53.05 Hospital Competition and Adoption of New Procedures in Low Volume Settings

Posted on December 22, 2014

H. L. Hill1, T. R. Grenda1,2, J. R. Thumma2, J. B. Dimick1,2, C. P. Scally1,2  1University Of Michigan,Department Of Surgery,Ann Arbor, MI, USA 2University Of Michigan,Center For Healthcare Outcomes And Policy,Ann Arbor, MI, USA

Introduction:  Concerns have been raised regarding the effect of market influences on the adoption of new surgical procedures. In an effort to identify non-clinical factors driving utilization of laparoscopic colectomy, we evaluated the impact of hospital market competition on utilization, focusing on low volume providers where premature adoption may threaten patient outcomes.

Methods:  We used Medicare claims data (2009-2012) to identify patients undergoing elective colectomy for cancer resection (laparoscopic n = 60,261, open n = 104,386). Patients were assigned to Core Based Statistical Areas (CBSA) in which they were treated, each CBSA representing a market area. Competition by CBSA was quantified using an econometric measure of market saturation, the Hirschman-Herfindahl index (HHI). CBSAs were stratified into HHI tertiles and the proportion of laparoscopic colectomies performed was evaluated across tertiles. We further stratified the cohort by surgeon volume to specifically analyze the impact of inter-hospital competition on low volume providers. 

Results: A total of 36% (n = 164,647) of colon resections from 2009 to 2012 were performed laparoscopically. We identified no trend toward increasing utilization across tertiles of increasing competition with 33% of colon resections performed laparoscopically in high competition settings compared to 36% in mid-competition settings and 31% in low-competition settings. Moreover, our findings identified no association between increasing hospital competition and adoption by low volume surgeons. The proportion of laparoscopic procedures performed within the low volume group was similar across competition levels at 15%, 18% and 18% respectively for markets of high-, mid- and low-competition.

Conclusion: These findings indicate that utilization of a laparoscopic approach for colon cancer resection was not highly sensitive to market pressures. Furthermore, market forces did not affect colectomy procedure choice in low volume settings.

 

53.06 The Result of Surgeon Aging on Patient Selection and Surgical Outcomes

Posted on December 22, 2014

T. E. Newhook1, C. A. Guidry1, F. Turrentine1, R. Sawyer1, R. S. Jones1  1University Of Virginia,Surgery,Charlottesville, VA, USA

Introduction:
Studies suggest that increasing surgeon age worsens surgical outcomes following selected highly complex procedures. The influence of surgeon age outcomes in a diverse population of surgical patients remains uncertain. For that reason, we aimed to investigate the influence of surgeon time from initial board certification on a population of general and vascular surgery patients.  

Methods:
We used the American College of Surgeons National Surgical Quality Improvement Program (NSQIP) database established at a tertiary care university hospital over a period of 10 years. The risk of NSQIP-defined complications or mortality was estimated for each patient by multivariable logistic regression utilizing variables identified by the NSQIP Risk Calculator, as well as albumin and operative year. Years of experience after American Board of Surgery (ABS) certification were calculated for each surgeon at the time of each operation. Multivariate linear regression with mixed effects modeling was used to determine the association between years of surgical practice and pre-operative risk of complications and mortality.  Additional multivariate logistic regression was also used to evaluate the influence of surgeon years post-board certification on risk-adjusted patient outcomes. Statistical significance was set at a p < 0.01. 

Results:
A total of 21,985 operations were included during the study period. Models for predicted complication and mortality rates performed well with C- statistics of 0.78 to 0.89. Years elapsed from initial ABS certification was predicted of both risk of any complication, as well as 30-day mortality (p < 0.001). There was a reduced operative risk as years from initial ABS certification increased. Moreover, an inverse correlation was observed between years after board certification and both predicted risk for any complication and 30-day mortality (Fig. 1). Importantly, after adjusting for risk, no association was observed between years from board certification and any complication or mortality.

Conclusion:
As surgeons aged they operated on patients having progressively fewer pre-operative risks for morbidity and mortality. There was no evidence that aging of surgeons resulted in increased morbidity or mortality. 
 

53.07 Rethinking Regionalization for Pyloromyotomy

Posted on December 22, 2014

S. Selvarajah1, E. B. Schneider1, E. R. Hammond1, M. Arafeh1, H. N. Alshaikh1, N. Nagarajan1, F. Gani1, H. Alturki1, C. K. Zogg1, A. Najafian1, A. H. Haider1  1Johns Hopkins University School Of Medicine,Baltimore, MD, USA

Introduction: Hypertrophic pyloric stenosis (HPS) affects 1-3 per 1,000 live births in the United States and is classically managed with pyloromyotomy. It has been proposed that complications occur less frequently when pyloromyotomy is performed at specialized children’s hospitals (CH), supporting the potential need for regionalized care. However, the majority of infants with HPS who undergo pyloromyotomy are admitted emergently, making regionalization arduous. We sought to evaluate the association between hospital type and in-hospital complications and cost of care after emergent pyloromyotomy among infants in a nationwide sample.

Methods: The 2006 and 2009 Kids Inpatient Database was utilized to identify all infants (age ≤1 year) admitted emergently for HPS that underwent pyloromyotomy. Weighted descriptive analysis was performed comparing the occurrence of post-operative complications at CH, general hospitals with children’s units (GHCU) and general hospitals without children’s units (GHNCU). Post-operative complications included hematoma, hemorrhage, shock, seroma, wound infection, and wound dehiscence. Cost per admission was calculated by multiplying hospital-specific cost-to-charge ratios with patient charges. Gamma regression was performed to estimate adjusted average cost per admission (in 2009 USD$), controlling for patient and hospital characteristics, and length of stay (LOS).

Results:Of the 18,703 infants who met inclusion criteria, 36.7% (n=6,858) were operated on in CH, 34.7% (n=6,487) in GHCU, and 28.7% (n=5,358) in GHNCU. The majority of infants were male (83.0%) and had government insurance (62.9%). Infants managed at GHNCU (31.5%) were more likely to have dehydration or metabolic derangement compared with infants at CH (24.3%, P=0.011) and GHCU (26.0%, P=0.006). However, infants at CH were more likely to undergo surgery on the day of admission compared with infants at GHNCU and GHCU (42.1 vs. 27.9% and 27.2% respectively, P<0.001 both). There was no significant difference in the occurrence of post-operative complications, as well as LOS across hospital types (Table). However, cost of care was $1,951 and $2,177 greater at CH compared with GHCU and GHNCU respectively. 

Conclusion:The cost of managing infants admitted emergently for pyloromyotomy is substantially greater at specialized children’s hospitals. Further research is necessary to determine whether regionalizing care for common pediatric emergency procedures, such as pyloromyotomy, is necessary and cost-effective.

