Cone MM, Herzig DO, Diggs BS, Dolan JP, Rea JD, Deveney KE, Lu KC. Dramatic Decreases in Mortality From Laparoscopic Colon Resections Based on Data From the Nationwide Inpatient Sample. Arch Surg. 2011;146(5):594-599. doi:10.1001/archsurg.2011.79
To determine the mortality rate and associated factors for laparoscopic and open colectomy as derived from the Nationwide Inpatient Sample database.
Nationwide Inpatient Sample database.
Between 2002 and 2007, the Nationwide Inpatient Sample estimated 1 314 696 patients underwent colectomy in the United States. Most (n = 1 231 184) were open, but 83 512 were laparoscopic. Patients who underwent a laparoscopic procedure that was converted to open were analyzed within the laparoscopic group on an intention-to-treat basis.
Main Outcome Measure
Mortality rate. Using a logistic regression model, patient and institutional characteristics were analyzed and evaluated for significant associations with in-hospital mortality.
In a multivariate analysis, significant predictors of increased mortality included older age, male sex, lower socioeconomic status, comorbidities, and emergency or transfer admission. Additionally, a laparoscopic approach was an independent predictor of decreased mortality when compared with open colectomy (relative risk, 0.51; P < .001).
Even when controlling for comorbidities, socioeconomic status, practice setting, and admission type, laparoscopy is an independent predictor of decreased mortality for colon resection.
Colon resection is a common operation that is associated with a mortality rate of 2% to 6%.1- 6 Multiple studies have evaluated factors related to mortality after colectomy, including surgeon factors or hospital volume,7- 9 while others have derived risk stratification models using patient comorbidities.6,10 Other models select colectomies for specific disease processes such as Clostridium difficile colitis,11,12 ischemic colitis,13 and colon cancer5,14 or by patient factors, including age or socioeconomic status.15 In this study, we sought to examine individual predictors of increased mortality among all colectomies, inclusive of all admission status, diagnosis, age, surgical approach, and socioeconomic class using a large national database.
The patient data were extracted from the Nationwide Inpatient Sample (NIS) compiled by the Healthcare Cost and Utilization Project for 2002 to 2007. This sample represents the all-payer inpatient experience via a 20% stratified probability sample of American nonmilitary, nonfederal hospitals for each year under consideration. Each individual discharge abstract for this population of patients is statistically weighted to provide a national representation of diagnoses and procedure volume. The NIS database includes inpatient data only, so deaths after hospital discharge were not included in this analysis.
Using the appropriate procedure and diagnosis codes as defined by the International Classification of Diseases, Ninth Revision, Clinical Modification, patients undergoing right, left, transverse, sigmoid, partial, multiple-segment, or total colectomy (codes 45.71-45.8) for each year under review were identified and their data were extracted from the overall data set. Cases were further classified as laparoscopic (additional procedure code 54.21 with or without diagnosis code V64.4 for conversion) or open (neither procedure code 54.21 nor diagnosis code V64.4 also present). Cases were included for analysis if they had any of the following diagnosis codes: 008.45 (C difficile colitis), 555.1 (Crohn colitis), 556 (ulcerative colitis), 558.9 (microscopic colitis), 560.2 (sigmoid or cecal volvulus), 562.1 (colon diverticulitis), 564.01 (constipation), 578.9 (acute gastrointestinal bleed), 235.5 (neoplasm of uncertain behavior), 211.3 (colon polyp), and 153 (colon cancer). Patient comorbidities were analyzed individually using a subset of the comprehensive list defined by the Elixhauser Comorbidity Index, which was developed by the Agency for Health Research and Quality for use with the NIS database.16 Comorbidities present in at least 5% of cases and those deemed clinically relevant (anemia, congestive heart failure, chronic pulmonary disease, diabetes mellitus, hypertension, fluid and electrolyte disorders, and obesity) were included. For the purpose of our analysis, converted cases were considered to be part of our laparoscopic cohort on an intent-to-treat basis.
