• Home
  • 2018 Research Competion Abstracts

2018 Kentucky Research Competition Abstracts

Scroll down to view all 2018 Abstracts

Challenges Applying the AAST Grading Scale to Acute Mesenteric Ischemia
Presenting Author: Sindall, Morgan, University of Kentucky College of Medicine

Acute mesenteric ischemia (AMI) is a highly morbid disease with a complex and diverse etiology that often leads to large healthcare costs. The American Association for the Surgery of Trauma (AAST) proposed a grading scale based on anatomic severity intended to predict outcomes based on clinical, imaging, operative and pathologic findings. This grading scale has not been yet been validated. The goal of this study was to evaluate the grading scale as an easily assigned predictor of outcomes including cost, hospital course, complication severity and mortality.

Patients for this retrospective cohort analysis were identified using diagnosis codes (ICD10-K55.0, ICD9–557.0). Inpatients >18 y/o from the years 2008-2015 were included, outpatients were excluded. AAST grade (1-5) was assigned in each category during a medical chart review. The highest overall AAST grade was recorded.
Results: 221 patients were analyzed and graded. Clinical grade 1 (abdominal pain and anorexia) was difficult to differentiate from clinical grade 2 (abdominal pain out of proportion to the exam). Clinical grades 4 and 5 had identical definitions. The difficulty with interpretation lead to a low clinical grade inter-rater reliability. Most patients were given a clinical grade 3 (n=105). There were no patients given a clinical grade 4, and few given a clinical grade 2 (n=6). The average clinical grade (3.5) was higher than the average radiologic (2.3), operative (2.9) and pathologic (2.8) grades. The clinical grade was often the determinant of the overall grade (correlation of 0.84), meaning this variability will have a large impact on the overall predictive value of the grading scale.
Most patients that underwent a bowel resection were given an operative grade 3 (n=103). This was very rarely directly described in the operative reports, which instead often described the bowel as “ischemic” or “necrotic”. It was assumed that the surgeon was examining the serosal surface and therefore any necrosis or ischemia mentioned was transmural. An operative grade 2 (mucosal ulceration) was only able to be assigned following endoscopy. This funneled a lot of patients into an operative grade 3. The grading scale would benefit from modification of the operative grading criteria to allow for a greater stratification.

In its current state, the AAST AMI grading scale is difficult to apply and will require clarification. Changes should be made to the grading scale to make it more suitable for clinical use.
Validation and Extension of the Ventral Hernia Repair Cost Prediction Model
Presenting Author: Nisiewicz, Michael John, University of Kentucky

Repair of ventral and incisional hernias (VHR) remains a costly challenge for healthcare systems. In a prior study of a single surgeon’s elective inpatient open VHR practice, a cost model was developed which predicted over 70% of hospital cost variation, and included CDC wound class, hernia defect size, age, ASA class, number of mesh pieces, and use of biologic mesh. The purpose of the current study was to evaluate the ventral hernia cost model with multiple surgeons’ elective inpatient open VHR cases and to extend to include non-elective/urgent/emergent, outpatient and laparoscopic VHR.

With IRB approval, elective and emergent cases of open and laparoscopic VHR (CPT codes 49560, 49561, 49565, 49566, 49654, 49655, 49656, 49657) performed at a single facility by multiple surgeons from October 1, 2014 to December 31, 2017 were identified. Cases in which VHR was done as a secondary procedure were excluded. Demographics, comorbidity status, ASA class, CDC wound class, length of stay, and 30 day outcomes were obtained from the local NSQIP database. Medical record review determined hernia defect size. Hospital cost data was obtained from the hospital cost accounting system. Forward multivariable regression of log transformed costs identified independent drivers of cost (p for entry < .05, for exit > .10).

Of the 387 VHRs, 74% were open repairs, 35.4% included separation of components, and 14.7% were non-elective. Mean age was 55 years, and 52% of patients were female. The base cost for an outpatient primary small VHR without mesh implantation was $4114. Including only the open, elective VHR cases, the previously reported six-factor cost model predicted 50% of the total cost variation. With all VHRs included, ten variables were found to independently drive costs, predicting 60% of the total cost variation from the base cost. Biggest cost drivers (≥ 15% increase) were preoperative open wound (+$1207), preoperative SIRS/sepsis (+$740), hernia defect size (+$616), inpatient status (+$875), use of absorbable mesh vs. synthetic or no mesh (+$752), use of biologic mesh (+$1000), and utilization of multiple mesh pieces (+$795). Other cost drivers included age, obesity, morbid obesity, and recurrent hernia.
Elective hernia repair cost variability may be predicted utilizing a six-factor model. In the broader context of all VHR repair at our institution, recurrent hernia, inpatient and non-elective surgery are greater cost drivers than wound class. Obesity, the presence of an open wound and systemic inflammation, relatively rare in the elective group due to optimization but more common in urgent/emergent cases, replaced ASA class as cost drivers.  Age, defect size, mesh type and number of meshes utilized were common to both models. A hernia cost model utilizing readily identifiable preoperative factors can be utilized to predict resource utilization