 

53.08 Assessing the association between insurance and management of pediatric blunt splenic injuries

Posted on December 22, 2014

M. Cerullo1, M. Michailidou1,2, Z. G. Hashmi1, A. T. Lwin1, E. B. Schneider1, A. H. Haider1  1Johns Hopkins University School Of Medicine,Baltimore, MD, USA 2University Of Arizona,Tucson, AZ, USA

Introduction:  Prior research has demonstrated the efficacy and safety of non-operative management for splenic injuries in children. In addition, repeated studies have documented an association between insurance status and hospital course. Though every state has set up its own State Children’s Health Insurance Program (SCHIP), enrollment rates have varied. Our study aims to determine how enrollment influences the effect of insurance status on in-hospital management of pediatric splenic injury (PSI).

Methods:  PSI surgical intervention (splenectomy and splenorrhaphy) was evaluated as the primary outcome measure. Enrollment into SCHIP from 1999 to 2011 was calculated using state-level data from the US Census Bureau’s Current Population Survey. The Savitzky-Golay smoothing algorithm was applied to yearly ratios of uninsured to publicly insured children to calculate the rate of enrollment in SCHIP by state. To evaluate the effect of insurance status on interventions, discharge records of patients aged less than 18 years with ICD-9 codes for non-penetrating splenic injury (865.-) from the Kids’ Inpatient Database for 2009 were used. Design weights were applied at the discharge level to produce national-level estimates. Logistic regression was used to evaluate the effect of insurance status on management of PSI after adjusting for confounders including age, sex, race, new injury severity score (NISS), income quartile, abdominal anatomic injury severity (AIS), and hospital characteristics (e.g. urbanicity, teaching status, management, children’s hospital designation). A hierarchical model was then constructed to evaluate the latent effect of enrollment rates on the effect of insurance status. 

Results: A total of 2843 patients with pediatric splenic injury in 39 states met inclusion criteria. Mean age of children with PSI was 13.4 years (SE: 4.4). Most were male (71.4%), white (76.5%), above median income (52.9%), and treated in urban teaching hospitals (72.0%). Adjusted odds of mortality was higher in children who underwent operative management compared to children who underwent non-operative management (OR=7.96, 95% CI: 4.51-14.07). Lack of insurance was not associated with mortality, (OR=1.37, 95% CI: 0.62-3.05), though uninsured children had 1.4 (95% CI: 1.16-2.91) times greater adjusted odds of undergoing operative management compared to insured children, and they continued to demonstrate greater odds of undergoing operative management even when clustering by state was controlled for in the hierarchical model (OR: 1.83; 95% CI: 1.09-3.10). Faster state-specific enrollment rate resulted in lower odds of operative management (p<0.05). 

Conclusion: Uninsured children are more likely than insured children to receive operative management for blunt splenic injury, regardless of their state’s rate of enrollment into SCHIP. However, children in faster enrolling states are less likely to undergo operative management overall, highlighting the importance of measures aimed at increasing insurance uptake.

 

53.09 Methodological Considerations in Outcomes Research: Timing of Surgery

Posted on December 22, 2014

K. D. Simmons1, L. E. Kuo1, R. L. Hoffman1, E. K. Bartlett1, D. N. Holena1, R. R. Kelz1  1University Of Pennsylvania,Department Of Surgery,Philadelphia, PA, USA

Introduction:

Outcomes research methodology is being used to rank hospital quality.  The usability of this information by patients and providers is dependent upon appropriate risk adjustment and transparency of the statistical methods.   For surgical outcomes, the timing of the procedure relative to the admission date is seldom considered or described in methods sections.  As surgery performed during an admission for another reason is often high-risk, this may be a source of unmeasured confounding, thereby limiting the utility of the ranking data.  We aimed to examine the association between surgical timing and outcomes as well as the subsequent effect on hospital rankings in a model of mortality following colectomy.

Methods:

Claims data from three state databases (California, Florida, and New York) were used to analyze in-hospital mortality following colectomy.  Two cohorts were identified using different selection criteria:  Cohort 1 included all patients who underwent a colectomy at any time during the admission; Cohort 2 was restricted to patients who underwent a colectomy on the day of admission only. Logistic regression was used in each cohort to adjust for patient and hospital characteristics associated with in-patient mortality and to estimate risk-adjusted mortality rates.  Bootstrapping was performed to control for differences in sample size.  Model performance was evaluated using the C statistic and the Hosmer-Lemeshow goodness-of-fit test.  Hospitals were ranked by observed-to-expected (O/E) mortality ratio in each cohort.

Results:

Cohort 1 (all colectomies) included 209,515 colectomies, of which 118,387 (56.5%) were included in Cohort 2 (day of admission only). Overall mortality rates were 6.3% for Cohort 1 and 3.2% for Cohort 2.  Models fit to mortality rates in each cohort fit the data well (Hosmer-Lemeshow statistics 105.0 in Cohort 1 and 37.6 in Cohort 2, p<10^-5 for each; C statistics 0.794 in Cohort 1 and 0.834 in Cohort 2). 

Using the resulting risk-adjusted mortality rates, we ranked 675 hospitals by O/E ratio in each cohort.  The two ranking systems had a correlation coefficient of 0.63 (p<.0001).  Nevertheless, 391 hospitals (57.9%) changed rank by at least 10% between the rankings, and 370 hospitals (54.8%) were in different quintiles when ranked by Cohort 1 versus Cohort 2.

Bootstrapping confirmed that the difference in risk-adjusted mortality between Cohort 1 and Cohort 2 was not due to the smaller number of patients in Cohort 2.

Conclusion:

Time between admission and surgery can influence the likelihood of death following colectomy and thereby alter hospital rankings in the absence of real differences in the care provided.  For maximally effective quality improvement, it is important that hospitals have information regarding the cohort of patients used to derive the ranking.  Moreover, given the influence of surgical timing on outcomes, its use in patient selection should be delineated within the methods sections of any subsequent reports.

53.10 Efficacy of a Multifaceted Program to Reduce Readmission after Cardiac Surgery

Posted on December 22, 2014

I. E. McElroy1, H. Wu1, E. Gee1, N. Satou1, R. Shemin1, P. Benharash1  1University Of California – Los Angeles,Cardiac Surgery,Los Angeles, CA, USA

Introduction:
Readmission rates after surgery have become commonly-used surrogate markers for quality of care. The Affordable Care Act mandates Medicare to penalize hospitals with “excessive” readmissions for diagnoses including myocardial infarction and heart failure. Beginning in 2015, the penalty will increase to three percent and will encompass more surgical diagnoses including coronary artery bypass grafting (CABG). Because cardiac surgical patients are at a higher risk of readmission due to comorbidities as well as surgical complications, we hypothesized that a multifaceted readmissions reduction initiative (RRI) would reduce the rate of unplanned hospitalization at our institution.

