Laparoscopic vs Open Gastric Bypass Surgery: Differences in Patient Demographics, Safety, and Outcomes

Objective To determine national outcome differences between laparoscopic Roux-en-Y gastric bypass (LRYGB) and open Roux-en-Y gastric bypass (ORYGB).

Design Retrospective cohort study.

Setting The Nationwide Inpatient Sample.

Patients Patients undergoing ORYGB and LRYGB.

Main Outcome Measures Outcome measures were number of procedures performed, patient and hospital characteristics, patient complications, mortality, length of stay, resource use, and Agency for Healthcare Research and Quality Patient Safety Indicators. Both demographic and outcomes variables were compared by either t test or χ² analysis, with regression analysis adjusting for confounding variables.

Results The ORYGB and LRYGB cohorts consisted of 41 094 and 115 177 cases, respectively. From 2005 to 2007, LRYGB was more commonly performed than ORYGB (72% vs 28%; P < .001) and at high-volume hospitals (69% vs 61%; P < .001). A higher percentage of ORYGB compared with LRYGB patients were Medicare (9.3% vs 7.1%) and Medicaid (10.4% vs 5.9%; P < .01) beneficiaries. More ORYGB patients compared with LRYGB patients were discharged with nonroutine dispositions (7.7% vs 2.4%; P = .005), died (0.2% vs 0.1%; P < .001), and had 1 or more complications (18.7% vs 12.3%; P < .001). All Patient Safety Indicator rates were higher for ORYGB. Patients who had ORYGB compared with LRYGB also had longer median lengths of stay (3.5 vs 2.4 days; P < .001) and higher total charges ($35 018 vs $32 671; P < .001). Patients who had LRYGB had a lower odds ratio than patients who had ORYGB for both mortality (odds ratio, 0.54; P < .001) and having 1 or more complications (odds ratio, 0.66; P < .001) even after adjusting for confounding variables.

Conclusion In this population-based study, LRYGB provided greater safety than ORYGB even after adjusting for patient-level socioeconomic and comorbidity differences.

Obesity is a major public health concern as rates of obesity have significantly increased. In the United States, prevalence of obesity in adults increased from 15% in 1980 to 32% in 2004 and 6% to 19% in children during the same period.^1-3 According to Centers for Disease Control and Prevention reports, 34% of US adults older than 20 years are obese. In addition, the Centers for Disease Control and Prevention reports 66% of US adults are overweight or obese and 5 million are extremely obese.⁴ The 2008 F as in Fat report indicates that 25% of adults are obese in 28 states and that no state has had a decrease in the numbers of obese individuals.⁵

Studies have also shown troubling sociodemographic disparities in the United States because higher proportions of African American individuals, Hispanic individuals, Medicaid patients, and those of low education and income status are obese.⁶ Obesity concerns are not limited to the United States, as obesity has become an international issue, even in less developed countries.^7-9

Quiz Ref IDObesity has been associated with numerous adverse health conditions including diabetes mellitus, cardiovascular disease, nonalcoholic liver disease, increased risk of disability, hypertension, dyslipidemia, some forms of cancers, gallstones, and musculoskeletal disorders.^10-12 These adverse health conditions lead to a significant increase in early mortality and reduction in life years in obese individuals.^13,14 In addition, obese individuals report lower quality of life and greater discrimination, bias, and stigma.^15,16 Treating obesity and its complications has been estimated to cost approximately $92 billion a year.¹⁷

Quiz Ref IDBariatric surgery has proven to be the most effective and enduring option in treating the morbidly obese. In the 10-year Swedish Obesity Study, gastric bypass surgery patients showed greater long-term weight loss, health-related quality-of-life improvements, and reduction in various risk factors compared with controls receiving conventional treatment. The Swedish Obesity Study trial found that the decrease in weight and risk factors was greater in those treated by gastric bypass than those treated by medical intervention, banding, or vertical banded gastroplasty.^18,19

