- Email: [email protected]
Inpatient survival after gastrectomy for gastric cancer in the 21st century
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 x x x ( 2 0 1 4 ) 1 e7
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.JournalofSurgicalResearch.com
Inpatient survival after gastrectomy for gastric cancer in the 21st century Han Wang, MD, MPH, Timothy M. Pawlik, MD, MPH, PhD, Mark D. Duncan, MD, Xuan Hui, MD, ScM, Shalini Selvarajah, MD, MPH, Joseph K. Canner, MHS, Adil H. Haider, MD, MPH, Nita Ahuja, MD, and Eric B. Schneider, PhD* Department of Surgery, Center for Surgical Trials and Outcomes Research, Johns Hopkins School of Medicine, Baltimore, Maryland
article info
abstract
Article history:
Background: Surgical treatment for gastric cancer has evolved substantially. To understand
Received 4 January 2014
how changes in patient- and hospital-level factors are associated with outcomes over the
Received in revised form
last decade, we examined a nationally representative sample.
26 February 2014
Methods: Retrospective cross-sectional discharge data from the 2001e2010 Nationwide Inpa-
Accepted 5 March 2014
tient Sample were analyzed using cross tabulation and multivariable regression modeling.
Available online xxx
Patients with a primary diagnosis of gastric cancer undergoing gastrectomy as primary procedure were included. We examined relationships between patient- and hospital-level factors,
Keywords:
surgery type, and outcomes including in-hospital mortality and length of stay (LOS).
Gastric cancer
Results: A total of 67,327 patients with gastric cancer undergoing gastrectomy nationwide
Variation
with complete information were included. Compared with patients treated in 2001, pa-
Survival
tients in 2010 were younger, more likely admitted electively, treated in a teaching hospital,
Gastrectomy
or at an urban center. There was no difference in the type of procedure performed over time. Factors associated with an increased risk of in-hospital mortality included older age, male gender, and nonelective admission (P < 0.05). In multivariable analysis, patients undergoing gastrectomy in 2010 demonstrated 40% lower odds of in-hospital mortality (odds ratio, 0.60; P ¼ 0.008). Overall mean LOS was 13.9 d (standard error, 0.1) without change over time. Factors associated with longer LOS included procedure type, hospital location, nonelective admission, and comorbid disease (all P < 0.05). Conclusions: The adjusted odds of in-hospital mortality among surgically treated patients with gastric cancer decreased >40% between 2001 and 2010. Further research is warranted to determine if these findings are due to better patient selection, regionalization of care, or improvement of in-hospital quality of care. ª 2014 Elsevier Inc. All rights reserved.
1.
Background
Gastric cancer is a leading cause of cancer death worldwide [1]. Over the past four decades, gastric cancer decreased from
being the most common cancer worldwide in 1975 to being the fifth most common neoplasm in 2012 [1]. Mortality because of gastric cancer has also decreased, but it remains the third leading cause of cancer death worldwide, contributing to
* Corresponding author. Department of Surgery, Center for Surgical Trials and Outcomes Research, Johns Hopkins School of Medicine, 600 N. Wolfe Street, Blalock 688, Baltimore, MD 21287. Tel.: þ1 410 502 2601; fax: þ1 410 955 8101. E-mail address: [email protected] (E.B. Schneider). 0022-4804/$ e see front matter ª 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jss.2014.03.015
2
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 x x x ( 2 0 1 4 ) 1 e7
723,000 deaths annually [2]. Endoscopic procedures have emerged as a therapeutic alternative to surgical intervention for early-stage disease, especially in Asian countries where there is a high prevalence of gastric cancer and where patients undergo routine screening. On the contrary, in Western countries, surgery remains the main curative treatment used because most cases of gastric cancer in the West are diagnosed at an advanced stage [3]. The epidemiology of surgically treated patients with gastric cancer has been described by Smith et al. [4], who showed a relatively stable in-hospital mortality of 6% from 1998e2003, with older age, male sex, and total gastrectomy independently associated with higher in-hospital mortality. In the past 10 y, evidence-based medicine has changed many aspects in the management of gastric cancer, especially in terms of surgical methods. Previous studies found that a more conservative approach, involving less extensive resection and reduced lymph node dissection, achieved similar mortality and morbidity compared with more aggressive techniques [3]. Over the same period, there has also been a reported change in the histology and location of gastric cancer [5]. For this study, we were interested in examining possible changes in the patterns of intervention for gastric cancer and hospital-based outcomes among patients surgically treated for gastric cancer in the past decade in the United States. To our best knowledge, this issue has not been well addressed in the current literature. To understand how changes in patientand hospital-level factors, as well as surgical treatment, are associated with outcomes over the last decade, we examined a nationally representative sample of patients surgically treated for gastric cancer.
2.
Methods
2.1.
Data source
Using the Nationwide Inpatient Sample (NIS) database, a part of the Healthcare Cost and Utilization Projects, we performed a retrospective cross-sectional analysis of surgically treated patients with gastric cancer from 2001e2010. The NIS is the largest publicly available all-payer inpatient care database of the United States, which is the discharge data composed of an approximate 20% sample of the US community hospitals [6]. By design, data from the NIS can be weighted to approximate the US national population [6].
2.2.
Study population
Participants with International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) procedure codes for gastrectomy (43.5e43.9, 43.91, or 43.99) as their primary procedure and diagnosis codes for gastric cancer (151.0e151.6, 151.8, or 151.9) as their primary diagnosis were considered to be surgically treated patients with gastric cancer and were included in the study (Table 1). Patients with missing information on demographic factors or key covariates were excluded.
2.3.
Covariates
Age, race, gender, hospital bed size and annual surgical volume, procedure type (total versus partial gastrectomy), emergent admission, and comorbidities were examined across the study period and were included as covariates in multivariate analysis as they were considered potential confounders based on literature review [7e14]. Differences in these factors were also examined between patients undergoing surgery in 2001 and those undergoing surgery in 2010 using Student t-test and chi-square test. Procedure type was identified using ICD-9-CM procedure codes. Patients were considered to have had a total gastrectomy if their primary procedure code was 43.9, 43.91, or 43.99; otherwise, patients were considered to have undergone a partial gastrectomy. Hospital bed size was categorized into small, medium, or large based on the hospital region [15]. To define the burden of comorbid illness among patients undergoing gastrectomy, a Charlson comorbidity index score was calculated for each individual using a previously published algorithm, which derives comorbidities (a total of 17 conditions) from the ICD-9-CM codes associated with each discharge record [16].
2.4.
Outcome
The primary outcome of interest was in-hospital mortality, specifically death before discharge from acute inpatient care. The secondary outcome of interest was length of stay (LOS), which was calculated by subtracting the admission date from the discharge date [15].
2.5.
Statistical analysis
Descriptive analysis was done to compare patients from 2001e2010. Differences in selected characteristics were determined using Student t-test for continuous variables and chi-square test for categorical variables. A P value of <0.05 was considered statistically significant. In-hospital mortality rate was calculated by year, as were the proportions of patients undergoing partial versus total gastrectomy. Age, race, gender, hospital bed size, surgical volume, emergent admission, and patient comorbidities were included in multivariate analysis. Annual hospital procedure-specific volume was calculated by summarizing the number of gastrectomies done in each reporting hospital in a given year and then categorizing the hospitals into tertiles for that year. Logistic regression (both unadjusted and adjusted) was used to calculate the odds of mortality. Negative binomial regression was used to examine LOS because of its nature as counts. All results reported are based on data weighted to be representative at the level of the US population. All analyses were performed using Stata 12.1/MP (StataCorp LP, College Station, TX).
