AAST 2012 PLENARY PAPER

How many sunsets? Timing of surgery in adhesive small bowel

obstruction: A study of the Nationwide Inpatient Sample

Dean Schraufnagel, BS, Sean Rajaee, BS, and Frederick Heaton Millham, MD, Newton, Massachusetts

BACKGROUND: Best practices promulgated by the Eastern Association for the Surgery of Trauma suggest that delay in surgery for adhesive
small bowel obstruction (ASBO) should not exceed 5 days. This study aimed to probe the relationship between operative delay
and adverse outcomes, defined as occurrence of a complication, requirement for bowel resection, prolonged postoperative stay,
or death in ASBO using the Nationwide Inpatient Sample.
METHODS: We used the Nationwide Inpatient Sample for 2009. The relationship among days to surgery (preoperative days) and defined as
occurrence of a defined set of complications, death during hospitalization, resection, and postoperative length of stay greater
than 7 days (postoperative days 9 7) was assessed, taking into account potential confounding factors using regression analysis.
RESULTS: A total of 27,046 patients were identified with small bowel obstruction; 4,826 (18%) of these required surgery, and the re-
mainder did not, staying a mean of 4 days (median, 3 days). Of the surgical group, 1,208 patients (25.0%) had Rsx, 1,527 (32%)
had postoperative days of greater than 7, 138 (2.86%) died, 3,216 (66.7%) were female. Mean age was 62.2 years, mean total
length of stay was 8.51 days, mean preoperative days was 1.94 days. Odds ratio (OR) of death for operated patients was 1.64
(95% confidence interval [CI], 1.11Y2.19) when preoperative days was 4 or more. Postoperative days of greater than 7 was
more likely if surgery preoperative days were 4 or more (OR, 1.26; 95% CIs, 1.07Y1.48). No relationship between
complication and preoperative days was observed.
CONCLUSION: Delay in management of small bowel obstruction is associated with death and longer postoperative stays. Delay was not
associated with complication or bowel resection. These data lend support to a policy encouraging observation of ASBO for no
more than 5 days. (J Trauma Acute Care Surg. 2013;74: 181Y189. Copyright * 2013 by Lippincott Williams & Wilkins)
LEVEL OF EVIDENCE: Epidemiologic study, level III; therapeutic study, level IV.
KEY WORDS: Bowel obstruction; complication, timing of surgery; Nationwide Inpatient Sample.

‘‘Never before in the history of medicine has the subject of
intestinal obstruction attracted more widespread attention than at
present.’’
VCharles Scudder, 19081
S mall bowel obstruction caused by abdominal adhesions
(ASBO) is a vexing consequence of abdominal surgery,
known to every surgeon who works within the peritoneal
cavity. ASBO emerged as an important clinical syndrome in the
last quarter of the 19th century, hand in hand with the enabling
technological developments of anesthesia and antisepsis.2,3
ASBO remains a significant clinical problem for general and
gynecologic surgeons today, occurring at a rate of 20 per
100,000 in the US population.4 Despite the contributions by
many of the leaders of general surgery during the last century,
there remains considerable debate over the indications for
surgery in patients who present with obstruction but without
overt evidence of strangulation or peritonitis. Much of the
Submitted: September 4, 2012, Revised: September 22, 2012, Accepted: September
25, 2012.
From the Department of Surgery, Newton Wellesley Hospital, Newton, MA.
Supplemental digital content is available for this article. Direct URL citations appear
in the printed text, and links to the digital files are provided in the HTML text of
this article on the journal’s Web site (www.jtrauma.com).
Address for reprints: Frederick Heaton Millham, MD, Newton Wellesley Hospital,
Newton, MA; email: fmillham@partners.org.
DOI: 10.1097/TA.0b013e31827891a1
J Trauma Acute Care Surg
Volume 74, Number 1
original surgical thinking on management of ASBO was the
product of an era when patients presented with more advanced
disease and surgeons worked with considerably less informa-
tion than is true today.1,5Y8 It was this early experience with
ASBO, which led to the dictum that surgeons should never let
the sun rise and set on a small bowel obstruction. Experience
during the past 30 years however has resulted in an increasing
use of nonoperative management of ASBO.9Y12 Yet, there re-
main proponents of early surgery.13,14 Two recent compre-
hensive evidence-based best practice guidelines have proposed
that nonoperative therapy is appropriate in many cases of ASBO
but should be limited to periods between 3 and 5 days.15,16
We wanted to look at current practices in the manage-
ment of ASBO in the United States and how these practices
compare with the proposed best practice guidelines. We wanted
to look at timing of surgery and for relationships between the
time from hospital admission to surgery and patient outcomes,
specifically the requirement for bowel resection, occurrence of
complications, prolonged postoperative length of stay (LOS), and
death. We hypothesize that preoperative delay exceeding that
recommended by existing best practice guidelines, 4 or 5 days,
would increase the incidence of complication, death, need for
small bowel resection, or prolonged postoperative hospital stay.
We used the Nationwide Inpatient Sample (NIS), a database of
inpatient hospitalizations created as part of the Healthcare Cost
and Utilization Project of the Agency for Healthcare Research
and Quality an agency of the US Department of Health and
Human Services.17 The 2009 edition of the NIS is widely used
in clinical research and contains demographic, diagnosis,
181

Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

Schraufnagel et al.
complication and procedure data on more than 8 million
patients admitted to hospitals in 44 states.18
PATIENTS AND METHODS
We used the NIS for 2009 as provided by Healthcare Cost
and Utilization Project. Data were managed and analyzed using
SAS statistical software (SAS Institute, Cary, NC). We included
for analysis any records with a primary diagnosis DRG Inter-
national Classification of DiseasesV9th Rev. (ICD-9) code
value of 560.81; intestinal or peritoneal adhesions with ob-
struction or 560.9; and unspecified intestinal obstruction.
Patients were considered to have had surgery if any of the ICD-
9 procedure codes for enterolysis 54.11, 54.21, 54.50, 54.51,
54.59, 45.61, or 45.62 were present in any of the primary op-
erative procedure fields. We identified patients as having had
a bowel resection if ICD-9 codes 54.61, multiple segmental
resection of small intestine, or 54.62, partial resection of
small intestine, were present in any of the procedure code fields.
We excluded records with ICD-9 codes for causes of ob-
struction other than adhesions in any of the NIS 14 secondary
diagnosis fields (Appendix 1 Supplemental Digital Content 1,
http://links.lww.com/TA/A223). We sought to exclude diagno-
ses other than postoperative adhesion that cause bowel obstruction
to focus the analysis on ASBO. We excluded patients who died
from the analyses of LOS in both the nonoperative management
and the operative management group.
The difference between the recorded date of admission
and the recorded date of first operative procedure defined
preoperative hospital days. These data were not normally dis-
tributed (Fig. 1); therefore, dummy variables were created to
categorize patients into groups who had surgery on or after the
fourth and the fifth hospital day. We excluded records where
the calculated delay in surgery was less than 0 or greater than
10 days. In the first case, we considered records with an ad-
mission to surgery days less than 0 to be erroneous records.
Figure 1. Pareto chart of number of preoperative days for
ASBO patients managed surgically.
182
Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
J Trauma Acute Care Surg
Volume 74, Number 1
There were very few of these. Moreover, more than 95% of
patients were operated on by the 10-day point; we defined those
exceeding this period as outliers. We defined postoperative
hospital days as the difference between the recorded date of
surgery and the recorded discharge date. We defined prolonged
postoperative LOS as that exceeding 7 days. We chose this
value because it represents the number of days of inanition at
which our hospital requires surgeons to consider providing
total parenteral nutrition. We excluded records with postop-
erative hospital days of 0 or greater than 20 because these likely
represented erroneously coded or outlying cases. We modified
the complications indicator provided with the NIS to better
reflect postoperative complications. The diagnoses included in
the definition of postoperative complication for this study are
listed in Appendix 2, (see Supplemental Digital Content 1,
http://links.lww.com/TA/A224) available on line. The demo-
graphic and comorbid disease data provided in the NIS defined
potential confounders for use in multivariate analyses; we
converted these data to categorical elements for this purpose.
We excluded patients younger than 18 years.
For patients undergoing surgery, we used death, re-
quirement for bowel resection, occurrence of complications,
and postoperative hospital stay greater than 7 days as outcome
variables. Relationships between these outcomes and other
factors were investigated using t test for continuous variables
such as age and W2 with Fisher’s exact test for categorical
variables. Factors with univariate statistical significance of less
than 0.10 were included in logistic regression analyses to assess
the relationships between the outcome variables, preoperative
days, and other potential confounding variables. We created a
number of models to look for the most parsimonious logistic
constructs with the greatest explanatory power. The final models
for each outcome, therefore, contained unique covariates.
RESULTS
A total of 27,046 patients were identified with ASBO,
meeting our inclusion criteria. Only 4,826 patients (18%) re-
quired surgery (Table 1). Patients managed nonoperatively were
hospitalized an average of 4 days (median, 3 days) (Fig. 2).
Eighty-one percent of patients avoiding surgery were dis-
charged within 5 days; 90% were discharged within 7. LOS
for patients having surgery averaged 8.5 days. Women repre-
sented 60% of the overall group but required surgery in 19% of
cases (p G 0.001). The mean age of all patients was 63.4 years;
the mean age of patients when operation was performed was
62 years (p G 0.001). White and African-American patients
were more often operated on (18% and 21%, respectively; both
p G 0.001). Native Americans (8%) all had lower rates of op-
erative management (p = 0.004). Of other demographic factors
reported in the NIS, Medicare patients were operated on less
frequently, having a 16% rate of operation (p G 0.001). Pri-
vately insured patients had a higher rate of operation of 19%
( p G 0.001). Patients treated at hospitals in smaller metropol-
itan areas of population 50,000 to 250,000 had a significantly
higher rate of operation (20%, p G 0.001). Nonmetropolitan or
rural counties had a lower percentage of patients managed
operatively (15%, p G 0.001).
* 2013 Lippincott Williams & Wilkins
J Trauma Acute Care Surg
Volume 74, Number 1 Schraufnagel et al.

and less frequent in patients on Medicaid (16%, p G 0.001).

TABLE 1. Demographic Characteristics of Patients With
ASBO in the NIS
Small hospitals had a lower rate of bowel resection (21%,
p = 0.01) as did urban nonteaching facilities (24%, p = 0.04).
All Patients Operative p Women had a lower rate of death during the admission
Total, n (%) 27,046 4,826 (18) (2%, p = 0.03), as did patients with private insurance (1%, p G
Age, mean (SD) 63.4 (16.9) 62 (17.5) G0.001 0.001) and self-pay status (G1%, p = 0.02). Patients admitted
LOS, mean (SD), d 5.3 (5.8) 8.5 (4.7) G0.001 over a weekend had a higher mortality rate (4%, p = 0.003).
Female, n (%) 16,167 (60) 3,216 (20) G0.001 Patients in medium-sized hospitals had a higher death rate (4%,
Race, n (%) p = 0.02), so too did those in rural hospitals (5%, p = 0.02).
Caucasian 17,553 (65) 3,253 (19) G0.001 Postoperative LOS of greater than 7 days was more fre-
African American 2,523 (9) 554 (22) G0.001 quent in black patients (36%), those admitted over a weekend
Hispanic 1,856 (7) 301 (16) 0.06 (33%, p = 0.01), and those classified as emergency admissions
Asian or Pacific Islander 593 (2) 104 (18) 0.87 (33%, p = 0.004). Income less than $38,000 per year was as-
Native American 139 (1) 12 (9) 0.00 sociated with higher percentage of patient’s having a postop-
Other 635 (2) 115 (18) 0.88 erative LOS exceeding 7 days (37%, p G 0.001). Conversely,
Missing 3,747 (14) 487 (13) income greater than $68,000 was associated with a lower rate
Median household income, n (%) of long postoperative stay of 27% ( p G 0.001). Medicare in-
G$1Y38,999 6,235 (23) 1,082 (17) 0.26 sured patients had a higher rate of long stays (42%, p G 0.001),
$39,000Y47,999 7,126 (26) 1,318 (18) 0.10 while privately insured patients had a lower rate of long stays
$48,000Y62,999 6,578 (24) 1,165 (18) 0.75 (19%, p G 0.001) as did patients whose payment source was
9$63,000 6,505 (24) 1,148 (18) 0.64 listed as ‘‘other’’ (22%, p = 0.02). Small hospitals had fewer
Missing 602 (2) 113 (19) patients who had long stays (26%, p G 0.001), and large hos-
Primary payer, n (%) pitals had more patients who had long stays (33%, p = 0.05).
Medicare 14,118 (52) 2,328 (16) G0.001 A number of comorbid conditions were associated with
Medicaid 1,801 (7) 303 (17) 0.25 the adverse outcomes. These are listed in Table 2.
Private or HMO 9,017 (33) 1,816 (20) G0.001 We constructed a number of multivariate models using
Self pay 1,188 (4) 223 (19) 0.39 a categorical variable (dummy variable) to identify patients
No charge 116 (0) 22 (19) 0.72 having preoperative days of 4 or more to evaluate the effect of
Other 735 (3) 120 (16) 0.30 delay in surgery, if any, on the four adverse outcomes once
Missing 71 (0) 14 (20) potentially confounding variables had been accounted for. When
Patient location, n (%) account was taken of all available potentially confounding
Central metro area Q 1 million people 7,095 (26) 1,277 (18) 0.69 variables, the relationship between preoperative days and
Fringe metro area Q 1 million people 6,518 (24) 1,169 (18) 0.82 complication and resection disappeared. The models for ex-
Metro areas 250,000Y999,000 5,149 (19) 936 (18) 0.49 cessive postoperative LOS, presented in Table 3, and that for
Metro Areas 50,000Y250,000 2,260 (8) 466 (21) G0.001 death, presented in Table 4, suggested a more robust rela-
Micropolitan counties 3,290 (12) 555 (17) 0.12 tionship between preoperative delay and these outcomes.
Not metrocounties or microcounties 2,116 (8) 329 (16) 0.004 These models had reasonable calibration and goodness of fit
Missing 618 (2) 94 (15)
Demographic characteristics of all patients satisfying inclusion criteria for ASBO and
the characteristics of those who did and did not require surgical intervention. p values are
the result of t test in the case of continuous variables, age, and LOS. For all other cate-
gorical variables, p values are result of W2 test.