Data extraction for the primary procedures, calculation of the national averages, and statistical analysis were performed using SAS/STAT software (release 9.1; SAS Institute Inc, Cary, North Carolina), taking into account the sampling weights and sampling structure of the data. Variables felt to be clinically relevant were included in a multivariate logistic regression model with mortality as the dependent variable. Because this is a nonrandomized study, we also evaluated the data with propensity matching to confirm findings from our multivariate analysis. To correct for potential confounding, we tested for effect measure modification with interaction terms.
Over the 6-year period between 2002 and 2007, a weighted estimate of 1 314 696 patients underwent a colectomy. Table 1 shows patient demographics. Of the 11 International Classification of Diseases, Ninth Revision diagnosis codes used as indications for undergoing a colectomy, colon cancer was the single most common diagnosis (43.8%), followed by diverticulitis (32.6%), and benign neoplasm (11.5%). Overall mortality was 3.7% (n = 48 637). In the open group, 3.9% died (n = 47 863), and mortality in the laparoscopic group was 0.9% (n = 774).
The multivariate mortality analysis is reported in Table 2. We found a substantially lower risk of death for all undergoing laparoscopic colectomy when compared with open colectomy (odds ratio [OR], 0.51; 95% confidence interval [CI], 0.43-0.60). Since this was nonrandomized, we used propensity matching to evaluate each decile of patients with the same likelihood of receiving a laparoscopic approach. In each decile, the mortality was lower for a laparoscopic approach. From this analysis, the overall mortality from laparoscopy carried an OR of 0.512 (95% CI, 0.43-0.60) compared with open.
Other factors associated with increased survival include younger age, female sex (OR, 0.82; 95% CI, 0.78-0.86), and hospitalization at a rural hospital (OR, 0.87; 95%, CI 0.81-0.95). Medicare and Medicaid insurance coverage were associated with decreased survival (OR, 1.85; 95% CI, 1.70-2.02 and OR, 2.07; 95% CI, 1.82-2.36, respectively). With respect to diagnoses, hemorrhage involving the colon (OR, 3.48; 95% CI, 3.06-3.94) and C difficile colitis (OR, 3.70; 95% CI, 3.32-4.12) had the most notable increased risk of death.
To further evaluate confounding, we looked carefully at effect measure modification. For in-hospital mortality, the multivariate model did not show any clinically significant interactions between approach (laparoscopic vs open) and diagnosis, age, or segment resected.
In this study, we used a national database to evaluate mortality-related outcomes for all patients undergoing a colectomy during 2002 to 2007. Laparoscopic colectomy was shown to be an independent predictor of a markedly decreased mortality rate when compared with open colectomy. Additionally, several factors were found to be associated with an increased risk of death, including older age, male sex, nonelective admission, diagnosis, Medicare insurance, lower income, and operations done at a teaching facility.
In establishing laparoscopic colectomy as a beneficial operation, perioperative mortality must be one of the first factors considered. Several large randomized trials evaluated laparoscopic treatment of colon cancer, including the Clinical Outcomes of Surgical Therapy Study Group (COST) trial, the multicenter European Colon Cancer Laparoscopic or Open Resection Study Group (COLOR) trial, the multicenter UK trial supported by the Medical Research Council (CLASSIC), and the Barcelona trial. None found a significant difference in early mortality between the open and laparoscopic groups.17- 20 Several other studies evaluated an existing database to look at mortality as a primary outcome, but the results vary from no difference to improved perioperative survival in the laparoscopic group.3,21- 24 These database studies, however, limited their patient selection by colon cancer, diverticulitis, nonemergent cases, or single institutions. Our analysis differs from these previous reports in that it reviews a national patient sample of all patients who underwent a colectomy during a 6-year period for all diagnoses, all admission types, and all levels of surgeon experience. These demographics allow conclusions to be made about laparoscopic colectomy as a whole, and the results are widely generalizable.
Our results show a significantly decreased risk of death with laparoscopic colectomy. The OR of 0.51 and a tight 95% CI (0.43-0.60) support the validity of this finding, even after accounting for comorbidities, demographics, and institutional characteristics in a multivariate analysis. These results are powered by the substantial patient numbers from NIS, the largest all-payer, inpatient care database,25 which may account for the lack of significance in the smaller studies. Also, the inclusion of emergency admissions in this data set may explain the higher mortality than what has been reported from the various randomized controlled trials. Our findings support a marked reduction of mortality in patients who are treated with a laparoscopic approach.
