Emergency general surgery outcomes at safety net hospitals

j o u r n a l o f s u r g i c a l r e s e a r c h 1 9 6 ( 2 0 1 5 ) 1 1 3 e1 1 7

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.JournalofSurgicalResearch.com

Emergency general surgery outcomes at safety net hospitals Charles Patrick Shahan, MD,a Teresa Bell, PhD,b Elena Paulus, MD, MS,a and Ben L. Zarzaur, MD, MPHb,1,* a

Department of Surgery, University of Tennessee Health Science Center, Memphis, Tennessee Department of Surgery, Center for Outcomes Research in Surgery, Indiana University School of Medicine, Indianapolis, Indiana

b

article info

abstract

Article history:

Background: The United States hospital safety net is defined by the Agency for Healthcare

Received 6 October 2014

Research and Quality as the top decile of hospitals, which see the greatest proportion of

Received in revised form

uninsured patients. These hospitals provide important access to health care for uninsured

16 February 2015

patients but are commonly believed to have worse outcomes. The aim of this study was to

Accepted 18 February 2015

compare the outcomes of emergency general surgery procedures performed at safety net

Available online 21 February 2015

and nonsafety net hospitals. Material and methods: The Healthcare Cost and Utilization Project Nationwide Inpatient

Keywords:

Sample from 2008e2010 was used to create a cohort of inpatients who underwent emer-

Safety net

gency appendectomy, cholecystectomy, or herniorrhaphy. Outcomes measured included

Emergency general surgery

length of stay, charge, cost, death in hospital, complications, and failure to rescue (FTR).

Health care policy

Univariate and logistic regression analysis was performed to associate variables with outcomes. Results: A total of 187,913 emergency general surgery cases were identified, 11.5% of which were performed at safety net hospitals. The safety net cohort had increased length of stay but lower mean charge and cost. Age, comorbidity score, black race, male gender, and Medicaid and Medicare insurance were associated with mortality, complication, and FTR. Lower socioeconomic status was associated with mortality and complication. Safety net status was positively associated with complication but not mortality or FTR. Conclusions: Safety net hospitals had higher complication rates but no difference in FTR or mortality. This may mean that the hospitals are able to effectively recognize and treat patient complications and do so without increased cost. ª 2015 Elsevier Inc. All rights reserved.

1.

Introduction

In the United States, safety net hospitals provide essential access to care for the nation’s uninsured. The Agency for Healthcare Research and Quality defines safety net hospitals as those in the top decile of hospitals, which provide the

largest proportion of care to the uninsured [1]. At this time, hospitals that provide more than 8.7% of care to self-pay patients are considered safety net hospitals. Because of significant financial pressures, nonsafety net hospitals have reduced the amount of uninsured patient care provided, while many safety net hospitals have been forced to close [2e4]. As a

* Corresponding author. Department of Surgery, Center for Outcomes Research in Surgery, Indiana University School of Medicine, Indianapolis, IN. Tel.: 317-274-7422; fax: 317-274-7876. E-mail address: [email protected] (B.L. Zarzaur). 1 Present address: 702 Rotary Circle Room 022B, Indianapolis, IN 46202. 0022-4804/$ e see front matter ª 2015 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jss.2015.02.044

114

j o u r n a l o f s u r g i c a l r e s e a r c h 1 9 6 ( 2 0 1 5 ) 1 1 3 e1 1 7

result, this further reduces access to care for uninsured patients who often experience delays in care because of difficulty finding practitioners and hospitals willing to treat them. Further complicating the future of care for the uninsured is the evolving funding mechanisms for safety net care implemented as part of the health care reform. Planned reductions in the Medicaid Disproportionate Share Hospital Program, as part of the Affordable Care Act, could result in further financial stress on the safety net [5]. Studies are beginning to show that such changes in reimbursement are associated with decreases in the quality of surgical care based on performance on global Centers for Medicare and Medicaid Services quality measures [5]. At present, few studies have examined surgical outcomes at safety net hospitals and those that have been published are limited to either a single institution or within one regional hospital system [6]. As surgical outcomes are increasingly tied to hospital quality evaluations and reimbursement, this is an important aspect of determining the future of the health care safety net. The purpose of our study was to determine whether safety net hospitals have poorer emergency general surgery outcomes than those of nonsafety net hospitals using a national administrative database over a 3-y period.

