2

0h

The American Journal of Surgery (2012) 204, 434–440

Clinical Science

Influence of obesity on complications and costs afterintestinal surgeryHeather Wakefield, M.D.a, Mary Vaughan-Sarrazin, Ph.D.b, Joseph J. Cullen, M.D.a,b,*aSurgical Service, Iowa City Veterans Affairs Medical Center, 1529 JCP, 200 Hawkins Drive, University of Iowa Hospitals andClinics, Iowa City, IA 52242, USA; bThe Center for Research in the Implementation of Innovative Strategies in Practice, Iowa

City Veterans Affairs Medical Center, University of Iowa Carver College of Medicine, Iowa City, IA, USA

AbstractBACKGROUND: Obesity is a risk factor for many comorbid conditions that increase the cost of

health care. We sought to examine the effect of obesity on surgical complications and cost in a groupof patients undergoing intestinal surgery.

METHODS: Using the Veterans Affairs Surgical Quality Improvement Program (VASQIP), whichincludes clinical data abstracted from medical records for Veterans Affairs (VA) surgical patients, andthe VA Decision Support System, which provides the costs of individual patient encounters on the basisof relative values assigned to intermediate products, we examined surgical complications and costs ofcare in 4,881 patients undergoing intestinal surgery in 2006. Patients were classified into 4 groups basedon body mass index (BMI): malnourished (�18), normal weight (18–30), obesity class I to II (30–40),and obesity class III (�40). Patient endpoints included the occurrence of any complication and surgicalcosts incurred within 30 days of surgery. Endpoints were compared across the 4 BMI categories inunadjusted analyses and risk-adjusted analyses and hospital-level variation using multivariable models.

RESULTS: After controlling for patient risk factors and hospital-level variation, patients in obesity classI to II were 1.21 times more likely to have any complication and patients in obesity class III were 1.41 timesmore likely to have any complication when compared with normal-weight patients. Similarly, patients inobesity class I to II were 1.44 times more likely to develop a wound complication compared withnormal-weight patients, and patients in class III were 1.84 times more likely to develop a wound compli-cation and 1.55 times more likely to develop a respiratory complication compared with normal-weightpatients. In contrast, costs were greatest for malnourished patients at $45,000 compared with normal-weightpatients at $37,000. However, after controlling for patient risk factors and variation in costs attributable tothe admitting hospital, there were no significant cost differences between the 4 BMI categories.

CONCLUSIONS: Obesity leads to increased wound and respiratory complications in intestinal surgery.Nevertheless, obesity alone is not an independent risk factor for increased costs in intestinal surgery.Published by Elsevier Inc.

KEYWORDS:Obesity;Intestinal surgery;Costs

Presented at the 35th annual meeting of the Association of VA Sur-geons, Irvine, CA, April 10–12, 2011.

Supported by a VA HSR&D merit review grant (IIR-07-151-1).The opinions expressed are those of the authors and not necessarily

those of the Department of Veterans Affairs or the US Government.* Corresponding author. Tel.: �1-319-353-8297; fax: �1-319-356-8378.E-mail address: joseph-cullen@uiowa.eduManuscript received August 10, 2011; revised manuscript January 17,

012

002-9610/$ - see front matter Published by Elsevier Inc.ttp://dx.doi.org/10.1016/j.amjsurg.2012.01.013

Obesity is a risk factor for many comorbid conditionsthat increase the cost of health care. These include but arenot limited to cardiovascular disease, diabetes, metabolicsyndrome, osteoarthritis, obstructive sleep apnea, and poly-cystic ovary syndrome. The United States is facing an obe-sity epidemic.1 In a recent epidemiologic review, it wasfound that the prevalence of obesity (defined as a body massindex [BMI] greater than 30) in adults increased from 13%

in the 1960s to 32% in 2004. Currently, 66% of adults are

cb4

awcdsvttvaBtmiar

icttfP

ioahch

pcmrBCaswmefitn

c

435H. Wakefield et al. Obesity and complications after intestinal surgery

either overweight (BMI �25) or obese. Sixteen percent ofhildren and adolescents are overweight. It is estimated thaty 2015, 75% of adults will be overweight or obese, and1% will be obese.