Other studies have looked at different operations and found similar outcomes with mortality. A recent publication by Dolan et al26 looked at laparoscopic vs open cholecystectomy and found laparoscopy to be an independent predictor of decreased mortality in all age groups younger than 80 years. Singla et al27 also looked at outcomes in complex laparoscopic abdominal operations and found decreased mortality in laparoscopic nephrectomy and colectomy groups compared with open.
The reason for our findings of a dramatic reduction in mortality in the laparoscopic colectomy group is difficult to elucidate. Confounding factors may exist within the database itself, including our inability to account for unmeasured comorbidities or assess cause of death. However, there is some evidence supporting the conclusion that laparoscopy itself is the factor associated with the lower mortality. Major open abdominal surgery is associated with a period of relative immunosuppression, while laparoscopic procedures minimize this cytokine response.28,29 Whether this decrease could lead to a mortality difference has yet to be shown. Another argument, as suggested by a recent National Surgical Quality Improvement Program study, is that the patients chosen for laparoscopic colectomy were overall at lower risk.30 The authors of that study did not, however, examine the relationship between mortality and patient variables in a multivariate model. The mortality reduction reported in our study persists after taking into account recorded comorbidities and other patient and institutional factors in a multivariate analysis.
Female sex was associated with a decreased risk of perioperative death when undergoing colectomy. In contrast, a study looking at the outcomes of sigmoid colectomy showed no difference in survival by sex, and a second study of colon cancer resections found no early survival difference by sex but found improved long-term survival in women.31,32 Dent et al33 found poorer survival in men undergoing colectomy for stage II cancer. They postulated that men experience greater postoperative morbidity than women because of clinical factors, such as comorbidities, or lack of protective biologic characteristics that women possess, including immunologic, hormonal, and molecular factors. There may also be unrecognized comorbidities in the male group or the severity of comorbidities is worse, which cannot be accounted for in this administrative database.
Another factor associated with improved survival was undergoing colectomy in rural hospitals. Similar findings were shown in a National Surgical Quality Improvement Program study looking at surgical outcomes between teaching and nonteaching centers. These findings were attributed to the fact that risk adjustment did not completely eliminate the differences in the patient populations, including a higher prevalence of comorbidities, longer operation times, and more complex operations in teaching institutions.34
There are several limitations to this study. The NIS is an administrative database and may be subject to coding errors. Selection bias is also difficult to eliminate without using randomization. Although we included comorbidities in our multivariate analysis, there is no way to quantify disease severity. We also cannot account for surgeon experience with laparoscopy, which undoubtedly contributed to the choice of surgical approach and to the conversion rate. Finally, this is a nonrandomized prospective cohort study. We attempted to compensate for that with our propensity matching analysis.
In conclusion, this article demonstrates that laparoscopic colectomy is an independent factor predictive of decreased mortality based on a representative sample of the US patient population representing all types of admissions, all payers, all income levels, and all types of treating institutions.
Correspondence: Kim C. Lu, MD, Oregon Health and Science University, 3181 SW Sam Jackson Park Rd, L223, Portland, OR 97239 (firstname.lastname@example.org).
Accepted for Publication: April 15, 2010.
Author Contributions:Study concept and design: Cone, Herzig, Dolan, Rea, Deveney, and Lu. Acquisition of data: Cone. Analysis and interpretation of data: Cone, Herzig, Diggs, Rea, and Lu. Drafting of the manuscript: Cone, Herzig, and Lu. Critical revision of the manuscript for important intellectual content: Herzig, Diggs, Dolan, Rea, Deveney, and Lu. Statistical analysis: Diggs. Administrative, technical, and material support: Cone, Herzig, and Rea. Study supervision: Herzig, Dolan, Deveney, and Lu.
Financial Disclosure: None reported.
Previous Presentation: This paper was presented as a poster at the 81st Annual Meeting of the Pacific Coast Surgical Association; February 14, 2010; Maui, Hawaii.