National Outcomes After Gastric Resection for Neoplasm

POSTER SESSION National Outcomes After Gastric Resection for Neoplasm

Jillian K. Smith, MPH; James T. McPhee, MD; Joshua S. Hill, MD; Giles F. Whalen, MD; Mary E. Sullivan, MS; Demetrius E. Litwin, MD, MBA; Frederick A. Anderson, PhD; Jennifer F. Tseng, MD

Hypothesis: That factors affecting outcomes of surgi- In-hospital mortality for patients undergoing surgery cal resection in the treatment of gastric cancer can be iden- was 6.0%, without significant change from 1998 through tified using a large US database. 2003. Factors predictive of significantly increased in- hospital mortality included low annual hospital surgical Design: Retrospective observational study. volume (lowest [Յ4 gastrectomies per year] vs highest [Ն11 gastrectomies per year], 6.8% vs 4.9%; adjusted Setting: The Nationwide Inpatient Sample from Janu- odds ratio [OR], 1.5; 95% confidence interval [CI], 1.2- ary 1, 1998, through December 31, 2003. 1.8]), older patient age (50-69 vs Ͻ50 years, 4.0% vs 2.1%; adjusted OR, 1.5; 95% CI, 1.1-2.2) (Ն70 vs Ͻ50 years, Patients: We included 13 354 patient discharges (ap- 8.6% vs 2.1%; adjusted OR, 2.9; 95% CI, 2.0-4.3), male proximately 66 096 nationally by weighted analysis) who sex (male vs female, 6.7% vs 5.0%; adjusted OR, 1.3; 95% underwent gastric resection for neoplasm. CI, 1.1-1.5), and procedure type (total gastrectomy vs all other resections, 8.0% vs 5.3%; adjusted OR, 1.4; 95% CI, Main Outcome Measure: In-hospital mortality. Uni- 1.2-1.7). variate analyses were performed by means of ␹2 tests. A multivariate logistic regression was performed to deter- Conclusions: Higher annual surgical volume is predic- mine which variables were independently predictive of tive of lower in-hospital mortality for patients undergo- in-hospital mortality. ing gastric resection for neoplasm. Other factors significantly associated with superior outcomes after gastric Results: During the study period, 50 738 patients (ap- resection included diagnosis type, procedure type, younger proximately 250 420 nationally) were discharged with the age, female sex, and fewer comorbid conditions. diagnosis of gastric neoplasm. Of those, 13 354 (26.3%) underwent gastric resection during their hospitalization. Arch Surg. 2007;142:387-393

ASTRIC CANCER IS THE tient’s comorbid conditions. Gastrectomy second leading cause of confers the greatest likelihood for sur- cancer-specific mortal- vival in the treatment of invasive gastric ity worldwide and the cancer.2 The type of gastrectomy remains eighth leading cause in an area of some debate; recent investiga- theG United States. The American Cancer tions have suggested that total gastrec- Society estimated that stomach cancer tomy offers no survival advantage over dis- would have an incidence of 22 280 cases tal or proximal subtotal resections based and would cause 11 430 deaths in the on the anatomic location of the tumor.3 A United States in 2006.1 regional study noted an overall decrease in operative mortality rates for gastric can- CME course available at cer resections, including total gastrectomy, extended total gastrectomy, distal www.archsurg.com gastrectomy, and proximal esophagogastrectomy, during the 20 years from 1976 Surgical resection is currently the only through 1995.4 In contrast, a more re- treatment option for gastric cancer with cent national analysis of operative mor- Author Affiliations: curative potential. Options for surgical tality trends from 1994 through 1999 Department of Surgery, treatment may depend on factors includ- found that the operative mortality rates of University of Massachusetts ing the anatomic location of the tumor, his- gastrectomy did not decline during the pe- Medical School, Worcester. tologic type, stage of disease, and pa- riod studied.5

