- Email: [email protected]
Community health indicators associated with outcomes of pancreatectomy
The American Journal of Surgery xxx (2017) 1e5
Contents lists available at ScienceDirect
The American Journal of Surgery journal homepage: www.americanjournalofsurgery.com
Community health indicators associated with outcomes of pancreatectomy Lauren Slakey Pointer, MD a, *, Zaid Al-Qurayshi, MB, ChB, MPH b, David Taylor Pointer Jr., MD b, Emad Kandil, MB, BCh, FACS b, Douglas Philip Slakey, MD, FACS b a b
Department of Internal Medicine, Tulane University School of Medicine, USA Department of Surgery, Tulane University School of Medicine, USA
a r t i c l e i n f o
a b s t r a c t
Article history: Received 22 August 2016 Received in revised form 26 January 2017 Accepted 5 February 2017
Purpose: This study evaluates the association of environmental, social and health risk factors in relation to outcomes of pancreatic surgery. Methods: Patients who underwent pancreatectomy with a 30 day postoperative follow up in Florida, New York and Washington states were identified using the State Inpatient Databases (SID) from 2010 to 2011. This data was merged with community health indicators complied from the County Health Ranking database. Fourteen community health indicators were used to determine higher risk communities. Communities were then divided into low and high risk communities based on a scoring system using accumulative community risk. Results: Among 3494 patients included recipients in high-risk communities were more likely African American (p < 0.001), younger (age 40e59; p ¼ 0.001), and had Medicaid as primary insurance (p ¼ 0.001). Management of patients in high-risk communities was associated with increased risk of postoperative complications (p < 0.001), ICU admissions (p < 0.001), increased length of stay (p < 0.001). Conclusion: Health indicators from patients' communities are predictors of increased risk of perioperative complications for individuals undergoing pancreas surgery. Synopsis: Environmental, social and health risk factors are likely contributors to healthcare outcomes. This study compares post-operative outcomes for patients who underwent pancreatic resection based on community risk factors. © 2017 Elsevier Inc. All rights reserved.
Keywords: Pancreatic surgery Health indicators Outcome disparities
1. Introduction Hospital morbidity and mortality has declined for patients undergoing pancreatic surgery and long term survival rates have improved in recent decades;1 however pancreatic resection is considered one of the most complex surgical procedures. While mortality rates for benign and malignant pre-ampullary and pancreatic disease resection have decreased to 2% at experienced centers,2e4 overall morbidity is still greater than 30%.5 Many
* Corresponding author. Department of Surgery, #8622, 430 Tulane Avenue, New Orleans, LA 70112, USA. E-mail addresses: [email protected] (L.S. Pointer), [email protected] (Z. AlQurayshi), [email protected] (D.T. Pointer), [email protected] (E. Kandil), [email protected] (D.P. Slakey).
studies have shown improved survival for patients undergoing pancreatic surgery at high volume institutions and by high volume surgeons,6e11 but few studies have evaluated the effects of patient selection on postoperative morbidity and mortality. Historically, comorbidities and decreased physiological reserves have reduced surgeon and oncologist willingness to operate on these patients.12 Many papers have suggested that careful patient preoperative and surgical risk assessment are important to reduce overall morbidity and mortality.11e13 These studies have emphasized known surgical risk factors such as obesity, cardiac status, renal failure and other medical diagnoses. But little specific information about patient factors and community influences have been specifically identified as important for preoperative assessment. In this study we evaluated the association of environmental, social and health risk factors of patients and their communities in relation to outcomes for pancreatic surgery. We hypothesized that
http://dx.doi.org/10.1016/j.amjsurg.2017.02.010 0002-9610/© 2017 Elsevier Inc. All rights reserved.
Please cite this article in press as: Pointer LS, et al., Community health indicators associated with outcomes of pancreatectomy, The American Journal of Surgery (2017), http://dx.doi.org/10.1016/j.amjsurg.2017.02.010
2
L.S. Pointer et al. / The American Journal of Surgery xxx (2017) 1e5
patient health indicators may be a strong predictor of prognosis prior to pancreatic resection. Our aim was to evaluate whether clinical outcomes were associated with the presence of risk factors in patients' communities in hopes that this risk assessment may help surgeons and clinicians better assess patient risks prior to pancreatic surgery. 2. Materials and methods A cross-sectional analysis was performed using the State Inpatient Database (SID) for the states of Florida, New York, and Washington, 2010e2011. SID is part of the Healthcare Cost and Utilization Project, sponsored by the Agency for Healthcare Research and Quality.14 SID includes inpatient discharge records from community hospitals for a given state. The data includes all patients, regardless of payer, providing a unique view of inpatient care in a defined market or state over time. The selection of the aforementioned states was based on their year of availability and completeness of recorded variables. SID is publicly available, de-identified data that is exempt from approval of Institutional Review Board. International Classification of Disease, 9th revision (ICD-9), was used to define parameters of the study. SID was merged with County Health Rankings Data based on patients' county Federal