[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Download PDF
Figure 1.
Total Hospital Costs for Liver and Pancreatic Resections Performed Between 2002 and 2011 by Individual Hospital
Total Hospital Costs for Liver and Pancreatic Resections Performed Between 2002 and 2011 by Individual Hospital
Figure 2.
Forest Plot of Results of Multivariable Hierarchical Linear Regression Model Quantifying the Effects of Patient, Hospital, and Operative Characteristics on Total Hospital Costs for Patients Undergoing Liver or Pancreatic Surgery Between 2002-2011
Forest Plot of Results of Multivariable Hierarchical Linear Regression Model Quantifying the Effects of Patient, Hospital, and Operative Characteristics on Total Hospital Costs for Patients Undergoing Liver or Pancreatic Surgery Between 2002-2011

Error bars indicate SD; LOS, length of stay.

Table 1.  
Patient, Hospital and Operative Characteristics of Study Populationa
Patient, Hospital and Operative Characteristics of Study Populationa
Table 2.  
Postoperative Outcomes of Study Populationa
Postoperative Outcomes of Study Populationa
Table 3.  
Multivariable Hierarchical Linear Regression Analysis for Total Hospital Costs
Multivariable Hierarchical Linear Regression Analysis for Total Hospital Costs
1.
Muñoz  E, Muñoz  W  III, Wise  L.  National and surgical health care expenditures, 2005-2025. Ann Surg. 2010;251(2):195-200.PubMedArticle
2.
Wennberg  JE, Peters  PG  Jr.  Unwarranted variations in the quality of health care: can the law help medicine provide a remedy/remedies? Spec Law Dig Health Care Law. 2004;(305):9-25.PubMed
3.
Wennberg  JE.  Practice variations and health care reform: connecting the dots. Health Aff (Millwood). 2004;suppl variation:VAR140-VAR144.PubMed
4.
Hussey  PS, Huckfeldt  P, Hirshman  S, Mehrotra  A.  Hospital and regional variation in Medicare payment for inpatient episodes of care. JAMA Intern Med. 2015;175(6):1056-1057.PubMedArticle
5.
Newhouse  JP, Garber  AM.  Geographic variation in health care spending in the United States: insights from an Institute of Medicine report. JAMA. 2013;310(12):1227-1228.PubMedArticle
6.
Congressional Budget Office. Budget Options, Volume 1: Health Care. https://www.cbo.gov/publication/41747?index=9925. Accessed May 26, 2015.
7.
Zhang  Y, Baik  SH, Fendrick  AM, Baicker  K.  Comparing local and regional variation in health care spending. N Engl J Med. 2012;367(18):1724-1731.PubMedArticle
8.
Birkmeyer  JD, Gust  C, Baser  O, Dimick  JB, Sutherland  JM, Skinner  JS.  Medicare payments for common inpatient procedures: implications for episode-based payment bundling. Health Serv Res. 2010;45(6, pt 1):1783-1795.PubMedArticle
9.
Delisle  DR.  Big things come in bundled packages: implications of bundled payment systems in health care reimbursement reform. Am J Med Qual. 2013;28(4):339-344.PubMedArticle
10.
Mavros  MN, de Jong  M, Dogeas  E, Hyder  O, Pawlik  TM.  Impact of complications on long-term survival after resection of colorectal liver metastases. Br J Surg. 2013;100(5):711-718.PubMedArticle
11.
Mayo  SC, Heckman  JE, Shore  AD,  et al.  Shifting trends in liver-directed management of patients with colorectal liver metastasis: a population-based analysis. Surgery. 2011;150(2):204-216.PubMedArticle
12.
Mayo  SC, Pulitano  C, Marques  H,  et al.  Surgical management of patients with synchronous colorectal liver metastasis: a multicenter international analysis. J Am Coll Surg. 2013;216(4):707-716.PubMedArticle
13.
Short  MN, Aloia  TA, Ho  V.  The influence of complications on the costs of complex cancer surgery. Cancer. 2014;120(7):1035-1041.PubMedArticle
14.
Patel  AS, Bergman  A, Moore  BW, Haglund  U.  The economic burden of complications occurring in major surgical procedures: a systematic review. Appl Health Econ Health Policy. 2013;11(6):577-592.PubMedArticle
15.
Vollmer  CM  Jr.  The economics of pancreas surgery. Surg Clin North Am. 2013;93(3):711-728.PubMedArticle
16.
Kachare  SD, Liner  KR, Vohra  NA, Zervos  EE, Hickey  T, Fitzgerald  TL.  Assessment of health care cost for complex surgical patients: review of cost, re-imbursement and revenue involved in pancreatic surgery at a high-volume academic medical centre. HPB (Oxford). 2015;17(4):311-317.PubMedArticle
17.
Nathan  H, Atoria  CL, Bach  PB, Elkin  EB.  Hospital volume, complications, and cost of cancer surgery in the elderly. J Clin Oncol. 2015;33(1):107-114.PubMedArticle
18.
Avritscher  EBC, Cooksley  CD, Rolston  KV,  et al.  Serious postoperative infections following resection of common solid tumors: outcomes, costs, and impact of hospital surgical volume. Support Care Cancer. 2014;22(2):527-535.PubMedArticle
19.
Charlson  ME, Pompei  P, Ales  KL, MacKenzie  CR.  A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40(5):373-383.PubMedArticle
20.
Amini  N, Spolverato  G, Kim  Y, Pawlik  TM.  Trends in hospital volume and failure to rescue for pancreatic surgery. J Gastrointest Surg. 2015;19(9):1581-1592.PubMedArticle
21.
Bureau of Labor Statistics USD of L. CPI Inflation Calculator. http://www.bls.gov/data/inflation_calculator.htm.
22.
Kilic  A, Shah  AS, Conte  JV,  et al.  Understanding variability in hospital-specific costs of coronary artery bypass grafting represents an opportunity for standardizing care and improving resource use. J Thorac Cardiovasc Surg. 2014;147(1):109-115.PubMedArticle
23.
Dimick  JB, Pronovost  PJ, Cowan  JA, Lipsett  PA.  Complications and costs after high-risk surgery: where should we focus quality improvement initiatives? J Am Coll Surg. 2003;196(5):671-678.PubMedArticle
24.
Brooks  GA, Li  L, Uno  H, Hassett  MJ, Landon  BE, Schrag  D.  Acute hospital care is the chief driver of regional spending variation in Medicare patients with advanced cancer. Health Aff (Millwood). 2014;33(10):1793-1800.PubMedArticle
25.
Haut  ER, Pronovost  PJ, Schneider  EB.  Limitations of administrative databases. JAMA. 2012;307(24):2589.PubMedArticle
Original Investigation
February 2016

