Customize your JAMA Network experience by selecting one or more topics from the list below.
Zingmond DS, McGory ML, Ko CY. Hospitalization Before and After Gastric Bypass Surgery. JAMA. 2005;294(15):1918–1924. doi:10.1001/jama.294.15.1918
Author Affiliations: Division of General Internal
Medicine and Health Services Research, Department of Medicine (Dr Zingmond)
and Department of Surgery (Drs McGory and Ko), The David Geffen School of
Medicine at the University of California Los Angeles, Los Angeles; University
of California Los Angeles Center for Surgical Outcomes and Quality (Drs Zingmond,
McGory, and Ko); and West Los Angeles VA Medical Center, Los Angeles, Calif
Context The use of Roux-en-Y gastric bypass (RYGB) has been reported to be effective
in the treatment of obesity and its related comorbidities. Utilization of
inpatient services after RYGB is less well understood.
Objective To determine the rates and indications for inpatient hospital use before
and after RYGB.
Design, Setting, and Participants Retrospective study of Californians receiving RYGB in California hospitals
from 1995 to 2004.
Main Outcome Measure Hospitalization in the 1 to 3 years after RYGB.
Results In California from 1995 to 2004, 60 077 patients underwent RYGB—11 659
in 2004 alone. The rate of hospitalization in the year following RYGB was
more than double the rate in the year preceding RYGB (19.3% vs 7.9%, P<.001). Furthermore, in the subset of patients (n = 24 678)
with full 3-year follow-up, a mean of 8.4% were admitted a year before RYGB
while 20.2% were readmitted in the year after RYGB, 18.4% in the second year
after RYGB, and 14.9% in the third year after RYGB. The most common reasons
for admission prior to RYGB were obesity-related problems (eg, osteoarthritis,
lower extremity cellulitis), and elective operation (eg, hysterectomy), while
the most common reasons for admission after RYGB were complications often
thought to be procedure related, such as ventral hernia repair and gastric
revision. In multivariate logistic regression models predicting 1-year readmission
after RYGB, increasing Charlson Comorbidity Index score, and hospitalization
in the 3-year period prior to RYGB were significantly associated with readmission
within a year.
Conclusions Increases in hospital use after surgery appear to be related to RYGB.
Payers, clinicians, and patients must consider the not-inconsequential rate
of rehospitalization after this type of surgery.
Bariatric surgical procedures are an increasingly common treatment for
morbid obesity. More than 100 000 Roux-en-Y gastric bypasses (RYGBs)—the
primary bariatric procedure now done—are performed annually in the United
States.1,2 Weight loss after RYGB
is reportedly both significant and sustained—many individuals lose more
than 40 kg in the initial period after the procedure without significant rebound
in weight after this period.3 A recent systematic
review and meta-analysis of bariatric procedures determined that the mean
percentage of excess weight loss after operation was 61%, with rates of resolution
or improvement for the following comorbidities: diabetes 86%, hyperlipidemia
70%, hypertension 79%, and obstructive sleep apnea 84%.4 In
the Swedish Obese Subjects (SOS) cohort study, 10-year follow-up after bariatric
procedure showed increased weight loss, improved lifestyle, and amelioration
of risk factors (eg, diabetes, hypertriglyceridemia, and hyperuricemia) compared
with conventional treatment for obesity.5
Medical interventions for weight loss do not have such dramatic and
persistent weight loss associated with them, making surgical intervention
attractive to individuals who have been unable to successfully lose weight.6 Based on a review of the current data, the American
College of Physicians recommends operation as a treatment option for patients
with a body mass index greater than 40; obesity-related comorbid conditions;
and inability to lose weight with a combination of a diet, exercise, or drug
therapy program.7,8 Body mass
index is calculated as weight in kilograms divided by the square of height
Previous research has addressed some of the economic issues related
to bariatric procedures. In a Canadian cohort, overall costs at 5 years appear
lower among those receiving a bariatric operation.9,10 In
contrast, patients in the SOS study undergoing an operation were more likely
than those not undergoing an operation to be hospitalized in the 6 years after
operation and medication costs did not differ between the groups.11,12 Procedure costs are in excess of
$10 000 per person with an unknown cost for postoperative complications.3 It remains unclear from this literature whether after
bariatric operation, the use of inpatient services, which routinely makes
up a large proportion of direct costs in the general population,13 is
less than before a bariatric operation.