However, mortality and other complications are serious risks associated with bariatric surgery and are inversely correlated with the volume of cases performed by the surgeon.²⁰Quiz Ref IDThe Centers for Medicare & Medicaid Services determined that bariatric procedures should only be performed at high-volume facilities that are either certified by the American College of Surgeons or the American Society for Metabolic and Bariatric Surgery/Surgery Review Corporation as a Bariatric Surgery Center of Excellence.^21,22 The Centers for Medicare & Medicaid Services defines high-volume practices as performing 125 or more weight-loss operations annually.²³

The efficacy of gastric bypass surgery has made it the most commonly performed bariatric surgery in the United States. In 2002, 88% of all bariatric surgeries were gastric bypass procedures.²⁴ Laparoscopic Roux-en-Y gastric bypass (LRYGB) was first introduced by Wittgrove et al²⁵ in 1994 and since then has become more commonly performed than open Roux-en-Y gastric bypass (ORYGB). Weller and Rosati²⁶ report that 75% of gastric bypass procedures in 2005 were performed laparoscopically.

The safety of LRYGB vs ORYGB has been addressed in single-center trials, but to our knowledge, no national data exist comparing the 2 approaches. Two randomized controlled trials have shown better outcomes for LRYGB compared with ORYGB.^27,28 Prospective randomized controlled trials are used to establish the efficacy of interventions, but population-based studies are critical to determine the effectiveness and adoption of interventions. However, few population-based studies have considered the effectiveness of LRYGB compared with ORYGB in terms of patient safety because International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) laparoscopic codes for gastric bypass surgery were not available until 2004 and not fully adopted until 2005.²⁹

Data were derived from the 2005-2007 Nationwide Inpatient Sample (NIS), a hospital discharge database created as part of the Agency for Healthcare Research and Quality Healthcare Cost and Utilization Project. The NIS is the largest publicly available, all-payer, inpatient database in the United States. For each year, the NIS contains information on 5 to 8 million inpatient stays from about 1000 hospitals from 37 different states (approximately 20% of all community hospitals in the United States), stratified by geographic region, hospital size, urban vs rural location, and teaching vs nonteaching status. Teaching status of the hospital was determined by hospital affiliation with either a medical school or an Accreditation Council for Graduate Medical Education residency program. Hospital size (small, medium, and large) has differing definitions depending on urban or rural and teaching status. The NIS includes sample weights, designated by the sampling design, to make nationally representative estimates of inpatient use. The NIS is based on a stratified probability sample of all nonfederal US hospitals, which provides a national estimate of inpatient health services. The data set contains information on primary and secondary diagnoses, demographic characteristics, procedure use, length of stay, payer, total charges, and admission and discharge status. The NIS does not have unique patient identifiers, and as a result, patients cannot be followed up longitudinally. Additional information about the NIS is available from the Agency for Healthcare Research and Quality.³⁰

In this retrospective cohort study, patients were identified by the ICD-9-CM procedure codes for LRYGB (44.38) and ORYGB (44.31 and 44.39).²⁹ To ensure homogeneity, cases were restricted to adult nontransfer elective patients with a diagnosis-related group code 288, obesity. Patients were also excluded if they had malignant neoplasm of digestive organs or peritoneum (ICD-9-CM codes 150-159), inflammatory bowel disease (ICD-9-CM codes 555.0-556.9), or noninfectious colitis (ICD-9-CM codes 557.0-558.9). The period studied was from January 1, 2005, through December 31, 2007. This period was chosen so that the results would be sufficiently recent to include the new ICD-9 LRYGB code (44.38), which was first introduced in 2004 and not readily used in practice until 2005.

Patient age, sex, accompanying diagnoses, and comorbidities were examined. The Deyo modification of the Charlson Comorbidity Index³¹ (0-3, with 3 indicating greatest comorbidity) was calculated for each patient based on ICD-9-CM diagnosis codes. In addition, Elixhauser comorbidity measures were calculated.³² To adjust for potential differences in delivery of care that might bias outcomes, structure of care was assessed by examining payer status, hospital size, hospital volume, and teaching status.