The obesity epidemic suggests that a significant percent-ge of patients requiring operations are going to be over-eight or obese. Obesity may lead to increased respiratory

omplications caused by obesity-related pulmonary disor-ers including obstructive sleep apnea and hypoventilationyndrome. Obesity is associated specifically with decreasedital capacity, functional residual capacity, forced expira-ory volume, and arterial oxygen tension. Obesity may leado cardiac complications caused by changes in the atrial andentricular structure and ventricular diastolic function andlso has been identified as a risk factor for atrial fibrillation.ecause adipose tissue is relatively avascular, leading to

issue hypoperfusion and decreased oxygen tension, obesityay lead to more wound complications including wound

nfections and impaired wound healing. Insulin resistancend impaired glucose metabolism leading to impaired mac-ophage function may lead to an increased risk of infection.

With the physiological changes associated with obesityn mind, we hypothesized that obesity would lead to in-reased surgical complications and the costs associated withhese complications. We analyzed a specific group of pa-ients undergoing intestinal surgery using data gatheredrom the Veterans Affairs Surgical Quality Improvementrogram (VASQIP) in the 2006 fiscal year.

Methods

To determine the effects of obesity on complicationsafter intestinal surgery, we used the VASQIP, which in-cluded 118 acute care Veterans Affairs (VA) hospitals thatperformed inpatient surgery during the 2006 fiscal year andare represented in the VASQIP.2,3 VASQIP collects a widerange of preoperative risks factors, information about theoperative procedures and anesthesia, and postoperativecomplications and mortality. Data are abstracted from eachVA medical center for patients undergoing surgery undergeneral, spinal, or epidural anesthesia using standard defi-nitions to ensure data reliability. For hospitals with largesurgical volumes, operations are sampled for the VASQIPon an 8-day cycle by selecting only the first 36 cases withinan 8-day period for inclusion in the VASQIP. If there aremore than 36 eligible cases in an 8-day period, the 37th caseis excluded although some sites require their nurse to in-clude all eligible cases.

We then combined VASQIP with the VA Decision Sup-port System (DSS) cost data, which provide details on costsat the level of individual encounters. DSS is an “activity-based” costing algorithm in which the actual quantity ofhealth care products used for specific health care encounters(eg, intensive care unit days of care, radiology tests, and

blood products) is extracted from electronic databases (eg,

blood units, chest x-rays, and consultation), and the costsassociated with the intermediate products are based on rel-ative resource units (RRUs) that represent the relative costsof the resources required to produce a given intermediateproduct. RRUs are generated using a standard set of weightsderived nationally, which are then modified within eachfacility to reflect local factors. The use of RRUs is anadvantage of the VA DSS system. Unlike hospital charges,RRUs explicitly represent the relative cost of producingdifferent patient care products. Thus, 2 patients undergoingthe exact same operation may have different estimates ofcosts because the costs of all intermediate products used bya given patient are assigned directly to the patient. Overheadcosts associated with non–patient care departments (ie, se-curity and administration) are distributed to patient caredepartments and then to intermediate products using a step-down method based on the square feet of space of thepatient care department. The total cost for a specific patientencounter is calculated by aggregating costs associated withintermediate products. Activity-based cost estimates havebeen used in a number of studies of the cost of individualhospitals and in some multicenter studies.4 A potential lim-tation of activity-based cost estimates is that the assignmentf relative values for specific products is often idiosyncraticnd may affect the comparability of costs from differentospitals. We have included random effects for hospitals toontrol for possible idiosyncratic assignment of cost acrossospitals.

There were 5,386 total intestinal surgeries defined asrincipal current procedural terminology (CPT) procedureode 44140–44147 or 44,204 to 44,208 (partial colon re-oval); 44150–44160 or 44,210 to 44,213 (total colon

emoval); or 44,005 to 44,799 (other intestinal procedures).ariatric operations were not included and were removed byPT code. Operations that were preceded within 30 days byprevious operation were excluded, leaving 4,991 inpatient

urgeries that were not preceded by a previous surgeryithin 30 days. Of the 4,991, 69 were excluded because ofissing values on some demographic variables, and 41 were

xcluded because they were cost outliers, which were de-ned as greater than 3 standard deviations below or above

he mean log cost. The final sample included 4,881 intesti-al operations.