3.
Results
From 2001e2010, a total of 79,871 patients who underwent gastrectomy for gastric cancer were identified. Among these patients, 12,544 were missing demographic information or key
3
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 x x x ( 2 0 1 4 ) 1 e7
Table 1 e List of ICD-9-CM diagnosis and procedure codes used.
Table 2 e Characteristics of surgically treated patients with gastric cancer in 2001 and 2010.
ICD-9-CM
Patient characteristics
151.0 151.1 151.2 151.3 151.4 151.5 151.6 151.8
Diagnosis Malignant Malignant Malignant Malignant Malignant Malignant Malignant Malignant stomach Malignant
151.9
neoplasm neoplasm neoplasm neoplasm neoplasm neoplasm neoplasm neoplasm
gastric cardia gastric pylorus gastric antrum gastric fundus gastric body gastric lesser curvature gastric greater curvature other specified sites of
neoplasm of stomach, unspecified
ICD-9-CM 43.5 43.6 43.7 43.8 43.81 43.89 43.9 43.91 43.99
of of of of of of of of
Procedure Partial gastrectomy with anastomosis to esophagus Partial gastrectomy with anastomosis to duodenum Partial gastrectomy with anastomosis to jejunum Other partial gastrectomy Partial gastrectomy with jejunal transposition Partial gastrectomy with bypass gastrogastrostomy Total gastrectomy Total gastrectomy with intestinal interposition Other total gastrectomy
covariates and were excluded, leaving 67,327 patients with gastric cancer who underwent gastrectomy nationwide were included in the analysis. Demographic and patient- and hospital-level characteristics appear in Table 2. Overall, compared with patients treated in 2001, patients treated in 2010 were younger (2001: mean age, 68.1; standard error [SE], 0.34 versus 2010: mean age, 66.7; SE, 0.36; P < 0.001), more likely admitted electively (2001: 68.9% versus 2010: 77.3%; P < 0.001), treated in a teaching hospital (2001: 57.8% versus 2010: 70.5%; P < 0.001), or at an urban center (2001: 91.3% versus 2010: 95.3%; P < 0.001).
3.1. Temporal trends in the use of total gastrectomy and in-hospital mortality The proportions undergoing total versus partial gastrectomy did not change significantly across the 10-y study period, with 31.7% of patients undergoing total gastrectomy in 2001 versus 30.2% in 2010 (P ¼ 0.422) (Fig. 1). Across the same 10-y study period, in-patient mortality fluctuated, but an overall reduction in mortality was observed, with a total of 7.2% of surgically treated patients with gastric cancer dying before index hospital discharge in 2001 versus 3.3% in 2010 (P < 0.001) (Fig. 2).
3.2.
2001, N ¼ 7292
2010, N ¼ 6568
P value
Mean age 68.1 0.34 66.7 0.36 Male (%) 59.0 61.3 Elective admission (%) 68.9 77.3 Chronic obstructive pulmonary 16.5 14.7 disease (%) Diabetes (%) 11.7 18.6 Congestive heart failure (%) 9.5 5.7 Renal disease (%) 0.5 4.6 Mild liver disease (%) 0.6 1.2 Mean weighted Charlson score 4.7 0.06 4.2 0.06 Moderate/severe liver disease 0.5 0.4 (%) Metastatic disease (%) 53.8 36.0 Urban hospital (%) 91.3 95.3 Teaching hospital (%) 57.8 70.5 Proportion undergoing total 31.7 30.2 gastrectomy (%)
<0.001 0.229 <0.001 0.181 <0.001 <0.001 <0.001 0.0933 <0.001 0.803 <0.001 <0.001 <0.001 0.422
1-y increase in calendar year is associated with 7% reduction of the odds of mortality (P < 0.001). Being Hispanic or Asian or Pacific Islander was associated with lower odds of mortality compared with being white, and female gender was associated with lower odds of mortality compared with male gender. Compared with nonelective admission, elective admission was associated with 59% decrease in the odds of mortality. Being treated in an urban hospital was associated with lower odds of mortality, as was being treated in a teaching hospital and being treated in a hospital with high-surgical volume (Table 3). In univariate analysis, several factors were associated with increased odds of mortality. For example, every 1-y increase in age was associated with 4% increase in the odds of in-hospital mortality (OR, 1.04; P < 0.001). Having more comorbidities was also associated with higher odds of mortality (OR, 1.2; P < 0.001 per 1-U increase in weighted Charlson comorbidity score). Hospital size, based on an overall count of beds, was not associated with in-hospital mortality (P ¼ 0.773).
Univariate analysis
In univariate analysis for mortality, certain temporal patientand hospital-level factors were found to be protective. For example, being treated in 2008, 2009, or 2010 was associated with statistically significant decrease in the odds of mortality (odds ratio [OR], 0.71 for 2008 with P ¼ 0.044; OR, 0.45 for 2009 with P < 0.001; and OR, 0.44 for 2010 with P < 0.001) compared with having been treated in 2001 (Table 3). We further analyze the unadjusted temporal trends of in-hospital mortality, every
Fig. 1 e Partial versus total gastrectomy (2001e2010).
4
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 x x x ( 2 0 1 4 ) 1 e7
mean, 15.1 d; 2009 versus 2001: P ¼ 0.001 and 2010 versus 2001: P ¼ 0.019). However, overall across the study period, Asian race, female gender, elective admission, and larger hospital bed size were associated with shorter LOS, whereas older age and having acute myocardial infarction, congestive heart failure, or chronic obstructive pulmonary disease were associated with longer LOS (Table 3).
3.3.
Fig. 2 e In-hospital mortality (2001e2010).
In terms of LOS, patients treated in 2009 and 2010 experienced significantly shorter LOS compared with those treated in 2001 (2009: mean, 13.5 d; 2010: mean, 13.9 d; and 2001:
Multivariate analysis
In multivariable analysis, there was, on average, a 3% annual decrease in the odds of in-hospital mortality after adjusting for age, race, gender, hospital bed size, annual surgical volume, emergency, and comorbidity scores (P ¼ 0.012). Overall, being treated in 2010 was associated with a 40% decrease in the odds of mortality (OR, 0.60; P ¼ 0.008) compared with being treated in 2001, after adjusting for possible confounders. Older age (OR, 1.05 per additional year; P < 0.001), male gender (OR, 1.32; P ¼ 0.002), and nonelective admission (OR, 2.28; P < 0.001) were associated with increased odds of in-hospital mortality. Being treated in a high-surgical-volume hospital was associated with a 35% decrease in the odds of mortality (OR, 0.65; P < 0.001) (Table 3).