Our outcomes of interests were number of procedures performed, patient and hospital characteristics, patient complications, mortality, length of stay, resource use, and Agency for Healthcare Research and Quality Patient Safety Indicators (PSIs). The ICD-9-CM codes for complications were gathered by referencing Santry et al²⁴ and performing our own review of ICD-9-CM codes (see codes in the eTable). The PSIs were adjusted for age, sex, diagnosis-related group, and Elixhauser comorbidity index.³² High-volume hospitals were defined as being hospitals that performed 125 or more bariatric surgery cases per year.

The PSIs were developed by the Agency for Healthcare Research and Quality Evidence-Based Practice Center at the University of California, San Francisco and Stanford University. The PSIs are a set of indicators providing information on potential inpatient adverse events following surgeries, medical procedures, and childbirth. The PSIs were developed through a comprehensive literature review, ICD-9-CM code study, review by a physician panel, risk adjustment measures, and empirical analyses. The PSIs screen for problems that patients experience as a result of exposure to the health care system and that are likely amenable to prevention by changes at the system or provider level.³³ The PSI software version 3.2 (March 2008) is available as module programs for use with SAS software version 9.1 (SAS Institute Inc).

We applied the sampling weights designated by the NIS sampling design to make nationally representative estimates of the number of procedures performed between 2005 and 2007. We established statistically significant differences in the categorical data using the Rao-Scott χ² and t test for continuous variables.³⁴ All point estimates, standard error estimates, and tests of significance accounted for the weighted nature of the data. Regression analyses were applied to complication and mortality outcomes to correct for potential confounders. All analyses were performed with SAS software version 9.1.

Characteristics of patients undergoing elective gastric bypass procedures from 2005 to 2007 appear in Table 1. The final cohorts of ORYGB and LRYGB were 41 094 and 115 177 weighted discharges, respectively (P < .001). The median age of the population was 42.7 years. The majority of patients were white (75%) and female (82.5%). We found that a higher percentage of ORYGB than LRYGB patients were covered by Medicare (9.3% vs 7.1%; P = .26) and Medicaid (10.4% vs 5.9%; P = .001).

The Figure describes characteristics of hospitals performing elective bariatric surgery procedures from 2005 to 2007. A majority of the gastric bypass procedures were performed at nonprofit (57%), teaching (61.3%), and high-volume hospitals (60.7%). There were no significant differences between LRYGB vs ORYGB for nonprofit or teaching status, but there was a significant difference with regard to high-volume status. A higher proportion of LRYGBs (69%) were performed at high-volume hospitals than ORYGB (61%) cases (P < .001).

A comparison of patient severity of illness is presented in Table 2. A higher percentage of ORYGB patients had Charlson Comorbidity Index scores of 2 or greater than LRYGB patients (7.8% vs 7.1%; P = .24). A greater proportion of ORYGB compared with LRYGB patients had comorbidities of diabetes mellitus (31.4% vs 29.6%; P = .03) and obstructive sleep apnea (28.6% vs 26.2%; P < .001), but a greater proportion of LRYGB compared with ORYGB patients had venous stasis disease (1.3% vs 0.7%; P < .001). There were no significant differences with hypertension or lysis of adhesion (surrogate marker for previous surgery) between the 2 cohorts. In terms of Elixhauser comorbidity measures, we found slightly higher rates in ORYGB than LRYGB cases for the following: cardiac arrhythmia (5.76% vs 4.26%; P = .008), congestive heart failure (2.03% vs 1.43%; P = .002), diabetes complicated (1.74% vs 1.20%; P = .01), pulmonary circulatory disorders (0.92% vs 0.47%; P = .001), and peripheral vascular disorders (0.46% vs 0.30%; P = .07). There were no significant differences between the 2 cohorts with the 25 other Elixhauser comorbidity measures (Table 3).