Patient endpoints included the occurrence of any of 18omplications captured in the VASQIP5 as well as surgical

costs incurred within 30 days of surgery. Complicationswere divided into 6 broad categories: cardiac, respiratory,central nervous system, urinary tract, wound complications,and other complications. Cardiac complications includedcardiac arrest and acute myocardial infarction. Respiratorycomplications included failure to wean from ventilationwithin 48 hours, postoperative pneumonia, pulmonary em-bolism, and reintubation for respiratory or cardiac failure.Central nervous system complications included cerebralvascular accident, coma lasting more than 24 hours, and

peripheral nerve injury. Urinary complications included

w

toffglmtcbfrrtaftcacats

faablahmdI

h.Oh

436 The American Journal of Surgery, Vol 204, No 4, October 2012

acute renal failure, renal insufficiency, and urinary tractinfection. Wound complications included deep wound sur-gical site infection, superficial surgical site infection, andwound dehiscence or disruption. Other complications in-cluded bleeding requiring more than 4 U, systemic sepsis,and deep vein thrombosis or thrombophlebitis.

Obesity was defined according to patients’ BMI. Patientswere classified based on BMI according to the followinginternational classification from the World Health Organi-zation: malnourished (less than 18), normal weight (BMI18–30), obesity class I (BMI 30–35), obesity class II (BMI35–40), and obesity class III (BMI greater than 40).6 Forstatistical analysis, class I and II were combined, resultingin 4 BMI categories.

Additional sociodemographic patient characteristics ob-tained from the VASQIP data included age, sex, race, an-nual income, and means test category, which indicates thereason for VA eligibility and was categorized as serviceconnected disability, low income, and other reason. Patientcomorbidity included indicators for specific preoperativeconditions (eg, congestive heart failure, weight loss, orbleeding disorder). Other risk factors included preoperativewound classification (ie, clean, clean/contaminated, con-taminated, or infected), smoking status, functional status (ie,independent, partially dependent, or totally dependent),American Society of Anesthesiology Physical Status class(ie, healthy, mild disease, severe, or life-threatening/mori-bund), and an indicator for emergency surgery. Finally,several preoperative laboratory values are available fromthe VASQIP. Multiple imputations were used to imputelaboratory values when missing.7 All analyses were run

ith and without the imputed laboratory values.Statistical analyses compared unadjusted endpoints across

he 4 BMI categories using a chi-square test for the occurrencef complications and using analysis of variance for log-trans-ormed costs. Subsequently, multivariable models were used tourther control for preoperative patient risk factors (eg, demo-raphics, comorbidities, functional status, and preoperativeaboratory values) and hospital fixed effects. Multivariableodels were developed after identifying patient characteris-

ics that were significant predictors of complications and/orosts using 4 primary steps. First, bivariate relationshipsetween complications or costs and individual patient riskactors were examined. For complication models, bivariateelationships were examined using the chi-square test forisk factors measured as dichotomous variables, and the test was used for risk factors measured as continuous vari-bles. Similarly, bivariate relationships between log-trans-ormed costs and patient risk factors were examined using tests for dichotomous risk factors and correlation coeffi-ients for continuous risk factors. Second, correlationsmong candidate variables were evaluated, and redundantollinear variables were eliminated. Third, candidate vari-bles that were significantly related (P � .01) to complica-ions or costs were entered into regression models with

tepwise elimination, and selected variables were inspected w

or consistency with clinical expectation and previous liter-ture based on VASQIP data. Final models were estimateds generalized linear regression models with a logit link andinomial distributed errors for complications and with a logink and gamma distributed errors for surgical costs. Modelslso included random effects for hospitals to account forospital-level variation in complication rates and cost esti-ation. All multivariable regression models were con-

ucted using SAS version 9.2 PROC GLIMMIX (SASnstitute, Cary, NC).