In the current research, we assessed the impact of RYGB on use of inpatient
care. We examined rates of inpatient hospitalization before and after RYGB
performed in California from 1995 to 2004. In this study, we addressed the
following 3 questions: (1) does utilization of inpatient hospitalization decrease
after RYGB; (2) are the reasons for inpatient hospitalization before and after
RYGB different; and (3) does the level of patient comorbidity and prior hospitalization
predict postoperative utilization of inpatient care?
Roux-en-Y gastric bypass cases in California from 1995 to 2004 were
identified using the annual hospital Patient Discharge Database, maintained
by the California Office of Statewide Health Planning and Development. The
database contains discharge abstracts for hospitalizations at all acute care
hospitals in California. Each discharge record abstract reports a unique record
linkage number (encrypted Social Security number), sex, age, race and ethnicity
categories (white, black, Asian, Native American, other, and unknown) expected
source of payment, ZIP code of residence, unique hospital identifier, diagnosis
related group, primary and secondary (up to 24) diagnoses, primary and secondary
(up to 20) procedures, dates of admission and discharge, total charges, and
patient disposition at discharge. All diagnoses and procedures were coded
using the International Classification of Diseases, 9th
Revision, Clinical Modification (ICD-9-CM).
Admissions for patients before and after RYGB were identified through
the record linkage number. Data on out-of-hospital deaths were identified
via linkage of Patient Discharge Database records with the most up-to-date
death records available (1995-2003) reported in the California Death Statistical
Master File. Thus, the most recent year of data with 365-day mortality is
2002 while the most recent data with 30-day mortality is through December
1, 2003. A description of the linkage methodology and its validation are described
elsewhere.14 Patient income was represented
by a proxy measure created by linking reported patient ZIP code of residence
to the 2000 US Census. Data on characteristics of individual hospitals were
obtained from Office of Statewide Health Planning and Development annual
hospital financial reports.
We identified cases of RYGB by selecting all records with diagnosis
related group equal to 288 (operating room procedure for obesity) and procedure
identifiable as being for gastric bypass—ICD-9-CM codes 44.31 and 44.39. Unique codes for laparoscopic RYGB do not exist
and alternative codes for laparoscopic procedures (eg, secondary procedure
codes for laparoscopy [54.21], laparoscopic cholecystectomy [51.23], or laparoscopic
appendectomy [47.01 or 47.11] appear to dramatically under code for laparoscopic
RYGB (only 39% were coded at our own institution) and thus a comparison of
laparoscopic and open cases was not performed. For each patient, the index
RYGB was the first operation reported between 1995 and 2004. Non-Californians
represented 6.1% of cases and were excluded from outcome analyses because
out-of-state readmission and mortality cannot be identified for these patients.
The occurrence and type of admissions for a year before and after the
index RYGB were characterized for all patients undergoing RYGB from 1995-2003.
Admission rates excluding pregnancy-related admissions were calculated for
each year before and after the index admission. The RYGB admission was not
included in the calculation of these measures. Using primary diagnosis and
procedure codes, we identified the prevalent reported reasons for hospitalization.
Three-year follow-up was available for individuals receiving RYGB through
2001, and these results are presented as a subgroup analysis. A cumulative
3-year admission rate was calculated and defined as any patient in the 1995-2001
subgroup with an admission in the 3 years before or after RYGB. We also calculated
total reported charges for admissions before and after RYGB.
Patient characteristics studied include sex; age category; combined
race and ethnicity categories (non-Latino black, non-Latino white, Latino,
Asian, other); insurance status, estimated median income (based on residence
ZIP code); Charlson Comorbidity Index;15 and
annual admission history in the 1, 2, and 3 years preceding RYGB. Race and
ethnicity was studied given the diverse nature of the California population
and the high prevalence of obesity in minority populations. Hospital characteristics
included hospital ownership, hospital size, academic medical center, and hospital
Analyses were constructed to answer the 3 research questions. First,
to determine whether the utilization of inpatient services decreased following
RYGB, we compared 1-year admission rates before and after RYGB for the cohort
undergoing RYGB between 1995 and 2003. In addition, 3-year admission rates
before and after RYGB were compared for the cohort with available 3-year follow-up
Second, we identified the reasons for inpatient hospitalization before
and after RYGB by comprehensive review of primary diagnosis and procedure ICD-9-CM codes for all admissions occurring before and
after RYGB. Diagnosis codes and procedure codes were grouped into 5 major
categories: (1) admission for a gastrointestinal tract procedure or gastric
bypass–related complication (eg, ventral hernia repair, gastric revision,
endoscopy); (2) elective admission unrelated to RYGB (eg, hysterectomy, plastic
surgery, knee surgery); (3) obesity-related admission (eg, osteoarthritis,
lower extremity cellulitis); (4) cardiac admission (eg, cardiac catheterization,
cardiac angioplasty, chest pain); and (5) other admission.