Differential Expression of RNA and the identification of enriched pathways in a
paired colon adenocarcinoma RNA-sequence dataset
Presenting Author: O'Brien, Stephen, University of Louisville

Colon adenocarcinoma is the fourth most common cancer in the United States and the third leading cause of cancer-related mortality. High throughput genomic sequencing has led to a number of significant advances in the understanding of tumor biology and the discovery of new molecular signaling pathways.

The raw RNA-sequence files for 40 patients with colon adenocarcinoma and paired normal colon epithelium samples were downloaded from The Cancer Genome Atlas. Clinical and demographic data for these patients were also downloaded. Sequencing files were aligned using STAR (Spliced Transcripts Alignment to a Reference) to the most recent genome annotation. A Binomial regression model was used to calculate the differential gene expression between the paired colon adenocarcinoma and normal colon epithelium. Differential gene expression was examined using a pathway analysis software (Qiagen® Ingenuity Pathway Analysis) in order to identify pathways which are significantly enriched in the dataset.

Using differential expression cut off values of a log fold change >2 or <-2 and a false discovery rate of <0.05, 33,514 genes were identified from the comparative analysis, of which 543 were upregulated and 1822 were downregulated. The top 20 most dysregulated genes are shown in Table 1. The most significantly enriched pathways were cAMP-mediated signaling and G-protein coupled receptor signaling.

This analysis of RNA sequence data from the Cancer Genome Atlas has identified significantly dysregulated RNA molecules which have not previously been described in human colon adenocarcinoma. The most enriched pathways are indicative of the increased metabolic activity in cancer biology. These data will be further explored in in vitro studies.
Table 1:

Gene Name

Gene Function

Log Fold Change

False Discovery Rate






Oncogene- proliferation in CRC




Oncogene- proliferation in CRC




Collagen-structural protein




Manganese ion transporter




Guanylate cyclase activator




Carbonic anhydrase 7




Carbonic anhydrase 4




Loss associated with progression of CRC




Cellular adhesion molecule




Neutrophil signaling




Guanylate cyclase activator




Peptide YY- neuro-endocrine molecule




Reduced expression inhibits differentiation in CRC




Chloride channel- tumor suppressor




Adhesion molecule




Aquaporin molecule




Carbonic anhydrase



Fibrinolysis shutdown and mortality in elderly trauma patients
Presenting Author: Quattrone, McKell, University of Kentucky

Fibrinolysis shutdown is a coagulopathy that prevents clot breakdown following injury, which can be identified using thromboelastography (TEG). Shutdown occurs at high rates within the trauma population, and is associated with high mortality rates. However, shutdown has not been studied within the elderly trauma population, specifically those with a traumatic brain injury (TBI). This study aimed to determine the relationship between incidence of shutdown coagulopathy in elderly trauma patients with TBI and its association with mortality in these patients.

Retrospective medical record review, including TEG results, of elderly (>65yo) tier 1 trauma activations from 2013–2018 at a level 1 academic trauma center. Patients were assigned groups based injury location, and data recorded for injury characteristics, hospital course, transfusion therapy, disability and mortality. Patient TEG charts were analyzed and assigned a coagulation phenotype based on the LYS30 value as follows: physiologic fibrinolysis (LYS30=1.0-2.9%), hyperfibrinolysis (LYS30>3%), or fibrinolysis shutdown (LYS30 ≤ 0.9%).

349 cases with 10 undocumented TEG results excluded, leaving 339 for TEG analysis. Of the 339 patients, 242 (71%) presented with shutdown coagulopathy, compared to 37 and 60 patients with normal and hyperfibrinolysis phenotypes for all injury locations. Four hour volumes of PRBCs, platelets, cryoprecipitate, and fresh frozen plasma were all higher than normal TEG result patients in the hyperfibrinolysis group (all p < .05) and no different from normal in the shutdown group (all p > .69). Mortality varied significantly by TEG result group for all injury locations (p<.013), with the highest mortality occurring in the hyperfibrinolysis group (Figure 1). Risk of mortality increased across all TEG groups in patients with head injuries compared to those without (Figure 2). However, TEG coagulation phenotype results did not vary between TBI and non-TBI injuries (p=.827).