Of the 4,826 who required surgical intervention, 922

patients (19%) had a complication, 1,208 (25.0%) required bowel
resection, 138 (3%) died, 1,527 (32%) remained hospitalized
for more than 7 postoperative days. The average preoperative days
was 1.94 days overall. Patients older than 60 years and those for
whom surgery occurred on the fourth postadmission day or later
were more likely to have all of the adverse outcomes, such as,
complication, requirement for bowel resection, death, and greater
than 7 postoperative days. Complication was more frequent in
patients with Medicare and less frequent in those with private
insurance (both p G 0.001). Complication was less likely in private
investorYowned hospitals (15%, p = 0.03).
The need for bowel resection was more common in Black
patients (19% p = 0.05), patients with Medicare and private Figure 2. Frequency distribution of hospital LOS for patients in
health insurance (28% and 22%, respectively; both p e 0.001) NIS with small bowel obstruction who do not require surgery.

* 2013 Lippincott Williams & Wilkins 183

Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
 J Trauma Acute Care Surg
Schraufnagel et al. Volume 74, Number 1

TABLE 2. Rates of Adverse Event After Operation for SBO by Delay Duration, Demographic Factors and Co-Morbidity
97 d Postoperative
All Patients Complication Resection Died LOS
n n % p n % p n % p n % p
Total 4,826 922 19 1,208 25 138 3 1,527 32
Age Q 60 y 2,726 604 22 G0.001 769 28 G0.001 127 5 G0.001 1,109 41 G0.001
Female 3,216 601 19 0.29 802 25 0.80 80 2 0.03 992 31 0.10
Days from admission to OR
4 or more preoperative days 972 239 25 G0.001 274 28 0.001 50 5 G0.001 420 43 G0.001
5 or more preoperative days 634 153 24 G0.001 199 31 G0.001 33 5 G0.001 288 45 G0.001
Race
White 3,253 631 19 0.48 794 24 0.16 89 3 0.20 1,008 31 0.17
Black 554 100 18 0.53 158 29 0.05 16 3 0.89 202 36 0.01
Hispanic 301 49 16 0.23 74 25 0.89 6 2 0.47 88 29 0.37
Native American 12 4 33 0.26 2 17 0.74 0 0 1.00 4 33 1.00
Other race 115 24 21 0.63 30 26 0.83 2 2 0.77 40 35 0.47
Admission type
Weekend admission 1,182 230 19 0.73 293 25 0.85 49 4 0.003 412 35 0.01
Admission type, emergent 3,124 597 19 1.00 773 25 0.53 97 3 0.2 1,033 33 0.004
Admission type, urgent 763 151 20 0.62 185 24 0.62 21 3 0.9 232 30 0.45
Income
Income G 38,000 1,082 206 19 0.97 280 26 0.47 35 3 0.41 397 37 G0.001
Income 39,000Y48,000 1,318 251 19 0.97 329 25 0.97 47 4 0.08 421 32 0.78
Income 48,000Y63,000 1,165 230 20 0.52 310 27 0.16 25 2 0.11 366 31 0.86
Income 9 63,000 1,148 215 19 0.73 270 24 0.18 30 3 0.61 314 27 G0.001
Urban/rural
Central metro area Q 1 million people 1,277 246 19 0.87 344 27 0.07 35 3 0.84 431 34 0.06
Fringe metro area Q 1 million people 1,169 239 20 0.19 301 26 0.53 30 3 0.55 348 30 0.12
Metro areas 250,000Y999,000 936 167 18 0.29 220 24 0.24 30 3 0.51 295 32 0.94
Metro areas 50,000Y250,000 466 85 18 0.66 115 25 0.91 11 2 0.66 150 32 0.79
Micropolitan counties 555 105 19 0.95 142 26 0.75 23 4 0.06 168 30 0.50
Not metrocounties or microcounties 329 66 20 0.66 73 22 0.24 9 3 1.00 114 35 0.24
Insurance status
Medicare 2,328 526 23 G0.001 651 28 G0.001 119 5 G0.001 987 42 G0.001
Medicaid 303 50 17 0.26 49 16 G0.001 4 1 0.11 88 29 0.34
Private Insurance 1,816 276 15 G0.001 408 22 0.001 13 1 G0.001 354 19 G0.001
Self-pay 223 40 18 0.73 61 27 0.43 1 0 0.02 60 27 0.12
No charge 22 3 14 0.79 9 41 0.09 0 0 1 8 36 0.65
Other payment source 120 23 19 1.00 26 22 0.46 0 0 0.05 26 22 0.02
Hospital size
Small 567 111 20 0.78 118 21 0.01 14 2 0.69 147 26 G0.001
Medium 1,267 233 18 0.48 325 26 0.57 48 4 0.02 403 32 0.89
Large 2,905 560 19 0.74 734 25 0.66 75 3 0.16 950 33 0.05
Hospital type
Government or federal 289 45 16 0.12 59 20 0.07 6 2 0.58 92 32 0.95
Private nonprofit 933 185 20 0.55 227 24 0.61 24 3 0.66 285 31 0.43
Private Investor owned 494 76 15 0.03 114 23 0.30 20 4 0.11 151 31 0.61
Private 209 44 21 0.47 50 24 0.74 9 4 0.20 63 30 0.70
Rural hospital 577 110 19 1.00 141 24 0.76 26 5 0.02 171 30 0.29
Urban nonteaching 2,098 416 20 0.27 494 24 0.04 54 3 0.34 682 33 0.26
Urban teaching 2,064 378 18 0.24 542 26 0.09 57 3 0.79 647 31 0.71
Geographic location
Northeast 1,000 195 20 0.72 259 26 0.49 20 2 0.07 324 32 0.57
Midwest 1,044 214 20 0.20 253 24 0.52 31 3 0.83 342 33 0.39
Southwest 1,855 333 18 0.11 473 25 0.56 57 3 0.48 579 31 0.63
West 927 180 19 0.78 223 24 0.47 30 3 0.44 282 30 0.39
(Continued on next page)