Reasons for admissions were attributed first to the major procedure
categories and then to the major diagnosis categories. Hierarchical assignments
were given to the procedure groupings because these tend to be more specific
and sensitive than corresponding diagnostic codes. Varying the order of assignment
did not significantly change the final classification results.
Finally, we performed staged multivariate logistic regression analyses
to identify factors associated with 1-year readmission following RYGB. We
specifically focused on the effect of the Charlson Comorbidity Index and prior
hospitalization within the 3 years preceding RYGB on the outcome of hospital
utilization a year after surgery for individuals having RYGB through 2001.
The model was controlled for patient case-mix (sex, age category, race, insurance
status, estimated median income), and hospital characteristics (hospital volume
and academic medical center). A clustering correction on hospital of surgery
was used to account for within-hospital clustering of cases and to improve
standard error estimates. A log likelihood test was performed to calculate
whether the Charlson Comorbidity Index score, prior hospitalization, or both
were significant predictors in the model. Because of multiple comparisons,
we chose a conservative threshold for statistical significance (P<.01).
All analyses were performed using Intercooled Stata 7.0 SE (College
Station, Tex). The study protocol was reviewed and approved by the internal
review board at the University of California Los Angeles. Due to the nature
of these secondary data analyses, waiver of patient consent was granted.
In California from 1995 to 2004, a total of 60 077 California residents
underwent RYGB for obesity. The number of procedures increased from 887 performed
in 1995 to 13 637 in 2003. In 2004, the number decreased to 11 659.
The 3694 non-California residents excluded from these analyses were similar
to California residents with respect to age, sex, and Charlson Comorbidity
Index but were more likely to be white, have self-pay insurance, and undergo
RYGB at for-profit, non-academic, and high-volume hospitals. Table 1 provides demographics of the overall cohort of California
residents who underwent RYGB from 1995 to 2004. The average age was 42.2 years,
83.9% of patients were women, and 88% were privately insured or self-pay.
The mean Charlson Comorbidity Index score was 0.62, with 43.6% of patients
having a score greater than 0.
Outcomes are also shown in Table 1.
Average length of stay was 3.5 days. Overall in-hospital mortality was 0.18%,
30-day mortality was 0.33%, and 1-year mortality was 0.91%. The in-hospital
mortality for Medicare beneficiaries (not shown in the table) was 0.64% and
higher than for the rest of the cohort (P<.001).
For patients with a year of follow-up (1995-2003), 19.3% were readmitted within
the first year compared with 7.9% being admitted in the year before surgery.
In a subset analysis of all patients who underwent RYGB with complete
3-year follow-up, we also examined the rate of hospitalization in the 3 years
before and after RYGB. Of these 24 678 patients, the mean percentage
of patients admitted in the year prior to RYGB was 8.4%. In each of the 3
years following RYGB, the rates of hospitalization remained increased, with
20.2% of patients readmitted in the first year after RYGB, 18.4% in the second
year after, and 14.9% in the third year after. The cumulative admission rate
for the 3-year period prior to RYGB was 20.2% compared with the cumulative
3-year admission rate after RYGB of 40.4%. For persons with 3 years of follow-up,
mean hospital charges were $33 672 for RYGB, $4970 for hospitalizations
in the 3 years before RYGB, and $20 651 for hospitalizations in the 3
years after RYGB. Finally, in the subset of patients with a full 5 years of
follow-up (1995-1999), postoperative admission rates remained elevated (mean
[SD], 13.3% [−0.3%]) in the fifth year after operation.