Information Processing Standard (FIPS) codes;15 The County Health Rankings measure the health of nearly all counties in the United States and rank them within states. The rankings are compiled using county-level measures from a variety of national and state data sources. These measures are standardized and combined using scientifically-informed weights.15 The study population consisted of adult (18 years) inpatients who underwent partial pancreatectomy (ICD-9: 52.5) or total pancreatectomy (ICD-9: 52.6) or Whipple procedure (ICD-9: 52.7) as the primary procedure. Re-admission data were included for 30 days after the operation. The independent factor of interest was the health risk status of the community where the patient lived. In creating an overall community health indicator from the County Health Rankings data we adopted a model proposed by Schold et al.16 Nine health measures were selected and each was scored from 0 (lowest risk) to 4 (highest risk) based on quintile classification. Those community measures included: premature death, poor reported general health, poor reported physical health days, poor reported mental health days, adult smoking, adult obesity, preventable hospital stay, high school graduation rate, and income. Subsequently the community health indicators were combined to create an accumulative community health risk indicator with a score range of (0e36), and categorized based on the median score16 into low health risk communities (score 16) and high health risk communities (score> 16). The main study outcomes included postoperative complication, intensive care unit (ICU) admission, hospital length of stay (LOS), and hospital charges. Dichotomized complications were defined as the presence of one or more of the general or specific complications based on the secondary diagnoses made during the hospital stay.17 LOS was categorized into short stay (75th percentile, 14 days) and long stay (>14 days). Charges were adjusted for inflation to reflect 2015 dollar value and categorized as low charges (75th percentile, $153,311.77) or high charges (>$153,311.77). Other independent factors that were considered included: (i) patient demographics: age (<40, 40e70, >70 years old), gender, and race (White, Black, Hispanic, Asian/Pacific Islander, Native American, other); (ii) socioeconomic factors: main payer of health service (Medicare, Medicaid, private insurance, self-pay, no charge, other); (iii) clinical factors: diagnosis (pancreatic malignancy, benign pancreatic disorders, other gastrointestinal (GI) malignancy, other benign GI disorders); pancreatectomy type (partial, total,
Table 1 Descriptive statistics of the study population in relation to the community health risk status. Study population Community Health Risk Status (%) (N ¼ 4272) High Risk (%) Low Risk (%) P valuea (N ¼ 1971) (N ¼ 2301) Age(year) <40 5.5 40 e 70 61.7 >70 32.9 Gender Male 48.3 Female 51.7 Race White 74.7 Black 9.2 Hispanic 8.8 Asian/Pacific Islander 2.5 Native American 0.2 Other 4.6 Service payer Medicare 49.4 Medicaid 8.2 Private insurance 38.9 Self-pay 1.6 No charge 0.4 Other 1.5 Charlson Comorbidity Index Low: 0 e 1 70.2 Medium-low: 2 e 3 27.5 Intermediate: 4 e 5 2.3 High: 6 0.02 BMI 25 kg/m2 No 87.7 Yes 12.3 Diagnosis Pancreatic malignancy 53.9 Benign pancreatic dx 24.7 Other GI malignancy 16.5 Benign other GI dx 4.9 Pancreatectomy Partial 36.0 Total 3.3 Whipple 60.8 Surgeon volume (surgeries/yr) Low: 4 25.8 Intermediate: 5 e 38 49.6 High: 39 24.6 Postoperative complications Absent 86.1 Preset 13.9 ICU admission No 55.9 Yes 44.1 In hospital death Not reported 96.7 Reported 3.3 Readmission No 87.2 Yes 12.9 Length of stay > 14 days No 75.5 Yes 24.5 Total charges > $153,311.77 No 75.0 Yes 25.0
5.8 63.5 30.7
5.1 60.1 34.8
0.016
47.9 52.1
48.7 51.3
0.59
76.6 11.5 6.6 1.6 0.3 3.5
73.1 7.2 10.6 3.3 0.2 5.6
<0.001
49.8 9.2 36.6 2.1 0.2 2.1
49.1 7.3 40.9 1.2 0.6 1.0
<0.001
68.4 28.7 2.9 0.00
71.6 26.6 1.8 0.04
0.020
87.2 12.8
88.1 11.9
0.36
54.1 25.3 15.7 4.9
53.8 24.2 17.3 4.8
0.54
36.7 4.0 59.3
35.3 2.7 62.1
0.019
28.7 52.0 19.3
23.4 47.5 29.1
<0.001
83.7 16.3
88.1 11.9
<0.001
50.6 49.4
58.7 41.3
<0.001
96.5 3.6
97.0 3.0
0.35
87.8 12.2
86.6 13.4
0.22
71.2 28.8
79.2 20.8
<0.001
71.8 28.2
77.7 22.3
<0.001
Abbreviations: dx, diagnosis; GI, gastrointestinal; BMI, body mass index. a Chi-square test.
Whipple); surgeon volume [low-volume (25th percentile, 4 pancreatectomies/year), intermediate-volume (>25th e 75th percentiles, 5e38 pancreatectomies/year), high-volume (>75th percentile, 39 pancreatectomies/year)]; body mass index (BMI < 25 kg/m2 vs. BMI 25 kg/m2); a modification of the
Please cite this article in press as: Pointer LS, et al., Community health indicators associated with outcomes of pancreatectomy, The American Journal of Surgery (2017), http://dx.doi.org/10.1016/j.amjsurg.2017.02.010
L.S. Pointer et al. / The American Journal of Surgery xxx (2017) 1e5 Table 2 The adjusted odds ratio of living in high health risk communities for patients who underwent pancreatectomy. Variable
Age(year) <40 40 e 70 >70 Gender Male Female Race White Black Hispanic Asian/Pacific Islander Native American Other Service payer Private insurance Medicare Medicaid Self-pay
% High Risk Community
Multivariate modela aOR
95% CI
P value
49.4 47.5 43.1
1.39 1.36 Reference
1.01, 1.91 1.16, 1.59
0.045 <0.001
45.7 46.5
Reference 1.04
0.92, 1.18
0.53
47.4 57.9 34.6 29.3 55.6 35.1
Reference 1.38 0.56 0.40 1.01 0.57
1.11, 0.44, 0.26, 0.26, 0.42,
1.72 0.71 0.62 3.87 0.77
0.004 <0.001 <0.001 0.99 <0.001
43.4 46.5 51.9 60.3
Reference 1.31 1.62 2.24
1.12, 1.54 1.27, 2.07 1.34, 3.75
<0.001 <0.001 <0.001
a
Model includes the following variables: age, gender, race, service payer, charlson comorbidity index, and type of pancreatectomy.