Table 3 e Univariate and multivariate logistic regression model of in-hospital mortality. Patient characteristics
Year 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Race White Black Hispanic Asian or Pacific Islander Native American Other Age Male Hospital bed size Small Medium Large Urban hospital Teaching hospital Nonelective admission Surgical volume tertile First Second Third Summed weighted Charlson score Total gastrectomy CI ¼ confidence interval.
Univariate
Multivariate
OR (95% CI)
P value
OR (95% CI)
P value
Reference 0.72 (0.51e1.02) 0.83 (0.69e1.15) 0.75 (0.53e1.05) 0.80 (0.57e1.13) 0.85 (0.60e1.20) 0.72 (0.52e1.01) 0.71 (0.51e0.99) 0.45 (0.30e0.68) 0.44 (0.31e0.64)
0.066 0.263 0.098 0.206 0.354 0.056 0.044 <0.001 <0.001
Reference 0.72 (0.51e1.04) 0.90 (0.64e1.26) 0.80 (0.56e1.14) 0.87 (0.61e1.23) 0.89 (0.62e1.27) 0.89 (0.63e1.25) 0.81 (0.57e1.14) 0.56 (0.37e0.84) 0.60 (0.41e0.87)
0.079 0.526 0.212 0.439 0.516 0.5 0.232 0.005 0.008
Reference 0.87 (0.68e1.10) 0.73 (0.54e0.98) 0.56 (0.36e0.87) 0.67 (0.21e2.13) 0.96 (0.61e1.52) 1.04 (1.04e1.05) 1.23 (1.46e1.03)
0.236 0.034 0.011 0.496 0.876 <0.001 0.019
Reference 0.98 (0.77e1.26) 0.82 (0.61e1.11) 0.73 (0.45e1.16) 1.02 (0.32e3.20) 1.21 (0.76e1.92) 1.05 (1.04e1.05) 1.32 (1.58e1.08)
0.9 0.209 0.177 0.973 0.407 <0.001 0.002
Reference 1.07 (0.77e1.48) 0.96 (0.71e1.29) 0.86 (0.80e0.93) 0.81 (0.77e0.85) 1.57 (1.49e1.66)
0.702 0.773 <0.001 <0.001 <0.001
Reference 1.05 (0.75e1.49) 1.18 (0.84e1.66) 1.01 (0.73e1.41) 1.02 (0.83e1.26) 2.08 (1.75e2.48)
0.777 0.338 0.945 0.829 <0.001
Reference 0.85 (0.80e0.90) 0.68 (0.64e0.72) 1.20 (1.18e1.21) 1.68 (1.42e1.98)
<0.001 <0.001 <0.001 <0.001
Reference 0.96 (0.78e1.19) 0.65 (0.51e0.82) 1.12 (1.07e1.16) 2.20 (1.84e2.63)
0.729 <0.001 <0.001 <0.001
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 x x x ( 2 0 1 4 ) 1 e7
Table 4 e Adjusted negative binomial model for LOS (>14 d). Patient characteristics Year 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Race White Black Hispanic Asian or Pacific Islander Native American Other Age Male Hospital bed size Small Medium Large Urban hospital Teaching hospital Nonelective admission Surgical volume tertile First Second Third Summed weighted Charlson score Total gastrectomy
Coefficient*
95% CI
Reference 0.03 0.02 0.10 0.01 0.01 0.02 0.02 0.07 0.04
0.10 0.03 0.05 0.08 0.16 0.03 0.06 0.08 0.08 0.06 0.05 0.09 0.09 0.04 0.13 0.00 0.10 0.03
Reference 0.00 0.06 0.16 0.14 0.02 0.01 0.05
0.04 0.04 0.925 0.10 0.01 0.023 0.23 0.09 <0.001 0.39 0.12 0.301 0.11 0.07 0.625 0.00 0.01 <0.001 0.08 0.01 0.004
Reference 0.03 0.10 0.12 0.04 0.41
0.03 0.05 0.05 0.01 0.38
0.08 0.311 0.15 <0.001 0.18 0.001 0.08 0.018 0.44 <0.001
Reference 0.02 0.04 0.02
0.06 0.09 0.01
0.02 0.348 0.01 0.111 0.02 <0.001
0.21
0.28 <0.001
0.24
P value
0.322 0.623 0.003 0.751 0.807 0.521 0.490 0.042 0.267
CI ¼ confidence interval. * Compared with the reference group, factors with a negative coefficient are associated with reduced LOS and factors with a positive coefficient are associated with increased LOS.
Mean LOS for all patients was 13.9 d (SE, 0.1), with no significant change observed across the 10-y study period; however, procedure type (total versus partial gastrectomy), larger hospital bed size, increasing age, nonelective surgery, and comorbid disease were associated with longer hospitalization (P < 0.05) (Table 4).
4.
Discussion
Between 2001 and 2010, there was a 54% reduction in inhospital mortality among surgically treated patients with gastric cancer in the United States. Possible differences in year-specific patient- or hospital-level factors may contribute to the tremendous decrease in in-hospital mortality observed in this study. Specifically, patients undergoing gastrectomy for gastric cancer in 2010 were younger with fewer comorbidities, and proportionally more patients undergoing gastrectomy had been admitted electively, compared with 2001.
5
These year-specific differences in patient characteristics may be due to better patient selection as many researchers have reported worse early postoperative outcomes in patients who are older [17] or who have more underlying comorbidities [18e20]. In our study, we also observed a significant association between undergoing gastrectomy on a nonelective basis and increased odds of in-hospital mortality. Improvements in patient selection may result from a number of factors including the following: improved surgeon effort in carefully selecting patients who have higher chances of benefiting from surgery, a change in the nature and bodily location of the disease, and improvements in gastric cancer screening strategies. Further study is needed to evaluate whether there is an actual change in the nature of gastric cancer as a disease in the United States or whether changes in screening and/or the selection of patients for surgery make this appear to be the case in the administrative data on which this study is based. However, it is important to recall that, at present, there is no standard for routine screening for stomach cancer in the United States, so the influence of disease screening might be insubstantial. Compared with surgically treated patients with gastric cancer in 2001, more patients in 2010 were treated in urban hospitals, teaching hospitals, hospitals with larger bed size, and in hospitals with high-surgical volume. Although clear relationships between procedure-specific surgeon and hospital volumes and surgical outcomes have been demonstrated for other disorders and procedures [21], studies in gastric cancer surgery have shown conflicting results regarding the association between volume and surgical outcomes. Some studies have found no association between procedure volume and patient outcomes [22]; some have reported a trend toward an association but without statistical significance [23], whereas other studies have demonstrated better surgical outcomes in high-volume hospitals [24]. Again, our study found that undergoing surgery in a highvolume hospital was associated with reduced in-hospital mortality but that hospital volume had no significant influence on LOS. Our findings from a large nationally representative data set collected across a 10-y period has the advantage of including a very large number of cases, which may support the trend toward better outcome in previous studies. [23] The regionalization of care for patients undergoing surgery for gastric cancer, that is, pooling patients in more “experienced” hospitals, may also contribute to the reduction in mortality observed over time in this study. The lack of a significant change across time in hospital LOS might be explained by the relatively stable proportions of different surgical procedures as the recovery time needed would be also relatively stable. Factors associated with longer LOS were similar to those associated with higher odds of mortality, except for surgical volume. Interestingly, higher surgical volume was associated with longer LOS, which might be because of more difficult cases being clustered in the higher volume hospitals. On a large scale, our findings suggest that the surgical treatment of patients with gastric cancer has improved substantially across the decade from 2001e2010. Further research is warranted to fully elucidate the reasons for the observed
6
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 x x x ( 2 0 1 4 ) 1 e7
improvements, whether the improvement results from patient-, surgeon-, procedure-, hospital-level factors, individually, or in combination.