(REPRINTED) ARCH SURG/ VOL 142, APR 2007 WWW.ARCHSURG.COM 387

ICD-9-CM Code Definition DATABASE Diagnostic To evaluate trends in gastric resection for neoplastic disease at 150.2 Malignant primary of esophagus (abdominal) the national level, we used the NIS for January 1, 1998, through 150.9 Malignant primary of esophagus December 31, 2003. The NIS is the largest national all-payer hos- 151.0 Malignant primary of GE junction pital inpatient care database in the United States. Data exist for 151.1 Malignant primary of prepylorus/pylorus approximately 7 million hospital discharges per year from a strati- 151.2 Malignant primary of antrum fied sample of 20% of nonfederal US community hospitals from 151.3 Malignant primary of fundus participating states (n=37 in 2003), including academic and spe- 151.4 Malignant primary of corpus cialty hospitals.13 151.5 Malignant primary of lesser curvature of the stomach 151.6 Malignant primary of greater curvature of the stomach The NIS is supported by the Healthcare Cost and Utiliza- 151.8 Malignant primary of contiguous sites of the stomach tion Project and contains all-payer discharge information for 151.9 Malignant primary of stomach wall 100% of patient discharges from participating hospitals. A 197.8 Malignant secondary gastric neoplasm weighting strategy is implemented by the NIS to allow popu- 211.0 Benign neoplasm of esophagus (abdominal) lation-based estimates to be drawn at the national level. Sampled and fundus hospitals are given appropriate weights based on the number 211.1 Benign gastric neoplasm of hospitals they represent in the database for a given year.13 230.1 Malignant neoplasm in situ of esophagus (abdominal) and fundus DIAGNOSIS AND PROCEDURE CODES 230.2 Malignant gastric neoplasm in situ 235.2 Gastric neoplasm of uncertain behavior To identify all patient discharges with the principal diagnosis 239.0 Unspecified gastric neoplasm of gastric neoplasm, we used selected codes from the Interna- Procedural tional Classification of Diseases, Ninth Revision, Clinical Modi- 43.5 Gastrectomy (partial/proximal) with anastomosis to fication (ICD-9-CM).14 We limited the initial cohort to patients esophagus undergoing gastric resections based on ICD-9-CM procedure 43.6 Gastrectomy (partial/distal) with anastomosis to codes. The diagnosis and procedure codes used in the analysis duodenum are listed in Table 1. 43.7 Gastrectomy (partial) with anastomosis to jejunum 43.8 Gastrectomy (partial) 43.81 Gastrectomy (partial) with jejunal transposition VOLUME ANALYSIS 43.89 Gastrectomy (partial) with gastrogastrostomy 43.9 Gastrectomy (total) To evaluate annual surgical volume, we divided the participat- 43.91 Gastrectomy (total) with intestinal interposition ing hospitals into 3 groups. Hospitals that performed a mean 43.99 Gastrectomy (total), other of 4 or fewer gastric resections per year were classified as low- volume centers; more than 4 to less than 11 gastric resections Abbreviations: GE, gastroesophageal; ICD-9-CM, International Classification per year, as medium-volume centers; and 11 or more gastric of Diseases, Ninth Revision, Clinical Modification. resections per year, as high-volume centers. The cutoff values of 4 and 11 resections were chosen because they evenly divided the national distribution of annual hospital volume for Hospital volume has been identified as having a sig- gastrectomy into thirds, a technique previously described.15,16 nificant influence on surgical outcome; several studies have demonstrated an inverse relationship between operative OUTCOME ANALYSIS mortality and hospital surgical volume for high-risk procedures, including major cancer resections.6-9 In light of The primary measured outcome for this study was in-hospital this relationship, it has been proposed that implementa- mortality. Mortality was defined as death from any cause prior tion of volume standards could prevent many surgical to discharge, regardless of the time from the operation. deaths.10 Others have argued that the volume-outcome evidence is still incomplete, and a better understanding is STATISTICAL ANALYSIS needed of the factors that contribute to the apparent re- 11 We analyzed all data using the advanced survey procedures in lationship between high volume and better outcome. SAS statistical software (version 9.1; SAS Institute Inc, Cary, Other predictive factors of operative mortality that have NC). Univariate analyses of categorical variables were per- 12 been identified include age and type of resection. formed using Rao-Scott ␹2 tests, with PϽ.05 considered statis- Many studies examining surgical outcomes for gas- tically significant. Trend analyses were performed using the Man- tric cancer are limited by the number of subjects or by a tel-Haenszel ␹2 test. A multivariate logistic regression was single-institution design. The goal of this study was to constructed with in-hospital mortality as the dependent vari- use the Nationwide Inpatient Sample (NIS), a national able. Independent variables included patient age group (Ͻ50, Ն administrative database, to analyze contemporary in- 50-69, and 70 years), sex, year of resection, type of resec- hospital mortality rates for gastric resections for neo- tion, neoplasm diagnosis, and various comorbid medical con- plasm. We examined the relative effects of predictors of ditions (congestive heart failure [CHF], renal failure, diabetes mellitus, and liver disease).17 Although a total of 30 comorbid in-hospital mortality, including hospital volume, pa- conditions are listed for use with the administrative data sets,17 tient age and sex, diagnosis, procedure, and comorbid many are not readily applicable or important to surgical pa- conditions. We further examined the contribution of these tients or were only recorded in a few patients. Therefore, CHF, factors to possible outcome differences among high-, me- renal failure, diabetes mellitus, and liver disease were chosen dium-, and low-volume hospitals. to be analyzed as a representative sample with clinical appli-