Results

Table 1 shows characteristics of the 4,881 patients eval-uated in this study. Overall, 69% were categorized as havinga normal BMI (BMI 18–30), whereas malnourished patients(BMI �18) represented 2%, obesity class I to II (BMI30–40) 25%, and obesity class III (BMI �40) 3% (Table1). Unadjusted analysis (Table 2) suggested a higher com-plication rate in the higher BMI categories although thisresult was not statistically significant (P � .06). Of the 4categories, patients in obesity class III were most likely tohave any complication, 38%, compared with 29%, 29%, and31% for normal, malnourished, and class I to II, respec-tively. Obesity class III patients are 1.5 times more likely todevelop any complication compared with patients with nor-mal BMIs (P � .015).

When comparing the individual complication categoriesin unadjusted analyses, obese patients were more likely tohave wound complications and respiratory complications(Table 2); 12.4% of normal weight patients developedwound infection compared with 6.3% for malnourished,17.4% for class I to II, and 21.2% for class III patients (P �.001). Normal-weight patients developed respiratory com-plications 12.5% of the time compared with 13.7% formalnourished, 8.8% for class I to II, and 17.6% for class IIIpatients (P � .05).

Patient risk factors and associated coefficients in the finalrisk adjustment models for any complication and for costsare shown in Table 3. After controlling for patient riskfactors and hospital-level variation, patients in obesity classI to II were still 1.21 times more likely to have any com-plication (P � .01) and patients in obesity class III were1.41 times more likely to have any complication (P � .04)when compared with normal-weight patients (Table 4). Pa-tients in obesity class I to II were still 1.44 times more likelyto develop a wound complication (P � .0001) comparedwith normal-weight patients (Table 4), whereas patients inclass III were 1.84 times more likely to develop a woundcomplication (P � .003). Malnourished patients appear toave less risk of wound complication with an odds ratio of46; however, this was not statistically significant (P � .08).besity class III patients were still 1.55 times more likely toave a respiratory complication compared with normal-

eight patients with a P value approaching significance

437H. Wakefield et al. Obesity and complications after intestinal surgery

(.06). In contrast, patients in obesity class I to II weresignificantly less likely to develop a respiratory complica-tion (P � .052). Although obesity did lead to greater woundand respiratory complications, it did not lead to greatercomplications in the following categories: cardiac, central

Table 1 Characteristics of 4,881 patients undergoing intestin

% (n)Malnourished(n � 95)

TotalMale 2.0 (94)Female .6 (1)

Age categoryAge 0–49 2.3 (8)Age 50–59 2.3 (30)Age 60–69 2.2 (29)Age 70–79 1.8 (23)Age 80 plus .8 (5)

RaceBlack 2.9 (21)White 1.8 (63)Missing 1.4 (4)Other 2.2 (7)

Patient risk factorsPreoperative sepsis present 7.4 (7)COPD 23.2 (22)Paraplegia 2.1 (2)Quadraplegia 2.1 (2)Hemiplegia 1.0 (1)Bleeding disorder 6.3 (6)Living independently 75.8 (72)Diabetes 7.4 (7)DyspneaNone 72.9 (74)

Dyspnea with minimal exertion 18.9 (18)Dyspnea at rest 3.2 (3)

Ascites 2.1 (2)Preoperative acute renal failure 0 (0)Emergency surgery 26.3 (25)ASA class

Class I–II 5.3 (5)Class III 67.4 (64)Class IV–V 27.4 (26)

Preoperative wound classClean/contaminated 82.1 (78)Contaminated/infected 17.9 (17)

COPD � chronic obstructive pulmonary disease; ASA � American So

Table 2 Percent complications in weight categories (not risk

ComplicationMalnourished (%)(n � 95)

Normal w(n � 3,3

Any complication 29.5 28.8Cardiac 2.1 2.6Central nervous system 1.05 .65Urinary 7.37 6.75Wound 6.32 12.4Respiratory 13.7 12.5Other 12.6 7.94

nervous system, and urinary complication or other compli-cations.