The hierarchically assigned rates and indications for hospitalization
for the year before and the 3 years after RYGB performed through 2001 are
shown in the Figure. Details of all
indications for admission are presented in Table
2. Prior to RYGB, the 2 most common reasons for admission were elective
operation (eg, hysterectomy and cholecystectomy) and obesity-related admission
(eg, osteoarthritis and lower extremity cellulitis). In contrast, the 2 most
common reasons for admission in the 1 to 3 years following RYGB were a gastrointestinal
(GI) tract or complication-related procedure or diagnosis (eg, gastric revision,
ventral hernia repair, wound infection, small bowel obstruction, and hypovolemia)
and elective surgery (eg, knee arthroplasty, plastic surgery). The 2 indications
for admission demonstrated differing trends. In the first year after RYGB,
the percentage of admissions for GI tract or complication-related reasons
was significantly greater than elective surgery. In the second year after
RYGB the percentage of admissions for GI tract or complication-related reasons
remains slightly greater than elective surgery, while in the third year after
RYGB, there were more admissions for elective procedures than for GI tract
or complication-related reasons.
Finally, we performed staged multivariate logistic regression models
predicting readmission in the year following RYGB (Table 3). A log-likelihood test for the addition of Charlson Comorbidity
Index (P<.001) and prior hospitalization (P<.001) to the baseline model (other patient-level and
hospital covariates) were both significant. After controlling for patient
demographics and hospital characteristics, patients who had a greater Charlson
Comorbidity Index score or were admitted in the 3 years before RYGB had greater
odds of 1-year readmission. Among the other covariates, being a woman was
associated with lower odds of readmission; and having Medicare or Medicaid
insurance or receiving RYGB at lower-volume hospitals was associated with
higher odds of readmission.
The number of gastric bypass operations performed in California increased
over 10-fold between 1995 and 2004 and will likely remain high due to the
epidemic problem of obesity and the lack of other efficacious long-term interventions.
An important factor supporting RYGB for the treatment of obesity is the resolution
of comorbid disease after RYGB, such as diabetes, sleep apnea, and hypertension.4
Our results also show that relative to utilization, mortality appears
to be a rare occurrence. While mortality is not a primary focus of our study,
we note that overall mortality rates within the hospital, at 30 days, and
at 1 year are low but are consistent with most other reports in the literature.4,16-18 Although
there is at least 1 population-based study that had rates as high as 1.7%
at 1 year,17 the large number of cases and
the accuracy of the California death linkage14 give
confidence to these findings. Relatively low mortality rates for RYGB in this
cohort likely reflect improving surgical skill (including the use of laparoscopic
approaches) in recent years, improved patient selection, and the concentration
of operations at high-volume referral centers in California.
A working hypothesis in our study was that use of health care services
should likewise improve, namely that inpatient care should decrease after
RYGB. However, we found significant and sustained increases in the rates of
hospital admission for morbidly obese patients after RYGB. Annual rates of
hospital admission after RYGB are double that prior to operation and are sustained
beyond a year in this population-based study. Most published studies in this
regard have been performed outside the United States but are inconclusive
about inpatient utilization after RYGB.9,10,12,19 For
example, in an observational cohort at a medical center in Canada, Christou
et al10 demonstrated that RYGB patients had
a lower number of hospitalizations, hospital days, physician visits, and total
direct costs over a 5-year period after operation vs age- and sex-matched
obese controls. In contrast, findings from the SOS study support the contention
that inpatient costs are higher in surgical patients postoperatively and that
these costs are directly related to RYGB.11 The
challenge to reconcile these differences underlies the importance of studying
inpatient utilization after RYGB in a population-based US sample. The current
study includes bariatric procedures performed at over 100 hospitals across
all of California, including a range of hospital types. The increase in utilization
after RYGB reflects the “real world” experience of these procedures
across surgical centers.