(REPRINTED) ARCH SURG/ VOL 142, APR 2007 WWW.ARCHSURG.COM 388

©2007 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/28/2021 cability to surgical procedures. In addition, certain hospital- level characteristics were included, such as hospital surgical vol- Table 2. Patient Demographics for All Patients Undergoing ume (low, medium, or high), hospital type (teaching or Gastrectomy for Neoplasm* nonteaching), and geographic region. Adjusted odds ratios (ORs) were calculated to determine the effects of the identified co- Overall No. Nationally variates on the outcome measure of in-hospital mortality. (%) of Study Weighted No. Factor Patients of Patients RESULTS No. of patients undergoing resection 13 354 66 096 Age group, y† Ͻ50 1542 (11.6) 7609 PATIENT DEMOGRAPHICS 50-69 5256 (39.4) 26 046 Ն70 6545 (49.1) 32 384 From 1998 through 2003, 50 738 patient discharges (rep- Sex resenting 250 420 patients nationally) occurred with the Male 7996 (59.9) 39 605 Female 5351 (40.1) 26 455 principal diagnosis of gastric neoplasm. All further data Neoplasm diagnosis type references list weighted numbers parenthetically to also GE junction tumor 2903 (21.7) 14 466 highlight the data representative of the national trends and Gastric body tumor 6147 (46.0) 30 461 differences. Of this initial cohort of patients, 13 354 (26.3%) Esophageal tumor (abdominal) 32 (0.2) 158 (representing 66 096 patients nationally) underwent gas- Antrum/pyloric/prepyloric 2569 (19.2) 12 649 tric resection during their hospitalization. Of the patients Metastatic disease 205 (1.5) 1005 Benign/unspecified 1326 (9.9) 6478 undergoing surgery, 7996 (59.9%) (representing 39 605 Cancer in situ, any site 172 (1.3) 834 patients nationally) were male. The mean age for patients Procedure type undergoing surgery was 67 (range, 18-105) years; 11.6% Total gastrectomy including 3735 (28.0) 18 506 were younger than 50 years; 39.4% were aged 50 to 69 esophagogastrectomy years; and 49.1% were 70 years or older. Patient demo- Partial gastrectomy 2166 (16.2) 10 719 graphic characteristics are summarized in Table 2. Proximal gastrectomy 1556 (11.7) 7767 with esophagostomy Distal gastrectomy/ 1033 (7.7) 5081 UNIVARIATE ANALYSIS gastropylorectomy/Billroth I Gastrectomy/Billroth II 4864 (36.4) 24 024 For all patients undergoing gastric resection, in-hospital Comorbid conditions mortality was 6.0%. Male patients were found to have sig- CHF 1046 (7.8) 5153 nificantly increased in-hospital mortality. Increasing age Diabetes mellitus 1793 (13.4) 8874 Renal failure 223 (1.7) 1084 was also associated with increased mortality. Significant dif- Liver disease 250 (1.9) 1233 ferences in in-hospital mortality were also found across di- agnoses. Table 3 summarizes the types of procedures pa- Abbreviations: CHF, congestive heart failure; GE, gastroesophageal. tients underwent based on ICD-9-CM diagnosis codes. The *Because of missing data, numbers within each category may not total the highest rate of mortality was found among patients under- number of patients undergoing resection. Percentages have been rounded and might not total 100. going resection who had been diagnosed as having ab- †Mean age was 67 (SE, 0.18) years. dominal esophageal tumors (15.5% in-hospital mortality) compared with patients who had benign neoplasms (1.1% in-hospital mortality) (overall PϽ.001) (Table 4). tric body tumors (43.5% vs 47.4% and 47.5%), antrum/ Gastrectomy/BillrothIIwasthemostfrequentlyperformed pyloric/prepyloric neoplasms (16.0% vs 22.9% and 18.9%), operation (36.4%). The highest rates of in-hospital mortal- and benign neoplastic disease (8.8% vs 11.4% and 9.2%) ity were found among patients undergoing total gastrectomy (overall PϽ.001). High-volume hospitals performed a with or without intestinal transposition, including esopha- greater number of total gastrectomies (32.7% vs 23.1% and gogastrectomy (8.0%), gastrectomy/Billroth II (6.0%), and 28.0%) and proximal gastrectomies (14.5% vs 9.0% and proximalgastrectomywithesophagostomy(5.9%),compared 11.6%) and fewer gastrectomies/Billroth II (30.8% vs 42.5% with patients undergoing distal gastrectomy/gastropylo- and 36.0%) (overall PϽ.001) (Table 5). rectomy-Billroth I (3.2%) (overall PϽ.001) (Table 4). Certain comorbid conditions were also found to be MULTIVARIATE ANALYSIS predictive of increased in-hospital mortality, including CHF, renal failure, and liver disease. Using a multivariate logistic regression model, we evaluated Comparisons of patient characteristics across surgical independent predictors of in-hospital mortality (Figure). volume levels demonstrated that patients treated at hos- Lower annual surgical volume was found to be predictive pitals with higher surgical volumes were slightly younger of significantly increased in-hospital mortality for patients compared with the 2 lower-volume groups (mean ages, 65.6 undergoinggastricresection(lowestvshighestvolume,6.8% vs 68.5 and 67.2 years; PϽ.001). Patients at high-volume vs 4.9%; adjusted OR, 1.5; 95% confidence interval [CI], 1.2- hospitals were also found to have a lower percentage of cer- 1.8). Male sex was associated with significantly increased tain comorbid conditions, specifically CHF. Compared with in-hospital mortality (male vs female, 6.7% vs 5.0%; adjusted the 2 lower-volume groups, high-volume centers surgi- OR, 1.3; 95% CI, 1.1-1.5). Increasing age was also predic- cally treated a higher proportion of patients diagnosed as tiveofincreasedmortality(50-69vsϽ50years,4.0%vs2.1%; having gastroesophageal junction tumors (28.5% vs 15.5% adjusted OR, 1.5; 95% CI, 1.1-2.2) (Ն70 vs Ͻ50 years, 8.6% and 21.4%) and a lower proportion of patients with gas- vs 2.1%; adjusted OR, 2.9; 95% CI, 2.0-4.3). Among pro-

(REPRINTED) ARCH SURG/ VOL 142, APR 2007 WWW.ARCHSURG.COM 389

Procedure

Total Partial Proximal Gastrectomy Distal Gastrectomy/ Gastrectomy/ Neoplasm Type Gastrectomy† Gastrectomy With Esophagostomy Gastropylorectomy/Billroth I Billroth II GE junction 1709 (58.4)/8448 166 (5.8)/839 993 (34.6)/5002 7 (0.3)/37 28 (1.0)/139 Gastric body 1753 (28.7)/8728 963 (15.7)/4781 359 (5.8)/1776 430 (7.0)/2130 2642 (42.8)/13 046 Esophageal (abdominal) 20 (62.0)/98 2 (6.4)/10 10 (31.6)/50 0 0 Antrum/pyloric/ 124 (4.8)/613 241 (9.5)/1200 33 (1.2)/155 362 (14.0)/1780 1809 (70.5)/8947 prepyloric Metastatic disease 23 (11.2)/113 75 (36.5)/367 9 (4.3)/43 17 (8.1)/82 81 (39.8)/400 Benign/unspecified 44 (3.3)/212 696 (52.6)/3409 121 (9.1)/588 209 (15.6)/1013 256 (19.4)/1255 Cancer in situ, any site 62 (35.2)/293 23 (13.4)/112 31 (18.3)/153 8 (4.6)/39 48 (28.4)/237