In Table 4, in-hospital outcomes are presented: total charges, length of stay, and rates of complication, mortality, and nonroutine disposition. Quiz Ref IDA significantly greater proportion of ORYGB compared with LRYGB patients were discharged with nonroutine dispositions (7.7% vs 2.4%), died (0.2% vs 0.1%), and had 1 or more complications (18.7% vs 12.3%). Patients who had ORYGB compared with LRYGB also had longer median lengths of stay (3.5 vs 2.4 days; P < .001) and higher total charges ($35 018 vs $32 671; P < .001).

Rates and odds ratios (ORs) of specific complications for LRYGB vs ORYGB are presented in Table 5. A higher percentage of ORYGB than LRYGB patients had blood transfusion (2.7% vs 1.9%; P < .001), abscess (0.7% vs 0.4%; P < .001), pulmonary embolism (0.3% vs 0.1%; P < .001), pneumonia (1.1% vs 0.5%; P < .001), other pulmonary complications (2.6% vs 1.2%; P < .001), wound complications (0.8% vs 0.4%; P < .001), splenic complications (0.5% vs 0.04%; P < .001), genitourinary tract complications (2.3% vs 1.3%; P < .001), cardiac arrhythmia (4.3% vs 2.8%; P < .001), cardiac complication not otherwise specified (1.2% vs 0.7%; P < .001), other and unspecified effects of external causes (9.2% vs 6.1%; P < .001), gastrointestinal leak (1.9% vs 1.2%; P < .001), sepsis (0.9% vs 0.4%; P < .001), and unexpected reoperations for surgical complications (1.1% vs 0.3%; P < .001). The OR for LRYGB vs ORYGB was 0.76 for blood transfusion (P < .001), 0.57 for abscess (P < .001), 0.24 for pulmonary embolism (P < .001), 0.50 for pneumonia (P < .001), 0.55 for other pulmonary complications (P < .001), 0.53 for wound complications (P < .001), 0.07 for spleen complications (P < .001), 0.64 for genitourinary tract complications (P < .001), 0.69 for cardiac arrhythmia (P < .001), 0.56 for cardiac complications not otherwise specified (P < .001), 0.65 for other and unspecified effects of external causes (P < .001), 0.66 for gastrointestinal leak (P < .001), 0.29 for unexpected reoperations for surgical complications (P < .001), 0.70 for small-bowel obstruction (P < .001), and 0.51 for sepsis (P < .001).

Selected PSI rates appear in Table 6. All PSI rates were significantly higher for ORYGB than LRYGB. Rates were more than twice as high for ORYGB vs LRYGB for failure to rescue, selected infections due to medical care, postoperative hemorrhage or hematoma, postoperative respiratory failure, and postoperative pulmonary embolism or deep vein thrombosis.

Data were entered into a multivariate logistic regression analysis to determine adjusted ORs for LRYGB vs ORYGB (Table 7). Laparoscopic gastric bypass was highly protective for mortality (OR, 0.54; P < .001), 1 or more complications (OR, 0.66; P < .001), and nonroutine disposition (OR, 0.43; P < .001). The model for 1 or more complications is presented in detail. Using patients who underwent ORYGB as a referent group, univariate analysis indicated that LRYGB had a reduced risk for in-hospital complications (OR, 0.657; P < .001) as did high-volume status (OR, 0.785; P = .003). The OR for complications was higher for age 60 years and older (OR, 2.174; P < .001), age between 50 and 59 years (OR, 1.818; P < .001), age between 40 and 49 years (OR, 1.430; P < .001), presence of lysis of adhesion (OR, 1.576; P < .001), and Medicare insurance status (OR, 1.422; P < .001). Calculation of incremental ORs indicated that the laparoscopic approach had the largest protective effect for complications, and there was little additive interaction between other known risk factors (Table 8).