Factors Associated With Interhospital Variability in Inpatient Costs of Liver and Pancreatic Resections

Author Affiliations
  • 1Department of Epidemiology, Johns Hopkins University Bloomberg School of Public Health, Baltimore, Maryland
  • 2Department of Surgery, Johns Hopkins University School of Medicine, Baltimore, Maryland
  • 3Department of Surgery, University of Illinois Hospital and Health Sciences System, Chicago
JAMA Surg. 2016;151(2):155-163. doi:10.1001/jamasurg.2015.3618
Abstract

Importance  In an era of accountable care, understanding variation in health care costs is critical to reducing health care spending.

Objective  To identify factors associated with increased hospital costs and quantify variations in costs among individual hospitals in patients undergoing liver and pancreatic surgery in the United States.

Design, Setting, and Participants  Retrospective analysis of total costs among 42 480 patients undergoing hepatopancreaticobiliary surgery from January 1, 2002, through December 31, 2011, using a nationally representative data set (Nationwide Inpatient Sample of the Healthcare Cost and Utilization Project). Analysis was conducted in May 2015.

Main Outcomes and Measures  Total inpatient costs and proportional variation in inpatient costs among individual hospitals.

Results  Among the 42 480 patients who underwent liver or pancreatic resection, the median age was 62 years, 52.4% were female, and 72.9% had a Charlson Comorbidity Index of 2 or higher. The median cost for the entire cohort was $21 535 (interquartile range, $15 373-$31 104), varying from $3320 to $279 102 among individual hospitals. On multivariable analysis, increasing patient comorbidity (coefficient, 2000.30; 95% CI, 1363.33-2637.27; P < .001) and operative characteristics (total pancreatectomy: coefficient, 12 742.31; 95% CI, 10 063.66-15 420.94; P < .001; lobectomy: coefficient, 6336.42; 95% CI, 3934.61-8737.24; P < .001) were associated with higher hospital costs. The development of postoperative complications, such as sepsis (coefficient, 30 571.25; 95% CI, 29 308.96-31 833.54; P < .001) or stroke (coefficient, 8925.34; 95% CI, 2801.38-15 049.30; P = .004), and a longer length of stay were most strongly predictive of higher inpatient cost (length of stay >14 days: coefficient, 44 162.24; 95% CI, 43 125.56-45 198.92; P < .001). After adjusting for patient and hospital characteristics, the overall cost of hepatopancreaticobiliary surgery varied by $9000 among individual hospitals.