Methods:
Starting in March 2014, we implemented a RRP consisting of post-discharge follow-up phone calls, formal education about postoperative care, and coordination of post-discharge care by creation of appointments before discharge. Using our institutional Society of Thoracic Surgeon’s database, we compared readmission rates between two cohorts: patients discharged prior to institution of the program (NRPP) and those discharged after (RPP).  Patients in the RRP group were further categorized as high risk for readmission if they had greater than four risks factors identified by the California CABG Outcomes Reporting Program. High risk patients were contacted via telephone within 72 hours of discharge. If the calls lead to an intervention that prevented a readmission, the action was recorded.

Results:
During the study period, a total of 164 (NRPP=51) CABG operations were performed with successful discharge, 25 (15%) of whom were readmitted within 30-days. The baseline characteristics (Table 1) and readmission rates did not significantly differ between the groups (RPP 15.6%, NRPP 15%, P>0.5). Of the 113 RPP patients, 65 (57%) were identified as high risk and received follow-up phone calls (66% successful calls), leading to seven interventions and five readmission preventions. Despite a significantly different calculated risk, no difference in readmission was seen between the high and normal risk subgroup (high=12%, normal=13%, P=0.93).

Conclusion:
We postulated that a program that would educate patients about postoperative care, coordination of post-discharge care, and targeted phone calls after discharge would reduce the rates of readmission after CABG.  While the RRI was successful in reducing readmission rates, addition of follow-up phone calls did not seem to significantly affect outcome. Major barriers including staffing and inability to communicate with patients existed and may have contributed to the apparent lack of efficacy. Healthcare organizations should carefully consider the true benefits of individual readmission reduction programs before permanent implantation. 
 

54.01 Readmission Destination and Risk of Mortality Following Major Surgery

Posted on December 22, 2014

B. S. Brooke1, P. P. Goodney3,4, L. W. Kraiss1, D. J. Gottlieb4, S. R. Finlayson1  1University Of Utah,Department Of Surgery,Salt Lake City, UT, USA 3Dartmouth-Hitchcock Medical Center,Department Of Surgery,Lebanon, NH, USA 4The Dartmouth Institute For Health Policy & Clinical Practice,Lebanon, NH, USA

Introduction:  Readmissions are common following major surgery, however, it is unknown whether the readmission destination (either the index hospital where surgery was performed, or a different hospital) affects mortality risk in the period following surgery.  The objective of this study was to determine if patients acheive better outcomes if they are readmitted to the index hospital where their major surgical procedure was performed.  

Methods:  We performed a retrospective cohort study among Medicare patients who required hospital readmission within 30-days following 12 major surgical procedures (open abdominal aortic aneurysm repair, infra-inguinal arterial bypass, aorto-bifemoral bypass, coronary artery bypass surgery, esophagectomy, colectomy, pancreatectomy, cholecystectomy, ventral hernia repair, craniotomy, hip replacement, and knee replacement) at nationwide acute-care and critical access hospitals between 2001 and 2011.  Readmission destination was categorized as to the index hospital (including readmission to a different hospital and transfer back to the index hospital within 24 hours) or readmissions to a non-index hospital.  We used generalized estimating equations with inverse probability weighting to determine associations between readmission destination and risk of 90-day mortality among patients requiring readmission following major surgical procedures.  

Results:  The proportion of patients readmitted or transferred to the index hospital where their operation occurred varied from 66% to 83% following all major surgeries, and was more likely to occur if the readmission was for a surgical versus a medical complication (P<.001).  When compared to patients readmitted to non-index hospitals, risk-adjusted 90-day mortality was significantly lower for patients who returned to the index hospital where their surgical procedure was performed.  This effect was significant (p<.001) for all procedures (figure), and was largest for patients readmitted after pancreatectomy (OR 0.58; 95% CI:0.47-0.68) and aorto-bifemoral bypass (OR 0.69: 95% CI:0.61-0.77).  The mortality benefits associated returning to the index hospital was evident for both medical and surgical readmissions.

Conclusion:  Patients readmitted following a wide range of major operations consistently achieve improved survival if they return to the hospital where their index operation took place.  These findings may have important implications for value-based regionalization of surgical care.

 

54.02 Insurance Status Influences Emergent Transfer Designation in Emergency Surgical Transfers

Posted on December 22, 2014

K. L. Kummerow1, S. Phillips1, R. M. Hayes1, J. M. Ehrenfeld1, M. D. Holzman1, K. Sharp1, R. Pierce1, W. Nealon1, B. K. Poulose1  1Vanderbilt University Medical Center,General Surgery,Nashville, TN, USA

Introduction:  There is a persistent perception that hospitals disproportionately transfer unfavorably insured patients to referral centers for emergency surgical care.  Given both recent changes in payer mix and decreased federal subsidization of referral centers under the Patient Protection and Affordable Care Act, we sought to understand whether unfavorably insured patients are more likely to be designated as emergent transfers. 

Methods:  A retrospective cohort study was performed of patient transfers from acute care facilities to a general, thoracic, urologic, or vascular surgery service at a tertiary referral center from 2011-2013.  Individuals insured by a commercial, Medicare, or federal (VA/Tricare) payer were defined as having favorable insurance.  Unfavorable insurance included Medicaid and uninsured.  The primary outcome of referring provider transfer designation as emergent versus non-emergent was evaluated using chi-squared test.  A multivariable logistic regression model was created to adjust for patient demographics, Elixhauser Comorbidity Score, Acute Physiology Score, and reason for transfer.  Intensity of pre-transfer care, measured by the proportion of patients in each group that underwent any procedure at the referring hospital prior to transfer, was evaluated as a secondary outcome. 

Results: The study cohort included 1,253 patient transfers.  Eighty-three percent were favorably insured while 17% had unfavorable insurance.  Favorably insured patients were older (mean age 60 years vs 44, p<0.01) with fewer non-white patients (8% vs 14%, p<0.01) and higher Elixhauser Scores (median 5 (interquartile range 0-13) vs 3 (0-7), p<0.01).  More favorably insured patients were transferred for continuity of care (27% vs 16%) and had undergone prior related procedures at the referral center (20% vs 13%, p=0.03).  Acute Physiology Scores did not differ in the groups (median 3 (2-5) in favorable, 4 (2-6) in unfavorable, p=0.22).  More unfavorably insured patients were designated as emergency transfers (72% vs 59%, p<0.01).  The association between unfavorable insurance and emergency transfer designation persisted after adjustment for demographics, comorbidities, acute physiology, and reason for transfer (odds ratio 1.7, 95% CI 1.2-2.5).  There was no difference in the proportion of patients that underwent a procedure prior to transfer (17% in favorable insurance group and 12% in unfavorable group, p=0.08).