Abbreviation: GE, gastroesophageal. *Data are expressed as unweighted number (percentage) of study patients/nationally weighted number of patients. †Includes esophagogastrectomy.

cedures, total gastrectomy was found to be predictive of sig- Table 4. Univariate Analysis of In-Hospital Mortality nificantly increased in-hospital mortality (total gastrectomy for All Patients Undergoing Gastrectomy for Neoplasm vs all other resections, 8.0% vs 5.3%; adjusted OR, 1.4; 95% CI, 1.2-1.7). Comorbid conditions found to be associated Mortality P with significantly increased in-hospital mortality included Factor Rate, % Value CHF (adjusted OR, 2.4; 95% CI, 2.0-2.9), renal failure (ad- Overall 6.0 justed OR, 6.3; 95% CI, 4.5-8.8), and liver disease (adjusted Sex Ͻ.001 Table 6 Male 6.7 OR, 2.5; 95% CI, 1.6-3.9) ( ). Female 5.0 Age group, y Ͻ.001 COMMENT Ͻ50 2.1 50-69 4.0 Ն70 8.6 In this study, we used representative US data from the NIS Neoplasm diagnosis type Ͻ.001 to demonstrate that the overall in-hospital mortality after GE junction tumor 8.1 gastric resection for neoplasm was 6.0% from 1998 through Gastric body tumor 6.2 2003. Our analysis shows that a number of factors affect Esophageal tumor (abdominal) 15.5 the in-hospital mortality rate for gastrectomy. Factors pre- Antrum/pyloric/prepyloric 5.5 dictive of significantly increased mortality included male Metastatic disease 7.1 Benign/unspecified 1.1 sex, older patient age, diagnosis type, gastrectomy type, Cancer in situ, any site 8.4 and low annual hospital surgical volume. On univariate Procedure type Ͻ.001 analysis, men had an approximately one-third excess risk Total gastrectomy, including 8.0 of in-hospital mortality compared with women (6.7% vs esophagogastrectomy 5.0%). Patients in the oldest group (Ն70 years) had a more Partial gastrectomy 4.1 than 4-fold risk of death compared with the youngest pa- Proximal gastrectomy 5.9 Ͻ with esophagostomy tients (8.6% vs 2.1% for age 50 years). Patients in the Distal gastrectomy/ 3.2 intermediate age group (50-69 years) had an intermedi- gastropylorectomy/Billroth I ate mortality rate (4.0%). In our study, patients undergo- Gastrectomy/Billroth II 6.0 ing total gastrectomy had the highest in-hospital mortal- Comorbid conditions ity (8.0%) of any gastric resection type, in contrast to CHF 16.2 Ͻ.001 Diabetes mellitus 5.4 .2 patients who underwent distal gastrectomy/gastropylorec- Renal failure 29.0 Ͻ.001 tomy/Billroth I (3.2%). Patients with certain comorbid con- Liver disease 12.1 Ͻ.001 ditions had much higher in-hospital mortality, including Hospital surgical volume Ͻ.001 individuals noted to have renal failure (29.0%) and CHF (gastrectomies per year) (16.2%). A hospital’s annual gastrectomy volume corre- Low (Յ4) 6.8 Medium (Ͼ4toϽ11) 6.4 lated with decreasing mortality, with low-, medium-, and High (Ն11) 4.9 high-volume hospitals having crude mortality rates of 6.8%, Year .4 6.4%, and 4.9%, respectively (PϽ.001). In the present study, 1998 6.5 we found that in-hospital mortality rates did not signifi- 1999 6.2 cantly change from 1998 through 2003, and that rates 2000 5.4 ranged from 5.3% to 6.5% during this era (P=.40). 2001 6.5 2002 5.3 To examine possible reasons for the mortality differences 2003 6.4 seen across hospital volume strata, we performed comparisons of selected covariates across low-, medium- and high- Abbreviations: CHF, congestive heart failure; GE, gastroesophageal. volume hospitals. These analyses demonstrated consider-

(REPRINTED) ARCH SURG/ VOL 142, APR 2007 WWW.ARCHSURG.COM 390

Hospital Volume, Gastrectomies per Year

Low Medium High Associated Characteristic (Յ4) (Ͼ4toϽ11) (Ն11) Overall Mortality, % No. of study patients (nationally weighted 4482 (21 799) 4403 (21 552) 4469 (22 745) 13 354 (66 096) 6.0 No. of patients) Sex, %† Male 57.5 60.6 61.7 60.0 6.7 Female 42.5 39.4 38.3 40.0 5.0 Mean age, y† 68.5 67.2 65.6 67.1 Age group, % Ͻ50 y 10.0 11.9 12.6 11.5 2.1 50-69 y 35.8 38.6 43.7 39.5 4.0 Ն70 y 54.2 49.5 43.6 49.0 8.6 Neoplasm diagnosis type, %† GE junction 15.5 21.4 28.5 21.9 8.1 Gastric body 47.4 47.5 43.5 46.1 6.2 Esophageal (abdominal) 0.3 0.3 0.2 0.2 15.5 Antrum/pyloric/prepyloric 22.9 18.9 16.0 19.2 5.5 Metastatic disease 1.6 1.2 1.7 1.5 7.1 Benign/unspecified 11.4 9.2 8.8 9.8 1.1 Cancer in situ, any site 1.0 1.4 1.4 1.3 8.4 Procedure type, %† Total gastrectomy including 23.1 28.0 32.7 28.0 8.0 esophagogastrectomy Partial gastrectomy 16.1 16.7 15.9 16.2 4.1 Proximal gastrectomy with esophagostomy 9.0 11.6 14.5 11.8 5.9 Distal gastrectomy/gastropylorectomy/ 9.3 7.7 6.1 7.7 3.3 Billroth I Gastrectomy/Billroth II 42.5 36.0 30.8 36.3 6.0 Comorbid conditions, % CHF† 9.0 8.2 6.3 7.8 16.2 Diabetes mellitus‡ 14.2 13.5 12.6 13.4 5.4 Renal failure§ 1.8 1.8 1.3 1.6 29.0 Liver disease|| 1.9 1.9 1.8 1.9 12.1