To our knowledge, this is the first national, all-payer, all-hospital, and largest comparison of LRYBG and ORYGB. We found that the majority of patients in our study sample underwent LRYGB (73.7%). This is consistent with 2 recent national estimates^26,35 but much higher than estimates done prior to the introduction of ICD-9-CM codes specific to LRYGB.³⁶ A greater proportion of LRYGB cases were performed at high-volume hospitals and this may be reflective of the additional experience required to perform LRYGB. A few significant differences in comorbidities between ORYGB and LRYGB patients existed consistent with results from a population-based study looking at data from academic centers across the United States from 2004 to 2006.³⁵

In addition, Livingston and Ko⁶ reported that of those eligible for bariatric surgery, 16% were covered by Medicare and 12%, by Medicaid. Among those who received bariatric surgery, only 5% were Medicare patients and 8%, Medicaid patients. Quiz Ref IDThis study shows that Medicaid (7.2%) or Medicare (7.8%) coverage remains low among those receiving bariatric surgery. In addition, a higher percentage of ORYGB (10.4%) compared with LRYGB (5.9%) patients were covered by Medicaid (P < .01). These disparities among Medicaid and nonwhite patients may be caused by the low socioeconomic status associated with these 2 groups. Greater efforts should be made to educate these groups about the benefits of gastric bypass surgery and ensure they have access to appropriate bariatric procedures, especially by improving coverage and reimbursement for bariatric surgery in these groups.

Livingston and Ko⁶ also reported that males composed 36% of the population eligible for gastric bypass surgery but only 14% of the population who received gastric bypass surgery. Consistent with studies by Livingston and Ko and Santry et al,²⁴ we also observed that only 17.5% of gastric bypass procedures were performed on men. This disparity between males and females may be due to greater societal pressures on females to be thin and/or reluctance from males to seek medical help for morbid obesity.

Most importantly and consistent with previous prospective randomized trials and population-based studies, LRYGB vs ORYGB had significantly lower rates of complications, length of stay, charges, and mortality.^26-28,35-37 To our knowledge, this is the first study to compare PSI rates between LRYGB and ORYGB procedures and we found that rates for LRYGB were lower for all PSIs compared with ORYGB.

Limitations of this study are reflective of the data source. Data are collected administratively, and as a result, important clinical information such as body mass index is not present. Furthermore, the data set only addresses inpatient complications and longer-term complications are not available. Furthermore, the generally longer length of stay associated with ORYGB may lead to increased complication rates, given the greater opportunity for iatrogenic harm while in the hospital.

The minimally invasive approach of LRYGB appears to allow greater safety and lower resource use than ORYGB. This large, nationally representative comparison confirms and replicates prior randomized trial evidence supporting the laparoscopic approach, indicating safe dissemination of this technology. For bariatric surgery, patient safety may be further enhanced by appropriate application of the laparoscopic approach.

Correspondence: John M. Morton, MD, MPH, Department of Surgery, Stanford University Medical Center, 300 Pasteur Dr, H3680, Stanford, CA 94305-5655 (morton@stanford.edu).

Accepted for Publication: January 6, 2012.

Author Contributions:Study concept and design: Banka, Woodard, and Morton. Acquisition of data: Hernandez-Boussard. Analysis and interpretation of data: Banka, Woodard, Hernandez-Boussard, and Morton. Drafting of the manuscript: Banka, Woodard, and Morton. Critical revision of the manuscrpt for important intellectual content: Banka, Woodard, Hernandez-Boussard, and Morton. Statistical analysis: Banka, Woodard, Hernandez-Boussard, and Morton. Obtained funding: Morton. Administrative, technical, and material support: Morton. Study supervision: Morton.

Funding/Support: This research was funded by the Stanford Center for Outcomes Research and Evaluation, Stanford University Medical School and Stanford University Hospitals and Clinics, Stanford, California.