Conclusion: The data demonstrate that unfavorably insured patients are more likely to be designated as emergent at the time of transfer request.  The difference is not related to comorbid conditions or severity of illness.  Transfer processes may be improved through clearer definitions of emergent transfer. 

 

54.03 Preliminary Impact of 2011 ACGME Duty Hour Regulations on Surgical Outcomes

Posted on December 22, 2014

C. Scally1, A. Ryan2, J. Thumma1, P. Gauger1, J. Dimick1  1University Of Michigan,Department Of Surgery,Ann Arbor, MI, USA 2Weill Cornell Medical College,Healthcare Policy & Research,New York, NY, USA

Introduction:  In July 2011, the Accreditation Council for Graduate Medical Education (ACGME) implemented an additional set of restrictions on resident work hours. The implications of these regulations for the education of surgical trainees has been a subject of significant scrutiny, but the effect on patient outcomes is unknown. 

Methods:  We used national Medicare claims data for patients undergoing common general (n = 1, 223, 815) and vascular (n = 475, 262) surgery procedures in two time periods, the 3 years immediately preceding the duty hour changes (January 2009 – June, 2011), and the 18 months following (July 2011 – December 2012). Hospitals were stratified into quintiles of teaching intensity using a resident-to-bed ratio. To account for other changes in surgical outcomes over this period, we used a control group of non-teaching hospitals. We then utilized a difference-in-differences analytic technique to compare risk adjusted 30-day mortality, morbidity, reoperation, and readmission rates prior to and following the duty hour changes.

Results: Following duty hour reform, no significant changes were seen in any of our measured outcomes when comparing teaching to non-teaching hospitals. Even when hospitals were stratified by teaching intensity there were no significant differences. For example, at the highest intensity teaching hospitals (resident to bed ratio ≥ .6), mortality rates before and after the duty hour changes were 4.2% and 4.0% respectively, compared to 4.7% and 4.4% for non-teaching hospitals (OR .98, 95% CI .89-1.07). Similarly, complication (OR 1.01, 95% CI .97-1.06), reoperation (OR 1.01, 95% CI .80-1.26), and readmission (OR .99, 95% CI .95-1.04) rates were unchanged between the two time periods.

Conclusion: In the immediate period following the 2011 ACGME duty hour changes, no significant improvement in outcomes was seen among Medicare beneficiaries undergoing common general and vascular surgery operations at teaching hospitals. 

 

54.04 Discordance between Perceived and Measured Frailty

Posted on December 22, 2014

N. Gupta1, M. L. Salter1,2,3, A. Massie1, M. A. McAdams-Demarco1,2, A. H. Law2, B. G. Jaar2, J. D. Walston3, D. L. Segev1,2  1Johns Hopkins University School Of Medicine,Department Of Surgery,Baltimore, MD, USA 2Johns Hopkins University School Of Public Health,Department Of Epidemiology,Baltimore, MD, USA 3Johns Hopkins University,Center On Aging And Health,Baltimore, MD, USA

Introduction:  Frailty, a novel domain of risk originally validated in the geriatrics literature, captures poor physiologic reserve. It is a construct that is independent of comorbidity and disability, and it has been shown to be an independent predictor of poor outcomes in the general surgery, transplant, and end-stage renal disease (ESRD) populations. However, frailty is not routinely measured in the clinical setting; instead, we rely on perceptions of frailty to influence clinical decision-making. Whether perceived frailty accurately reflects measured frailty remains unknown. 

Methods:  We conducted a cross-sectional study of 145 ESRD patients at a single dialysis center in Baltimore, Maryland. We used the Linda Fried criteria of shrinking, weakness, exhaustion, low physical activity, and slow walking speed to measure frailty and classify patients as non-frail, intermediately frail, and frail. We measured provider perceptions of frailty by informing providers (nephrologists and nurse practitioners [NPs]) of the components of frailty and asking them to rate their patients as non-frail, intermediately frail, or frail; we measured patient perceptions of frailty by asking patients to rate their own frailty on this categorical scale. Weighted kappa statistic was used to assess the relationship between measured and perceived frailty. Patient characteristics influencing measured or perceived frailty were evaluated using ordered logistic regression.

Results: Among non-frail participants, 34.4%, 30.0%, and 31.6% were misperceived as intermediately frail or frail by a nephrologist, NP, and themselves. The agreement between measured and perceived frailty was, at best, only slightly better than what would be expected by chance alone (nephrologists: kappa = 0.24; NPs: kappa = 0.27; patients: kappa = 0.07). Older age (adjusted OR [aOR]=1.36, 95% CI: 1.11-1.68, p=0.003 per 5-year increase in age) and greater comorbidity (aOR=1.49, 95% CI: 1.27-1.75, p<0.001 per one additional comorbidity) were associated with greater likelihood of being perceived as frail by a nephrologist. Being non-African American was associated with greater likelihood of being perceived as frail by an NP (aOR=5.51, 95% CI: 3.21-9.48, p=0.003) and by the patient themselves (aOR=4.20, 95% CI: 1.61-10.9, p=0.003). Obesity was associated with less likelihood of being perceived as frail by nephrologists (aOR 0.21, 95% CI: 0.16-0.29, p<0.001) and NPs (aOR 0.44, 95% CI: 0.27-0.72, p=0.001). Disability was the only patient characteristic associated with measured frailty (aOR = 1.47, 95% CI: 1.04-2.08, p=0.029 for each additional ADL difficulty). 

Conclusion: Perceived frailty is an inadequate proxy for measured frailty. Patient characteristics associated with perceived frailty differ by rater and are different than those associated with measured frailty. If frailty is to be used to assess risk for adverse outcomes, quantitative frailty assessments rather than perceived frailty should be considered.
 

54.06 Overtriage Rates Continue to Burden a Mature Trauma Center

Posted on December 22, 2014

M. Soult1, J. N. Collins1, T. J. Novosel1, L. D. Britt1, L. J. Weireter1  1Eastern Virginia Medical School,Norfolk, VA, USA

Introduction:  As the number of emergency department (ED) visits continues to increase, trauma systems can become a resource for busy EDs to allow for the rapid evaluation of patients.  Over time at a mature trauma center, the rate of trauma alerts being discharged home from the trauma bay continued to rise.  The objective of this study was to investigate the cause of the increased over triage rate and determine if the over triage rate could be reduced by adherence to activation protocols.