Conclusions and Relevance  Significant variability was noted in hospital costs among patients undergoing pancreatic and liver surgery. Future policies should focus on reducing variations in costs by promoting payment paradigms that support a better quality of care and lower costs.

Introduction

An estimated 17% of the annual gross domestic product is spent each year in health care costs, making the US health care system one of the most expensive worldwide.1 Reducing unwanted and therefore wasteful resource utilization has recently been targeted as a potential area for decreasing health care spending.2 Defined as variations in care that are unexplained by disease characteristics, patient preferences, or the dictates of evidence-based medicine, unwanted resource utilization has been reported among a wide variety of diseases and procedures.3 Although previously thought to be largely owing to regional differences in practices and payment structures, previous reports47 suggest that a significant proportion of this variation is more likely because of differences among providers, including physicians and hospitals. Policymakers have proposed provider-focused interventions, such as bundled-payment systems, accountable care organizations, and value-based purchasing, to reduce escalating health care costs while maintaining a high quality of care.8,9

Hepatic and pancreatic resections represent complex surgical procedures that remain the mainstay of treatment for benign and malignant conditions of the hepatopancreaticobiliary (HPB) system. Although mortality after these procedures has steadily decreased over time, HPB operations continue to represent complex surgical procedures associated with high postoperative morbidity.1012 Therefore, the limited research assessing costs associated with HPB surgery have not surprisingly revealed that hospital costs are primarily driven by postoperative complications and postdischarge care.1315 For example, Kachare et al16 noted that total hospital costs were almost 5 times greater among patients developing a postoperative complication after pancreatic surgery, whereas Nathan and colleagues17 found that hospital reimbursements were nearly 50% higher among patients who developed a postoperative complication after HPB surgery.16,17 In contrast, another study18 has suggested that higher hospital volume, improved clinical pathways, and surgeon experience are associated with decreased hospital expenses. Although providing valuable insight into the variations in costs, these studies are limited to single-center reports or report hospital charges that are an inaccurate representation of hospital financials and confounded by policies of insurance practitioners and institutions. The current study aimed to describe variations in costs after HPB surgery using a nationally representative data set. In particular, we sought to identify factors associated with increased hospital costs and quantify variations in costs among individual hospitals across the United States.

Methods
Data Source and Patient Population

This cross-sectional study was performed in May 2015 using data from the Nationwide Inpatient Sample (NIS) of the Healthcare Cost and Utilization Project collected from January 1, 2002, through December 31, 2011. Created and maintained by the Agency for Healthcare Research and Quality, the NIS represents the largest in-patient, all-payer database in the United States. Per year, the database includes information from 8 million inpatient admissions collected from more than 1000 hospitals in 46 states. Using a stratified sampling method based on hospital-level characteristics (geographic region, teaching status, hospital size, and urban vs rural location), the NIS is a 20% representative sample of all inpatient hospital visits in the United States. Using deidentified data, this study was deemed exempt by the Johns Hopkins Hospital Institutional Review Board.

International Classification of Disease, Ninth Revision, Clinical Manifestation (ICD-9-CM) procedure codes were used to identify patients older than 18 years undergoing liver (codes 50.22 and 50.3) and pancreatic (codes 52.51, 52.52, 52.53, 52.59, 52.6, and 52.7) resections. To ensure homogeneity of the patient population, patients undergoing any other concomitant surgery or surgery performed on an emergency basis were excluded from the study population. Patient comorbidity was classified according to the Charlson Comorbidity Index (CCI), categorizing patients into 3 groups accordingly (CCI of 0, 1, or ≥2).19 Individual hospitals were identified by a unique hospital identifier for further comparison. Hospital-level variables included teaching status, location (geographic region and urban vs rural location), hospital bed size, and annual surgical volume. Postoperative complications were described using previously validated ICD-9-CM diagnosis codes.20 Postoperative complications included respiratory failure, pneumonia, acute myocardial infarction, cardiac arrest, acute renal failure, sepsis, stroke, venous thromboembolism, gastrointestinal hemorrhage, and surgical site infections.