2.

Materials and methods

The Healthcare Cost and Utilization Project Nationwide Inpatient Sample (NIS) from years 2008e2010 was used to construct a retrospective cohort. The NIS is designed to approximate a 20% stratified sample of Unites States hospitals, composed of data from 41 states [7]. Over 24 million inpatient encounters are included in the NIS data set from 2008e2010. The study cohort was constructed using International Classification of Diseases Ninth Revision (ICD-9) procedure codes to identify patients who had one of the three emergency surgical procedures being studied. The procedures studied were appendectomy (47.0, 47.01, and 47.09), cholecystectomy (51.2, 51.22, 51.23, and 51.24), and herniorrhaphy (53.0, 53.2, and 53.5e53.6). These operations were selected based on the assumption that every hospital with basic general surgery capabilities should be able to perform them. The cohort was further narrowed to include only unplanned admissions. Patients aged <18 y were excluded. The study data set included patient information on age, sex, race and/or ethnicity, and comorbidities. Comorbidity was quantified using the Elixhauser score, a validated modification of the Charlson comorbidity index designed for use with large administrative databases [8]. A score is assigned to each comorbidity based on its association with mortality. The sum of these points for each patient equals the Elixhauser score. Race and/or ethnicity data were reclassified from the standard database as white, black, Hispanic, or other. A binary variable was added to the data set to identify each case as either insured or uninsured based on the coded insurance status. Those with commercial insurance, Medicare, or Medicaid were designated as insured. Patients were designated as uninsured if the insurance status was listed as selfpay or other. Every hospital in the database was assigned

safety net status based on the proportion of uninsured patients treated at that specific hospital. The top decile of hospitals treating the highest proportion of uninsured patients each year were identified as safety net hospitals in keeping with the Agency for Healthcare Research and Quality definition [1]. Each patient encounter was subsequently assigned safety net status based on the hospital where they received treatment. Where available, cost data were obtained by multiplying the total charges for each case by the supplemental cost-tocharge ratio for each hospital provided by the NIS. Socioeconomic status was based on the mean household income from the patient’s zip code. These were classified as lowest ($1e38,999), low ($39,000e47,999), high ($48,000e62,999), and highest ($63,000þ). Outcomes measured included length of stay, cost, charge, death, failure to rescue (FTR), and complications. FTR is defined as the presence of a complication and subsequent hospital death. FTR has been shown to be less influenced by patient characteristics and is more sensitive to hospital quality of care characteristics compared with that of mortality alone [9]. Finally, complications were identified using ICD-9 codes. These included surgery specific complications and generalized complications. The complications observed are listed in Table 1. Statistical analysis was performed using SPSS (version 19.0; IBM Inc, Armonk, NY) and SAS (version 9.3; SAS Institute Inc, Cary, NC). Continuous variables were expressed as means or percentages where appropriate. Interquartile ranges were calculated for length of stay, cost, and charge. Multivariable regression and hierarchical analysis were performed where appropriate to evaluate the relationship between all available patient and hospital variables and outcomes. These models were chosen in an effort to account and control for both patient- and hospital-level characteristics. Any result with P < 0.05 was considered statistically significant. This study was approved and deemed exempt by The University of Tennessee Health Science Center Institutional Review Board.

3.

Results

The study included 187,913 emergency general surgery cases in the cohort. Approximately 11.5% of cases were treated at safety net hospitals. The study sample included cases from

Table 1 e Complications observed. Complication Postoperative shock Hemorrhage/hematoma Accidental puncture Wound disruption Retained foreign body Postoperative infection Nervous system Cardiac Respiratory Gastrointestinal Urinary

