vs $12,803 [$7,356-$25,026], p<0.01). After controlling for confounding factors through multivariable logistic regression, motorcycle collision was associated with an increased risk for reinjury (OR 1.19, p<0.01) compared to non-motorcycle collisions. Risk factors for reinjury included: leaving against medical advice (OR 2.26, p<0.01), age 25-41 (OR 1.47, p<0.01), and ISS > 15 (OR 1.32, p<0.01). Protective factors included: admission to large (OR 0.75, p<0.01) or medium-sized hospitals (OR 0.82, p=0.01) and private insurance (OR 0.80, p<0.01).

 

Conclusions:

This is the first nationwide study representing the largest evaluation of readmission and reinjury after motor vehicle collision, including readmission to different hospitals. Motorcyclists have higher reinjury rates. The identified risk factors offer areas for further study in injury and reinjury prevention.

 

29.03 Keeping Your Head: Making the Case for a Hospitalist Neurosurgeon

Posted on January 18, 2018

E. Turner1, K. A. Hollenbach1, W. D’Angelo1, B. Chung1, J. Rappold1  1Maine Medical Center,Acute Care Surgery/Surgery/Tufts University School Of Medicine,Portland, MAINE, USA

Introduction: Traumatic brain injury (TBI) affects patients of all ages and genders and often results in significant morbidity and mortality. This is particularly true for patients with moderate to severe TBI (GCS 3-12) who often require emergent neurosurgical (NS) interventions (ICP monitoring, EVD, craniotomy and/or craniectomy). In rural states, this access is often limited or requires transfer to higher level of care facilities, resulting in a significant delay to intervention. To address this issue, our rural ACS verified Level I trauma center (TC) instituted a program whereby a single, dedicated neurosurgeon was available Monday through Friday for emergent NS consultations and operative procedures as indicated.

Methods: A retrospective cohort study was conducted at a rural Level I trauma center utilizing the institution’s trauma registry. Information on all trauma admissions from 1 October 2012 through 30 September 2016 with TBI were included. Standard demographics and injury related variables were abstracted. Survival by period of care (pre-hospitalist NS; 1 October 2012-30 September 2014 to post-hospitalist NS: 1 October 2014-30 September 2016) was analyzed using logistic regression to control for patient age and injury severity. Subsequent analyses were conducted by whether care was provided Monday – Friday versus Saturday – Sunday across the entire study period and for each of the two periods of interest.

Results:

A total of 7005 patients were admitted to the trauma service of which 1968 TBI patients were identified: 959 pre-hospitalist NS (PRE) and 1009 post-hospitalist NS (POST) with mortalities of 8.76% and 7.04%, respectively.  Patients were slightly older and had significantly greater ISS score in POST group. After adjusting for the confounding effect of age and ISS, POST patients were significantly less likely to die than PRE patients (OR = 0.62; 95% CI = 0.42, 0.90).  Stratification by weekend or weekday treatment identified an even stronger protective effect among patients with TBI during the weekdays when the dedicated neurosurgeon was available (OR = 0.55; 95% CI = 0.34, 0.89).    

Conclusion: This study demonstrates significant decreased mortality after instituting a designated hospitalist neurosurgeon readily available for emergent consultation and rapid operative intervention for patients with TBI. When restricted to weekday treatment, the effect was more pronounced, lending support for expanding the designated NS hospitalist role within our hospital as well as encouraging other TCs to explore the potential benefit of a dedicated hospitalist NS at their centers.

 

29.04 The Burden And Epidemiology Of Gun-related Hospital Admissions In The United States: 2003-2014

Posted on January 18, 2018

S. C. Gale1, J. Kocik1, J. S. Murry1, V. Y. Dombrovskiy1  1East Texas Medical Center,Acute Care Surgery,Tyler, TX, USA

Introduction:
Despite the important social, political, and financial implications, the nationwide annual burden and trends for gun-related hospital admissions in the US are not well described. We examined the epidemiology of all gunshot wound (GSW) related hospital admissions in the US over a 12-year period

Methods:
The Nationwide Inpatient Sample was queried (2003-2014) for GSW-related admissions using all pertinent ICD-9-CM codes. Demographics, injury circumstances, and mortality were compiled and examined. Chi square, t-test and Armitage trend test were used. P<0.05 was significant.