Statistical Analysis

Continuous variables were described as means (SDs) or medians (interquartile ranges [IQRs]) and compared using the t test or the nonparametric Kruskal-Wallis test as applicable. Categorical variables were similarly expressed as whole numbers and proportions and compared using the Pearson χ2 test.

Total hospital charges as provided within the data set were inflation adjusted to and reported as 2012 US dollars using the Consumer Price Index maintained by the US Department of Labor.21 As previously noted, hospital charges may be confounded by payer policies and other factors unrelated to resource utilization. Therefore, total hospital costs were calculated for each hospital using hospital-specific cost to charge ratios developed by the Agency for Healthcare Research and Quality.

To examine factors associated with total hospital costs after liver or pancreatic surgery, we built a mixed-effects multivariable linear regression model adjusting for patient- and hospital-level characteristics. Following the natural hierarchical structure whereby patients are clustered within hospitals each with their own respective characteristics and policies, a random-effects intercept was specified at the hospital level. Selection of patient- and hospital-specific variables (fixed effects) was based on results from univariable analysis. In particular, variables that demonstrated statistically significant associations, defined as P < .05, were included in multivariable analysis. Results from the multivariable analysis are presented as coefficients with corresponding 95% CIs and represent the difference in total in-hospital costs between patients groups. Using previously described methods, we quantified interhospital variability in total costs by the SD of the hospital effect calculated from multivariable analysis.22 Statistical significance for all tests was defined as P < .05. All analyses were performed using STATA software, version 12.0 for Windows (StataCorp).

Results
Patient and Hospital Characteristics

A total of 42 480 patients underwent a liver or pancreatic resection and met inclusion criteria (Table 1). The median age of the cohort was 62 years (IQR, 51-71 years), and 52.4% were female (n = 22 225 [52.4%]). Comorbidities were common, with more than two-thirds of patients presenting with a CCI of 2 or higher (n = 30 957 [72.9%]). Chronic obstructive disease (n = 4831 [11.4%]), congestive heart failure (n = 1196 [2.8%]), and chronic renal disease (n = 969 [2.3%]) were the most common comorbidities. Of note, nearly half of all patients were insured by private insurers (n = 20 603 [48.6%]), whereas two-fifths of patients were Medicare enrollees (n = 17 169 [40.5%]). As expected, most operations were performed at urban centers (n = 41 313 [97.7%]) with an affiliated teaching facility (n = 35 636 [84.3%]).

Several patient and hospital characteristics differed between patients undergoing a liver compared with pancreatic resection (Table 1). Among the 18 923 patients (44.5%) who underwent a liver resection, 5874 (31.0%) underwent a hepatic lobectomy, whereas a partial hepatectomy was performed in 13 049 (69.0%). Of the 23 557 patients (55.5%) in whom a pancreatic resection was performed, more than half (n = 12 968 [55.1%]) underwent a pancreaticoduodenectomy and a third of patients underwent a distal pancreatectomy (n = 7573 [32.2%]). The remaining 3016 patients underwent a proximal (n = 340 [1.4%]), subtotal (n = 203 [0.9%]), partial (n = 1427 [6.1%]), or total pancreatectomy (n = 1176 [5.0%]). Patients undergoing liver surgery were younger (median age, 59 years [IQR, 49-69 years] vs 63 years [IQR, 53-72 years]; P < .001) and proportionately more likely to be privately insured (10 044 [53.1%] vs 10 559 [44.9%], P < .001), whereas patients undergoing pancreatic surgery were more likely to be covered by Medicare (6652 [35.3%] vs 10 517 [44.7%], P < .001). Although commonly noted among all patients, patients undergoing liver surgery were proportionally more likely to present with comorbidity (CCI ≥2: 15 172 [80.3%] vs 15 782 [67.0%]; P < .001). Of interest, however, chronic obstructive pulmonary disease (1886 [10.0%] vs 2945 [12.5%]), congestive heart failure (443 [2.3%] vs 753 [3.2%]), and chronic renal disease (380 [2.0%] vs 589 [2.5%]) were all more commonly noted among patients undergoing pancreatic surgery (all P < .05). Compared with pancreatic surgery, liver resections were more likely to have been performed at a teaching hospital (16 195 [86.0%] vs 19 441 [82.8%], P < .001) and less likely to have been performed at a hospital in a rural setting (401 [2.1%] vs 572 [2.4%], P = .03).