ICD-9 code 998.0 998.1 998.2 998.3 998.4 998.5 997.0 997.1 997.3 997.4 997.5

j o u r n a l o f s u r g i c a l r e s e a r c h 1 9 6 ( 2 0 1 5 ) 1 1 3 e1 1 7

1898 hospitals, of which 208 (11.0%) were designated as safety net facilities. The characteristics of the cohort are shown in Table 2. The patients seen at the safety net hospitals were younger, more likely to be male (48% versus 51%, P < 0.001), more likely to be a minority (52.7% versus 71.0% white, P < 0.001), and from areas of lower socioeconomic status (41.5% versus 23.3% lowest income group, P < 0.001). Fewer safety net patients had commercial insurance (26.9% versus 48.2%) and a greater percentage of patients were uninsured (20.9% versus 8.7%, P < 0.001) or had other types of insurance (16.8% versus 4.1%, P < 0.001). There was a lower average Elixhauser comorbidity score in patients seen at safety net hospitals (1.04 versus 1.29, P < 0.001). Age, Elixhauser score, black race, male gender, Medicaid, Medicare, or other insurance status, complication, and decreasing socioeconomic status were all positively associated with increased mortality. Hispanic ethnicity was protective against mortality. Age, Elixhauser score, black race, male gender, Medicaid or Medicare insurance, and decreasing socioeconomic status were all positively associated with the presence of complication. Nonsafety net status was protective against complication (odds ratio 0.91 [0.85e0.98]). Age, Elixhauser score, black race, male gender, Medicaid, Medicare, and other insurance status were positively associated with FTR. Safety net status was not associated with death or FTR. Full regression analysis results are presented in Table 3. In regard to outcome measures, safety net hospitals had an increased mean length of stay (5.7 versus 5.3 d, standard deviation (SD) ¼ 8.81, P < 0.001) but lower mean cost ($16,380 versus 16,689, SD ¼ $25,557, P < 0.001) and charges ($50,200 versus 50,757, SD ¼ $86,351, P <

Table 2 e Characteristics of patients undergoing emergency general surgery 2008e2010. Characteristics

Overall (N ¼ 187,913)

Hospital safety net status Safety net

Gender (%) Female Age in years (mean) Race (%) White Black Hispanic Other Income ($) 1e38,999 39,000e47,999 48,000e62,999 63,000þ Elixhauser score Insurance (%) Medicare Medicaid Commercial Self-pay Other

49.9 50.4

Nonsafety P net value

48 51 47.6  18.7 50.8  20.0

<0.001 <0.001 <0.001

68.8 9.7 14.5 7.0

52.7 13.8 25.0 8.5

71.0 9.2 13.1 6.7 <0.001

24.6 25.6 25.1 24.8 1.24  2.53 28.0 10.6 45.7 10.1 5.5

35.1 23.3 29.8 25.0 23.0 25.3 12.1 26.4 1.29  2.56 1.04  2.29 14.3 26.9 21.1 20.9 16.8

10.2 48.2 28.9 8.7 4.1

<0.001 <0.001

115

0.001). Although means were very similar between safety net and nonsafety net hospitals, the differences in SDs were quite large (Table 4).

4.

Discussion

This is the first retrospective nationwide study of emergency general surgery outcomes comparing safety net and nonsafety net hospitals. The patient profiles of the two hospital groups were not surprising; we found that the patients treated with emergency general surgery at safety net hospitals are younger, more likely to be male, minority, from areas of lower socioeconomic status, less likely to have commercial insurance, and have fewer comorbidities. The safety net hospitals have increased length of stay but decreased costs and charges. Patient age, Elixhauser score, male gender, and Medicaid or Medicare insurance are all positively associated with increased complications, mortality, and FTR. One of the surprising findings was that the percentage of patients with comorbidities was lower in the safety net group. This could be explained by the younger average age of this group. There are also other possible explanations for this. First, patients may have comorbidities that are simply undiagnosed because they presumably do not have regular access to medical care. Another possibility is that because these patients are younger they may actually have fewer comorbidities. A finding that we cannot explain well is the independent association of Medicare and Medicaid insurance status with increased complications, mortality, and FTR even while our models accounted for age and comorbidity. Because insurance status was independently associated with all three outcomes, other unidentified factors may have influenced the results. The most likely explanation is that uncoded patient variables not readily captured in administrative data are responsible for these findings. The findings from the cost analysis were the most significant finding of this study as they revealed that safety net hospitals had increased complications, yet provided care for patients at a lower cost. Although the mean difference in costs between safety net and nonsafety net hospitals was only $309, this equates to roughly a 3% reduction, which could yield large cost savings at the national level. For instance, if all the patients treated at nonsafety net hospitals had a $309 cost reduction, then that would extrapolate to over $51 million for the cohort. Although we cannot explain the cost difference based on the data available, it is possible that lower operating costs at safety net hospitals may result in reduced costs for patients. There are several limitations of this study, most of which relate to the use of administrative data. First, no clinical data were available in the database so that the acuity of patients was unknown. Limitations associated with reliance on ICD-9 coding accuracy cannot be assessed, but coding errors are likely present to some extent in a database of this size. The extent to which this may impact our results is unknown. Additionally, not every state includes data for each of the elements in the database, and socioeconomic data are missing in a small number of cases. There were also missing data