Charlson comorbidity index score (CCIS) used to assess patient comorbidities (low score: 0e1, medium-low: 2e3, intermediate: 4e5, high score: 6); and inpatient death.18 Cross-tabulation and Chi-square tests were used to examine the association between each of the independent factors and the outcomes of interest. Factors with significant association were considered confounders and were included in multivariate logistic regression models. Multivariate logistic regression models were used for calculating the odds ratio (OR) and 95% confidence interval (CI). Significance level was set as (a ¼ 0.05). All data analyses were performed using SAS 9.3 for Windows (SAS Institute, Cary, NC, USA). 3. Results A total of 4272 discharge records were identified for the years 2010e2011 (Table 1). The mean age of the study population was 63.5 (±13.1) years and females formed 51.7% of the sample. The majority of the study sample was White (74.7%), and had Medicare coverage (49.8%). Pancreatic malignancy was the most common diagnosis (53.9%), while benign pancreatic disorders and other GI conditions represented 24.7% and 21.4%, respectively. Whipple procedure was performed in 60.8% of the cases, while total pancreatectomy was the least performed procedure (3.3%). Most
3
patients had one or no comorbidities at time of admission (70.2%). Postoperative complications were reported in 595 (13.9%) of the patients and 1196 (44.1%) required ICU admission. A total of 549 (12.9%) patients were readmitted in the 30 days that followed the operations and a total of 140 (3.3%) patients died during the follow up period. Patients living in the high health risk communities were more likely to be < 70 years old, Black, and with Medicaid or Medicare health coverage (p < 0.05 each) (Table 2) (Fig. 1). The risk of postoperative complication was significantly higher for patients living in high health risk communities [OR: 1.41, 95%CI: (1.17, 1.69), p < 0.001] (Table 3) (Fig. 1). Additionally, they were at a higher risk ICU admission [OR: 1.32, 95%CI: (1.11, 1.57), p < 0.001] and a hospital stay of more than 14 days [OR: 1.50, 95%CI: (1.29, 1.76), p < 0.001]. Hospitals that treated patients coming from communities of poor health measures, charged them significantly more than $153,311.77 compared to patients living in low health risk communities [OR: 1.40, 95%CI: (1.20, 1.63), p < 0.001]. 4. Discussion In our analysis several important outcomes for pancreatic surgery were impacted by patient community risk factors. An increase in community risk factors is associated with an increase postoperative complications, ICU admissions, longer length of hospital stay and health care costs. Our study suggests that clinicians and healthcare providers should include community and social determinants of health to assess patient operative risks, and to improve delivery of care, and outcomes for pancreatic surgery. It has been suggested that factors beyond clinical and physiological conditions are associated with morbidity and mortality in the general public.19e26 Health lifestyle and behavioral factors including smoking cigarettes, obesity, drinking alcohol and physical inactivity have been cited as major determinants of premature and preventable morbidity and mortality.19,27 Differences in health outcomes by socioeconomic position is a persistent and increasing public health problem. We found that differences in health indicators by communities may explain variation in outcomes for surgical procedures and other clinical interventions. Few studies have looked at patient specific health risks in relation to pancreatic surgery outcomes. Cheung et al. studied pancreatic cancer care and found that patients of lowest socioeconomic status consistently had lower median survival even when being treated at high volume hospitals.28 In their analysis patients from less affluent areas tended to be African American and Hispanic, have Medicare or Medicaid for insurance coverage, consume more alcohol and were younger at diagnosis. In comparison, our study found that patients in high health risk communities were young, black and had Medicaid or Medicare as their primary insurance. Other studies have found that black patient and patients
Fig. 1. Pancreatectomy risk based on patients' community health risk status.
Please cite this article in press as: Pointer LS, et al., Community health indicators associated with outcomes of pancreatectomy, The American Journal of Surgery (2017), http://dx.doi.org/10.1016/j.amjsurg.2017.02.010
4
L.S. Pointer et al. / The American Journal of Surgery xxx (2017) 1e5
Table 3 The adjusted odds ratio of selected risk factors in relation of patients' community health risk status. Variable
%
Postoperative complications Low risk 11.9 High risk 16.3 ICU admission Low risk 41.3 High risk 49.4 Length of stay > 14 days Low risk 20.8 High risk 28.8 Total charges > $153,311.77 Low risk 22.3 High risk 28.2
Multivariate modela aOR
95% CI
P value
Reference 1.41
1.17, 1.69
<0.001
Reference 1.32
1.11, 1.57
0.001
Reference 1.50
1.29, 1.76
<0.001
Reference 1.40
1.20, 1.63
<0.001
a
Model includes the following variables: community health risk status, age, gender, race service payer, charlson comorbidity index. type of pancreatectomy, diagnosis, and surgeon volume.
of lower socioeconomic status tend to obtain their care in hospitals with higher morbidity and mortality rates,29,30 but our study suggests that inherent environmental factors may play a role in these measured outcomes. For hospitals that treat more high risk patients, outcomes may be worse irrespective of volume. Schold et al. looked at community risk factors in association with kidney transplant outcomes. They used multiple databases to create a community risk score which was merged with data from the Scientific Registry of Transplant Recipients. Increased community risk was found to be associated with increased mortality and decreased likelihood of transplantation. Higher risk communities had higher comorbid conditions that they believed reflected underlying illnesses not routinely identified. Additionally, patients from these communities often received limited access to care.16,31 A different study by Schold, found that health indicators are associated with post-transplant mortality independent of traditional risk factors. In their analysis, community health indicators were felt to explain more variation in outcomes than factors that are traditionally considered clinically relevant. While they note that it is inappropriate to assign risk directly to individuals within a high risk community, conditions of the community where a patient lives may reflex exposure risks. Community indicators such as low birth weight, preventable hospitalization, increased inactivity rate and smoking were associated with overall graft loss and post-transplant mortality. They pointed out that in current hospital reimbursement based on performance assessments, variability in health indicators may lead to decreased access of care for patients from high risk communities.32 Similar to these kidney transplant studies by Schold, our analysis showed a potential impact of community risks on quality assessment of hospitals and surgical centers. Currently. community health indicators are not accounted for in performance evaluations. Hospitals that treat patients from high risk communities are more likely to have poor outcomes unrelated to traditional measures of quality of care. Thus the current reimbursement system may further divide access to high quality care for patients in high risk communities as community factors are not current accounted for in measurements used by regulatory agencies.31 Current performance evaluations based on outcomes alone may potentiate patient care disparities as providers may avoid treatment of patients from higher risk communities as these patients may have been showed to have poorer outcomes, and thus provide poorer reimbursement, irrespective of quality of care provided. The limitations of our study include the administrative nature of the database which lacks clinical details such as histopathological