Postoperative Outcomes

Overall, postoperative complications were observed in 13 580 patients (32.0%) in the study cohort, with patients undergoing a pancreatic resection more likely to develop a postoperative complication vs patients who underwent liver surgery (5158 [27.3%] vs 8422 [35.8%], P < .001). Respiratory complications (respiratory failure: 1091 [5.8%] vs 1927 [8.2%]; pneumonia: 679 [3.6%] vs 1351 [5.7%]), sepsis (571 [3.0%] vs 1199 [5.1%]), surgical site infections (793 [4.2%] vs 1930 [8.2%]), and cardiac complications (605 [3.2%] vs 1067 [4.5%]) were more commonly noted among patients after pancreatic surgery (all P < .001; Table 2). This pattern was observed throughout the 10-year duration of the study. In particular, postoperative complications after pancreatic surgery were observed in 31.9% to 39.1% of patients compared with 22.9% to 30.1% of patients undergoing a liver resection (eFigure 1 in the Supplement). Therefore, patients undergoing liver surgery not surprisingly had a shorter length of stay (LOS) than patients discharged after pancreatic surgery (median LOS, 6 days [IQR, 5-8 days] vs 7 days [IQR, 6-10 days]; P < .001). Although less than 20% of patients undergoing liver surgery had an LOS greater than 9 days (n = 3651 [19.1%]), nearly a third of patients who had a pancreatic resection had an LOS greater than 9 days (n = 3093 [28.9%], P < .001).

Total Hospital Costs

The median total hospital cost for surgery for the entire cohort was $24 460 (IQR, $17 077-$36 067), with notably higher costs associated with pancreatic surgery compared with liver surgery (median cost, $21 535 [IQR, $15 373-$31 104] vs $27 096 [IQR, $19 013-$40 467]; P < .001). This trend in total hospital costs was observed across the study period from 2002 to 2011 (eFigure 2 in the Supplement). Significant differences were similarly noted in total costs among individual hospitals, ranging from $3320 to $279 102 (Figure 1). Furthermore, specific hospital characteristics were associated with a higher total cost. For example, total hospital costs were noted to be higher at teaching hospitals (median cost, $24 792 [IQR, $17 527-$36 071] vs $22 203 [IQR, $14 629-$35 977]; P < .001) and centers located in an urban setting (median cost, $24 528 [IQR, $17 135-$36 174] vs $20 852 [IQR, $15 123-$30 723]; P < .001).

To further explore variations in total hospital costs, a multivariable hierarchical linear regression model adjusting for patient- and hospital-level characteristics was built with a random-effects intercept at the hospital level. On multivariable analysis, several patient characteristics were associated with increased total hospital costs (Table 3). For example, female patients and patients with preexisting comorbidity were associated with a higher total cost (all P < .05). Of note, although no differences in cost were noted among patients with a CCI of 1 and patients without any comorbidity, patients with a CCI of 2 or higher had more than $2000 higher in-patient costs (coefficient, 2000.30; 95% CI, 1363.33-2637.27; P < .001). Differences in cost were similarly noted by procedure, with patients undergoing a hepatic lobectomy (coefficient, 6336.42; 95% CI, 3934.61-8738.24; P < .001), a partial pancreatectomy (coefficient, 6664.79; 95% CI, 3982.67-9346.92; P < .001), or a total pancreatectomy (coefficient, 12 742.31; 95% CI, 10 063.66-15 420.94; P < .001) having higher total cost compared with patients undergoing other HPB procedures. As expected, patients who developed a postoperative complication had higher costs vs patients who had an uncomplicated postoperative course. In particular, patients who developed postoperative sepsis (coefficient, 30 571.25; 95% CI, 29 308.96-31 833.54; P < .001) or acute renal failure (coefficient, 15 863.47; 95% CI, 14 794.04-16 932.90; P < .001) after surgery had the highest incremental increase in total hospital costs. Furthermore, compared with patients with an LOS less than 5 days, total hospital costs among patients with an LOS greater than 14 days were more than $44 000 higher (coefficient, 44 162.24; 95% CI, 43 125.56-45 198.92; P < .001). In addition, several hospital parameters were also associated with a higher total hospital cost. Although no differences in costs were noted between urban and rural hospitals on multivariable analysis, teaching hospitals were noted to have a $1900 higher cost for patients undergoing elective HPB procedures (coefficient, 1904.22; 95% CI, 579.98-3228.46; P = .001). Regional differences were also noted in total costs, with Western centers having approximately $9300 higher total cost for surgery (coefficient, 9358.53; 95% CI, 7482.05-11 235.00; P < .001). Total costs were similarly noted to vary by annual hospital volume. Of interest, compared with high-volume centers, total hospital costs were noted to be almost $4000 and $3000 lower at intermediate and low-volume hospitals, respectively (intermediate volume: coefficient, −4150.86; 95% CI, −5787.10 to −2514.62; P < .001; low volume: coefficient, −2883.10; 95% CI, −4034.92 to −1731.28; P < .001).