116

j o u r n a l o f s u r g i c a l r e s e a r c h 1 9 6 ( 2 0 1 5 ) 1 1 3 e1 1 7

Table 3 e Regression analysis. Characteristics

Age Elixhauser score Race White Black Hispanic Other Gender Female Male Insurance Commercial Self-pay Medicaid Medicare Other Complication Location Rural Urban Income status, $ >63,000 48,000e62,999 39,000e47,999 1e38,999 Safety net Yes No

Mortality

Complication

FTR

OR (95% CI)

P value

OR (95% CI)

P value

OR (95% CI)

P value

1.05 (1.04e1.06) 1.12 (1.11e1.13)

<0.0001

1.03 (1.03e1.03) 1.07 (1.06e1.08)

<0.0001 <0.0001

1.04 (1.03e1.05) 1.09 (1.07e1.10)

<0.0001 <0.0001

Reference 1.35 (1.18e1.4) 0.81 (0.69e0.96) 0.95 (0.78e1.15)

<0.0001

Reference 1.27 (1.20e1.34) 0.96 (0.91e1.01) 1.05 (0.98e1.12)

<0.0001

Reference 1.43 (1.18e1.73) 0.95 (0.75e1.21) 0.98 (0.75e1.29)

0.00206

Reference 1.19 (1.09e1.29)

<0.0001

Reference 1.15 (1.11e1.82)

<0.0001

Reference 1.14 (1.01e1.28)

0.0419

Reference 1.19 (0.90e1.57) 1.96 (1.61e2.39) 1.65 (1.44e1.89) 1.42 (1.07e1.89) 2.95 (2.71e3.21)

<0.0001

Reference 1.01 (0.95e1.08) 1.37 (1.29e1.45) 1.18 (1.13e1.23) 1.03 (0.95e1.12) N/A

<0.0001

Reference 1.15 (0.74e1.78) 1.74 (1.30e2.34) 1.57 (1.29e1.92) 1.66 (1.10e2.50) N/A

<0.0001

Reference 1.2 (0.90e1.57)

0.0465

Reference 0.99 (0.92e1.06)

0.694

Reference 1.28 (0.96e1.70)

0.094

Reference 1.16 (1.02e1.32) 1.30 (1.14e1.47) 1.45 (1.27e1.65)

<0.0001

Reference 1.08 (1.03e1.13) 1.12 (1.07e1.18) 1.13 (1.08e1.19)

<0.0001

Reference 1.05 (0.87e1.26) 1.13 (0.95e1.36) 1.17 (0.97e1.41)

0.3332

Reference 0.85 (0.72e1.01)

<0.0001

Reference 0.91 (0.85e0.98)

0.058

0.0217

Reference 0.97 (0.73e1.27)

0.7623

CI ¼ confidence interval; N/A ¼ not applicable; OR ¼ odds ratio.

regarding costs and charges. There were 7.9% of cases that had missing cost data and 2.1% of cases that had missing charge data. This could lead to inaccurate estimates for the cost and charge data. Other studies have reported that uninsured trauma patients have higher rates of FTR but that safety net status of hospitals was not associated with worse outcomes [10]. Although a different subset of surgery patients, these findings along with our results demonstrate that safety net hospitals are presently capable of providing similar quality of care in spite of caring for more vulnerable populations with fewer

resources. As the financial future of these hospitals remains unclear, we think the quality of care provided by safety net hospitals is critical to monitor as reimbursement policies continue to evolve. One recent study has already found that public hospitals perform worse on Centers for Medicare and Medicaid Services quality measures and that they are disadvantaged with regard to the Medicare value-based purchasing program, which serves to tie quality to reimbursement [5]. Our results demonstrate that these hospitals do have a higher complication rate, but this does not result in increased cost or mortality. The combination of planned reductions in funding