Abbreviations: CHF, congestive heart failure; GE, gastroesophageal. *Percentages have been rounded and may not total 100. †PϽ.001. †P = .09. ‡P = .14. §P = .94. ||P = .9.

able heterogeneity in the patient populations undergoing gastric resection at different-volume hospitals, with signifi- Increased In-Hospital Mortality Surgical Volume cant differences in mean patient age, the presence of major Low vs High comorbidities, diagnosis types, and types of surgical resec- Medium vs High tion. Patients treated at hospitals with higher surgical volumes were slightly younger compared with the 2 lower- Total Gastrectomy vs All Other Procedures volume groups (mean age, 65.6 vs 68.5 and 67.2; PϽ.001); these patients were also found to have a lower percentage Age of certain comorbid conditions, specifically CHF. This may 50-69 vs <50 Years implythathigh-volumecentersaretreatingyounger,healthier ≥70 vs <50 Years

patients. However, the data also show that high-volume cen- Male vs Female ters are treating less benign neoplastic disease compared with the 2 lower-volume groups (8.8% vs 11.4% and 9.2%; 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 Adjusted Odds Ratio PϽ.001), are more often performing the higher-risk procedures,andhavethehighestratesfortotalgastrectomies(32.7% vs 23.1% and 28.0%; PϽ.001), the procedure found to have Figure. Multivariate logistic regression for in-hospital mortality after gastric resection for neoplasm. Data are represented as adjusted odds ratios the highest in-hospital mortality. These data suggest a com- (triangles) with surrounding 95% confidence intervals. plex relationship between the patient type and hospital volume. After adjustment for these and other known factors patients undergoing resection at medium-volume hospitals in our multivariate model, patients undergoing gastric re- had an intermediate OR for death at 1.3. section at low-volume hospitals had odds for perioperative Our results, which showed an in-hospital mortality death 1.5-fold those of patients at higher-volume hospitals; rate of 6.0% for gastric resections for neoplasm, com-