Using results from the multivariable analysis, we quantified interhospital variability in total costs. Despite adjusting for patient and hospital characteristics, total inpatient costs varied by approximately $9000 (coefficient, 8987.06; 95% CI, 8374.91-9643.96; P < .001) among hospitals in patients undergoing HPB surgery (Table 3). Of note, among patients who did not develop a postoperative complication, hospital characteristics, such as hospital region and annual volume, were more predictive of a higher cost compared with patient-specific characteristics, such as age, comorbidity, and type of surgery (Figure 2).

Discussion

Variations in health care provision not explained by disease, patient preference, or the dictates of evidence-based practices are frequently cited as evidence of unnecessary or wasteful resource utilization.2,3 On the basis of a report by the Institute of Medicine that suggests that variations in care were strongly associated with differences in physician and hospital characteristics and practice, policymakers have proposed provider-focused policies, including bundled-payment systems, accountable care organizations, and value-based purchasing, as potential means to curtail increasing health care costs.4,5 Despite this increasing emphasis on cost control, there are limited data describing variations in and factors associated with cost among the surgical population. With the use of a nationally representative data set, the current study examined and quantified variations in total hospital costs among 42 480 patients undergoing either a liver or pancreatic resection in the United States. Of note, we found that although patient-specific factors, such as patient age, sex, and comorbidity, were important drivers of cost, a larger proportion of the variation in inpatient costs was noted at the level of the hospital. In particular, regional differences were noted in total hospital costs and differences in teaching status. Of interest, the major drivers of differences in cost were the development of postoperative complications and an increased LOS. However, after adjusting for patient- and hospital-level characteristics, total costs were noted still to vary by approximately $9000 among individual hospitals.

Although a previous study23 highlighted increased resource utilization among patients who develop a postoperative complication after major surgery, few studies provide data pertaining to the relative economic implications of postoperative complications. The current study noted that patients who developed a postoperative complication had a marked incremental increase in costs that ranged from $6000 to $30 000. Although the development of postoperative complications may reflect differences in quality of care, they are likely also a consequence of differences in patient characteristics, such as comorbidity or severity of illness. Supporting this notion, the current study found that patients with greater comorbidity (CCI ≥2) had a $2000 higher hospital cost. Furthermore, patients undergoing more complex procedures, such as total pancreatectomy or hepatic lobectomy, had the highest inpatient costs of care. Similar to these findings, the study by Kachare et al16 found that patients undergoing a pancreaticoduodenectomy had the highest hospital costs compared with patients undergoing other pancreatic resections. Taken together, findings of the current study suggest that quality improvement strategies should not only target practitioners and processes but also account for the inherent variations in patient case mix and disease severity.