4.1.
Limitations
Our study is subject to certain limitations, many of which are inherent to working with large administrative data sets. For example, the NIS relies on ICD-9 codes to identify both disorders and procedures. These codes may lack sufficient granularity to enable detection of important differences between patients. For example, in this study, comorbidities were defined as binary variables, which may not capture the whole spectrum of disease severity adequately, thereby limiting our possible understanding of the difference between patients with mild and severe comorbid diseases of the same nature. Also, because the NIS does not contain information regarding neoadjuvant therapy, tumor staging, or en-block resection of adjacent organs, we were unable to account for these potentially important factors. Our outcomes of interest, in-hospital mortality, and LOS are short-term outcomes. Indicators for long-term outcomes or long-term quality of life measures that may be very important to patients were not available in the NIS database. Also, this study examined only patients admitted to inpatient care with a primary diagnosis gastric cancer who underwent surgical resection and does not provide information on the natural history of the disease in patients whose cancer was not resected.
5.
Conclusion
The proportion of patients undergoing total versus partial gastrectomy remained stable across the 10-y study period; however, after controlling for age, comorbidity, hospital-level factors, and nonelective admission, the odds of in-hospital mortality among surgically treated patients with gastric cancer decreased >40% between 2001 and 2010. Further research is warranted to determine if these findings are due to better patient selection, regionalization of care, or improvements in quality of care.
Acknowledgment Author contributions: H.W., E.B.S., and T.M.P. contributed to the conception and design. H.W., E.B.S., T.M.P., and J.K.C. did the analysis and interpretation. H.W., E.B.S., and T.M.P. did the writing of the article. M.D.D., X.H., S.S., J.K.C., N.A., and A.H.H. did the revisions to the article. There was no funding for the study.
Disclosure All authors report that they do not have any conflicts with this article or its content.
references
[1] Ferlay J, Shin HR, Bray F, Forman D, Mathers C and Parkin DM. Globocan 2008 v2.0: cancer incidence and mortality worldwide: IARC CancerBase No. 10. 2010 [cited 2013 Oct 18, 2013]; Available from: http://globocan.iarc.fr. [2] Howlader N, Noone AM, Krapcho M, et al(eds). SEER Cancer Statistics Review, 1975-2010. 2012 [cited 2013 Oct18, 2013]; Available from: http://seer.cancer.gov/csr/1975_2010/. [3] National Comprehensive Cancer Network. Gastric cancer, V. 2.2013. 2013 [cited 2014 Jan. 3]; Available from: http://www. nccn.org/professionals/physician_gls/pdf/gastric.pdf. [4] Smith JK, McPhee JT, Hill JS, et al. National outcomes after gastric resection for neoplasm. Arch Surg 2007;142:387. [5] Powell J, McConkey CC. Increasing incidence of adenocarcinoma of the gastric cardia and adjacent sites. Br J Cancer 1990;62:440. [6] Healthcare Cost and Utilization Project (HCUP): overview of the Nationwide Inpatient Sample (NIS) [online]. [cited 2013 Oct 18th]; Available from: http://www.hcup-us.ahrq.gov/ nisoverview.jsp. [7] Wang CS, Hsieh CC, Chao TC, et al. Resectable gastric cancer: operative mortality and survival analysis. Chang Gung Med J 2002;25:216. [8] Hundahl SA, Phillips JL, Menck HR. The National Cancer Data Base report on poor survival of U.S. gastric carcinoma patients treated with gastrectomy: fifth edition American Joint Committee on Cancer staging, proximal disease, and the “different disease” hypothesis. Cancer 2000; 88:921. [9] Sah BK, Zhu ZG, Wang XY, et al. Post-operative complications of gastric cancer surgery: female gender at high risk. Eur J Cancer Care (Engl) 2009;18:202. [10] Coupland VH, Lagergren J, Luchtenborg M, et al. Hospital volume, proportion resected and mortality from oesophageal and gastric cancer: a population-based study in England, 2004-2008. Gut 2013;62:961. [11] Mocan L, Tomus C, Bartos D, et al. Long term outcome following surgical treatment for distal gastric cancer. J Gastrointestin Liver Dis 2013;22:53. [12] Blansfield JA, Clark SC, Hofmann MT, Morris JB. Alimentary tract surgery in the nonagenarian: elective vs. emergent operations. J Gastrointest Surg 2004;8:539. [13] Kim W, Song KY, Lee HJ, Han SU, Hyung WJ, Cho GS. The impact of comorbidity on surgical outcomes in laparoscopyassisted distal gastrectomy: a retrospective analysis of multicenter results. Ann Surg 2008;248:793. [14] Seo SH, Hur H, An CW, et al. Operative risk factors in gastric cancer surgery for elderly patients. J Gastric Cancer 2011;11: 116. [15] Healthcare Cost and Utilization Project (HCUP). Description of data elements: Nationwide Inpatient Sample (NIS).. [cited 2013 October 18th]; Available from: https://http://www.hcupus.ahrq.gov/db/nation/nis/NISCORE_A-Lv3.pdf. [16] 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;43:1130. [17] Vural S, Civil O, Kement M, et al. Risk factors for early postoperative morbidity and mortality in patients underwent radical surgery for gastric carcinoma: a single center experience. Int J Surg 2013;11:1103. [18] Yu J, Hu J, Huang C, et al. The impact of age and comorbidity on postoperative complications in patients with advanced gastric cancer after laparoscopic D2 gastrectomy: results from the Chinese laparoscopic gastrointestinal surgery study (CLASS) group. Eur J Surg Oncol 2013;39:1144.
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 x x x ( 2 0 1 4 ) 1 e7
[19] Oh SY, Kwon S, Lee KG, et al. Outcomes of minimally invasive surgery for early gastric cancer are comparable with those for open surgery: analysis of 1,013 minimally invasive surgeries at a single institution. Surg Endosc 2014;28:789. [20] Mita K, Ito H, Hashimoto M, et al. Postoperative complications and survival after gastric cancer surgery in patients older than 80 years of age. J Gastrointest Surg 2013;17:2067. [21] Schneider EB, Hyder O, Wolfgang CL, et al. Provider versus patient factors impacting hospital length of stay after pancreaticoduodenectomy. Surgery 2013;154:152.
7
[22] Damhuis RA, Meurs CJ, Dijkhuis CM, Stassen LP, Wiggers T. Hospital volume and post-operative mortality after resection for gastric cancer. Eur J Surg Oncol 2002;28:401. [23] Reavis KM, Hinojosa MW, Smith BR, Wooldridge JB, Krishnan S, Nguyen NT, et al. Hospital volume is not a predictor of outcomes after gastrectomy for neoplasm. Am Surg 2009;75:932. [24] Kelly JF, Hellinger FJ. Physician and hospital factors associated with mortality of surgical patients. Med Care 1986;24:785.