A high prevalence of fibrinolysis shutdown existed in the elderly trauma population studied and in the subgroup of TBI patients; however, the prevalence of shutdown was not significantly increased in TBI patients. The increase in mortality observed in patients with coagulopathies across the entire elderly population suggests that shutdown coagulopathy has a similar effect on elderly trauma patients as previously recorded in the adult trauma population. The increase in mortality risk across all TEG phenotypes in TBI injury patients, along with the shutdown group approaching the risk of the lysis group in TBI suggests clotting may be more detrimental in head injury patients.Risk of Recurrence in Differentiated Thyroid Cancer: A Population-Based
Comparison of the 7th and 8th Editions of the American Joint Committee on
Presenting Author: Gan, Tong, University of Kentucky

Thyroid cancer incidence has steadily increased in the past decade. Differentiated thyroid cancer (DTC) survival remains excellent but requires lifelong surveillance. Historically, the AJCC 7th edition staging system has been a poor predictor of recurrence requiring adjunctive prediction models. We hypothesize that the new American Joint Committee on Cancer 8th edition improves upon the utility of the 7th edition in stratifying the risk of recurrence in DTC.
A population-based retrospective review compared the risk of recurrence in DTC patients according to the AJCC 7th and 8th editions using the Kentucky Cancer Registry data from 2004-2012. We excluded patients with metastatic disease. Kaplan-Meier plots and Cox-regression analysis were performed.

The study cohort included 3248 patients with DTC considered disease free after treatment. In 20% of cases, patients were down-staged from the 7th to the 8th edition. Most patients had stage I disease (80% in the 7th edition and 94% in the 8th edition). A total of 110 (3%) patients recurred after a median of 27 months. Risk of recurrence was significantly associated with stage for both editions (p < 0.001). In the 7th edition, there was poor differentiation of lower stages and better differentiation of higher stages (HR 0.91, 95% CI 0.39-2.11 stage II; HR 3.72, 95% CI 2.29-6.07 stage III; HR 11.66, 95% CI 7.10-19.15 stage IV; all compared with stage I). The 8th edition better differentiated lower stages but not higher stages (HR 4.06, 95% CI 2.38-6.93 stage II; HR 13.07, 95% CI 5.30-32.22 stage III; 11.88, 95% CI 3.76-37.59 stage IV; all compared with stage I).

The 7th edition had an overall better fit to the data with better differentiation between stage III and IV disease. The 8th edition was better at differentiating lower stage disease while having no clear advantage in stage III and IV disease. Limitations remain, however, emphasizing the importance of adjunctive strategies to estimate risk of recurrence.INFLUENCE OF THE OPIOID EPIDEMIC ON FIREARM VIOLENCE
Presenting Author: Dittmer, Sarah, University of Kentucky

The opioid crisis is a major public health emergency, killing more Americans than motor vehicle collisions and firearms combined. However, current data likely underestimates the full impact on mortality due to limitations in reporting and toxicology screening that have been previously described. Given the established relationship between illicit drug use and gun-related behaviors, we aimed to explore the relationship between opioid overdose ED visits (ODED) and firearm-associated ED visits (FAED).

For the years 2010 to 2017 we analyzed county-level emergency department visits in Kentucky for ODED (per 1,000) and FAED (per 10,000) using Office of Health Policy and US Census Bureau data. Additional variables analyzed included: insurance status, ethnicity, median household earnings, unemployment rate, and education level.

ODED and FAED visits were correlated (Rho = .178, p < .001) and both increased over the study period, remarkably so after 2013 (p < .001 for increase, Figure 1). FAED visits were higher in rural compared to urban counties (p < .001), while ODED visits were not (Figure 2). In multivariable analysis, FAED visits were associated with ODED visits (B= 0.17, p=.001), rural status (B = 0.33, p = .012), white race rate (B = -2.4, p = .012), and high school diploma rate (-6.45, p < .001) after adjustment for year. Unemployment and earnings were univariate correlates with FAED visits (rho = .19, p < .001 and -.15, p < .001 resp.) but were not significant in the multivariable model.

In addition to existing nonfatal consequences of the opioid crisis (e.g,. neonatal abstinence, burden on the criminal justice & foster care systems, incidence of opioid use disorder, etc.), firearm violence appears to be a corollary impact, particularly in rural counties. Future analyses should examine opioid use characteristics (e.g., prescription vs. illicit) as well as the impact of interventional models to reduce associated harm.