Another important finding of the current study was that total hospital costs were noted to vary by geographic region. In particular, despite accounting for patient and hospital characteristics, the cost of HPB operations was noted to be approximately $9300 higher at Western hospitals. Although regional variations in health care costs have been previously described, there is disagreement regarding the appropriate policy responses.4,5,7 Proposed regional policies include reducing Medicare payments to high-spending regions and limiting the supply of health care resources using certificate-of-need criteria.4 Supporting the implementation of regional policies, previous reports5,24 have suggested that decreasing regional variations for inpatient services would result in an overall 27% reduction in Medicare expenditures. In contrast, a previous study4 has called into question the implementation of such policies, indicating that doing so may unfavorably penalize high-performing hospitals and health care professionals in low-performing regions. Supporting this notion, the current study noted higher costs among patients treated at teaching hospitals and hospitals with a higher surgical volume. The implementation of traditional policies to penalize such hospitals caring for a large proportion of complex and uninsured patients may serve to widen health care disparities and in fact decrease the overall quality of care. Furthermore, other studies4,5 have suggested that rather than regional variations in health care, differences in hospital and physician characteristics better explain variations in health care. The current study similarly noted that although regional differences were important, the relative effect among individual hospitals was noted to be higher. In particular, we noted that independent of patient and hospital characteristics, the costs of surgery varied by approximately $9000 among hospitals. Hussey and colleagues4 similarly found that although hospital referral regions accounted for up to a third of variations in costs with different conditions, most variation in costs was attributable to differences among individual hospitals. Results from this study therefore collectively suggest the variations in costs would be best reduced by a reorientation of practice from traditional high-volume practice to a more deliberate value-oriented care.

The current study should be interpreted with several limitations. Using data collected from administrative claims, the study was bound by limitations inherent to administrative data.25 In particular, the data set lacked granular details pertaining to cost and resource utilization. Itemizing the individual components of the total cost, such as operating room costs, ward costs, pharmacy costs, and postoperative investigations or procedures, may assist in elucidating the specific reasons for the significant interhospital cost variability. Furthermore, because data were limited to the inpatient experience of patients, we could not comment on variations in the costs of postdischarge care, which is a significant contributor to variable health care costs. Despite these limitations, the use of a large, nationally representative data set facilitated a broad, generalizable comparison among US hospitals. In addition, unlike previous studies that assessed hospital charges that may be confounded by the preferences of insurers, this study examined total hospital costs, which allowed for a more accurate measure of health care spending and resource utilization.

Conclusions

Significant variability was noted in hospital costs among patients undergoing pancreatic and liver surgery. Although a proportion of this variability was accounted for by patient and operative factors, the most significant driver of increased costs was the development of postoperative complications. Furthermore, despite accounting for patient case mix and hospital characteristics, significant interhospital variations in costs were noted for patients undergoing HPB surgery. Newer, more robust policies and payment services are necessary to decrease anomalous variations in surgical costs while promoting a high-quality, cost-effective provision of care.

Back to top
Article Information

Accepted for Publication: July 6, 2015.

Corresponding Author: Timothy M. Pawlik, MD, MPH, PhD, Department of Surgery, Johns Hopkins Hospital, 600 N Wolfe St, Blalock 665, Baltimore, MD 21287 (tpawlik1@jhmi.edu).

Published Online: October 28, 2015. doi:10.1001/jamasurg.2015.3618.

Author Contributions: Dr Pawlik had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Nelson-Williams, Gani, Kilic, Spolverato, Pawlik.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Nelson-Williams, Gani, Spolverato, Amini.

Critical revision of the manuscript for important intellectual content: Nelson-Williams, Gani, Kilic, Spolverato, Kim, Wagner, Ejaz, Pawlik.

Statistical analysis: Nelson-Williams, Gani, Kilic, Kim, Ejaz.

Administrative, technical, or material support: Nelson-Williams, Spolverato, Wagner, Pawlik.

Study supervision: Nelson-Williams, Wagner, Amini, Pawlik.

Conflict of Interest Disclosures: None reported.