records or specifics of management. One shortcoming of the dataset is the possibility of missing patients who were readmitted in the year following the study period or readmitted to hospitals outside their state. Given the administrate nature of the data set, there are some limitations that may prevent capturing the complete clinical picture. However, we have used processes which have previously been reported and verified by other studies such as Murphy et al.17 On the other hand, the strengths of the study include the large and generalizable study sample, the availability of a wide array of factors that were assessed for their confounding effects, and the use of a robust and accumulative health indicator developed from twelve county-based health measures. In summary, understanding the impact of a patient's community health status is essential for population-wide health improvement and better delivery of hospital care.33 In recent years elimination of health disparities has become of focus of many public health officials to improve length and quality of life.34 But in current preoperative assessment, patients' behaviors and community health indicators are not routinely assessed or measured. From our analysis, these health indicators are in fact associated outcomes for pancreatic surgery and are an important consideration in healthcare decisions and measurement of healthcare outcomes. More research needs to be done in order to determine how best to use this knowledge in the clinical setting. Grant support None. Conflicts of interest None. References 1. Cameron JL, He J. Two thousand pancreaticoduodenectomies. J Am Coll Surg. 2015;220:530e536. 2. Crist DW, Sitzmann JV, Cameron JL. Improved hospital morbidity, mortality, and survival after the whipple procedure. Ann Surg. 1987;206:358e365. 3. Sohn TA, Yeo CJ, Cameron JL, et al. Resected adenocarcinoma of the pancreas616 patients: results, outcomes, and prognostic indicators. J Gastrointest Surg. 2000;4(6):567e579. 4. Hill JS, McPhee JT, Whalen GF, Sullivan ME, Warshaw AL, Tseng JF. In-hospital mortality after pancreatic resection for chronic pancreatitis: population-based estimates from the nationwide inpatient sample. J Am Coll Surg. 2009;209: 468e476. 5. Winter JM, Cameron JL, Campbell KA, et al. 1423 pancreaticoduodenectomies for pancreatic cancer: a single-institution experience. J Gastrointest Surg. 2006;10:1199e1210. 6. Gooiker GA, van Gijn W, Wouters MW, et al. Systematic review and metaanalysis of the volume-outcome relationship in pancreatic surgery. Br J Surg. 2011;98:485e494. 7. Kotwall CA, Maxwell JG, Brinker CC, et al. National estimates of mortality rates for radical pancreaticoduodenectomy in 25,000 patients. Ann Surg Oncol. 2002;9:847e854. 8. Ho V, Heslin MJ. Effect of hospital volume and experience on in-hospital mortality for pancreaticoduodenectomy. Ann Surg. 2003;237:509e514. 9. Wittel UA, Makowiec F, Sick O, et al. Retrospective analyses of trends in pancreatic surgery: indications, operative techniques, and postoperative outcome of 1,120 pancreatic resections. World J Surg Oncol. 2015;13, 102e0150525-6. 10. Gordon TA, Burleyson GP, Tielsch JM, et al. The effects of regionalization on cost and outcome for one general high-risk surgical procedure. Ann Surg. 1995;221: 43e49. 11. Eeson G, Chang N, McGahan CE, et al. Determination of factors predictive of outcome for patients undergoing a pancreaticoduodenectomy of pancreatic head ductal adenocarcinomas. HPB (Oxford). 2012;14:310e316. 12. Adham M, Bredt LC, Robert M, et al. Pancreatic resection in elderly patients: should it be denied? Langenbecks Arch Surg. 2014;399:449e459. 13. Yeo CJ, Cameron JL, Sohn TA, et al. Six hundred fifty consecutive pancreaticoduodenectomies in the 1990s: pathology, complications, and outcomes. Ann Surg. 1997;226:248e257. 14. Healthcare Cost and Utilization Project (HCUP). Overview of the state inpatient databases. http://hcup-us.ahrq.gov/sidoverview.jsp. Updated 2014.
Please cite this article in press as: Pointer LS, et al., Community health indicators associated with outcomes of pancreatectomy, The American Journal of Surgery (2017), http://dx.doi.org/10.1016/j.amjsurg.2017.02.010
L.S. Pointer et al. / The American Journal of Surgery xxx (2017) 1e5 15. Community Health Rankings and Roadmaps. Rankings. http://www. countyhealthrankings.org/our-approach. Updated 2014. 16. Schold JD, Buccini LD, Kattan MW, et al. The association of community health indicators with outcomes for kidney transplant recipients in the United States. Arch Surg. 2012;147:520e526. 17. Murphy MM, Knaus 2nd WJ, Ng SC, et al. Total pancreatectomy: a national study. HPB (Oxford). 2009;11:476e482. 18. Quan H, Sundararajan V, Halfon P, et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care. 2005;43: 1130e1139. 19. Lantz PM, House JS, Lepkowski JM, et al. Socioeconomic factors, health behaviors, and mortality: results from a nationally representative prospective study of US adults. JAMA. 1998;279:1703e1708. 20. Berkman LF, Breslow L. Health and Ways of Living. New York, NY: Oxford University Press; 1983. 21. US Department of Health and Human Services. Healthy People 2000: National Health Promotion and Disease Prevention Objectives. Washington, DC: US Dept of Health and Human Services; 1990. DHHS publication 91e50212. 22. McGinnis MJ, Foege WH. Actual causes of death in the United States. JAMA. 1993;270:2207e2212. 23. Wiley JA, Camacho TC. Life-style and future health: evidence from the alameda county study. Prev Med. 1980;9:1e21. 24. Wilson PS. Established risk factors and coronary artery disease: the Framingham study. Am J Hypertens. 1994;7:75e125.
5
25. Hirdes JP, Forbes WF. The importance of social relationships, socioeconomic status and health practices with respect to mortality among Ontario males. J Clin Epidemiol. 1992;554:175e182. 26. Patterson RE, Haines PS, Pokin BM. Health life- style patterns of US adults. Prev Med. 1994;23:453e460. 27. Galea S, Tracy M, Hoggatt KJ, et al. Estimated deaths attributable to social factors in the United States. Am J Public Health. 2011;101(8):1456e1465. 28. Cheung MC, Yang R, Byrne MM, et al. Are patients of low socioeconomic status receiving suboptimal management for pancreatic adenocarcinoma? Cancer. 2010;116:723e733. 29. Girotti ME, Shih T, Revels S, et al. Racial disparities in readmissions and site of care for major surgery. J Am Coll Surg. 2014;218:423e430. 30. Silverman DT, Hoover RN, Brown LM, et al. Why do black Americans have a higher risk of pancreatic cancer than white Americans? Epidemiology. 2003;14(1):45e54. 31. Slakey DP. Surgical outcomes beyond the individual. Arch Surg. 2012;147(6): 526e527. 32. Schold JD, Heaphy EL, Buccini LD, et al. Prominent impact of community risk factors on kidney transplant candidate processes and outcomes. Am J Transpl. 2013;13(9):2374e2383. 33. Russell MW, Campbell LA, Kisely S, et al. The development of community health indicators: a district-wide approach. Chronic Dis Can. 2011;31:65e70. 34. Metzler M, Kanarek N, Highsmith K, et al. Community health status indicators project: the development of a national approach to community health. Prev Chronic Dis. 2008;5:A94.