Methods:  A retrospective chart review was performed all patients who presented as a trauma alert to a single level I trauma center from January 2000 to February 2014.  Full activation was designated as an alpha alert, while partial activation was a bravo alert.  Using the Cribari method for over and under triage, yearly rates were calculated.  Additionally, discharge rates as a function of activation level were calculated.

Results: During the study period trauma activation was required for 25,348 patients.  From 2000-2006 there was an average of 1710 activations and the over triage rate was 63%.  The emergency department (ED) underwent expansion that was completed near the end of 2006.  From 2007-2012 there was an average of 1804 activations and the over triage rate of 74% (p<0.05).  Discharge rates mirrored the over triage rate with 14% of alpha and 41% of bravo alerts were discharged from the trauma bay from 2000-2006 and 22% of alpha and 50% of bravo alerts were discharged from the trauma bay from 2007-2012 (p<0.05).  In order to assess if adherence to defined activation criteria was being adhered to, activations were audited from May 2013-February 2014.  There were 2005 activations and the over triage rate underwent a significant decrease to 69%.  However, 24% of alphas and 52% of bravo alerts were discharged from the trauma bay.  The audit revealed that when criteria from the protocol were met and documented, the over triage rate further declined to 65%.  However, even with activation criteria being complied with, discharge rates remained unchanged with 22% of alphas and 56% of bravos being discharged.

Conclusion: Over utilization of resources remains an issue in this trauma system.  Through the maturation of a level I trauma center, comfort with trauma activations increased and there was a departure from identifying alert activation criteria and documenting reason for activation.  However, after identification of this departure, increased adherence to defined protocols caused a decrease in over triage rates.  While current criteria do not cause triage rates to meet national benchmarks, they provide significant improvement and further modification of activation criteria is being evaluated.

 

54.07 Falls In The Elderly: A Cause Of Death Analysis At A Level 1 Trauma Center

Posted on December 22, 2014

C. J. Allen1, W. M. Hannay1, C. R. Murray1, R. J. Straker1, J. P. Meizoso1, J. J. Ray1, M. Hanna1, C. I. Schulman1, A. S. Livingstone1, N. Namias1, K. G. Proctor1  1University Of Miami,Trauma And Critical Care,Miami, FL, USA

Introduction:  Fall-related mortality is projected to exceed both MVC and firearm mortality rates.  One in three elderly adults falls each year. Although tremendous resources are focused on this growing public health concern with many arguing we require more devoted trauma centers to address this issue, exactly how these patients die is unknown. To fill this gap, we analyzed all falls at a level I trauma center. We hypothesize there is large portion of elderly fatalities that do not die directly from the fall injury itself.

Methods:   From 01/2002 to 12/2012, 7,293 fall admissions were reviewed. Demographics, fall height, injury patterns, procedures and outcomes were obtained. Each elderly (≥65y) fatality was analyzed for root cause of death and categorized as either: direct result from fall, complication of prolonged hospital stay (indirectly related to injury of fall), or fatal event preceding fall. Physician notes were used to define cause of death, and medical examiner and autopsy reports were reviewed in ambiguous cases. Parametric data is represented as mean ± standard deviation and non-parametric data as median (interquartile range). Student’s t-test and Fisher’s Exact test were used as appropriate.

Results: In 2002-2007, 25% of all falls were elderly, but in 2008-2012 this proportion increased to 30% (p<0.001). When comparing non-elderly (n=5,187) to elderly (n=2,076), age was 38±17y vs 78±8y, 76% vs 50% M (p<0.001), ISS 10±10 vs 12±10 (p=NS), LOS 8±6d vs 12±6d (p<0.001), mortality 3.4% vs 13.7% (p<0.001). Non-elderly fatalities (n=180) were 82% M, and ISS 32±15, while elderly fatalities (n=285) were 58% M (p<0.001), 91% ground-level height, ISS 23±11 (p<0.001). Elderly cause of death is detailed in the figure below.

Conclusion: The fall population is becoming proportionately more elderly.  Although with similar injury severity, elderly are associated with a higher LOS and >4x the mortality rate compared to the non-elderly. As the vast majority of elderly fatalities fall from ground-level height, they have a significantly lower injury severity than non-elderly fatalities. Over one-third of elderly deaths are due to a fatal event preceding the fall or from complications from a prolonged hospital stay. Since fall mortality rates continue to rise despite public health concern, it is notable that a large portion of these deaths are not directly related to the fall injury. As focus on elderly fall mortality increases and as many believe we require more devoted trauma centers to combat this growing health concern, defining the true causes of death may be necessary to focus resources to the actual issues. 

 

54.08 Do High Malpractice Risk Environments Deter Poor Care?

Posted on December 22, 2014

C. Minami1, J. Chung1, J. Holl1, M. Mello2, K. Bilimoria1  1Northwestern University,Surgery,Chicago, IL, USA 2Stanford University,Law School,Palo Alto, CA, USA

Introduction: The current US malpractice system rests on the assumption that the constant threat of lawsuit will deter negligence and improve the quality of care.  However, the system may fail to function in its intended manner and result in unintended consequences, such as encouraging “defensive” medical practices including over-testing and over-treatment that, in turn, may lead to potential patient harm.  The objective of this study was to assess whether state malpractice environment is associated with hospital performance on national quality indicators.

Methods: From HospitalCompare (October 2011 public release), short-term, acute-care, general hospitals in the U.S. that publicly reported process-of-care, imaging efficiency, outcomes, and/or Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) measures were identified.    Hierarchical regression models were developed to estimate the associations between state malpractice environment and hospital performance on quality indicators.  State-level indicators of malpractice environment included rates of paid claims/physician, state average Medicare Malpractice Geographic Practice Cost Index (MGPCI; a reflection of premiums), 8 malpractice laws, and a composite measure that combines claims, premiums, and the laws.  The hospital quality indicators included publicly reported processes of care, outcomes, and patient experience (HCAHPS).

Results: Overall, there were few associations between measures of malpractice environment and quality indicators.  There were no significant associations between process-of-care measures and the number of paid claims, MGPCI, state malpractice laws, or malpractice environment composite measures.  Higher state malpractice costs/premiums were associated with increased use of certain diagnostic tests (e.g., brain CTs and cardiac stress tests).  Hospitals in high malpractice risk environments were, by some measures, associated with lower 30-day risk-adjusted mortality rates but had significantly higher 30-day readmission rates (p<0.05).  Higher malpractice cost/premium environments were associated with lower HCAHPS performance (effect size range:-2.06 to -4.92 percentage points, Bonferroni-corrected p<0.01).  