184 * 2013 Lippincott Williams & Wilkins

Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
J Trauma Acute Care Surg
Volume 74, Number 1 Schraufnagel et al.

TABLE 2. (Continued)
97 d Postoperative
All Patients Complication Resection Died LOS
n n % p n % p n % p n % p
Comorbidity
AIDS 9 3 33 0.39 4 44 0.24 1 11 0.2 4 44 0.48
Alcohol abuse 84 21 25 0.16 23 27 0.61 2 2 1.0 34 40 0.10
Iron deficiency anemia 752 196 26 G0.001 249 33 G0.001 47 6 G0.001 376 50 G0.001
Rheumatoid arthritis or collagen vascular disease 138 35 25 0.06 39 28 0.37 4 3 1.00 47 34 0.58
Chronic blood loss anemia 39 11 28 0.15 11 28 0.71 1 3 1.00 19 49 0.02
Congestive heart failure 375 112 30 G0.001 127 34 G0.001 43 11 G0.001 241 64 G0.001
Chronic lung disease 815 194 24 G0.001 207 25 0.79 46 6 G0.001 334 41 G0.001
Coagulopathy 155 48 31 G0.001 62 40 G0.001 20 13 G0.001 77 50 G0.001
Depression 499 102 20 0.43 112 22 0.17 15 3 0.78 166 33 0.42
Uncomplicated diabetes mellitus 660 125 19 0.96 178 27 0.23 22 3 0.45 256 39 G0.001
Diabetes with chronic complications 88 20 23 0.41 18 20 0.38 6 7 0.04 41 47 0.004
Drug abuse 70 21 30 0.03 17 24 1.00 3 4 0.45 25 36 0.52
Hypertension 2,363 492 21 0.003 610 26 0.22 71 3 0.60 891 38 G0.001
Hypothyroidism 606 110 18 0.54 151 25 0.96 13 2 0.30 214 35 0.04
Liver disease 89 24 27 0.08 21 24 0.81 6 7 0.04 26 29 0.73
Lymphoma 42 12 29 0.12 17 40 0.03 1 2 1.00 14 33 0.87
Electrolyte disorder 1,749 450 26 G0.001 513 29 G0.001 91 5 G0.001 790 45 G0.001
Metastatic cancer 87 26 30 0.01 29 33 0.08 5 6 0.10 50 57 G0.001
Neurological disorder 273 63 23 0.10 77 28 0.22 12 4 0.13 131 48 G0.001
Obesity 303 61 20 0.65 67 22 0.24 3 1 0.05 98 32 0.80
Paralysis 45 12 27 0.19 8 18 0.30 4 9 0.04 26 58 G0.001
Peripheral vascular disease 357 84 24 0.03 156 44 G0.001 27 8 G0.001 163 46 G0.001
Psychosis 168 42 25 0.06 26 15 G0.001 9 5 0.06 68 40 0.01
Pulmonary circulation disorder 89 28 31 0.01 24 27 0.71 11 12 G0.001 50 56 G0.001
Renal failure 306 95 31 G0.001 98 32 0.004 30 10 G0.001 156 51 G0.001
Solid tumor without metastasis 90 31 34 0.001 34 38 0.01 10 11 G0.001 41 46 0.01
Peptic ulcer disease 3 1 33 0.47 1 33 1.00 0 0 1.00 0 0 0.56
Valvular Disease 188 44 23 0.13 45 24 0.80 14 7 G0.001 85 45 G0.001
Weight loss 464 159 34 G0.001 187 40 G0.001 38 8 G0.001 297 64 G0.001
Univariate measures of association between demographic factors, surgical timing, site of care, and comorbid conditions and the four outcomes of interest. All p values are results of W2
with Fisher’s exact test, two-sided.

with c statistics (area under the curve) ranging from 0.74 to
0.85 and Hosmer-Lemeshow p 9 0.74.
DISCUSSION
In the absence of Class 1 prospective data, the timing of
surgery for ASBO remains a subject of some debate, despite
recent publication of two comprehensive best practice guide-
lines.15,16 Bizer et al.19 reported in 1981 that 46% of patients
admitted with ASBO for whom surgery was delayed more
than 24 hours were able to avoid surgery, beginning an era of
more conservative management of this syndrome. Since then
varying periods of observation of stable patients with ASBO
have been proposed, varying from as short as 2 days13 to as
long as 10.20 Nauta,21 in a retrospective analysis of more than
400 admissions found that 86% of patients admitted with
ASBO resolve without surgery. A study of population-based
administrative data from California found 76% of more than
32,000 patients admitted with ASBO resolve without surgery.9
The time to resolution in nonoperative cases is less well re-
ported but ranges from an average of 22 hours22 to 6 days.23
In more than 27,000 ASBO patients recorded in the 2009
NIS, we found that 82% recovered without surgical interven-
tion. However, while the majority of patients destined to have
theirs resolve, did so within 5 days of admission, the distri-
bution of LOS for these patients has a clinically significant
right-sided tail. While the LOS for patients who are discharged
nonoperated upon averaged 4 days (median 3 days), there
remain a number of patients for whom longer stays ended in
successful nonoperative management nonetheless. Unfortu-
nately the NIS data set does not contain clinical information
such as white blood cell counts, imaging results, or physical
examination findings. This makes parsing findings which were
associated with successful nonoperative management and
those that predicted surgery impossible. With the data avail-
able, we are limited to determining the impact of delay on
outcome of patients having surgery, regardless of indication.