Please cite this article in press as: Pointer LS, et al., Community health indicators associated with outcomes of pancreatectomy, The American Journal of Surgery (2017), http://dx.doi.org/10.1016/j.amjsurg.2017.02.010
Contents lists available at ScienceDirect
The American Journal of Surgery journal homepage: www.americanjournalofsurgery.com
Community health indicators associated with outcomes of pancreatectomy Lauren Slakey Pointer, MD a, *, Zaid Al-Qurayshi, MB, ChB, MPH b, David Taylor Pointer Jr., MD b, Emad Kandil, MB, BCh, FACS b, Douglas Philip Slakey, MD, FACS b a b
Department of Internal Medicine, Tulane University School of Medicine, USA Department of Surgery, Tulane University School of Medicine, USA
a r t i c l e i n f o
a b s t r a c t
Article history: Received 22 August 2016 Received in revised form 26 January 2017 Accepted 5 February 2017
Purpose: This study evaluates the association of environmental, social and health risk factors in relation to outcomes of pancreatic surgery. Methods: Patients who underwent pancreatectomy with a 30 day postoperative follow up in Florida, New York and Washington states were identified using the State Inpatient Databases (SID) from 2010 to 2011. This data was merged with community health indicators complied from the County Health Ranking database. Fourteen community health indicators were used to determine higher risk communities. Communities were then divided into low and high risk communities based on a scoring system using accumulative community risk. Results: Among 3494 patients included recipients in high-risk communities were more likely African American (p < 0.001), younger (age 40e59; p ¼ 0.001), and had Medicaid as primary insurance (p ¼ 0.001). Management of patients in high-risk communities was associated with increased risk of postoperative complications (p < 0.001), ICU admissions (p < 0.001), increased length of stay (p < 0.001). Conclusion: Health indicators from patients' communities are predictors of increased risk of perioperative complications for individuals undergoing pancreas surgery. Synopsis: Environmental, social and health risk factors are likely contributors to healthcare outcomes. This study compares post-operative outcomes for patients who underwent pancreatic resection based on community risk factors. © 2017 Elsevier Inc. All rights reserved.
Keywords: Pancreatic surgery Health indicators Outcome disparities
1. Introduction Hospital morbidity and mortality has declined for patients undergoing pancreatic surgery and long term survival rates have improved in recent decades;1 however pancreatic resection is considered one of the most complex surgical procedures. While mortality rates for benign and malignant pre-ampullary and pancreatic disease resection have decreased to 2% at experienced centers,2e4 overall morbidity is still greater than 30%.5 Many
* Corresponding author. Department of Surgery, #8622, 430 Tulane Avenue, New Orleans, LA 70112, USA. E-mail addresses: [email protected] (L.S. Pointer), [email protected] (Z. AlQurayshi), [email protected] (D.T. Pointer), [email protected] (E. Kandil), [email protected] (D.P. Slakey).
studies have shown improved survival for patients undergoing pancreatic surgery at high volume institutions and by high volume surgeons,6e11 but few studies have evaluated the effects of patient selection on postoperative morbidity and mortality. Historically, comorbidities and decreased physiological reserves have reduced surgeon and oncologist willingness to operate on these patients.12 Many papers have suggested that careful patient preoperative and surgical risk assessment are important to reduce overall morbidity and mortality.11e13 These studies have emphasized known surgical risk factors such as obesity, cardiac status, renal failure and other medical diagnoses. But little specific information about patient factors and community influences have been specifically identified as important for preoperative assessment. In this study we evaluated the association of environmental, social and health risk factors of patients and their communities in relation to outcomes for pancreatic surgery. We hypothesized that
http://dx.doi.org/10.1016/j.amjsurg.2017.02.010 0002-9610/© 2017 Elsevier Inc. All rights reserved.
Please cite this article in press as: Pointer LS, et al., Community health indicators associated with outcomes of pancreatectomy, The American Journal of Surgery (2017), http://dx.doi.org/10.1016/j.amjsurg.2017.02.010
2
L.S. Pointer et al. / The American Journal of Surgery xxx (2017) 1e5
patient health indicators may be a strong predictor of prognosis prior to pancreatic resection. Our aim was to evaluate whether clinical outcomes were associated with the presence of risk factors in patients' communities in hopes that this risk assessment may help surgeons and clinicians better assess patient risks prior to pancreatic surgery. 2. Materials and methods A cross-sectional analysis was performed using the State Inpatient Database (SID) for the states of Florida, New York, and Washington, 2010e2011. SID is part of the Healthcare Cost and Utilization Project, sponsored by the Agency for Healthcare Research and Quality.14 SID includes inpatient discharge records from community hospitals for a given state. The data includes all patients, regardless of payer, providing a unique view of inpatient care in a defined market or state over time. The selection of the aforementioned states was based on their year of availability and completeness of recorded variables. SID is publicly available, de-identified data that is exempt from approval of Institutional Review Board. International Classification of Disease, 9th revision (ICD-9), was used to define parameters of the study. SID was merged with County Health Rankings Data