Figure 1. Firearm ED Visits and Overdose Per Year

Figure 2. Overdose ED Visits and Firearm Visits by County Rurality


Presenting Author: Keeven, David, University of Kentucky

To identify the mortality, morbidity and resource utilization burden of non-trauma
transfers to emergency general surgery services.
Background: Emergency General Surgery (EGS) Patients require greater resources and have
increased rates of morbidity and mortality. Previous work isolating colectomy procedures has
shown increased mortality with respect to transferring institution; Emergency Department (ED),
Inpatient (IPT), Nursing Home (NH) and Direct Admissions (DA). With length of stay, time in
need of critical care measures and duration of operation being major factors in cost we
hypothesize that “Patient transfer status negatively effects morbidity, mortality and resource

Data was obtained for patients undergoing emergency general surgery, for the top ten
procedures by volume, using public files from the American College of Surgeons National
Surgery Quality Improvement Program for years 2012 and 2016 (Table 1). We analyzed risk
factors and 30-day outcomes by transfer status using chi-square analysis and multivariable
logistic regression. Significance was set at p < .001.
Results: A total of 101,244 procedures were identified. Transferred patients had increased
clinical risk, operative complexity, and poorer outcomes. Fewer transfers were initiated for less
technically sophisticated cases such as laparoscopic appendectomy and cholecystectomy while
more complex acute open cases were transferred. Transfer patients required longer operations,
more transfusions, incurred greater critical care complications identified as; prolonged and
unplanned intubations, renal failure, treated pulmonary embolisms, sepsis, cardiac arrest, infarct,
and stroke. Transfer patients returned to the operating room more often, had higher rates of
readmission and greater 30-day mortality (Table 2). These effects remained after adjusting for
procedure group, secondary procedures, age and ASA class.

Our study demonstrates significant differences in mortality, morbidity and
resource burden of EGS transfer patients dependent upon the transferring facility that were not
attributable to case mix, age and comorbid status alone. These data point to potential financial
and quality assessment challenges of tertiary referral centers.

Table1. The distribution of the top ten procedures (by volume) performed emergently on general surgery services varied with transfer status (chi-square p < .001)

Table 2. Emergency General Surgery Outcomes by Transfer Status
1Adjusted for age, sex, ASA class, primary procedure group and secondary procedure work RVUs. 2P-value
is for chi-square test of variation in outcome proportions across transfer status groups. *Adjusted odds
ratio different than 1.0, p < .001.

Management of enteroatmospheric fistulae using patient-matched 3D-printed

Presenting Author: Warwick, James, University of Kentucky

An enteroatmosperic fistula (EAF) is a dreaded complication of open abdomen (OA) that can develop following OA surgery. An EAF is an abnormal connection between the intestinal lumen and the outside environment. The development of an EAF greatly complicates wound management and is associated with a higher mortality rate. EAF leakage of intestinal contents into the surrounding wound bed make fistula isolation the top wound care priority in these patients. However, the high variability and volatility of EAF/OA presentation make effective wound care difficult, and no optimal approach has been standardized. Many current management plans involve the use of negative pressure wound therapy (NPWT), which has been indicated to accelerate general wound healing. However, in EAF/OA patients, utilizing NPWT efficiently can be challenging. The special dressings and air tight seals NPWT requires frequently fail because of device leakage and necessitate frequent dressing changes, which are tedious and time consuming. NPWT also restricts mobility and is expensive, which limits the ability of patients to manage their wound in the outpatient setting. Thus, the need exists for an EAF/OA management system that effectively isolates EAF contents, is cost-efficient, accessible for patient use, and can be customized to meet individual patient characteristics.

This study investigates the efficacy of a 3D-printed device to isolate fistula effluent independent of negative pressure. The device would act as a shunt, diverting fistula contents away from the OA wound bed and into an anteriorly placed ostomy bag. The 3D-printing manufacturing process allows detailed customization to account for individual EAF variants such as output, location, size, and depth; this customizability is anticipated to result in more effective isolation and minimal device dislodgement. Due to the relative rarity of EAF/OA patients, this clinical investigation will be a compilation of individual case studies.

Data will be collected while patients remain admitted to the University of Kentucky Hospital (with potential for outpatient continuation), regarding dressing change frequency/time, leakage rates, patient mobility, complication rates, technology acceptance (TAM model), and patient quality of life. We will assess the efficacy of device usage as a viable EAF management technique based on its overall effectiveness, its ease of use, and its cost relative to other techniques.