Conclusion: We found little evidence that state malpractice environments with greater levels of malpractice risk were associated with better hospital quality.  They were, however, associated with lower levels of patient satisfaction and potential increases in “defensive” medicine practices (e.g., imaging, readmissions).  These findings do not support the assumption that the current malpractice system deters poorer quality of care. 

 

52.02 Geographic Clustering of Guideline Adherence in Colon Cancer Care Using Spatial Autocorrelation

Posted on December 22, 2014

R. L. Hoffman1, K. D. Simmons1, G. C. Karakousis1, N. N. Mahmoud1, R. R. Kelz1  1Hospital Of The University Of Pennsylvania,Philadelphia, PA, USA

Introduction: Stage-specific, evidence-based guidelines to assist with the delivery of quality cancer care were first released in 1996. Adherence to guideline-based therapy has been shown to have an impact on overall and recurrence free survival for colon cancer patients. In order to identify high yield regions for educational campaigns regarding colon cancer guidelines, we analyzed the geographic distribution of guideline-adherent care using spatial autocorrelation.

Methods: Patients aged 65-84 years diagnosed with AJCC stage II and III colon cancer were identified within the SEER-Medicare database (2005-2009). High risk stage II disease was defined as those with a T4 tumor, poor differentiation and <12 lymph nodes examined. Guideline adherence was assigned using stage-specific NCCN guidelines. The percentage of adherence was calculated for each state and county within the SEER registry catchment areas containing more than 10 patients, and translated to a choropleth map. Tests of spatial autocorrelation with queen contiguity-based spatial weights were used to evaluate geographic clustering of guideline-concordant therapy delivery. The Moran’s I and the local indicator of spatial autocorrelation (LISA) were calculated.

Results:There were 17,715 patients identified. A total of 4,933 (28%) were low risk Stage IIA/IIB, 4,446 (34%) were high risk stage IIA/IIB, and 8,336 (47%) were stage IIIA/B/C. A total of 13,017 (73%) patients underwent colectomy, 3,549 (27%) had ≤12 lymph nodes removed, 3,988 (31%) received chemotherapy, and 485 (4%) were treated with radiation.  A total of 331 (3%) were treated with all modalities. Of all those who underwent surgery, 6,348 (49%) received guideline concordant treatment and 6,669 (51%) were discordant (5,837; 45% undertreated, 832; 6% overtreated). Stage IIA/B patients received concordant therapy 53% of the time and are undertreated in 39% of cases while Stage III patients receive concordant care 44% of the time and are undertreated in 52%. Twelve states were represented in the SEER registry and 73 counties. Adherence rates ranged from 43% to 56% by state and ranged from 16% to 87% by county. The value of Moran’s I was 0.187. Clustering of concordant care varied throughout the US (Fig 1).

Conclusion:There is variation in the treatment of stage II and III colon cancer across geographic locations within the United States. Variation seems to be unrelated to the presence of NCI Designated Comprehensive Cancer Centers. Graphical representation of NCCN guideline concordance patterns using spatial autocorrelation represents a new approach to identifying high yield regions for education campaigns. Issues related to cancer care access may be more specifically targeted using this knowledge.

 

52.03 Evaluation of Sonoelastography and Race in Benign and Malignant Breast Tumors

Posted on December 22, 2014

M. Martin1, N. Zaremba1, D. Anderson2, A. Davis1, S. Schafer2, H. Bumpers1  1Michigan State University,Department Of Surgery,Lansing, MI, USA 2Michigan State University,Department Of Radiology,East Lansing, MICHIGAN, USA

Introduction: Sonoelastography (SE) is an evolving breast imaging tool that has a theoretical and practical basis.  SE combines compression and ultrasonography to estimate tissue elasticity. Breast cancer is generally firmer than adjacent benign tissue.  A quantitative score (1-5) is assigned to an image created over the lesion. Benign lesions appear softer and more green (scores 1-3) and malignant lesions are usually less elastic and blue (scores 4-5).  SE may vary due to histolopathology.  Factors thought to contribute to disparities between White American (WA) and African Amercian (AA) women are late stage disease, characterized by large tumors and advanced disease.  Racial differences in SE has not been studied, and it is unknown if tumor elastic properties differ. The purpose was to evaluate SE in definining elastic properties of breast lesions in WA and AA women.

Methods: From January 2012 through January 2013,  patients’ charts were reviewed for demographics, pathology, and breast imaging.  SE scores were assigned according to the Tsukuba Elasticity Score (TES) system.  Correlation was made between two races (WA and AA) and tumors malignancy or benignity. Descriptive statistics were performed.  This study was approved by the Institutional Review Board of Michigan State University.

Results: SE was performed in 181 women with a mean age of 49.7±15.2 years (±SD). Of the 205 lesions identified, 183 were in WA women (62 malignant, 121 benign), and 22 in AA women (5 malignant, 17 benign).

The mean TES for malignant lesions in WA women was 3.1±1.1, and 2.4±1.7 for AA’s (p = 0.32).  The mean TES for benign lesions in WA women was 1.9±1.4, and 2.4±1.4 for AA women (p = 0.06). No significant racial differences were noted in elasticity for malignant or benign tumors.

Among all malignant lesions in both WA and AA women, the mean TES was 3.1±1.1, the mean BI-RADS was 4.4±0.6. This indicates that suspicious lesions are less elastic. ER/PR/Her2-neu characteristics, including triple negative tumors (TNT), did not influence SE.  All malignant lesions had a TES of 2 or higher, and 79.1% (53/67) were infiltrating ductal carcinoma.  In lesions coded BI-RADS 3 and 4, the mean TES in WA women was 2.5±1.3, and 3.0±0.7 in AA women (P = 0.12). A suggestion that AA women with suspicious BIRADS scores may have firmer tumors.

Benign lesions had a mean TES was 1.9±1.4, which was significantly less than the mean the TES of malignant lesions (3.1±1,1, p<0.001).  SE will differentiate malignancy and benignity.

Conclusion: SE can be a useful diagnostic aid in the detection and differentiation of malignant and benign breast tissues. There was no demonstration of elasticity difference between breast tumors in WA and AA women. This study is the first to characterize the elastic properties of malignant and benign breast tumors between races.

 

52.04 Disparities in Pediatric Gonadal Torsion: Does Gender, Race and Insurance Status Affect Outcomes?

Posted on December 22, 2014

S. S. Satahoo1, H. Hua1, J. E. Sola1, H. L. Neville1  1University Of Miami,Surgery,Miami, FL, USA

Introduction: Ovarian and testicular torsion are emergencies requiring prompt surgical treatment to preserve gonadal function. However, diagnosis in females is often delayed due to non-specific symptoms. We sought to assess disparities in management and outcomes between males and females with torsion, and evaluate the effect of race and insurance status.