based on patients' county Federal Information Processing Standard (FIPS) codes;15 The County Health Rankings measure the health of nearly all counties in the United States and rank them within states. The rankings are compiled using county-level measures from a variety of national and state data sources. These measures are standardized and combined using scientifically-informed weights.15 The study population consisted of adult (18 years) inpatients who underwent partial pancreatectomy (ICD-9: 52.5) or total pancreatectomy (ICD-9: 52.6) or Whipple procedure (ICD-9: 52.7) as the primary procedure. Re-admission data were included for 30 days after the operation. The independent factor of interest was the health risk status of the community where the patient lived. In creating an overall community health indicator from the County Health Rankings data we adopted a model proposed by Schold et al.16 Nine health measures were selected and each was scored from 0 (lowest risk) to 4 (highest risk) based on quintile classification. Those community measures included: premature death, poor reported general health, poor reported physical health days, poor reported mental health days, adult smoking, adult obesity, preventable hospital stay, high school graduation rate, and income. Subsequently the community health indicators were combined to create an accumulative community health risk indicator with a score range of (0e36), and categorized based on the median score16 into low health risk communities (score 16) and high health risk communities (score> 16). The main study outcomes included postoperative complication, intensive care unit (ICU) admission, hospital length of stay (LOS), and hospital charges. Dichotomized complications were defined as the presence of one or more of the general or specific complications based on the secondary diagnoses made during the hospital stay.17 LOS was categorized into short stay (75th percentile, 14 days) and long stay (>14 days). Charges were adjusted for inflation to reflect 2015 dollar value and categorized as low charges (75th percentile, $153,311.77) or high charges (>$153,311.77). Other independent factors that were considered included: (i) patient demographics: age (<40, 40e70, >70 years old), gender, and race (White, Black, Hispanic, Asian/Pacific Islander, Native American, other); (ii) socioeconomic factors: main payer of health service (Medicare, Medicaid, private insurance, self-pay, no charge, other); (iii) clinical factors: diagnosis (pancreatic malignancy, benign pancreatic disorders, other gastrointestinal (GI) malignancy, other benign GI disorders); pancreatectomy type (partial, total,
Table 1 Descriptive statistics of the study population in relation to the community health risk status. Study population Community Health Risk Status (%) (N ¼ 4272) High Risk (%) Low Risk (%) P valuea (N ¼ 1971) (N ¼ 2301) Age(year) <40 5.5 40 e 70 61.7 >70 32.9 Gender Male 48.3 Female 51.7 Race White 74.7 Black 9.2 Hispanic 8.8 Asian/Pacific Islander 2.5 Native American 0.2 Other 4.6 Service payer Medicare 49.4 Medicaid 8.2 Private insurance 38.9 Self-pay 1.6 No charge 0.4 Other 1.5 Charlson Comorbidity Index Low: 0 e 1 70.2 Medium-low: 2 e 3 27.5 Intermediate: 4 e 5 2.3 High: 6 0.02 BMI 25 kg/m2 No 87.7 Yes 12.3 Diagnosis Pancreatic malignancy 53.9 Benign pancreatic dx 24.7 Other GI malignancy 16.5 Benign other GI dx 4.9 Pancreatectomy Partial 36.0 Total 3.3 Whipple 60.8 Surgeon volume (surgeries/yr) Low: 4 25.8 Intermediate: 5 e 38 49.6 High: 39 24.6 Postoperative complications Absent 86.1 Preset 13.9 ICU admission No 55.9 Yes 44.1 In hospital death Not reported 96.7 Reported 3.3 Readmission No 87.2 Yes 12.9 Length of stay > 14 days No 75.5 Yes 24.5 Total charges > $153,311.77 No 75.0 Yes 25.0
5.8 63.5 30.7
5.1 60.1 34.8
0.016
47.9 52.1
48.7 51.3
0.59
76.6 11.5 6.6 1.6 0.3 3.5
73.1 7.2 10.6 3.3 0.2 5.6
<0.001
49.8 9.2 36.6 2.1 0.2 2.1
49.1 7.3 40.9 1.2 0.6 1.0
<0.001
68.4 28.7 2.9 0.00
71.6 26.6 1.8 0.04
0.020
87.2 12.8
88.1 11.9
0.36
54.1 25.3 15.7 4.9
53.8 24.2 17.3 4.8
0.54
36.7 4.0 59.3
35.3 2.7 62.1
0.019
28.7 52.0 19.3
23.4 47.5 29.1
<0.001
83.7 16.3
88.1 11.9
<0.001
50.6 49.4
58.7 41.3
<0.001
96.5 3.6
97.0 3.0
0.35
87.8 12.2
86.6 13.4
0.22
71.2 28.8
79.2 20.8
<0.001
71.8 28.2
77.7 22.3
<0.001
Abbreviations: dx, diagnosis; GI, gastrointestinal; BMI, body mass index. a Chi-square test.
Whipple); surgeon volume [low-volume (25th percentile, 4 pancreatectomies/year), intermediate-volume (>25th e 75th percentiles, 5e38 pancreatectomies/year), high-volume (>75th percentile, 39 pancreatectomies/year)]; body mass index (BMI < 25 kg/m2 vs. BMI 25 kg/m2); a modification of the
Please cite this article in press as: Pointer LS, et al., Community health indicators associated with outcomes of pancreatectomy, The American Journal of Surgery (2017), http://dx.doi.org/10.1016/j.amjsurg.2017.02.010
L.S. Pointer et al. / The American Journal of Surgery xxx (2017) 1e5 Table 2 The adjusted odds ratio of living in high health risk communities for patients who underwent pancreatectomy. Variable
Age(year) <40 40 e 70 >70 Gender Male Female Race White Black Hispanic Asian/Pacific Islander Native American Other Service payer Private insurance Medicare Medicaid Self-pay
% High Risk Community
Multivariate modela aOR
95% CI
P value
49.4 47.5 43.1
1.39 1.36 Reference
1.01, 1.91 1.16, 1.59
0.045 <0.001
45.7 46.5
Reference 1.04
0.92, 1.18
0.53
47.4 57.9 34.6 29.3 55.6 35.1
Reference 1.38 0.56 0.40 1.01 0.57
1.11, 0.44, 0.26, 0.26, 0.42,
1.72 0.71 0.62 3.87 0.77
0.004 <0.001 <0.001 0.99 <0.001
43.4 46.5 51.9 60.3
Reference 1.31 1.62 2.24
1.12, 1.54 1.27, 2.07 1.34, 3.75
<0.001 <0.001 <0.001
a
Model includes the following variables: age, gender, race, service payer, charlson comorbidity index, and type of pancreatectomy.