* 2013 Lippincott Williams & Wilkins 185

Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
 J Trauma Acute Care Surg
Schraufnagel et al. Volume 74, Number 1

longer postoperative LOS. Patients operated on the fourth day

TABLE 3. Multivariate Analysis of Factors Associated With
Postoperative LOS Greater Than 7 Days
or later had a 26% greater risk of staying more than 7 days
postoperatively ( p = 0.005). Those for whom surgery occurred
Variable OR 95% CI p on the fifth day or later had a 36% greater chance of remaining
4 or more preoperative days 1.26 1.07Y1.48 0.005 hospitalized more than 7 days postoperatively ( p = 0.002). The
Weekend admission 1.09 0.94Y1.27 0.27 data available in the NIS do not permit any greater precision
Age Q 60 y 1.62 1.35Y1.96 G0.0001 on the question as to why these patients stay longer. We surmise
Black 1.32 1.06Y1.63 0.01 that in many case, this outcome is caused by a delay in return
Income G 38,000 1.25 1.06Y1.47 0.008 of bowel function in these patients whose small intestine have
Income 963,000 0.90 0.76Y1.06 0.21 been distended for a longer period, although other, unmeasured
Central metro area Q 1 million people 1.09 0.93Y1.27 0.29 factors could also be at work.
Medicare 1.04 0.81Y1.34 0.77 Finally, our analysis also supports the notion that delay
Private insurance 0.69 0.54Y0.87 0.002 in surgery may increase the odds of in-hospital mortality. Delay
Other payment source 0.86 0.53Y1.42 0.56 of four or more days increased the chance of death by 64%
Small hospital 0.86 0.68Y1.09 0.21 ( p = 0.01), although the model with a longer delay of five or
Large hospital 1.11 0.95Y1.29 0.19 more days failed to meet statistical significance. Death was a
Emergency admission 1.09 0.95Y1.26 0.23 relatively uncommon outcome, with only 138 (3%) of 4,826
Alcohol abuse 1.34 0.83Y2.16 0.23 patients in this sample dying during admission. It may be that
Iron deficiency anemia 1.53 1.28Y1.83 G0.0001 potentially confounding clinical factors, not captured in our
Chronic blood loss anemia 1.24 0.62Y2.48 0.54 multivariate model, led to delay in patients who may have been
Congestive heart failure 2.47 1.93Y3.15 G0.0001 more ill than our models accounted for.
Chronic lung disease 1.15 0.97Y1.37 0.11 Several other findings in this analysis deserve mention.
Coagulopathy 1.24 0.87Y1.78 0.24 The significantly higher operative rate for women in the sample
Uncomplicated diabetes mellitus 1.11 0.92Y1.34 0.29
Diabetes with chronic complications 1.12 0.69Y1.79 0.65
Hypertension with complications 1.07 0.92Y1.24 0.37 TABLE 4. Multivariate Analysis of Factors Associated
With Death
Electrolyte disorder 1.59 1.38Y1.83 G0.0001
Metastatic cancer 2.12 1.33Y3.38 0.002 Variable OR 95% CI p
Neurological disorder 1.52 1.16Y1.99 0.002 More than 3 preoperative days 1.64 1.11Y2.40 0.01
Paralysis 1.69 0.88Y3.26 0.12 Weekend admission 1.46 0.99Y2.14 0.05
Peripheral vascular disease 1.19 0.94Y1.52 0.15 Age Q 60 y 3.08 1.47Y6.45 0.003
Psychosis 1.40 0.99Y1.98 0.06 Female 0.78 0.53Y1.13 0.19
Pulmonary circulation disorder 1.55 0.97Y2.47 0.06 Income39,000Y48,000 1.18 0.80Y1.75 0.41
Renal failure 1.06 0.81Y1.38 0.68 Micropolitan counties 1.16 0.59Y2.30 0.67
Solid tumor without metastasis 1.48 0.93Y2.35 0.10 Medicare 1.78 0.66Y4.81 0.25
Valvular disease 1.00 0.72Y1.40 0.99 Private insurance 0.80 0.27Y2.35 0.69
Weight loss 2.84 2.27Y3.52 G0.0001 Self-pay 0.65 0.07Y5.77 0.70
Odds of postoperative hospital stay exceeding 7 days. Table c statistic (area under the Medium-size hospital 1.67 1.13Y2.45 0.01
curve) is 0.74. Hosmer-Lemeshow statistic W2 = 8.75, p = 0.36. Rural hospital 1.85 0.96Y3.53 0.06
Northeast 0.70 0.42Y1.16 0.16
Iron deficiency anemia 1.29 0.86Y1.94 0.22
In univariate analyses, our study shows that all four
Congestive heart failure 2.54 1.66Y3.88 G0.0001
adverse outcomes, namely, complication, resection, prolonged
Chronic lung disease 1.54 1.04Y2.29 0.03
postoperative LOS, and death, were more likely in patients for
Coagulopathy 3.37 1.92Y5.93 G0.0001
whom surgery was delayed by 4 days or more. When potential
Diabetes with chronic complications 0.74 0.26Y2.09 0.57
confounding factors are taken into account, our analysis shows
Liver disease 2.68 1.01Y7.12 0.05
that there is no increase in the rate of complication owing to
Electrolyte disorder 1.54 1.04Y2.29 0.03
delay of 4 days or more. Bickell et al.,13 in a study of 141
Obesity 0.37 0.11Y1.20 0.10
patients found an increasing risk of small bowel resection in
Paralysis 2.20 0.71Y6.75 0.17
ASBO patients who had delays in surgery of more than 96 hours.
Peripheral vascular disease 1.52 0.93Y2.48 0.09
Our univariate analysis supports this notion although the
Psychosis 1.62 0.74Y3.58 0.23
difference was slight (28% vs. 25%, p = 0.001). This finding
Pulmonary circulation disorder 2.40 1.15Y5.03 0.02
did not withstand multivariate analysis, however. When op-
Renal failure 1.72 1.05Y2.81 0.03
eration occurred five or more days after admission, our model
Solid tumor without metastasis 3.13 1.44Y6.78 0.004
suggested a 22% higher likelihood of resection in the NIS
Valvular disease 1.21 0.64Y2.29 0.57
patient groups ( p = 0.04); however, the best model we could
Weight loss 1.77 1.15Y2.73 0.01
create had at best mediocre calibration.
The present analysis does present substantial support for Odds of death. Table c statistic (area under the curve) is 0.85. Hosmer-Lemeshow
statistic W2 = 4.37, p = 0.82.
the notion that delay in operation of four or more days leads to