Results:
Of 445,884,043 US admissions over 12 years, 354,043 (0.08%) sustained GSW. Mean GSW admissions were 29504 ± 3127 annually. Notwithstanding a 2010 spike, GSW admissions trended lower over time (Figure 1) and remained stable and very small proportion of total admissions (0.076% to 0.077%; p=0.50). Over 12 years, the population rate of GSW hospitalizations per 100000 decreased from 9.7 to 8.6 (p<0.0001). Demographically, patients were most likely male (89%), African-American (49%), in the South (43%), and uninsured (p<0.0001 for all). Mean age was 31±13.  Data on circumstances (assault (64.4%), unintentional (24.4%), suicide (9.3%), and legal (2%)) demonstrate that rates for assaults decreased significantly over time, while suicide attempts, unintentional injury and legal intervention all remained stable. Mortality was 8% with a slight increase over time (p<0.04) and was highest amongst suicides (34%), followed by legal intervention (7.8%). During the study period, costs increased by 24% (p<0.0001) while length of stay remained stable at 7 days. In 2014, the uninsured rate markedly dropped while rates for Medicaid enrollment sharply rose.

Conclusion:
National data demonstrate an overall decrease in GSW-related hospitalizations over the 12-year period – most attributable to a decrease in assault-related shootings. Suicide attempts were most likely associated with in-hospital mortality. Despite high-profile events, hospitalization data do not support publicized claims of increased gun-related violence in the United States.
 

29.01 Indicators of Shock In Trauma: Anything Is Better Than Nothing, But Lactate Is Best!

Posted on January 18, 2018

A. Jordan1, W. Terzian1, T. R. Wojda3, M. S. Cohen2, J. Luster4, J. Seoane4, P. Salen2, H. Stankewicz2, E. McCarthy3, S. P. Stawicki1,3  1St. Luke’s University Health Network,Department Of Surgery,Bethlehem, PA, USA 2St. Luke’s University Health Network,Department Of Emergency Medicine,Bethlehem, PA, USA 3St. Luke’s University Health Network,Department Of Research & Innovation,Bethlehem, PA, USA 4Temple University,St. Luke’s University Hospital Campus,Bethlehem, PA, USA

Introduction: Mortality prediction in trauma continues to be challenging, with unexpected deaths continuing despite better understanding of pathophysiology and clinical management of trauma-related shock. Several laboratory variables have been evaluated for their ability to quantitate mortality risk in injured patients. Despite individual drawbacks, popular indicators of physiologic stress are serum bicarbonate (SB), anion gap (AG), base deficit (BD), and lactate. The aim of this study was to compare the utility and mortaliy prediction for each of these variables in a large, single institution trauma patient sample.

Methods: After IRB approval, we queried our Level I Trauma Center registry records for patient sex, age, ISS, GCS, mortality, and initial (trauma bay) laboratory assessments (comprehensive metabolic panel + subcomponents, arterial + venous blood gases). Main outcome variable was 30-day mortality. Analyses included the examination of stratified AG (≤3, 6, 9, etc), BD (≥16, 12, 8, etc), SB (≤10, 14, 18, etc) and lactate (≤1, 2, 3, etc) versus 30-day mortality (adjusted for sex, age, and ISS). Additional comparisons evaluated the ability of each of the above variables to predict mortality using receiver operating characteristic (ROC) curves (DeLong method). Data are reported as mean±standard deviation (SD) or median with interquartile range (IQR). AUC values are reported as area±standard error (SE). Statistical significance was set at α<0.01.

Results: The study sample included 2,811 patients (70% male; median age 44 yrs with IQR 26-58 yrs, median ISS 9 with IQR 4-16, and 5% mortality). Available laboratory values included: mean serum lactate 2.83±2.51 (n=371), mean BD 1.27±5.01 (n=1,167), mean SB 24.8±5.29 (n=2,165), and AG 11.2±6.80 (n=2,128). Mortality increased with escalating physiologic stress, as reflected by each indicator corrected for age, sex, and ISS (Fig 1; all p<0.001). Overall, serum lactate was the best predictor for mortality (AUC, 0.75±0.04SE) followed by BD (0.724±0.03), SB (0.679±0.03) and AG (0.661±0.03). Combinations of the above parameters did not improve mortality prediction.

Conclusion: Although all of the variables examined in this study offer predictive value for trauma-related mortality, initial serum lactate and BD are superior to serum bicarbonate or AG. Initial serum lactates ≥3 are associated with doubling of mortality, while lactates ≥7 carry more than quadruple baseline mortality. For BD, mortality increases from <5% for BD <4 to >40% for BD >16. In the absence of lactate or BD assessments, serum bicarbonate and AG may be helpful in crude mortality risk stratification.

« Previous Page
Next Page »