Methods: The National Inpatient Sample was queried for pediatric patients with “emergent”, “urgent” or “trauma center” admission and ICD-9 codes for ovarian torsion and testicular torsion. Demographic data, operative repair, gonadal loss, length of stay (LOS), total charges (TC), and mortality were recorded. Student’s t-test, Χ2 test, and logistic regression were performed, where appropriate. Propensity score matching (PSM) using nearest neighbor 1:1 matching was implemented to further evaluate endpoints. A p-value <0.05 was considered significant.

Results:There were 2254 unweighted encounters. There were no mortalities in the cohort. Forty-eight percent were females. The average age for males was 11.6 ± 5.3 years and 12.6 ± 3.7 years for females (p<0.001). Among males, 90% underwent surgery (p<0.001), of which 40% required orchiectomy. Conversely, 73% of females had surgery (p<0.001), of which 78% had oophorectomy. Insurance status (p=0.012), race (p<0.001), hospital control (p=0.007), hospital location/teaching status (p=0.006), and hospital region (p<0.001) were significant and used for PSM, with age. Hospital bed size and calendar year were not significant. Females had longer LOS (2.4 ± 1.8 days vs. males 1.3 ± 2.2 days, p<0.001) with higher TC ($20,006.44 ± 13,500.45 vs. males $12,506.83 ± 12,586.50, p<0.001). After PSM, two gender cohorts of 1026 encounters each were obtained. For males, 90% had surgical repair (p<0.001), of which 37% had orchiectomy. Only 74% of females underwent surgery (p<0.001), of which 78% had oophorectomy. Females had longer LOS (2.4 ± 1.8 days vs. 1.2 ± 2.2 days for males, p<0.001) and had higher TC ($20,058.44 ± 13,420.82 vs. males $12,309.84 ± 12,377.18, p<0.001). Logistic regression revealed that males (Odds Ratio, OR 0.153 [0.114-0.206]) and older patients (age OR 0.944 [0.916-0.973]) were less likely to undergo gonadal loss. Compared to self pay patients, those with Medicare/Medicaid were more likely to have gonadal loss (2.169 [1.126-4.179]). Race was not significant (p=0.374).

Conclusion:Disparities exist in the management of torsion, primarily based on gender. Females were overall less likely to undergo surgery and when they did, they were more likely to have gonadal loss. Females also had higher charges and longer hospitalizations, which could be reflective of the invasive nature of surgery. In addition,public insurance and age are associated with increased gonadal loss.

52.05 Inequalities in the Use of Helmets by Race and Payer Status Among Pediatric Cyclists

Posted on December 22, 2014

B. C. Gulack1, B. R. Englum1, K. L. Rialon1, L. J. Talbot1, J. E. Keenan1, H. E. Rice1, J. E. Scarborough1, O. O. Adibe1  1Duke University Medical Center,Department Of Surgery,Durham, NC, USA

Introduction: Despite numerous nationwide campaigns to increase helmet use among pediatric cyclists, as well as data demonstrating a significant survival advantage for patients who wear a helmet, a substantial number of children are still injured while riding without a helmet.   There is further concern that the number of patients who ride without a helmet may be significantly different across racial and socio-economic divides.  To better understand this issue, we performed the following study to determine demographics associated with helmet use, particularly along racial and socioeconomic lines.

Methods: The National Trauma Data Bank (NTDB) was queried during the years of 2007, 2010, and 2011 for patients under the age of 16 who were involved in a bicycle accident (ICD-9 code 826.1).  Patients were excluded if they were transferred to another facility, or if their helmet status was unknown.  Patients were grouped based on whether or not they had a helmet on during the accident.  Groups were compared with regards to baseline characteristics.  Multivariable logistic regression was then used to determine factors significantly associated with helmet use.  Age, gender, race, region, and payer status were included in the model.

Results: Of the 7,678 patients included in the analysis, 1,695 (22.1%) were wearing a helmet during their cycling accident.  The overall median age was 11 years (Interquartile rage [IQR]: 7,13 years), and 1,741 (22.7%) patients were female. On univariable analysis, non-helmeted riders were significantly more likely to be Black/African-American (10.1% vs 3.7%, p<0.001), and more likely to be insured by Medicaid (32.8% vs 14.3%, p<0.001).  Following adjustment with multivariable logistic regression, Black/African-American patients continued to have a significantly decreased odds of having been wearing a helmet during their accident compared to White/Caucasian patients (adjusted odds ratio [AOR] (95% confidence interval [CI]): 0.40 (0.30, 0.53).  Patients on Medicaid also had a significantly decreased odds of having been wearing a helmet compared to patients on private insurance (AOR (95% CI): 0.33 (0.28, 0.38)).

Conclusion: Patients who are Black/African-American or who are on Medicaid are significantly less likely to be wearing a helmet when involved in a bicycle accident.   Further efforts in education to promote helmet use need to be directed towards these groups.

52.06 Penetrating Trauma in Adolescents: Sex and Socioeconomic Factors Define Injury Pattern

Posted on December 22, 2014

W. A. Young1, C. S. Muratore1, F. I. Luks1, W. G. Cioffi2, D. S. Heffernan2  1Hasbro Children’s Hospital, Rhode Island Hospital And Alpert Medical School Of Brown University,Division Of Pediatric Surgery,Providence, RHODE ISLAND, USA 2Rhode Island Hospital And Alpert Medical School Of Brown University,Division Of Trauma And Acute Care Surgery,Providence, RHODE ISLAND, USA

Introduction:  Adolescent patients are increasingly vulnerable to gang violence and penetrating trauma with lifelong physical and psychosocial consequences. Male and female adolescents experience different societal influences and peer pressures. Socioeconomic disparities have been reported in adult trauma victims. Specifically, we have shown that health care insurance, as a marker of Socioeconomic Status (SES), is associated with worse outcome following traumatic injuries. However, there is a paucity of data among pediatric patients. We reviewed socioeconomic and gender differences with respect to risk factors for, and patterns of, injury in adolescent penetrating trauma patients. 

Methods:  Retrospective review of adolescent penetrating trauma patients aged 12-18 inclusive over a 5 year period admitted to the states only Level 1 trauma center. Data was collected from the prospectively maintained trauma database for demographics including insurance status, location (street vs. home) and intent of injury, all injuries sustained and Injury Severity Score (ISS).  Hospital course included operations, complications, length of stay, mortality and disposition. 