(REPRINTED) ARCH SURG/ VOL 142, APR 2007 WWW.ARCHSURG.COM 391

©2007 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/28/2021 trectomy increase perioperative mortality validate in part Table 6. Logistic Regression of In-Hospital Mortality the previous work of others. In an analysis based on the for All Patients Undergoing Gastrectomy for Neoplasm Dutch Gastric Cancer Trial, investigators found that being older than 65 years and male were the most important OR (95% CI) risk factors for death, together with the extent of nodal Factor Adjusted Unadjusted dissection.26 Total gastrectomy has been found by some 19 Sex investigators to confer an additional risk of mortality, Male vs female 1.3 (1.1-1.5) 1.4 (1.2-1.6) whereas the national Veterans Affairs study did not find Age group, y a significance difference for morbidity or mortality by gas- 50-69 (vs Ͻ50) 1.5 (1.1-2.2) 1.9 (1.3-2.8) trectomy type.18 Ն70 (vs Ͻ50) 2.9 (2.0-4.3) 4.2 (2.9-6.1) An increasing body of evidence has documented the Neoplasm diagnosis type vs benign/unspecified inverse relationship between operative mortality and hos- 6-9 GE junction tumor 7.2 (4.2-12.2) 7.8 (4.6-13.2) pital surgical volume. However, unlike the results for Gastric body tumor 5.0 (3.0-8.2) 5.8 (3.5-9.7) procedures such as pancreatic resection, for which the Esophageal tumor 12.6 (3.9-40.6) 16.9 (5.7-50.0) volume-outcome relationship has been consistently es- (abdominal) tablished,16,27-32 the existing literature has painted an in- Antrum/pyloric/prepyloric 4.5 (2.7-7.5) 5.2 (3.1-8.7) determinate picture as to volume effects on mortality af- Metastatic disease 6.49 (3.0-13.9) 6.9 (3.3-14.5) Cancer in situ, any site 6.0 (2.8-12.5) 8.2 (4.0-16.5) ter gastrectomy. Investigators using the Medicare Claims Procedure type database and the NIS for 1994 through 1999 found that Total gastrectomy vs all other 1.4 (1.2-1.7) 1.6 (1.4-1.8) the mortality with gastrectomy for cancer ranged from resections 8.7% to 13% across volume strata.9 The same group,7 using Comorbid conditions the NIS for 1995 through 1997 to examine mortality with CHF vs none 2.4 (2.0-2.9) 3.5 (2.9-4.2) 8 types of cancer resections, found that differences for Diabetes mellitus vs none 0.8 (0.6-1.0) 0.9 (0.7-1.1) Renal failure vs none 6.3 (4.5-8.8) 6.9 (5.1-9.3) in-hospital mortality between low- and high-volume hos- Liver disease vs none 2.5 (1.6-3.9) 2.2 (1.5-3.2) pitals only were significant for pancreatectomy, esoph- Hospital surgical volume agectomy, and pulmonary lobectomy. Mortality rates for (gastrectomies per year) gastrectomy ranged from 8.7% in low-volume hospitals Low (Յ4) vs high (Ն11) 1.5 (1.2-1.8) 1.4 (1.2-1.7) (Ͻ9 procedures) to 6.9% in high-volume hospitals (Ͼ17 Medium (Ͼ4toϽ11) 1.3 (1.1-1.6) 1.3 (1.1-1.6) procedures), and after adjustment for confounders no dif- vs high (Ն11) ference was found in the adjusted OR for mortality. Abbreviations: CHF, congestive heart failure; CI, confidence interval; Previous data are conflicting regarding improvement 4,5 GE, gastroesophageal; OR, odds ratio. in operative mortality of gastric resections over time. Our data do not show a decrease in operative mortality for gastrectomy during the recent period from 1998 pare favorably with those reported previously in the lit- through 2003, although our overall mortality skews lower erature. Birkmeyer et al,9 who used the Medicare Claims than those of earlier studies. However, given known dif- database and the NIS for 1994 through 1999, found that ferences in databases such as the NIS, Medicare Claims the in-hospital or 30-day mortality for gastrectomy for Database, Medicare Provider Analysis and Review, and cancer ranged from 8.7% to 13%. Finlayson et al7 used Veterans Affairs National Surgical Quality Improve- the NIS from 1995 through 1997 to examine mortality ment Program, our results, while encouraging, are not after 8 types of cancer resections and reported a mortal- directly comparable to those of the previous studies. ity rate for gastrectomy ranging from 6.9% to 8.7%. Nei- Our study is by definition limited by the confines of ther of these studies presented an overall mortality rate the NIS data. Because the NIS is an administrative data- after gastrectomy, only ranges over volume strata. A na- base, the accurate assessment of case mix, including co- tional study from the Department of Veterans Affairs medi- morbidities, and the separation of preexisting comorbid cal centers showed a 7.6% overall 30-day mortality rate factors from postoperative complications can be diffi- after gastrectomy for cancer.18 Case series studies from cult.33,34 Stage information and preoperative patient per- single institutions in the United States have reported post- formance status measurements are not included in the NIS operative mortality rates as low as 4% after gastrectomy data set, although other investigators have concluded that for adenocarcinoma of the stomach19; however, these rates such patient-level characteristics are not likely to explain are not directly comparable to our national level data. Al- mortality differences across hospital volume levels.7,9,26 A though gastric neoplasms and their surgical treatment are further caveat is that the present study examines the out- known to vary widely by geographic region and by pa- come measure of in-hospital mortality. Other important tient race and ethnicity,20-23 international reports of in- outcomes, such as perioperative complications (eg, hem- terest include an operative mortality of 6.5% for cura- orrhage, infection, anastomotic leak, need for reopera- tive resections for gastric cancer at a single institution tion or intervention, and hospital readmission) and long- in Chile from 1982 to 1990.24 Gastric cancer results dif- term survival, are also important in understanding the fer most strikingly in Asian countries20; a recent study global risks after gastrectomy for neoplasm. from Taiwan reported in-hospital mortality ranging from Our data demonstrate that, in a large national database, 1.4% to 5.4%.25 patients undergoing gastrectomy at high-volume hospitals Our results showing that factors including male sex, have decreased perioperative mortality compared with pa- older patient age, and procedure types including total gas- tients undergoing these procedures at lower-volume hos-