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.JournalofSurgicalResearch.com
Inpatient survival after gastrectomy for gastric cancer in the 21st century Han Wang, MD, MPH, Timothy M. Pawlik, MD, MPH, PhD, Mark D. Duncan, MD, Xuan Hui, MD, ScM, Shalini Selvarajah, MD, MPH, Joseph K. Canner, MHS, Adil H. Haider, MD, MPH, Nita Ahuja, MD, and Eric B. Schneider, PhD* Department of Surgery, Center for Surgical Trials and Outcomes Research, Johns Hopkins School of Medicine, Baltimore, Maryland
article info
abstract
Article history:
Background: Surgical treatment for gastric cancer has evolved substantially. To understand
Received 4 January 2014
how changes in patient- and hospital-level factors are associated with outcomes over the
Received in revised form
last decade, we examined a nationally representative sample.
26 February 2014
Methods: Retrospective cross-sectional discharge data from the 2001e2010 Nationwide Inpa-
Accepted 5 March 2014
tient Sample were analyzed using cross tabulation and multivariable regression modeling.
Available online xxx
Patients with a primary diagnosis of gastric cancer undergoing gastrectomy as primary procedure were included. We examined relationships between patient- and hospital-level factors,
Keywords:
surgery type, and outcomes including in-hospital mortality and length of stay (LOS).
Gastric cancer
Results: A total of 67,327 patients with gastric cancer undergoing gastrectomy nationwide
Variation
with complete information were included. Compared with patients treated in 2001, pa-
Survival
tients in 2010 were younger, more likely admitted electively, treated in a teaching hospital,
Gastrectomy
or at an urban center. There was no difference in the type of procedure performed over time. Factors associated with an increased risk of in-hospital mortality included older age, male gender, and nonelective admission (P < 0.05). In multivariable analysis, patients undergoing gastrectomy in 2010 demonstrated 40% lower odds of in-hospital mortality (odds ratio, 0.60; P ¼ 0.008). Overall mean LOS was 13.9 d (standard error, 0.1) without change over time. Factors associated with longer LOS included procedure type, hospital location, nonelective admission, and comorbid disease (all P < 0.05). Conclusions: The adjusted odds of in-hospital mortality among surgically treated patients with gastric cancer decreased >40% between 2001 and 2010. Further research is warranted to determine if these findings are due to better patient selection, regionalization of care, or improvement of in-hospital quality of care. ª 2014 Elsevier Inc. All rights reserved.
1.
Background
Gastric cancer is a leading cause of cancer death worldwide [1]. Over the past four decades, gastric cancer decreased from
being the most common cancer worldwide in 1975 to being the fifth most common neoplasm in 2012 [1]. Mortality because of gastric cancer has also decreased, but it remains the third leading cause of cancer death worldwide, contributing to
* Corresponding author. Department of Surgery, Center for Surgical Trials and Outcomes Research, Johns Hopkins School of Medicine, 600 N. Wolfe Street, Blalock 688, Baltimore, MD 21287. Tel.: þ1 410 502 2601; fax: þ1 410 955 8101. E-mail address: [email protected] (E.B. Schneider). 0022-4804/$ e see front matter ª 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jss.2014.03.015
2
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 x x x ( 2 0 1 4 ) 1 e7
723,000 deaths annually [2]. Endoscopic procedures have emerged as a therapeutic alternative to surgical intervention for early-stage disease, especially in Asian countries where there is a high prevalence of gastric cancer and where patients undergo routine screening. On the contrary, in Western countries, surgery remains the main curative treatment used because most cases of gastric cancer in the West are diagnosed at an advanced stage [3]. The epidemiology of surgically treated patients with gastric cancer has been described by Smith et al. [4], who showed a relatively stable in-hospital mortality of 6% from 1998e2003, with older age, male sex, and total gastrectomy independently associated with higher in-hospital mortality. In the past 10 y, evidence-based medicine has changed many aspects in the management of gastric cancer, especially in terms of surgical methods. Previous studies found that a more conservative approach, involving less extensive resection and reduced lymph node dissection, achieved similar mortality and morbidity compared with more aggressive techniques [3]. Over the same period, there has also been a reported change in the histology and location of gastric cancer [5]. For this study, we were interested in examining possible changes in the patterns of intervention for gastric cancer and hospital-based outcomes among patients surgically treated for gastric cancer in the past decade in the United States. To our best knowledge, this issue has not been well addressed in the current literature. To understand how changes in patientand hospital-level factors, as well as surgical treatment, are associated with outcomes over the last decade, we examined a nationally representative sample of patients surgically treated for gastric cancer.
2.
Methods
2.1.
Data source
Using the Nationwide Inpatient Sample (NIS) database, a part of the Healthcare Cost and Utilization Projects, we performed a retrospective cross-sectional analysis of surgically treated patients with gastric cancer from 2001e2010. The NIS is the largest publicly available all-payer inpatient care database of the United States, which is the discharge data composed of an approximate 20% sample of the US community hospitals [6]. By design, data from the NIS can be weighted to approximate the US national population [6].
2.2.
Study population
Participants with International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) procedure codes for gastrectomy (43.5e43.9, 43.91, or 43.99) as their primary procedure and diagnosis codes for gastric cancer (151.0e151.6, 151.8, or 151.9) as their primary diagnosis were considered to be surgically treated patients with gastric cancer and were included in the study (Table 1). Patients with missing information on demographic factors or key covariates were excluded.
2.3.
Covariates
Age, race, gender, hospital bed size and annual surgical volume, procedure type (total versus partial gastrectomy), emergent admission, and comorbidities were examined across the study period and were included as covariates in multivariate analysis as they were considered potential confounders based on literature review [7e14]. Differences in these factors were also examined between patients undergoing surgery in 2001 and those undergoing surgery in 2010 using Student t-test and chi-square test. Procedure type was identified using ICD-9-CM procedure codes. Patients were considered to have had a total gastrectomy if their primary procedure code was 43.9, 43.91, or 43.99; otherwise, patients were considered to have undergone a partial gastrectomy. Hospital bed size was categorized into small, medium, or large based on the hospital region [15]. To define the burden of comorbid illness among patients undergoing gastrectomy, a Charlson comorbidity index score was calculated for each individual using a previously published algorithm, which derives comorbidities (a total of 17 conditions) from the ICD-9-CM codes associated with each discharge record [16].
2.4.
Outcome
The primary outcome of interest was in-hospital mortality, specifically death before discharge from acute inpatient care. The secondary outcome of interest was length of stay (LOS), which was calculated by subtracting the admission date from the discharge date [15].
2.5.
Statistical analysis
Descriptive analysis was done to compare patients from 2001e2010. Differences in selected characteristics were determined using Student t-test for continuous variables and chi-square test for categorical variables. A P value of <0.05 was considered statistically significant. In-hospital mortality rate was calculated by year, as were the proportions of patients undergoing partial versus total gastrectomy. Age, race, gender, hospital bed size, surgical volume, emergent admission, and patient comorbidities were included in multivariate analysis. Annual hospital procedure-specific volume was calculated by summarizing the number of gastrectomies done in each reporting hospital in a given year and then categorizing the hospitals into tertiles for that year. Logistic regression (both unadjusted and adjusted) was used to calculate the odds of mortality. Negative binomial regression was used to examine LOS because of its nature as counts. All results reported are based on data weighted to be representative at the level of the US population. All analyses were performed using Stata 12.1/MP (StataCorp LP, College Station, TX).