Results: Of 158 patients, 80% male, no difference existed between sexes in age (16 vs 16years;p=0.9), or rates of intoxication (95% vs 75%;p=0.32), but males had higher ISS (7.8 vs 3.4;p<0.05), more likely injured on the street (32.8% vs 18%;p<0.01) versus home (33.6% vs 69%;p<0.01), less likely from self-harm (11% vs 56%;p<0.001). No guns were used in either home or street for self-harm/suicide in either gender. Of the injuries occurring in the home, females were more likely to self-harm (78% vs 22%;p<0.001). There was no difference in street-related gunshot injuries between males and females (35% vs 33%;p=0.9). Self-harm was significantly more likely to occur among insured patients (59% vs 11.1%;p<0.001). However, within the insured adolescents, penetrating trauma among females, compared to males, was dramatically more likely to be from self-harm (91.6% vs 24%;p<0.0001). Overall, there were no differences in length of stay, ICU admission or mortality.

Conclusion:
Socioeconomic and social disparities exist among adolescent penetrating trauma victims. Despite perceptions that males predominate gang-related trauma, we have shown that males and females are equally victims of street related gun violence. Despite outreach interventions, gang related violence continues to increase, especially among uninsured (lower SES) adolescents. Our data argues for increased efforts developing activities and alternatives to street violence.  However, a distinct pediatric trauma population exists that is potentially neglected by current programs.  These are socially isolated, vulnerable female adolescents participating in self-harm at home.  This highlights the need for programs to develop coping mechanisms and self-esteem in the pre-teen years to off-set societal pressures in young females.
 

52.07 Discussions with a Nephrologist and Access to Kidney Transplantation

Posted on December 22, 2014

N. Gupta1, L. M. Kucirka2, M. L. Salter2, A. H. Law2, K. S. Balhara3, D. L. Segev1,2  1Johns Hopkins University School Of Medicine,Department Of Surgery,Baltimore, MD, USA 2Johns Hopkins University School Of Public Health,Department of Epidemiology,Baltimore, MD, USA 3Johns Hopkins University School Of Medicine,Department Of Emergency Medicine,Baltimore, MD, USA

Introduction:  Nephrologists play a key role in educating dialysis patients about kidney transplantation (KT). In fact, informed consent for dialysis should, by definition, involve a discussion of alternate therapies such as KT; furthermore, Medicare requires such a discussion about KT for reimbursement for dialysis care. However, little is known about (1) how much time nephrologists spend with their patients in these discussions, (2) the relationship between patient characteristics and time spent discussing KT, and (3) the relationship between time spent discussing KT and subsequent access to and pursuit of KT.

Methods:  We conducted a cross-sectional survey of prevalent dialysis patients recruited from four Davita dialysis centers across Maryland to assess patient-reported time spent discussing KT with a nephrologist. Eligibility criteria included no prior history of transplant. Subsequent access to KT was ascertained via linkage to SRTR and was defined as receiving a transplant from a live donor or registering for the deceased donor waitlist. We used modified Poisson regression to quantify the relationship between patient characteristics, including time spent discussing KT with a nephrologist, and access to KT. We used ordered logistic regression to quantify the relationship between patient characteristics and time spent discussing KT with a nephrologist.

Results: Of 254 patients surveyed, 40.9% reported 0 minutes, 37.4% reported 1- 20 minutes, 11.4% reported 21-30 minutes, 3.1% reported 31-45 minutes, and 7.1% reported greater than 45 minutes of KT discussion with a nephrologist. Age greater than 50 years (aRR 0.47, 95% CI: 0.25-0.89, p-value=0.02) was associated with decreased odds of discussing KT for >20 minutes. Overall, 16.1% achieved access to KT. Age greater than 50 years (adjusted RR [aRR] 0.49, 95% CI: 0.28-0.83, p-value=0.008) was associated with decreased access to KT. Reporting 21-30 minutes (aRR 6.67, 95% CI: 2.89-15.37, p-value<0.001), reporting 31-45 minutes (aRR 8.20, 95% CI: 3.13-21.49, p-value<0.001), and reporting >45 minutes (aRR 4.77, 95% CI: 1.79-12.66, p-value=0.002) were associated with increased access to KT. 

Conclusion: Nephrologist discussions of longer than 20 minutes were associated with increased access to and/or pursuit of KT. However, only 60% of patients reported discussing KT with their nephrologist, despite 100% requirement by Medicare. Facilitating increased discussion of KT between patients and nephrologists may increase access to surgical care. 

 

52.08 Geographic Disparities in Mortality Following Head Trauma

Posted on December 22, 2014

M. P. Jarman2, R. C. Castillo2, A. R. Carlini2, A. H. Haider1  1Johns Hopkins University School Of Medicine,Department Of Surgery,Baltimore, MD, USA 2Johns Hopkins Bloomberg School Of Public Health,Department Of Health Policy And Management,Baltimore, MD, USA

Introduction:  Treatment at a designated trauma center is proven to reduce mortality from traumatic brain injury, but the majority of US residents in rural areas do not have timely access to Level I or II trauma centers. Rural residents also face elevated risk of traumatic brain injury compared to non-rural residents, and may experience more severe injuries that their urban and suburban counterparts.  

Methods:  We performed a retrospective analysis of 2006-2011 National Emergency Department Sample data to determine if mortality following trauma brain injury differs across urban/rural classifications. Emergency department (ED) visits with ICD-9-CM codes for intracranial injury (ICD-9-CM 850-854) as the primary diagnosis were included in these analyses (N = 180,499). Odds of death in the ED were calculated using multiple logistic regression analyses with patient residential urban/rural status, Injury Severity Score, comorbidities, trauma center designation, patient age, and patient gender as covariates. All analyses were performed using Stata 12.1.

Results: Residents from rural communities were 21% (p = 0.001) more likely to die of traumatic brain injury that non-rural residents, when controlling for severity, comorbidities, trauma center designation, age, and gender. Rural residents treated at Level I trauma centers were 18% (p = 0.010) more likely to die of their injuries, compared to non-rural residents. There was no statistically significant difference in mortality between rural and non-rural residents with head injury treated at Level II or Level III centers (p = 0.092 and p = 0.465, respectively). Rural residents treated for head injury at Level IV centers were 76% more likely to die compared to non-rural residents (p < 0.001). 

Conclusion: People living in rural communities are significantly more likely than non-rural residents to die following traumatic brain injury. This disparity is present at Level I trauma centers and a Level IV centers, which typically serve as safety net providers in rural communities. The disparity is not present at Level II and Level III trauma centers. Distance and travel time to treatment likely play a significant role in brain injury outcomes for rural residents, but measures of distance and time were not available for these analysis. Future analyses should explore the interaction between time to treatment, level of care, and outcomes for rural residents.

« Previous Page
Next Page »