(REPRINTED) ARCH SURG/ VOL 142, APR 2007 WWW.ARCHSURG.COM 392

©2007 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/28/2021 pitals.Althoughpatientsatlow-volumehospitalshaveanum- 5. Goodney PP, Siewers A, Stukel T, et al. Is surgery getting safer? national trends ber of risk factors predisposing to poorer outcomes, such in operative mortality. J Am Coll Surg. 2002;195:219-227. 6. Killeen SD, O’Sullivan MJ, Coffey JC, Kirwan WO, Redmond HP. Provider vol- as older age and significant medical comorbidities, these ume and outcomes for oncological procedures. Br J Surg. 2005;92:389-402. volume-mortality relationships maintain significance after 7. Finlayson EV, Goodney PP, Birkmeyer JD. Hospital volume and operative mor- a logistic regression to control for these and other covari- tality in cancer surgery: a national study. Arch Surg. 2003;138:721-725. ates. Our overall in-hospital mortality rate of 6.0% for gas- 8. Begg CB, Cramer L, Hoskins W, et al. Impact of hospital volume on operative trectomy for neoplasm is comparable to the 5.9% rate of in- mortality for major cancer surgery. JAMA. 1998;280:1747-1751. 9. Birkmeyer JD, Siewers AE, Finlayson EV, et al. Hospital volume and surgical mor- hospital mortality for pancreatic resection for neoplasm in tality in the United States. N Engl J Med. 2002;346:1128-1137. the NIS during the same period, which we have recently re- 10. Birkmeyer JD, Finlayson E, Birkmeyer C. Volume standards for high-risk surgi- ported.16 Because gastric neoplasms carry a relatively poor cal procedures: potential benefits of the Leapfrog initiative. Surgery. 2001; prognosis, patients being considered for gastric resection for 130:415-422. 11. Lipscomb J. Transcending the volume-outcome relationship in cancer care. J Natl neoplasm may be able to markedly improve their odds of Cancer Inst. 2006;98:151-154. long-term survival by seeking care at hospitals that perform 12. Park DJ, Lee HJ, Kim HH, et al. Predictors of operative morbidity and mortality relatively high volumes of such surgery. in gastric cancer surgery. Br J Surg. 2005;92:1099-1102. 13. Healthcare Cost and Utilization Project. Overview of the Nationwide Inpatient Sample Accepted for Publication: December 4, 2006. (NIS). http://www.hcup-us.ahrq.gov/nisoverview.jsp. Accessed September 20, 2006. 14. International Classification of Diseases, Ninth Revision, Clinical Modification. Wash- Correspondence: Jennifer F. Tseng, MD, Department of ington, DC: Public Health Service, US Dept of Health and Human Services; 1998. Surgery, University of Massachusetts Medical School, 15. Fahy BN, Frey C, Ho H, et al. Morbidity, mortality, and technical factors of distal UMass Memorial Medical Center, 119 Belmont St, Swift pancreatectomy. Am J Surg. 2002;183:237-241. House, Worchester, MA 01605 ([email protected]). 16. McPhee JT, Hill JS, Whalen GF, et al. Perioperative mortality for pancreatec- Author Contributions: Dr Tseng had full access to all of tomy: a national perspective. Ann Surg. In press. 17. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use the data in the study and takes responsibility for the in- with administrative data. Med Care. 1998;36:8-27. tegrity of the data and the accuracy of the data analysis. 18. Grossmann EM, Longo WE, Virgo KS, et al. Morbidity and mortality of gastrec- Study concept and design: Tseng. Acquisition of data: tomy for cancer in Department of Veterans Affairs Medical Centers. Surgery. 2002; Anderson and Tseng. Analysis and interpretation of data: 131:484-490. 19. Soreide JA, van Heerden JA, Burgart LJ, et al. Surgical aspects of patients with Smith, McPhee, Hill, Whalen, Litwin, and Tseng. Draft- adenocarcinoma of the stomach operated on for cure. Arch Surg. 1996;131: ing of the manuscript: Smith and Hill. Critical revision of 481-486. the manuscript for important intellectual content: McPhee, 20. Hundahl SA, Stemmermann GN, Oishi A. Racial factors cannot explain superior Hill, Whalen, Sullivan, Litwin, Anderson, and Tseng. Sta- Japanese outcomes in stomach cancer. Arch Surg. 1996;131:170-175. tistical analysis: Smith, McPhee, Hill, and Tseng. Ob- 21. Stemmermann GN, Nomura AM, Kolonel LN, Goodman MT, Wilkens LR. Gas- tric carcinoma: pathology findings in a multiethnic population. Cancer. 2002; tained funding: Anderson and Tseng. Administrative, tech- 95:744-750. nical, and material support: Whalen, Sullivan, Anderson, 22. Theuer CP, Kurosaki T, Ziogas A, Butler J, Anton-Culver H. Asian patients with and Tseng. Study supervision: Litwin, Anderson, and gastric carcinoma in the United States exhibit unique clinical features and supe- Tseng. rior overall and cancer specific survival rates. Cancer. 2000;89:1883-1892. 23. Yao JC, Tseng JF, Worah S, et al. Clinicopathologic behavior of gastric adeno- Financial Disclosure: None reported. carcinoma in Hispanic patients: analysis of a single institution’s experience over Funding/Support: This study was supported by the Pancre- 15 years. J Clin Oncol. 2005;23:3094-3103. aticCancerAlliance(DrTseng)andaPanCAN–SamuelStrom 24. Cenitagoya GF, Bergh CK, Klinger-Roitman J. A prospective study of gastric can- YoungInvestigatorAwardfromtheAmericanSocietyofClini- cer: “real” 5-year survival rates and mortality rates in a country with high incidence. cal Oncology (Dr Tseng). Dr Tseng is also a Howard Hughes Dig Surg. 1998;15:317-322. 25. Lin HC, Xirasagar S, Lee HC, Chai CY. Hospital volume and inpatient mortality Medical Institute Early Career Awardee. after cancer-related gastrointestinal resections: the experience of an Asian country. Previous Presentations: This study was presented at the Ann Surg Oncol. 2006;13:1182-1188. 87th Annual Meeting of the New England Surgical So- 26. Dutch Gastric Cancer Study Group. Risk factors for surgical treatment in the Dutch ciety; September 16, 2006; Groton, Conn, and is pub- gastric cancer trial. Br J Surg. 1997;84:1567-1571. 27. Birkmeyer JD, Finlayson SR, Tosteson AN, Sharp SM, Warshaw AL, Fisher ES. lished after peer review and revision. The discussions that Effect of hospital volume on in-hospital mortality with pancreaticoduodenectomy. follow this article are based on the originally submitted Surgery. 1999;125:250-256. manuscript and not the revised manuscript. 28. Birkmeyer JD, Warshaw AL, Finlayson SR, Grove MR, Tosteson AN. Relation- Acknowledgment: We thank Theodore P. McDade, MD, ship between hospital volume and late survival after pancreaticoduodenectomy. for his contributions to the UMass Surgical Outcomes Surgery. 1999;126:178-183. 29. Gordon TA, Bowman HM, Bass EB, et al. Complex gastrointestinal surgery: im- Analysis and Research Group, including the careful re- pact of provider experience on clinical and economic outcomes. J Am Coll Surg. view of the manuscript. 1999;189:46-56. 30. Gordon TA, Bowman HM, Tielsch JM, Bass EB, Burleyson GP, Cameron JL. Statewide regionalization of pancreaticoduodenectomy and its effect on in- REFERENCES hospital mortality. Ann Surg. 1998;228:71-78. 31. Gouma DJ, van Geenen RC, van Gulik TM, et al. Rates of complications and death 1. American Cancer Society. Cancer Facts & Figures 2006. Atlanta, Ga: American after pancreaticoduodenectomy: risk factors and the impact of hospital volume. Cancer Society Inc; 2006. Ann Surg. 2000;232:786-795. 2. Mansfield PF, Yao JC, Crane CH, et al. Gastric cancer. In: Kufe D, Pollock R, 32. Kotwall CA, Maxwell JG, Brinker CC, Koch GG, Covington DL. National esti- Weichselbaum R, eds. Cancer Medicine. 6th ed. Hamilton, Ontario: BC Decker Inc; mates of mortality rates for radical pancreaticoduodenectomy in 25 000 patients. 2003. Ann Surg Oncol. 2002;9:847-854. 3. Brennan MF. Current status of surgery for gastric cancer: a review. Gastric Cancer. 33. Iezzoni LI, Foley SM, Daley J, et al. Comorbidities, complications, and coding 2005;8:64-70. bias: does the number of diagnosis codes matter in predicting in-hospital mortality? 4. Msika S, Benhamiche AM, Tazi MA, et al. Improvement of operative mortality JAMA. 1992;267:2197-2203. after curative resection for gastric cancer: population-based study. World J Surg. 34. Jencks SF, Williams DK, Kay TL. Assessing hospital-associated deaths from discharge 2000;24:1137-1142. data: the role of length of stay and comorbidities. JAMA. 1988;260:2240-2246.