3.
Results
From 2001e2010, a total of 79,871 patients who underwent gastrectomy for gastric cancer were identified. Among these patients, 12,544 were missing demographic information or key
3
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 x x x ( 2 0 1 4 ) 1 e7
Table 1 e List of ICD-9-CM diagnosis and procedure codes used.
Table 2 e Characteristics of surgically treated patients with gastric cancer in 2001 and 2010.
ICD-9-CM
Patient characteristics
151.0 151.1 151.2 151.3 151.4 151.5 151.6 151.8
Diagnosis Malignant Malignant Malignant Malignant Malignant Malignant Malignant Malignant stomach Malignant
151.9
neoplasm neoplasm neoplasm neoplasm neoplasm neoplasm neoplasm neoplasm
gastric cardia gastric pylorus gastric antrum gastric fundus gastric body gastric lesser curvature gastric greater curvature other specified sites of
neoplasm of stomach, unspecified
ICD-9-CM 43.5 43.6 43.7 43.8 43.81 43.89 43.9 43.91 43.99
of of of of of of of of
Procedure Partial gastrectomy with anastomosis to esophagus Partial gastrectomy with anastomosis to duodenum Partial gastrectomy with anastomosis to jejunum Other partial gastrectomy Partial gastrectomy with jejunal transposition Partial gastrectomy with bypass gastrogastrostomy Total gastrectomy Total gastrectomy with intestinal interposition Other total gastrectomy
covariates and were excluded, leaving 67,327 patients with gastric cancer who underwent gastrectomy nationwide were included in the analysis. Demographic and patient- and hospital-level characteristics appear in Table 2. Overall, compared with patients treated in 2001, patients treated in 2010 were younger (2001: mean age, 68.1; standard error [SE], 0.34 versus 2010: mean age, 66.7; SE, 0.36; P < 0.001), more likely admitted electively (2001: 68.9% versus 2010: 77.3%; P < 0.001), treated in a teaching hospital (2001: 57.8% versus 2010: 70.5%; P < 0.001), or at an urban center (2001: 91.3% versus 2010: 95.3%; P < 0.001).
3.1. Temporal trends in the use of total gastrectomy and in-hospital mortality The proportions undergoing total versus partial gastrectomy did not change significantly across the 10-y study period, with 31.7% of patients undergoing total gastrectomy in 2001 versus 30.2% in 2010 (P ¼ 0.422) (Fig. 1). Across the same 10-y study period, in-patient mortality fluctuated, but an overall reduction in mortality was observed, with a total of 7.2% of surgically treated patients with gastric cancer dying before index hospital discharge in 2001 versus 3.3% in 2010 (P < 0.001) (Fig. 2).
3.2.
2001, N ¼ 7292
2010, N ¼ 6568
P value
Mean age 68.1 0.34 66.7 0.36 Male (%) 59.0 61.3 Elective admission (%) 68.9 77.3 Chronic obstructive pulmonary 16.5 14.7 disease (%) Diabetes (%) 11.7 18.6 Congestive heart failure (%) 9.5 5.7 Renal disease (%) 0.5 4.6 Mild liver disease (%) 0.6 1.2 Mean weighted Charlson score 4.7 0.06 4.2 0.06 Moderate/severe liver disease 0.5 0.4 (%) Metastatic disease (%) 53.8 36.0 Urban hospital (%) 91.3 95.3 Teaching hospital (%) 57.8 70.5 Proportion undergoing total 31.7 30.2 gastrectomy (%)
<0.001 0.229 <0.001 0.181 <0.001 <0.001 <0.001 0.0933 <0.001 0.803 <0.001 <0.001 <0.001 0.422
1-y increase in calendar year is associated with 7% reduction of the odds of mortality (P < 0.001). Being Hispanic or Asian or Pacific Islander was associated with lower odds of mortality compared with being white, and female gender was associated with lower odds of mortality compared with male gender. Compared with nonelective admission, elective admission was associated with 59% decrease in the odds of mortality. Being treated in an urban hospital was associated with lower odds of mortality, as was being treated in a teaching hospital and being treated in a hospital with high-surgical volume (Table 3). In univariate analysis, several factors were associated with increased odds of mortality. For example, every 1-y increase in age was associated with 4% increase in the odds of in-hospital mortality (OR, 1.04; P < 0.001). Having more comorbidities was also associated with higher odds of mortality (OR, 1.2; P < 0.001 per 1-U increase in weighted Charlson comorbidity score). Hospital size, based on an overall count of beds, was not associated with in-hospital mortality (P ¼ 0.773).
Univariate analysis
In univariate analysis for mortality, certain temporal patientand hospital-level factors were found to be protective. For example, being treated in 2008, 2009, or 2010 was associated with statistically significant decrease in the odds of mortality (odds ratio [OR], 0.71 for 2008 with P ¼ 0.044; OR, 0.45 for 2009 with P < 0.001; and OR, 0.44 for 2010 with P < 0.001) compared with having been treated in 2001 (Table 3). We further analyze the unadjusted temporal trends of in-hospital mortality, every
Fig. 1 e Partial versus total gastrectomy (2001e2010).
4
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 x x x ( 2 0 1 4 ) 1 e7
mean, 15.1 d; 2009 versus 2001: P ¼ 0.001 and 2010 versus 2001: P ¼ 0.019). However, overall across the study period, Asian race, female gender, elective admission, and larger hospital bed size were associated with shorter LOS, whereas older age and having acute myocardial infarction, congestive heart failure, or chronic obstructive pulmonary disease were associated with longer LOS (Table 3).
3.3.
Fig. 2 e In-hospital mortality (2001e2010).
In terms of LOS, patients treated in 2009 and 2010 experienced significantly shorter LOS compared with those treated in 2001 (2009: mean, 13.5 d; 2010: mean, 13.9 d; and 2001:
Multivariate analysis
In multivariable analysis, there was, on average, a 3% annual decrease in the odds of in-hospital mortality after adjusting for age, race, gender, hospital bed size, annual surgical volume, emergency, and comorbidity scores (P ¼ 0.012). Overall, being treated in 2010 was associated with a 40% decrease in the odds of mortality (OR, 0.60; P ¼ 0.008) compared with being treated in 2001, after adjusting for possible confounders. Older age (OR, 1.05 per additional year; P < 0.001), male gender (OR, 1.32; P ¼ 0.002), and nonelective admission (OR, 2.28; P < 0.001) were associated with increased odds of in-hospital mortality. Being treated in a high-surgical-volume hospital was associated with a 35% decrease in the odds of mortality (OR, 0.65; P < 0.001) (Table 3).
Table 3 e Univariate and multivariate logistic regression model of in-hospital mortality. Patient characteristics
Year 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Race White Black Hispanic Asian or Pacific Islander Native American Other Age Male Hospital bed size Small Medium Large Urban hospital Teaching hospital Nonelective admission Surgical volume tertile First Second Third Summed weighted Charlson score Total gastrectomy CI ¼ confidence interval.