186 * 2013 Lippincott Williams & Wilkins

Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
J Trauma Acute Care Surg
Volume 74, Number 1
(60%, p G 0.001) is likely caused by the higher frequency of
pelvic surgery in women. Pelvic adhesions leading to ASBO
are believed to be more likely to require entrolysis than
extrapelvic adhesions.24
Patients identified as black had higher rates of surgery
(22% vs. 17% for non-black patients, p G 0.001) and higher
rates of small bowel resection (29% vs. 24% for non-black
patients, p = 0.05). In our multivariate model for resection
(data not shown), this disparity persisted, black patients having
an odds ratio of 1.27 for resection versus all other racial cat-
egories ( p = 0.02). Black patients also had a greater risk of
prolonged postoperative LOS (OR, 1.32; 95% CI, 1.06Y1.63,
p = 0.01). This finding deserves further inquiry.
Patients in the lowest income group, those making less
than $38,000 per year, also had significantly longer postop-
erative stay. One might speculate that these patients have access
to fewer resources to permit earlier hospital discharge. It might
be that with increasing access to home services resulting from
the Affordable Care Act, this disparity may resolve. If so, this
would represent a potentially testable hypothesis for future
researchers.
The NIS is a large administrative database, and as such,
several limitation attach to its use and to the interpretation of
the findings of this study. We limited our analysis to two
principle diagnosis ICD-9 codes to limit our analysis to patients
likely to have been experiencing from ASBO. There may be
other patients in the data set with ASBO in the setting of other
processes that were chosen as principle diagnoses by hospital
coders for nonclinical reasons. We believe the effect of this sort
of classification bias is minimal.
Our choice of diagnoses to define our aggregate com-
plication variable may have neglected relevant diagnoses in
ASBO, confounding our conclusion that delay does not in-
crease risk of complication. Our exclusion of patients who had
procedures other than entrolysis or small bowel resection may
also have introduced a bias in the analysis. These patients were
excluded to eliminate patients not experiencing ASBO from
the analysis; however, doing so may introduce a bias of ret-
rospective insight into a study of prospective surgical decision
making. However, the concordance of our surgical rate with
other large studies9,21,25 suggests that our inclusion/exclusion
criteria are valid.
Most patients with ASBO resolve without surgery. Most
who do so, resolve within 96 hours of admission. This study
supports the recommendations made by two independent
best practice guidelines, that surgery in ASBO should not be
delayed by more than 3 days to 5 days.15,16 This is the first
study to suggest that delay in surgery is an important factor in
postoperative LOS, an observation that may inform the care
of ASBO patients by surgeons wishing to develop more effi-
cient practice guidelines in ASBO. However, that a substantial
minority of patients, for whom surgery is delayed more than
5 days, are discharged home without surgery after much longer
hospital stays should provoke future research into better pre-
dictive models for the need for surgery. There are no physical
examination, laboratory, or radiologic findings for ASBO
patients listed in the NIS. Given its ubiquity, a prospective,
multicenter study of ASBO examining both indications and
timing of surgery should be considered. Nonetheless, it is clear
* 2013 Lippincott Williams & Wilkins
Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Schraufnagel et al.
that the sun may rise and set on bowel obstruction without harm
coming to patients. It seems, from this and other work, how-
ever, that to reduce risk of mortality or prolonged postoperative
stay, 4 to 5 sunsets may be the most patients with persistent
ASBO should see before seeing the lights of the operating
room. However, because of the limitations imposed by the
absence of clinical data from this analysis, surgeons caring for
any individual patient should be informed by these findings
to make a bedside judgment on the decision to operate on a
patient who remains obstructed but not clinically toxic on
Day 4 or 5. More certain recommendations await a prospective,
multicenter, trial including clinical data. We hope this work
will stimulate such a trial.
AUTHORSHIP
F.H.M. performed the literature search. F.H.M., S.R., and D.S contrib-
uted the study design. S.R. and D.S. performed the data collection. S.R.,
D.S., and F.H.M. performed the data analysis. F.H.M. performed the
data interpretation. F.H.M. and D.S. wrote the article. F.H.M. provided
critical revision.
DISCLOSURE
The authors declare no conflicts of interest.
REFERENCES
1. Scudder CL. The principles underlying the treatment of acute intestinal
obstruction: a study of one hundred and twenty-one cases of acute intes-
tinal obstruction from the massachusetts general hospital clinic. Boston
Med Surg J. 1908;159:491Y497.
2. Battle W. Intestinal obstruction coming on four years after the operation
of ovariotomy. Lancet. 1883;121:818Y819.
3. Treves F. Intestinal Obstruction, Its Varieties With Their Pathology,
Diagnosis and Treatement. Philadelphia, PA: Henry C. Lea’s Son & Co;
1884:558.
4. Geiger TM, Roberts PL, Read TE, Marcello PW, Schoetz DJ, Ricciardi R.
Has the use of anti-adhesion barriers affected the national rate of bowel
obstruction? Am Surg. 2011;77:773Y777.
5. Hartley J. On the early recognition and treatment of acute intestinal ob-
struction. BJM. 1909;20:1463Y1466.
6. Richardson EP. Acute intestinal obstruction: a study of a second series
of cases from the Massachusetts General Hospital. Boston Med Surg J.
1920;183:288Y296.
7. Miller CJ. A study fo three hundred forty-three surgical cases of intestinal
obstruction. Ann Surg. 1929;89:91Y107.
8. McKittrick LS, Sarris SP. Acute mechanical obstruction of the small bowel.
N Engl J Med. 1940;222:611Y622.
9. Foster NM, McGory ML, Zingmond DS, Ko CY. Small bowel obstruction:
a population-based appraisal. J Am Coll Surg. 2006;203:170Y176.
10. Rocha F, Theman T, Matros E, Ledbetter SM, Zinner MJ, Ferzoco SJ.
Nonoperative management of patients with a diagnosis of high-grade small
bowel obstruction by computed tomography. Arch Surg. 2009;144:
1000Y1004.
11. Fevang BT, Jensen D, Svanes K, Viste A. Early operation or conservative
management of patients with small bowel obstruction? Eur J Surg. 2002;
168:475Y481.
12. Gowen GF. Long tube decompression is successful in 90% of patients with
adhesive small bowel obstruction. Am J Surg. 2003;185:512Y515.
13. Bickell N, Federman A, Aufses A. Influence of time on risk of bowel resection
in complete small bowel obstruction. J Am Coll Surg. 2005;201:847Y854.
14. Hayanqa A, Bass-Wilkins K, Bulkley GB. Current management of small
bowel obstruction. Adv Surg. 2005;39:1Y33.
187