If obese patients have more complications, we hypoth-esized that costs for postsurgical care would be greatestfor obese patients. In contrast, unadjusted mean costs

ery in study population

l weight3,377)

BMI 30–40(n � 1,244)

BMI �40(n � 165) P value

(3273) 25.4 (1196) 3.2 (152) .004(104) 28.9 (48) 7.8 (13)

�.001(230) 28 (98) 4 (14)(839) 27.6 (356) 4.9 (63)(840) 30.4 (405) 4.4 (59)(905) 24.6 (311) 1.9 (24)(563) 11.4 (74) .8 (5)

.347(510) 24.0 (175) 3.0 (22)(2435) 26.0 (921) 3.6 (129)(200) 26.2 (75) 2.4 (7)(232) 22.9 (73) 2.2 (7)

(162) 4.1 (51) 7.3 (12) .176(575) 13.8 (171) 17.6 (29) .013(44) .7 (9) 1.2 (2) .335(14) .6 (7) 1.2 (2) .072(79) 2.2 (28) 1.8 (3) .835(205) 5.7 (71) 5.4 (9) .958(2843) 89.8 (1117) 83.6 (138) �.001(236) 10.8 (134) 21.8 (36) �.001

.029(2821) 82.5 (1026) 73.3 (121)(480) 15.5 (193) 23.0 (38)(76) 2.0 (25) 3.6 (6)(111) 1.6 (20) 0 (0) .002(41) 1.6 (20) 3.0 (5) .118(672) 15.4 (192) 18.8 (31) .002

�.001(640) 19.2 (239) 10.3 (17)(2187) 66.4 (826) 67.3 (111)(550) 14.4 (179) 22.4 (37)

.004(3018) 91.9 (1143) 91.5 (151)(359) 8.1 (101) 8.5 (14)

f Anethesiologists.

ed)

(%) BMI 30–40 (%)(n � 1,244)

BMI �40 (%)(n � 165) P value

31.1 37.6 .0642.1 4.24 .377.4 1.21 .52

7.32 9.09 .6517.4 21.2 �.0018.8 17.6 .0056.9 10.3 .115

al surg

Norma(n �

69.462.6

65.765.163.071.687.0

70.068.669.972.7

4.817.01.3.4

2.36.1

84.27.0

83.514.22.23.31.2

19.9

18.964.816.3

89.410.6

ciety o

adjust

eight77)

sPttacsr

anw3BagBgp

438 The American Journal of Surgery, Vol 204, No 4, October 2012

were greatest for malnourished patients at $45,735 (95%confidence interval [CI], $38,447–$53,023) comparedwith normal-weight patients at $37,689 (95% CI,$36,681–$38,896; P � .02). Obesity class I to II costslightly less than normal-weight patients at $33,965 (95%CI, $32,470 –$35,461; P � .001), and class III costlightly more at $42,143 (95% CI, $37,049 –$47,237;

� .08). Malnourished patients cost 1.25 times morehan normal BMI patients and patients in obesity class Io II cost significantly less (about 9% less). However,fter controlling for patient risk factors and variation inosts attributable to the admitting hospital, there were noignificant cost differences between the 4 BMI catego-ies.

Table 3 Risk adjustment model for the occurrence of any com

Effect

Odds ratio for thecomplication*(95% CI, P value)

Age (per 10 years) 1.10 (1.03–1.17,Diabetes 1.37 (1.10–1.72,COPD 1.41 (1.17–1.69,Ascites 1.84 (1.25–2.70,Renal failure 1.97 (1.09–3.55,ASA class 3 1.52 (1.24–1.86,ASA class 4 or 5 1.89 (1.45–2.47,Contaminated wound 1.59 (1.29–1.97,Dirty wound 1.38 (1.00–1.90,Hemiplegia 1.72 (1.15–2.57,Paraplegic —Quadriplegia —Bleeding disorder —Functional status � independent —Preoperative sepsis —Emergency 1.41 (1.18–1.69,Any secondary CPT code 1.17 (1.02–1.35,Surgical work RVU —Preoperative SGOT —Preoperative albumin .83 (.74–.92, P �Preoperative hematocrit 1.01 (1.00–1.03,Preoperative BUN 1.01 (1.01–1.02,Preoperative white blood count

COPD � chronic obstructive pulmonary disease; ASA � American Sooxaloacetic transaminase; BUN � blood urea nitrogen.