Our examination of the indications for admission shows that the reasons
for admission differ in the preoperative and postoperative periods. When classified,
diagnoses are much more likely to be obesity related (eg, osteoarthritis,
lower extremity cellulitis) or elective admissions (eg, hysterectomy and other
general surgery procedures such as cholecystectomy) in the years before RYGB
while following RYGB, particularly in the first year, the majority of admissions
appear to be procedure-related complications. In the third year following
RYGB, there was an encouraging reversal in that there were more admissions
for elective procedures than operation-related complications. Regarding elective
operations, it seems that after RYGB and successful weight loss, individuals
undergo plastic procedures, such as panniculectomy. Additionally, there is
probably a delayed demand for other interventions—elective procedures
postponed because of a patient’s obesity (eg, total knee arthroplasty,
The reasons for admission before RYGB differ from the chronic conditions
commonly highlighted as improving after RYGB. The recent meta-analysis and
systematic review of RYGB by Buchwald et al4 evaluated
the improvement or resolution of 4 obesity-related complications: diabetes,
hypertension, sleep apnea, and osteoarthritis, but none of these are highly
prevalent reasons for admission prior to RYGB in the current study. These
chronic conditions are much less likely to be considered as a primary reason
for admission, but rather as comorbid illnesses. In contrast, common reasons
for admission prior to RYGB are acute illnesses often associated with obesity—Pickwickian
Syndrome manifestations (chest pain, congestive heart failure, coronary artery
disease, and obstructive lung disease), lower extremity cellulitis, and cholelithiasis.
Our findings may have implications for payers and purchasers of health
care. Rather than expecting a decrease in inpatient health care utilization
after RYGB, the costs associated with inpatient hospitalization may remain
elevated for as many as 5 years following RYGB. Analysis of 3-year charges
before and after RYGB suggest that costs of post-RYGB–related procedures
and complications may be 40% to 60% of the costs of RYGB itself.
The research reported herein has a number of limitations. The study
is retrospective in nature and causality cannot be confirmed. Administrative
data may not accurately report all diagnoses and procedures. The California
Patient Discharge Database does not include the patient’s weight or
body mass index, which may be highly predictive of hospitalization both before
and after RYGB. We were only able to evaluate utilization of inpatient services
(vs outpatient visits or medication use) although these findings will likely
affect the overall costs of RYGB since inpatient hospitalization costs represent
the largest percentage (31%) of the US national health care expenditures.13 We were unable to reliably study laparoscopic vs
open RYGB procedures. Finally, analyses are for RYGB performed only in California
and may not be generalizable to other regions of the country or other periods.
Despite these limitations, the study presents the first comparison of hospitalization
rates for a 3-year period before and after RYGB with trends over time. The
current study demonstrates that there is not only an increase in hospitalization
rates after RYGB, but the reasons for admission are different after RYGB,
and this increase is apparent for at least 3 to 5 years after the procedure.
Lastly, this study demonstrates one of the limitations of using administrative
data for the purposes of comparing different types of procedures. We were
unable to distinguish laparoscopic and open procedures because we found that
the available coding scheme captured only 39% of cases performed laparoscopically
at our institution. As a result, differences in health care utilization found
between open and laparoscopic cases cannot be reliably detected.
The potential of RYGB for yielding long-term weight reduction and alleviation
of obesity-related comorbid illnesses has significantly increased the rates
of RYGB over the past decade. Despite these potential benefits, the current
study demonstrates that the rates of hospitalization double in the years after
operation and that many of these admissions are directly attributable to this
Corresponding Author: David S. Zingmond,
MD, PhD, Division of General Internal Medicine and Health Services Research,
Department of Medicine, The David Geffen School of Medicine at University
of California Los Angeles, 911 Broxton Plaza, Los Angeles, CA 90095-1736 (firstname.lastname@example.org).
Author Contributions: Dr Zingmond had full
access to all of the data in the study and takes responsibility for the integrity
of the data and the accuracy of the data analysis.
Study concept and design: Zingmond, Ko.
Acquisition of data: Zingmond, Ko.
Analysis and interpretation of data: Zingmond,
Drafting of the manuscript: Zingmond.
Critical revision of the manuscript for important
intellectual content: Zingmond, McGory, Ko.
Statistical analysis: Zingmond, McGory.
Obtained funding: Ko.
Administrative, technical, or material support:
Study supervision: Ko.
Financial Disclosures: None reported.
Funding Support: Dr Zingmond is funded by a
Mentored Clinical Scientist Award, grant K08 AG023024-01A1 from the National
Institute on Aging. Dr McGory is funded by the Robert Wood Johnson Clinical
Scholars Program at the University of California Los Angeles.
Role of the Sponsor: The funding sources had
no role in the development or synthesis of these analyses or in the writing
of the manuscript.