Schraufnagel et al.
15. Diaz JJ, Bokhari F, Mowery N, Acosta JA, Block EF, Bromberg WJ, et al.
Guidelines for management of small bowel obstruction. J Trauma. 2008;
64:1651Y1664.
16. Catena F, Di Saverio S, Kelly M, Biffl W, Ansaloni L, Mandala V, et al.
Bologna Guidelines for Diagnosis and Management of Adhesive Small
Bowel Obstruction (ASBO): 2010 Evidence-Based Guidelines of the
World Society of Emergency Surgery. World J Emerg Surg. 2011;6:5.
17. Overview of the National Inpatient Sample (NIS). Available at: http://
www.hcup-us.ahrq.gov/nisoverview.jsp#Whatis. Accessed July 31, 2012.
18. Mancini GJ, Petroski GF, Lin W-C, Sporn E, Miedema BW, Thaler K.
Nationwide impact of laparoscopic lysis of adhesions in the management
of intestinal obstruction in the US. J Am Coll Surg. 2008;207:520Y526.
19. Bizer L, Liebling R, Delaney H, Gliedman M. Small bowel obstruction:
the role of nonoperative treatment in simple intestinal obstruction and pre-
dictive criteria for strangulation obstruction. Surgery. 1981;89:407Y413.
20. Shih S, Jeng K, Lin C, Chou S, Wang H, Chang W, et al. Adhesive small
bowel obstruction: how long can patients tolerate conservative treatment?
World J Gastroenterol. 2003;9:603Y605.
21. Nauta RJ. Advanced abdominal imaging is not required to exclude
strangulation if complete small bowel obstructions undergo prompt lap-
arotomy. J Am Coll Surg. 2005;200:904Y911.
22. Seror D, Feigin E, Szold A, Allweis TM, Carmon M, Nissan S, et al.
How conservatively can postoperative small bowel obstruction be treated?
Am J Surg. 1993;165:121Y126.
23. Tanaka S, Yamamoto T, Kubota D, Matsuyama M, Uenishi T, Kubo S, et al.
Predictive factors for surgical indication in adhesive small bowel ob-
struction. Am J Surg. 2008;196:23Y27.
24. Ellis H. The clinical significance of adhesions: focus on intestinal ob-
struction. Eur J Surg. 1997;577:5Y9.
25. Ambiru S, Furuyama N, Kimura F, Shimizu H, Yoshidome H, Miyazaki M,
et al. Effect of hyperbaric oxygen therapy on patients with adhesive intes-
tinal obstruction associated with abdominal surgery who have failed to respond
to more than 7 days of conservative treatment. Hepatogastroenterology.
2008;55:491Y495.
DISCUSSION
Dr. Rao R. Ivatury (Richmond, Virginia): Adhesive
small bowel obstruction continues to be a common clinical
problem in acute care surgery. One of the frequent adages we
are constantly reminded of is that ‘‘never let the sun set or
rise on small bowel obstruction.’’ Now, recent comprehensive,
evidence-based practice guidelines have suggested that non-
operative therapy is appropriate in many cases but should be
limited to periods between 3 days and 5 days. The authors
propose to probe this relationship between operative delay and
adverse outcomes in acute small bowel obstruction using the
Nationwide Inpatient Sample (NIS).
The NIS is a unique and power database of hospital in-
patient stays. It contains data from approximately 8 million
people and the hospital stays each year. So many researchers
and policy makers have asked clinical questions about this da-
tabase. The authors, as we heard, have defined adverse outcomes
as complications, requirement for bowel resection, prolonged
postoperative stays, or death.
They looked at more than 27,000 patients staying with
nonoperative management for approximately 4 days. Of the
surgical group, 25% of the patients had resection, 32% of
patients had postoperative days more than 7, and 3% died, and
interestingly, a most patients were female.
When a count was taken of all the available potentially
confounding variables, the relationship between preoperative
days and complications of resection disappeared, and the
188
Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
J Trauma Acute Care Surg
Volume 74, Number 1
models for postoperative length of stay suggested a more ro-
bust relationship between preoperative delays and death and
longer postoperative stays. So these data lend support to your
policy encouraging observation for not more than 3 days to
5 days.
Now, the authors do caution about the administrative
nature of the database. I will respectfully submit that the last
slide they showed on the limitations of the study is probably
the most important slide. I congratulate the authors on the labor
of love, poring over this large database, but I must regretfully
conclude that I am not convinced about its validity. So my
questions for you are as follows.
Since you stated that the NIS database is missing in many
important elements of physical examination and laboratory
findings, how valuable is this 4-day to 5-day cutoff? What
do you suggest we should do when a patient has not im-
proved after 4 days to 5 days? See the light of the operating
room, as you suggest? Or should we proceed with more
investigations?
Finally, since the increased mortality was one of the poor
outcomes you noted in the study but your group did not have
increased rate of small bowel resection, strangulation was not
the cause of this increased mortality, so what was the cause of
the mortality? Were there comorbid conditions that the data-
base never accounted for?
I agree with you that these answers cannot come from a
large database, but I think you just laid the foundation for a very
good American Association for the Surgery of Trauma mul-
ticenter prospective study.
I would like to congratulate the authors and thank the
American Association for the Surgery of Trauma for the privilege
of the podium.
Dr. Demetrios Demetriades (Richmond, Virginia): This
is an excellent and practical study. The message is that we need
to identify those patients who do not respond to nonoperative
management and operate on them early. We have Class 1 evidence,
excellent evidence, that oral Gastrografin can reliably identify
these patients. Would you consider including in your manage-
ment protocol oral Gastrografin within 48 hours of admission?
Thank you.
Dr. Ajai Malhotra (Richmond, Virginia): I enjoyed the
presentation. One question, everything has a price to pay, as
Dr. Maier pointed out yesterday, so if we have followed your
protocol or your suggestion of everybody gets operated, how
many more would have been operated on that otherwise did
successful nonoperative management?
Dr. Samir M. Fakhry (Charleston, South Carolina):
Fred, that was a very nice presentation. I had a question and a