Table 4 e Comparing outcomes at safety net and nonsafety net hospitals. Outcome

Overall

Safety net status Safety net

Length of stay (d) Mean (SD) Median (IQR) Cost ($) Mean (SD) Median (IQR) Charge ($) Mean (SD) Median (IQR) IQR ¼ interquartile range.

5.37  8.81

5.7  8.6 3 (1e6)

Nonsafety net 5.3  8.6 3 (1e6)

16,652  25,557

16,380  26,135 9458 (6491e15,445)

16,689  25,478 9919 (7007e16,319)

50,692  80,017

50,200  86,351 27,692 (18,000e47,000)

50,757  79,133 29,802 (19,400e50,550)

P value <0.001

<0.001

0.0345

j o u r n a l o f s u r g i c a l r e s e a r c h 1 9 6 ( 2 0 1 5 ) 1 1 3 e1 1 7

for care of the uninsured and reduced payments based on complication-related quality measures may compromise the ability of safety net hospitals to provide safe, quality care in the future.

5.

Conclusions

In conclusion, we believe the most important finding of this study was that the safety net hospitals had higher complication rates but no difference in FTR or mortality. Because FTR is a measure more reflective of hospital characteristics, we interpret this to mean that although there are more complications at these hospitals, they are able to recognize them before they can progress to mortality and do so without increasing cost.

Acknowledgment The authors thank the University of Tennessee Health Science Center Department of Surgery for funding this research. Authors’ contributions: C.P.S. and B.L.Z. contributed to the study concept and design and acquisition of data. T.B., C.P.S., E.P., and B.L.Z. contributed to the analysis and interpretation. C.P.S. and E.P. drafted the article. T.B. and B.L.Z. performed a review and revision of the article.

Disclosure The authors report no proprietary or commercial interest in any product mentioned or concept discussed in this article.

117

references

[1] Andrews RM, Stull DE, Fraser I, et al. HCUP Fact Book No. 8. Agency for Healthcare Quality and Research. Available at: http://archive.ahrq.gov/data/hcup/factbk8/; 2006. Accessed 6/2013. [2] Siegel B. Re-engineering the public hospital system: saving the safety net. Bull N Y Acad Med 1996;73:357. [3] Andrulis DP, Duchon LM. The changing landscape of hospital capacity in large cities and suburbs: implications for the safety net in metropolitan America. J Urban Health 2007;84: 400. [4] Bazzoli GJ, Lindrooth RC, Kang R, Hasnain-Wynia R. The influence of health policy and market factors on the hospital safety net. Health Serv Res 2006;41:1159. [5] Dupree JM, Neimeyer J, McHugh M. An advanced look at surgical performance under Medicare’s hospital-inpatient value-based purchasing program: who is winning and who is losing? J Am Coll Surg 2014;218:1. [6] Moore EE. Acute care surgery: the safety net hospital model. Surgery 2007;141:297. [7] HCUP nationwide inpatient sample (NIS). Healthcare cost and utilization project (HCUP). 2008-2010. Agency for Healthcare Research and Quality, Rockville, MD. Available at: www.hcup-us.ahrq.gov/nisoverview.jsp. [8] van Walraven C, Austin PC, Jennings A, et al. A modification of the Elixhauser comorbidity measures into a point system for hospital death using administrative data. J Med Care 2006; 47:626. [9] Silber JH, Williams SV, Krakauer H, Schwartz JS. Hospital and patient characteristics associated with death after surgery. A study of adverse occurrence and failure to rescue. Med Care 1992;30:615. [10] Bell TM, Zarzaur BL. Insurance status is a predictor of failure to rescue in trauma patients at both safety net and nonsafety net hospitals. J Trauma Acute Care Surg 2013;75:728.