Charlson comorbidity index score (CCIS) used to assess patient comorbidities (low score: 0e1, medium-low: 2e3, intermediate: 4e5, high score: 6); and inpatient death.18 Cross-tabulation and Chi-square tests were used to examine the association between each of the independent factors and the outcomes of interest. Factors with significant association were considered confounders and were included in multivariate logistic regression models. Multivariate logistic regression models were used for calculating the odds ratio (OR) and 95% confidence interval (CI). Significance level was set as (a ¼ 0.05). All data analyses were performed using SAS 9.3 for Windows (SAS Institute, Cary, NC, USA). 3. Results A total of 4272 discharge records were identified for the years 2010e2011 (Table 1). The mean age of the study population was 63.5 (±13.1) years and females formed 51.7% of the sample. The majority of the study sample was White (74.7%), and had Medicare coverage (49.8%). Pancreatic malignancy was the most common diagnosis (53.9%), while benign pancreatic disorders and other GI conditions represented 24.7% and 21.4%, respectively. Whipple procedure was performed in 60.8% of the cases, while total pancreatectomy was the least performed procedure (3.3%). Most
3
patients had one or no comorbidities at time of admission (70.2%). Postoperative complications were reported in 595 (13.9%) of the patients and 1196 (44.1%) required ICU admission. A total of 549 (12.9%) patients were readmitted in the 30 days that followed the operations and a total of 140 (3.3%) patients died during the follow up period. Patients living in the high health risk communities were more likely to be < 70 years old, Black, and with Medicaid or Medicare health coverage (p < 0.05 each) (Table 2) (Fig. 1). The risk of postoperative complication was significantly higher for patients living in high health risk communities [OR: 1.41, 95%CI: (1.17, 1.69), p < 0.001] (Table 3) (Fig. 1). Additionally, they were at a higher risk ICU admission [OR: 1.32, 95%CI: (1.11, 1.57), p < 0.001] and a hospital stay of more than 14 days [OR: 1.50, 95%CI: (1.29, 1.76), p < 0.001]. Hospitals that treated patients coming from communities of poor health measures, charged them significantly more than $153,311.77 compared to patients living in low health risk communities [OR: 1.40, 95%CI: (1.20, 1.63), p < 0.001]. 4. Discussion In our analysis several important outcomes for pancreatic surgery were impacted by patient community risk factors. An increase in community risk factors is associated with an increase postoperative complications, ICU admissions, longer length of hospital stay and health care costs. Our study suggests that clinicians and healthcare providers should include community and social determinants of health to assess patient operative risks, and to improve delivery of care, and outcomes for pancreatic surgery. It has been suggested that factors beyond clinical and physiological conditions are associated with morbidity and mortality in the general public.19e26 Health lifestyle and behavioral factors including smoking cigarettes, obesity, drinking alcohol and physical inactivity have been cited as major determinants of premature and preventable morbidity and mortality.19,27 Differences in health outcomes by socioeconomic position is a persistent and increasing public health problem. We found that differences in health indicators by communities may explain variation in outcomes for surgical procedures and other clinical interventions. Few studies have looked at patient specific health risks in relation to pancreatic surgery outcomes. Cheung et al. studied pancreatic cancer care and found that patients of lowest socioeconomic status consistently had lower median survival even when being treated at high volume hospitals.28 In their analysis patients from less affluent areas tended to be African American and Hispanic, have Medicare or Medicaid for insurance coverage, consume more alcohol and were younger at diagnosis. In comparison, our study found that patients in high health risk communities were young, black and had Medicaid or Medicare as their primary insurance. Other studies have found that black patient and patients
Fig. 1. Pancreatectomy risk based on patients' community health risk status.
Please cite this article in press as: Pointer LS, et al., Community health indicators associated with outcomes of pancreatectomy, The American Journal of Surgery (2017), http://dx.doi.org/10.1016/j.amjsurg.2017.02.010
4
L.S. Pointer et al. / The American Journal of Surgery xxx (2017) 1e5
Table 3 The adjusted odds ratio of selected risk factors in relation of patients' community health risk status. Variable
%
Postoperative complications Low risk 11.9 High risk 16.3 ICU admission Low risk 41.3 High risk 49.4 Length of stay > 14 days Low risk 20.8 High risk 28.8 Total charges > $153,311.77 Low risk 22.3 High risk 28.2
Multivariate modela aOR
95% CI
P value
Reference 1.41
1.17, 1.69
<0.001
Reference 1.32
1.11, 1.57
0.001
Reference 1.50
1.29, 1.76
<0.001
Reference 1.40
1.20, 1.63
<0.001
a
Model includes the following variables: community health risk status, age, gender, race service payer, charlson comorbidity index. type of pancreatectomy, diagnosis, and surgeon volume.
of lower socioeconomic status tend to obtain their care in hospitals with higher morbidity and mortality rates,29,30 but our study suggests that inherent environmental factors may play a role in these measured outcomes. For hospitals that treat more high risk patients, outcomes may be worse irrespective of volume. Schold et al. looked at community risk factors in association with kidney transplant outcomes. They used multiple databases to create a community risk score which was merged with data from the Scientific Registry of Transplant Recipients. Increased community risk was found to be associated with increased mortality and decreased likelihood of transplantation. Higher risk communities had higher comorbid conditions that they believed reflected underlying illnesses not routinely identified. Additionally, patients from these communities often received limited access to care.16,31 A different study by Schold, found that health indicators are associated with post-transplant mortality independent of traditional risk factors. In their analysis, community health indicators were felt to explain more variation in outcomes than factors that are traditionally considered clinically relevant. While they note that it is inappropriate to assign risk directly to individuals within a high risk community, conditions of the community where a patient lives may reflex exposure risks. Community indicators such as low birth weight, preventable hospitalization, increased inactivity rate and smoking were associated with overall graft loss and post-transplant mortality. They pointed out that in current hospital reimbursement based on performance assessments, variability in health indicators may lead to decreased access of care for patients from high risk communities.32 Similar to these kidney transplant studies by Schold, our analysis showed a potential impact of community risks on quality assessment of hospitals and surgical centers. Currently. community health indicators are not accounted for in performance evaluations. Hospitals that treat patients from high risk communities are more likely to have poor outcomes unrelated to traditional measures of quality of care. Thus the current reimbursement system may further divide access to high quality care for patients in high risk communities as community factors are not current accounted for in measurements used by regulatory agencies.31 Current performance evaluations based on outcomes alone may potentiate patient care disparities as providers may avoid treatment of patients from higher risk communities as these patients may have been showed to have poorer outcomes, and thus provide poorer reimbursement, irrespective of quality of care provided. The limitations of our study include the administrative nature of the database which lacks clinical details such as histopathological