References
1.
Muñoz  E, Muñoz  W  III, Wise  L.  National and surgical health care expenditures, 2005-2025. Ann Surg. 2010;251(2):195-200.PubMedArticle
2.
Wennberg  JE, Peters  PG  Jr.  Unwarranted variations in the quality of health care: can the law help medicine provide a remedy/remedies? Spec Law Dig Health Care Law. 2004;(305):9-25.PubMed
3.
Wennberg  JE.  Practice variations and health care reform: connecting the dots. Health Aff (Millwood). 2004;suppl variation:VAR140-VAR144.PubMed
4.
Hussey  PS, Huckfeldt  P, Hirshman  S, Mehrotra  A.  Hospital and regional variation in Medicare payment for inpatient episodes of care. JAMA Intern Med. 2015;175(6):1056-1057.PubMedArticle
5.
Newhouse  JP, Garber  AM.  Geographic variation in health care spending in the United States: insights from an Institute of Medicine report. JAMA. 2013;310(12):1227-1228.PubMedArticle
6.
Congressional Budget Office. Budget Options, Volume 1: Health Care. https://www.cbo.gov/publication/41747?index=9925. Accessed May 26, 2015.
7.
Zhang  Y, Baik  SH, Fendrick  AM, Baicker  K.  Comparing local and regional variation in health care spending. N Engl J Med. 2012;367(18):1724-1731.PubMedArticle
8.
Birkmeyer  JD, Gust  C, Baser  O, Dimick  JB, Sutherland  JM, Skinner  JS.  Medicare payments for common inpatient procedures: implications for episode-based payment bundling. Health Serv Res. 2010;45(6, pt 1):1783-1795.PubMedArticle
9.
Delisle  DR.  Big things come in bundled packages: implications of bundled payment systems in health care reimbursement reform. Am J Med Qual. 2013;28(4):339-344.PubMedArticle
10.
Mavros  MN, de Jong  M, Dogeas  E, Hyder  O, Pawlik  TM.  Impact of complications on long-term survival after resection of colorectal liver metastases. Br J Surg. 2013;100(5):711-718.PubMedArticle
11.
Mayo  SC, Heckman  JE, Shore  AD,  et al.  Shifting trends in liver-directed management of patients with colorectal liver metastasis: a population-based analysis. Surgery. 2011;150(2):204-216.PubMedArticle
12.
Mayo  SC, Pulitano  C, Marques  H,  et al.  Surgical management of patients with synchronous colorectal liver metastasis: a multicenter international analysis. J Am Coll Surg. 2013;216(4):707-716.PubMedArticle
13.
Short  MN, Aloia  TA, Ho  V.  The influence of complications on the costs of complex cancer surgery. Cancer. 2014;120(7):1035-1041.PubMedArticle
14.
Patel  AS, Bergman  A, Moore  BW, Haglund  U.  The economic burden of complications occurring in major surgical procedures: a systematic review. Appl Health Econ Health Policy. 2013;11(6):577-592.PubMedArticle
15.
Vollmer  CM  Jr.  The economics of pancreas surgery. Surg Clin North Am. 2013;93(3):711-728.PubMedArticle
16.
Kachare  SD, Liner  KR, Vohra  NA, Zervos  EE, Hickey  T, Fitzgerald  TL.  Assessment of health care cost for complex surgical patients: review of cost, re-imbursement and revenue involved in pancreatic surgery at a high-volume academic medical centre. HPB (Oxford). 2015;17(4):311-317.PubMedArticle
17.
Nathan  H, Atoria  CL, Bach  PB, Elkin  EB.  Hospital volume, complications, and cost of cancer surgery in the elderly. J Clin Oncol. 2015;33(1):107-114.PubMedArticle
18.
Avritscher  EBC, Cooksley  CD, Rolston  KV,  et al.  Serious postoperative infections following resection of common solid tumors: outcomes, costs, and impact of hospital surgical volume. Support Care Cancer. 2014;22(2):527-535.PubMedArticle
19.
Charlson  ME, Pompei  P, Ales  KL, MacKenzie  CR.  A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40(5):373-383.PubMedArticle
20.
Amini  N, Spolverato  G, Kim  Y, Pawlik  TM.  Trends in hospital volume and failure to rescue for pancreatic surgery. J Gastrointest Surg. 2015;19(9):1581-1592.PubMedArticle
21.
Bureau of Labor Statistics USD of L. CPI Inflation Calculator. http://www.bls.gov/data/inflation_calculator.htm.
22.
Kilic  A, Shah  AS, Conte  JV,  et al.  Understanding variability in hospital-specific costs of coronary artery bypass grafting represents an opportunity for standardizing care and improving resource use. J Thorac Cardiovasc Surg. 2014;147(1):109-115.PubMedArticle
23.
Dimick  JB, Pronovost  PJ, Cowan  JA, Lipsett  PA.  Complications and costs after high-risk surgery: where should we focus quality improvement initiatives? J Am Coll Surg. 2003;196(5):671-678.PubMedArticle
24.
Brooks  GA, Li  L, Uno  H, Hassett  MJ, Landon  BE, Schrag  D.  Acute hospital care is the chief driver of regional spending variation in Medicare patients with advanced cancer. Health Aff (Millwood). 2014;33(10):1793-1800.PubMedArticle
25.
Haut  ER, Pronovost  PJ, Schneider  EB.  Limitations of administrative databases. JAMA. 2012;307(24):2589.PubMedArticle
×