This device is not approved by the FDA for commercial use. It is being studied under an abbreviated investigational device exemption (IDE) protocol. Upon IRB approval, this clinical study would be conducted and monitored solely within the University of Kentucky. Further study of this device involving other institutions or commercialized use would require more extensive FDA regulation, which may hinder the investigative process.

Incidence of and Risk Factors for Multiple Readmissions following Kidney

Presenting Author: Schucht, Jessica, University of Louisville

Unplanned readmission is increasingly being utilized as a surgical quality metric. A subset
of kidney transplant recipients undergo multiple readmissions(MR), though the incidence and risk
factors are not well described. This study was undertaken to evaluate the incidence and risk factors for
MR following kidney transplantation.

All patients undergoing deceased donor kidney transplant at a single center over a three year
period from were analyzed via retrospective chart review for factors associated with MR. P values <0.05
were considered significant.

Of 141 patients, the 30-day readmission rate was 26.2%. Multiple readmissions occurred in
43(30.5%) patients. Neither age, race, gender, initial organ function, nor dialysis vintage, were
associated with MR. Diabetic recipients were significantly more likely to have MR(44.9% vs.
22.8%;p=0.007), as were those who received basiliximab induction(47.2% vs. 25% for alemtuzumab
induction;p=0.025),those who suffered acute rejection(73.7% vs. 23.8%;p<0.001), and those who had
unplanned reoperations(60% vs. 28.2%;p=0.04). Patients whose initial readmission was beyond 30 days
were more likely to have MR versus those readmitted earlier(64.9% vs. 42.2%;p=0.041).
Diabetes(OR=5.4;p=0.02) and rejection(OR14.1;p<0.001) remained significant on multivariable analysis.
Infection was the most common reason for initial readmission in patients with MR(23.3%), followed by
immunologic reasons(14.0%). One year patient survival (90.2% vs. 98.0%;p=0.013) and death-censored
graft survival (90.0% vs. 100%;p=0.004) were significantly reduced for patients with MR.

Multiple readmissions are required for 30% of kidney transplant recipients, primarily due to
infection and immunologic causes. Recipients who with diabetes and those who have acute rejection are
at greatest risk, and warrant further studies for readmission prevention.

Unplanned Re-operation following Deceased Donor Renal Transplantation
Presenting Author: Watson, Ashley, University of Louisville

Re-operation in transplant has the potential for increased graft loss and/or graft dysfunction. The cause and timing of re-operation could signal a negative outcome as in any general surgery patient. This single center study looks at the return to the operating room as a quality measurement to determine the long-term effect on graft function in deceased donors.

A retrospective review of all DDRT performed at a single institution between 1/2015 and 11/2017 was performed. Donor and recipient demographics, indications for and timing of reoperation, and impact on length of stay and mortality were examined. Continuous and categorical variables were analyzed by Student’s t-test and Fisher’s exact test, respectively. P-values <0.05 were considered significant

There were 141 DDRT performed during the study period. A total of 10 (7.1%) patients required an unplanned return to the operating room within 90 days of kidney transplantation. The median day of reoperation post kidney transplant was 12.5(range 0-83). Reoperations were performed for immediate postoperative ultrasound findings concerning for vascular compromise, bleeding (one was from a percutaneous biopsy site, the other was from an intraoperative biopsy site). The remainder of reasons for reoperation included adhesive small bowel obstruction, acute cholecystitis with choledocholithiasis, perforated diverticulitis, hip fracture, and tracheostomy performed for post-transplant respiratory failure. Overall, 40% of reoperations were for acute general surgery indications. Four (40%) of the reoperations were unrelated to the transplant operation (the bowel obstruction was deemed transplant related as the point of obstruction was an adhesion to the transplanted kidney). There was a non-statistically significant trend towards longer length of stay in the patient undergoing reoperation (13.1 days vs. 3.9 days; p=0.061), while readmission was significantly more frequent in the reoperation group (70% vs. 35.9%; p=0.044). There was no significant difference in 90 day mortality or early graft function (as defined by serum creatinine) in patients who underwent reoperation versus those who did not. Three, six, and twelve month survival were 99.2%, 98.5%, and 96.2% for patients who did not require reoperation versus 90%, 80%, and 80% for patients who required reoperation (p=0.008).

Unplanned reoperation following kidney transplantation is a relatively uncommon event, with general surgery emergencies unrelated to the transplant itself as the most common indication. Although reoperation is associated with a higher rate of delayed graft function and longer hospitalization, there is no effect on perioperative graft loss or mortality.