Univariate
Multivariate
OR (95% CI)
P value
OR (95% CI)
P value
Reference 0.72 (0.51e1.02) 0.83 (0.69e1.15) 0.75 (0.53e1.05) 0.80 (0.57e1.13) 0.85 (0.60e1.20) 0.72 (0.52e1.01) 0.71 (0.51e0.99) 0.45 (0.30e0.68) 0.44 (0.31e0.64)
0.066 0.263 0.098 0.206 0.354 0.056 0.044 <0.001 <0.001
Reference 0.72 (0.51e1.04) 0.90 (0.64e1.26) 0.80 (0.56e1.14) 0.87 (0.61e1.23) 0.89 (0.62e1.27) 0.89 (0.63e1.25) 0.81 (0.57e1.14) 0.56 (0.37e0.84) 0.60 (0.41e0.87)
0.079 0.526 0.212 0.439 0.516 0.5 0.232 0.005 0.008
Reference 0.87 (0.68e1.10) 0.73 (0.54e0.98) 0.56 (0.36e0.87) 0.67 (0.21e2.13) 0.96 (0.61e1.52) 1.04 (1.04e1.05) 1.23 (1.46e1.03)
0.236 0.034 0.011 0.496 0.876 <0.001 0.019
Reference 0.98 (0.77e1.26) 0.82 (0.61e1.11) 0.73 (0.45e1.16) 1.02 (0.32e3.20) 1.21 (0.76e1.92) 1.05 (1.04e1.05) 1.32 (1.58e1.08)
0.9 0.209 0.177 0.973 0.407 <0.001 0.002
Reference 1.07 (0.77e1.48) 0.96 (0.71e1.29) 0.86 (0.80e0.93) 0.81 (0.77e0.85) 1.57 (1.49e1.66)
0.702 0.773 <0.001 <0.001 <0.001
Reference 1.05 (0.75e1.49) 1.18 (0.84e1.66) 1.01 (0.73e1.41) 1.02 (0.83e1.26) 2.08 (1.75e2.48)
0.777 0.338 0.945 0.829 <0.001
Reference 0.85 (0.80e0.90) 0.68 (0.64e0.72) 1.20 (1.18e1.21) 1.68 (1.42e1.98)
<0.001 <0.001 <0.001 <0.001
Reference 0.96 (0.78e1.19) 0.65 (0.51e0.82) 1.12 (1.07e1.16) 2.20 (1.84e2.63)
0.729 <0.001 <0.001 <0.001
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 x x x ( 2 0 1 4 ) 1 e7
Table 4 e Adjusted negative binomial model for LOS (>14 d). Patient characteristics Year 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Race White Black Hispanic Asian or Pacific Islander Native American Other Age Male Hospital bed size Small Medium Large Urban hospital Teaching hospital Nonelective admission Surgical volume tertile First Second Third Summed weighted Charlson score Total gastrectomy
Coefficient*
95% CI
Reference 0.03 0.02 0.10 0.01 0.01 0.02 0.02 0.07 0.04
0.10 0.03 0.05 0.08 0.16 0.03 0.06 0.08 0.08 0.06 0.05 0.09 0.09 0.04 0.13 0.00 0.10 0.03
Reference 0.00 0.06 0.16 0.14 0.02 0.01 0.05
0.04 0.04 0.925 0.10 0.01 0.023 0.23 0.09 <0.001 0.39 0.12 0.301 0.11 0.07 0.625 0.00 0.01 <0.001 0.08 0.01 0.004
Reference 0.03 0.10 0.12 0.04 0.41
0.03 0.05 0.05 0.01 0.38
0.08 0.311 0.15 <0.001 0.18 0.001 0.08 0.018 0.44 <0.001
Reference 0.02 0.04 0.02
0.06 0.09 0.01
0.02 0.348 0.01 0.111 0.02 <0.001
0.21
0.28 <0.001
0.24
P value
0.322 0.623 0.003 0.751 0.807 0.521 0.490 0.042 0.267
CI ¼ confidence interval. * Compared with the reference group, factors with a negative coefficient are associated with reduced LOS and factors with a positive coefficient are associated with increased LOS.
Mean LOS for all patients was 13.9 d (SE, 0.1), with no significant change observed across the 10-y study period; however, procedure type (total versus partial gastrectomy), larger hospital bed size, increasing age, nonelective surgery, and comorbid disease were associated with longer hospitalization (P < 0.05) (Table 4).
4.
Discussion
Between 2001 and 2010, there was a 54% reduction in inhospital mortality among surgically treated patients with gastric cancer in the United States. Possible differences in year-specific patient- or hospital-level factors may contribute to the tremendous decrease in in-hospital mortality observed in this study. Specifically, patients undergoing gastrectomy for gastric cancer in 2010 were younger with fewer comorbidities, and proportionally more patients undergoing gastrectomy had been admitted electively, compared with 2001.
5
These year-specific differences in patient characteristics may be due to better patient selection as many researchers have reported worse early postoperative outcomes in patients who are older [17] or who have more underlying comorbidities [18e20]. In our study, we also observed a significant association between undergoing gastrectomy on a nonelective basis and increased odds of in-hospital mortality. Improvements in patient selection may result from a number of factors including the following: improved surgeon effort in carefully selecting patients who have higher chances of benefiting from surgery, a change in the nature and bodily location of the disease, and improvements in gastric cancer screening strategies. Further study is needed to evaluate whether there is an actual change in the nature of gastric cancer as a disease in the United States or whether changes in screening and/or the selection of patients for surgery make this appear to be the case in the administrative data on which this study is based. However, it is important to recall that, at present, there is no standard for routine screening for stomach cancer in the United States, so the influence of disease screening might be insubstantial. Compared with surgically treated patients with gastric cancer in 2001, more patients in 2010 were treated in urban hospitals, teaching hospitals, hospitals with larger bed size, and in hospitals with high-surgical volume. Although clear relationships between procedure-specific surgeon and hospital volumes and surgical outcomes have been demonstrated for other disorders and procedures [21], studies in gastric cancer surgery have shown conflicting results regarding the association between volume and surgical outcomes. Some studies have found no association between procedure volume and patient outcomes [22]; some have reported a trend toward an association but without statistical significance [23], whereas other studies have demonstrated better surgical outcomes in high-volume hospitals [24]. Again, our study found that undergoing surgery in a highvolume hospital was associated with reduced in-hospital mortality but that hospital volume had no significant influence on LOS. Our findings from a large nationally representative data set collected across a 10-y period has the advantage of including a very large number of cases, which may support the trend toward better outcome in previous studies. [23] The regionalization of care for patients undergoing surgery for gastric cancer, that is, pooling patients in more “experienced” hospitals, may also contribute to the reduction in mortality observed over time in this study. The lack of a significant change across time in hospital LOS might be explained by the relatively stable proportions of different surgical procedures as the recovery time needed would be also relatively stable. Factors associated with longer LOS were similar to those associated with higher odds of mortality, except for surgical volume. Interestingly, higher surgical volume was associated with longer LOS, which might be because of more difficult cases being clustered in the higher volume hospitals. On a large scale, our findings suggest that the surgical treatment of patients with gastric cancer has improved substantially across the decade from 2001e2010. Further research is warranted to fully elucidate the reasons for the observed
6
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 x x x ( 2 0 1 4 ) 1 e7
improvements, whether the improvement results from patient-, surgeon-, procedure-, hospital-level factors, individually, or in combination.