records or specifics of management. One shortcoming of the dataset is the possibility of missing patients who were readmitted in the year following the study period or readmitted to hospitals outside their state. Given the administrate nature of the data set, there are some limitations that may prevent capturing the complete clinical picture. However, we have used processes which have previously been reported and verified by other studies such as Murphy et al.17 On the other hand, the strengths of the study include the large and generalizable study sample, the availability of a wide array of factors that were assessed for their confounding effects, and the use of a robust and accumulative health indicator developed from twelve county-based health measures. In summary, understanding the impact of a patient's community health status is essential for population-wide health improvement and better delivery of hospital care.33 In recent years elimination of health disparities has become of focus of many public health officials to improve length and quality of life.34 But in current preoperative assessment, patients' behaviors and community health indicators are not routinely assessed or measured. From our analysis, these health indicators are in fact associated outcomes for pancreatic surgery and are an important consideration in healthcare decisions and measurement of healthcare outcomes. More research needs to be done in order to determine how best to use this knowledge in the clinical setting. Grant support None. Conflicts of interest None. References 1. Cameron JL, He J. Two thousand pancreaticoduodenectomies. J Am Coll Surg. 2015;220:530e536. 2. Crist DW, Sitzmann JV, Cameron JL. Improved hospital morbidity, mortality, and survival after the whipple procedure. Ann Surg. 1987;206:358e365. 3. Sohn TA, Yeo CJ, Cameron JL, et al. Resected adenocarcinoma of the pancreas616 patients: results, outcomes, and prognostic indicators. J Gastrointest Surg. 2000;4(6):567e579. 4. Hill JS, McPhee JT, Whalen GF, Sullivan ME, Warshaw AL, Tseng JF. In-hospital mortality after pancreatic resection for chronic pancreatitis: population-based estimates from the nationwide inpatient sample. J Am Coll Surg. 2009;209: 468e476. 5. Winter JM, Cameron JL, Campbell KA, et al. 1423 pancreaticoduodenectomies for pancreatic cancer: a single-institution experience. J Gastrointest Surg. 2006;10:1199e1210. 6. Gooiker GA, van Gijn W, Wouters MW, et al. Systematic review and metaanalysis of the volume-outcome relationship in pancreatic surgery. Br J Surg. 2011;98:485e494. 7. Kotwall CA, Maxwell JG, Brinker CC, et al. National estimates of mortality rates for radical pancreaticoduodenectomy in 25,000 patients. Ann Surg Oncol. 2002;9:847e854. 8. Ho V, Heslin MJ. Effect of hospital volume and experience on in-hospital mortality for pancreaticoduodenectomy. Ann Surg. 2003;237:509e514. 9. Wittel UA, Makowiec F, Sick O, et al. Retrospective analyses of trends in pancreatic surgery: indications, operative techniques, and postoperative outcome of 1,120 pancreatic resections. World J Surg Oncol. 2015;13, 102e0150525-6. 10. Gordon TA, Burleyson GP, Tielsch JM, et al. The effects of regionalization on cost and outcome for one general high-risk surgical procedure. Ann Surg. 1995;221: 43e49. 11. Eeson G, Chang N, McGahan CE, et al. Determination of factors predictive of outcome for patients undergoing a pancreaticoduodenectomy of pancreatic head ductal adenocarcinomas. HPB (Oxford). 2012;14:310e316. 12. Adham M, Bredt LC, Robert M, et al. Pancreatic resection in elderly patients: should it be denied? Langenbecks Arch Surg. 2014;399:449e459. 13. Yeo CJ, Cameron JL, Sohn TA, et al. Six hundred fifty consecutive pancreaticoduodenectomies in the 1990s: pathology, complications, and outcomes. Ann Surg. 1997;226:248e257. 14. Healthcare Cost and Utilization Project (HCUP). Overview of the state inpatient databases. http://hcup-us.ahrq.gov/sidoverview.jsp. Updated 2014.
Please cite this article in press as: Pointer LS, et al., Community health indicators associated with outcomes of pancreatectomy, The American Journal of Surgery (2017), http://dx.doi.org/10.1016/j.amjsurg.2017.02.010
L.S. Pointer et al. / The American Journal of Surgery xxx (2017) 1e5 15. Community Health Rankings and Roadmaps. Rankings. http://www. countyhealthrankings.org/our-approach. Updated 2014. 16. Schold JD, Buccini LD, Kattan MW, et al. The association of community health indicators with outcomes for kidney transplant recipients in the United States. Arch Surg. 2012;147:520e526. 17. Murphy MM, Knaus 2nd WJ, Ng SC, et al. Total pancreatectomy: a national study. HPB (Oxford). 2009;11:476e482. 18. Quan H, Sundararajan V, Halfon P, et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care. 2005;43: 1130e1139. 19. Lantz PM, House JS, Lepkowski JM, et al. Socioeconomic factors, health behaviors, and mortality: results from a nationally representative prospective study of US adults. JAMA. 1998;279:1703e1708. 20. Berkman LF, Breslow L. Health and Ways of Living. New York, NY: Oxford University Press; 1983. 21. US Department of Health and Human Services. Healthy People 2000: National Health Promotion and Disease Prevention Objectives. Washington, DC: US Dept of Health and Human Services; 1990. DHHS publication 91e50212. 22. McGinnis MJ, Foege WH. Actual causes of death in the United States. JAMA. 1993;270:2207e2212. 23. Wiley JA, Camacho TC. Life-style and future health: evidence from the alameda county study. Prev Med. 1980;9:1e21. 24. Wilson PS. Established risk factors and coronary artery disease: the Framingham study. Am J Hypertens. 1994;7:75e125.
5
25. Hirdes JP, Forbes WF. The importance of social relationships, socioeconomic status and health practices with respect to mortality among Ontario males. J Clin Epidemiol. 1992;554:175e182. 26. Patterson RE, Haines PS, Pokin BM. Health life- style patterns of US adults. Prev Med. 1994;23:453e460. 27. Galea S, Tracy M, Hoggatt KJ, et al. Estimated deaths attributable to social factors in the United States. Am J Public Health. 2011;101(8):1456e1465. 28. Cheung MC, Yang R, Byrne MM, et al. Are patients of low socioeconomic status receiving suboptimal management for pancreatic adenocarcinoma? Cancer. 2010;116:723e733. 29. Girotti ME, Shih T, Revels S, et al. Racial disparities in readmissions and site of care for major surgery. J Am Coll Surg. 2014;218:423e430. 30. Silverman DT, Hoover RN, Brown LM, et al. Why do black Americans have a higher risk of pancreatic cancer than white Americans? Epidemiology. 2003;14(1):45e54. 31. Slakey DP. Surgical outcomes beyond the individual. Arch Surg. 2012;147(6): 526e527. 32. Schold JD, Heaphy EL, Buccini LD, et al. Prominent impact of community risk factors on kidney transplant candidate processes and outcomes. Am J Transpl. 2013;13(9):2374e2383. 33. Russell MW, Campbell LA, Kisely S, et al. The development of community health indicators: a district-wide approach. Chronic Dis Can. 2011;31:65e70. 34. Metzler M, Kanarek N, Highsmith K, et al. Community health status indicators project: the development of a national approach to community health. Prev Chronic Dis. 2008;5:A94.
Please cite this article in press as: Pointer LS, et al., Community health indicators associated with outcomes of pancreatectomy, The American Journal of Surgery (2017), http://dx.doi.org/10.1016/j.amjsurg.2017.02.010