(REPRINTED) ARCH SURG/ VOL 142, APR 2007 WWW.ARCHSURG.COM 393

©2007 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/28/2021 lution? Have the authors seen equivalent responses with alter- same effect. We have used peritoneal dialysis fluid because it native hyperosmotic solutions, for example, hypertonic saline? is available in the clinical setting. The mechanism of the os- 3. The authors conclude that adjunctive DPR may favor- motic effect appears to prevent the sodium and water trans- ably impact postresuscitative SIRS. To what extent have they port across the membrane in response to cellular acidosis of been able to confirm this hypothesis in a chronic version of their the shock state. We have looked at a more severe model of shock animal model? And finally, the “million dollar question” (al- with a focus on outcome, and DPR improves survival by about though perhaps in the current economic environment a mil- 30%. We have done experiments in a model of severe decom- lion dollars is too small an amount)—have they examined the pensated shock. impact of this approach in patients? The osmolarity of the 2.5% solution that was used in these studies is almost 400. You asked about our clinical experience I thank the Association for the privilege of discussing this with this technique. We have used it in 12 patients with vari- important paper. ous degrees of acute abdominal compartment syndrome or in Dr Garrison: Dr West, thank you for your questions. The shock patients where damage control laparotomy was used. Sev- DPR solution was started at the beginning of the intravenous eral patients in this experience died as a result of severe injury resuscitation rather than at the time of the hemorrhage. The early in care or from head injury. This small experience does resuscitation was then carried out during the ensuing 30 min- not allow an analysis as a function of severity of injury or de- utes. The vasodilatory effect of the peritoneal dialysis fluid is gree of blood loss, but we have noted that bowel edema was almost instantaneous, and the microvessels maximally dilate. minimal at the time of reexploration and pack removal, and If you topically apply nitroprusside, the vessels will not dilate we were able to close the abdominal fascia. further. We have done other studies focused on osmotic control of the dilation effect. For example, mannitol will have the Financial Disclosure: None reported.

Correction

Errors in Table. In the Poster Session article titled “National Outcomes After Gastric Resection for Neoplasm” by Smith et al, published in the April 2007 issue of the Archives (2007;142[4]:387-393), errors inadvertently occurred in Table 3 on page 390. Because this study involved an analysis of the Nationwide Inpatient Sample, no data involving 10 or fewer obser- vations should have been published. The corrected Table 3 is included herein in its entirety.

Table 3. Gastric Resection Procedure by Neoplasm Diagnosis Type*

Procedure

Total Partial Proximal Gastrectomy Distal Gastrectomy/ Gastrectomy/ Neoplasm Type Gastrectomy† Gastrectomy With Esophagostomy Gastropylorectomy/Billroth I Billroth II GE junction 1709 (58.4)/8448 166 (5.8)/839 993 (34.6)/5002 Յ10 (0.3)/37 28 (1.0)/139 Gastric body 1753 (28.7)/8728 963 (15.7)/4781 359 (5.8)/1776 430 (7.0)/2130 2642 (42.8)/13 046 Esophageal (abdominal) 20 (62.0)/98 Յ10 (6.4)/10 Յ10 (31.6)/50 0 0 Antrum/pyloric/ 124 (4.8)/613 241 (9.5)/1200 33 (1.2)/155 362 (14.0)/1780 1809 (70.5)/8947 prepyloric Metastatic disease 23 (11.2)/113 75 (36.5)/367 Յ10 (4.3)/43 17 (8.1)/82 81 (39.8)/400 Benign/unspecified 44 (3.3)/212 696 (52.6)/3409 121 (9.1)/588 209 (15.6)/1013 256 (19.4)/1255 Cancer in situ, any site 62 (35.2)/293 23 (13.4)/112 31 (18.3)/153 Յ10 (4.6)/39 48 (28.4)/237

Abbreviation: GE, gastroesophageal. *Data are expressed as unweighted number (percentage) of study patients/nationally weighted number of patients. †Includes esophagogastrectomy.

(REPRINTED) ARCH SURG/ VOL 144 (NO. 7), JULY 2009 WWW.ARCHSURG.COM 634