4.1.
Limitations
Our study is subject to certain limitations, many of which are inherent to working with large administrative data sets. For example, the NIS relies on ICD-9 codes to identify both disorders and procedures. These codes may lack sufficient granularity to enable detection of important differences between patients. For example, in this study, comorbidities were defined as binary variables, which may not capture the whole spectrum of disease severity adequately, thereby limiting our possible understanding of the difference between patients with mild and severe comorbid diseases of the same nature. Also, because the NIS does not contain information regarding neoadjuvant therapy, tumor staging, or en-block resection of adjacent organs, we were unable to account for these potentially important factors. Our outcomes of interest, in-hospital mortality, and LOS are short-term outcomes. Indicators for long-term outcomes or long-term quality of life measures that may be very important to patients were not available in the NIS database. Also, this study examined only patients admitted to inpatient care with a primary diagnosis gastric cancer who underwent surgical resection and does not provide information on the natural history of the disease in patients whose cancer was not resected.
5.
Conclusion
The proportion of patients undergoing total versus partial gastrectomy remained stable across the 10-y study period; however, after controlling for age, comorbidity, hospital-level factors, and nonelective admission, the odds of in-hospital mortality among surgically treated patients with gastric cancer decreased >40% between 2001 and 2010. Further research is warranted to determine if these findings are due to better patient selection, regionalization of care, or improvements in quality of care.
Acknowledgment Author contributions: H.W., E.B.S., and T.M.P. contributed to the conception and design. H.W., E.B.S., T.M.P., and J.K.C. did the analysis and interpretation. H.W., E.B.S., and T.M.P. did the writing of the article. M.D.D., X.H., S.S., J.K.C., N.A., and A.H.H. did the revisions to the article. There was no funding for the study.
Disclosure All authors report that they do not have any conflicts with this article or its content.
references
[1] Ferlay J, Shin HR, Bray F, Forman D, Mathers C and Parkin DM. Globocan 2008 v2.0: cancer incidence and mortality worldwide: IARC CancerBase No. 10. 2010 [cited 2013 Oct 18, 2013]; Available from: http://globocan.iarc.fr. [2] Howlader N, Noone AM, Krapcho M, et al(eds). SEER Cancer Statistics Review, 1975-2010. 2012 [cited 2013 Oct18, 2013]; Available from: http://seer.cancer.gov/csr/1975_2010/. [3] National Comprehensive Cancer Network. Gastric cancer, V. 2.2013. 2013 [cited 2014 Jan. 3]; Available from: http://www. nccn.org/professionals/physician_gls/pdf/gastric.pdf. [4] Smith JK, McPhee JT, Hill JS, et al. National outcomes after gastric resection for neoplasm. Arch Surg 2007;142:387. [5] Powell J, McConkey CC. Increasing incidence of adenocarcinoma of the gastric cardia and adjacent sites. Br J Cancer 1990;62:440. [6] Healthcare Cost and Utilization Project (HCUP): overview of the Nationwide Inpatient Sample (NIS) [online]. [cited 2013 Oct 18th]; Available from: http://www.hcup-us.ahrq.gov/ nisoverview.jsp. [7] Wang CS, Hsieh CC, Chao TC, et al. Resectable gastric cancer: operative mortality and survival analysis. Chang Gung Med J 2002;25:216. [8] Hundahl SA, Phillips JL, Menck HR. The National Cancer Data Base report on poor survival of U.S. gastric carcinoma patients treated with gastrectomy: fifth edition American Joint Committee on Cancer staging, proximal disease, and the “different disease” hypothesis. Cancer 2000; 88:921. [9] Sah BK, Zhu ZG, Wang XY, et al. Post-operative complications of gastric cancer surgery: female gender at high risk. Eur J Cancer Care (Engl) 2009;18:202. [10] Coupland VH, Lagergren J, Luchtenborg M, et al. Hospital volume, proportion resected and mortality from oesophageal and gastric cancer: a population-based study in England, 2004-2008. Gut 2013;62:961. [11] Mocan L, Tomus C, Bartos D, et al. Long term outcome following surgical treatment for distal gastric cancer. J Gastrointestin Liver Dis 2013;22:53. [12] Blansfield JA, Clark SC, Hofmann MT, Morris JB. Alimentary tract surgery in the nonagenarian: elective vs. emergent operations. J Gastrointest Surg 2004;8:539. [13] Kim W, Song KY, Lee HJ, Han SU, Hyung WJ, Cho GS. The impact of comorbidity on surgical outcomes in laparoscopyassisted distal gastrectomy: a retrospective analysis of multicenter results. Ann Surg 2008;248:793. [14] Seo SH, Hur H, An CW, et al. Operative risk factors in gastric cancer surgery for elderly patients. J Gastric Cancer 2011;11: 116. [15] Healthcare Cost and Utilization Project (HCUP). Description of data elements: Nationwide Inpatient Sample (NIS).. [cited 2013 October 18th]; Available from: https://http://www.hcupus.ahrq.gov/db/nation/nis/NISCORE_A-Lv3.pdf. [16] 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;43:1130. [17] Vural S, Civil O, Kement M, et al. Risk factors for early postoperative morbidity and mortality in patients underwent radical surgery for gastric carcinoma: a single center experience. Int J Surg 2013;11:1103. [18] Yu J, Hu J, Huang C, et al. The impact of age and comorbidity on postoperative complications in patients with advanced gastric cancer after laparoscopic D2 gastrectomy: results from the Chinese laparoscopic gastrointestinal surgery study (CLASS) group. Eur J Surg Oncol 2013;39:1144.
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 x x x ( 2 0 1 4 ) 1 e7
[19] Oh SY, Kwon S, Lee KG, et al. Outcomes of minimally invasive surgery for early gastric cancer are comparable with those for open surgery: analysis of 1,013 minimally invasive surgeries at a single institution. Surg Endosc 2014;28:789. [20] Mita K, Ito H, Hashimoto M, et al. Postoperative complications and survival after gastric cancer surgery in patients older than 80 years of age. J Gastrointest Surg 2013;17:2067. [21] Schneider EB, Hyder O, Wolfgang CL, et al. Provider versus patient factors impacting hospital length of stay after pancreaticoduodenectomy. Surgery 2013;154:152.
7
[22] Damhuis RA, Meurs CJ, Dijkhuis CM, Stassen LP, Wiggers T. Hospital volume and post-operative mortality after resection for gastric cancer. Eur J Surg Oncol 2002;28:401. [23] Reavis KM, Hinojosa MW, Smith BR, Wooldridge JB, Krishnan S, Nguyen NT, et al. Hospital volume is not a predictor of outcomes after gastrectomy for neoplasm. Am Surg 2009;75:932. [24] Kelly JF, Hellinger FJ. Physician and hospital factors associated with mortality of surgical patients. Med Care 1986;24:785.