comment.
My question is: I can t remember if you can tell from the
NIS how many previous surgeries these people have had or
what kind of surgeries these were.
My comment is I want to agree with Dr. Ivatury. In your
conclusion in the abstract, you say that there is an association
between delay in management and death and postoperative
days greater than 7, but then then you say optimal timing for
operation and nonresolving seems to be less than 5 days. I have
to agree that your study is excellent for hypothesis develop-
ment, but it is not appropriate for hypothesis testing.
* 2013 Lippincott Williams & Wilkins
J Trauma Acute Care Surg
Volume 74, Number 1
I think we have gone to a trend where we are using ad-
ministrative data sets to make conclusions about causation
that are not warranted. Being guilty of the same thing myself
on a couple of occasions, I would love to see what you think of
that. Thank you.
Dr. Martin Zielinski (Rochester, Minnesota): I would
like to congratulate the authors for attempting to limit the
hospital duration of stay for patients with adhesive small bowel
obstruction.
I would like to follow up on Dr. Demetriades’ comment.
There are several predictive models in the literature, which
answer the question posed by the authors. Our group has pub-
lished one such predictive model, which combines an easy-to-use
model with the Gastrografin challenge. Using this algorithm,
we can predict the need for operative exploration within
12 hours of emergency department consultation 89% of the
time. I would encourage the authors, if they continue in their
current work, to ensure that the Gastrografin challenge is
a focus. I would also caution the authors in using such a broad
base database. In developing our model, we used very specific
patient data points, namely, the duration of obstipation as well
as computed tomographic findings of the small bowel feces
sign and mesenteric edema. I do not believe that this level of
detail will be present in a large database.
Dr. Rachael Callcut (San Francisco, California): I have
a quick methodologic question. Frequently, when the NIS is
used, we actually control for comorbidities using one of the
comorbidity indices. I am curious if there was a specific reason
why you guys chose to look at the individual comorbidities
in your model rather than summing it into one of the more
commonly validated and used comorbidity indices.
Dr. Charles E. Lucas (Detroit, Michigan): Important
not to compare apples and oranges. How many people had
Stage IV ovarian cancer as opposed to those who had only a
simple appendectomy in the past?
Dr. Frederick H. Millham (Newton, Massachusetts):
Thank you, Dr. Ivatury and questioners. Dr. Ivatury, with regard
to how valuable a four- or five-day cutoff is; I agree that these
data are not strong enough to demand that patients be oper-
ated on by Day 5.
These are cases that give surgeons gray hair. That is a
surgical judgment that we all make at the bedside. Our intention
with this work was to try to inform the experienced surgeon
as to what the risk might be of additional delay as I think we
have all seen patients who have had 7 days or longer preop-
erative stay then have an even longer postoperative stay. We
have shown that the longer you wait to operate, the longer the
postoperative stay will be.
As for additional investigations, once computed to-
mography contrast is on board, most patients require only plain
* 2013 Lippincott Williams & Wilkins
Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Schraufnagel et al.
abdominal radiographs to follow their progress. On the fifth
day, if a KUB suggests continued obstruction and the patient is
unchanged, the thoughtful surgeon should use the data pre-
sented here to their clinical impression.
It is possible that there are a large number of bowel
resections performed on patients presenting with strangulation,
operated on the day of admission. We did not try to account for
that. Although we accounted for a number of comorbidities, it
may still be that poorer-risk patients experienced longer delay,
contributing to the risk of death. This should have been
accounted for with the regression model but may not have been
totally so.
With regard to Dr. Demetriades’ question, there are no data
in this data set about who got Gastrografin and who did not. I
agree, Gastrografin is a useful therapy for small bowel obstruction.
There are a lot of data both in the American and Japanese
literature to support the use of Gastrografin and sometimes
the therapeutic application of computed tomographic scan-
ning on Day 1 on these patients. We did not have these data to
look at.
Dr. Malhotra, with regard to how many more patients would
be operated on, we did not perform that calculation. The two
frequency histograms would suggest that early operation trades
off a number of patients who are likely to go home and late op-
eration may trade off a number of patients who are going to die.
I do not believe that this work is powerful enough to
definitively answer the question of the optimal day to operate
ondeveloping paper than a hypothesis-proving paper. Perhaps,
we should tone down our conclusion.
You are correct, the NIS does not tell us about previous
hospitalizations. This would be great information to have.
Dr. Zielinski, thank you for your suggestions. Use of
Gastrografin should be an important component of any pro-
spective work on adhesive small bowel obstruction.
Dr. Callcut, why not use a comorbidity index such as
the Charlson index? One could certainly do that. We chose to
adjust individually using the numerous comorbidity indicators
that are available in the NIS. Perhaps as a subsidiary work, we
could compare comorbidity indices to the more aggressive
modeling that we undertook.
Dr. Lucas, we screened the data for indicators of ma-
lignancy among the diagnosis fields available in the data set.
We eliminated cases with DRG International Classification
of DiseasesV9th Rev. codes for GI or GYN malignancy as
secondary admitting diagnoses. There may have been some
nonadhesive cases that slipped through, but to the extent the
data set permitted, we limited our analysis to adhesive small
bowel obstruction.
I thank the Association for the privilege of presenting this
work in this beautiful setting.
189