*A generalized linear model for the occurrence of any complicationfactors and the admitting hospital using fixed effects.

†A generalized linear model for the occurrence of any complication ufactors and the admitting hospital using fixed effects.

Table 4 Odds of complication by obesity class relative to pat

Malnutrition

Any complication .90 (.56–1.45, P � .666)Wound complication .46 (.20–1.08, P � .08)Respiratory complication .97 (.51–1.83, P � .925)

*Relative odds of complication based on multivariable model contro

CommentsMultiple studies have shown that obesity leads to an

increased operative time and increased intraoperative bloodloss. Khan et al8 analyzed 586 patients undergoing pancre-toduodenectomy and showed that operating time was sig-ificantly higher as BMI increased. Specifically, patientsith a BMI greater than 35 had an average operating time of80 minutes compared with 342 minutes for patients with aMI less than 25. There was also increased blood loss, withn 800-mL average blood loss for patients with a BMIreater than 35 compared with 500 mL for patients with aMI less than 25.8 In another study, Efron et al9 investi-ated performing restorative proctocolectomies in obeseatients. They defined obesity as a BMI greater than 30,

ion within 30 days of intestinal surgery

ence of any Relative cost ratio for costs within 30days of surgery†

(95% CI, P value)

04) 1.02(1.00–1.05, P � .046)06) —01) 1.12 (1.07–1.18, P � .001)02) —25) —01) 1.21 (1.15–1.27, P � .001)01) 1.42 (1.32–1.53, P � .001)01) 1.11 (1.04–1.18, P � .001)48) —08) —

1.50 (1.26–1.80, P � .001)1.62 (1.24–2.12, P � .001)1.15 (1.06–1.24, P � .001).75 (.71–.80, P � .001)

1.33 (1.21–1.47, P � .001)01) 1.21 (1.14–1.28, P � .001)23) 1.18 (1.14–1.23, P � .001)

1.02 (1.01–1.02, P � .0011.00 (1.00–1.00, P � .001)

)- .87 (.85–.90, P � .001)37) .99 (.98–.99, P � .001)01) 1.00 (1.00–1.01, P � .001)

1.01 (1.00–1.01, P � .001)

Anesthesiologists; RVU � relative value units; SGOT � serum glutamic

git link and binomial distributed errors and controlling for patient risk

logit link and gamma-distributed errors and controlling for patient risk

ith a normal BMI*

esity I–II Obesity III

21 (1.04–0.41, P � .012) 1.41 (1.01–2.03, P � .044)44 (1.20–1.74, P � .001) 1.84 (1.24–2.74, P � .003)79 (.62–1.00, P � .052) 1.55 (.98–2.45, P � .062)

r patient risk factors indicated in Table 3.

plicat

occurr

P � .0P � .0P � .0P � .0P � .0P � .0P � .0P � .0P � .0P � .0

P � .0P � .0

.001P � .0P � .0

ciety of

using lo

sing a

ients w

Ob

1.1..

lling fo

athpwar

sosvso

439H. Wakefield et al. Obesity and complications after intestinal surgery

which resulted in an operative time that was longer in theobese group (229 vs 196 min). In addition, other studieshave looked at the impact of obesity on resource utilizationfor general surgical procedures. Over 400 colectomies andother procedures were analyzed showing that comparedwith normal-weight patients, obese patients had an addi-tional 21 minutes of operating time.10 Thus, obesity mayinfluence morbidity and mortality after certain operations,which would directly increase the costs of taking care ofthese patients.

Our study shows that obesity leads to increased woundand respiratory complications after intestinal surgery as wellas an increased risk of developing any complication. Pa-tients with a BMI of 30 to 40 were 1.21 times more likelyto have any complication compared with normal-weightpatients, and patients with a BMI greater than 40 were 1.41times more likely to have any complication. Specifically,obese patients were more likely to have respiratory andwound complications. Patients with a BMI of 30 to 40 hada nearly 1.5-fold increase in wound complications and pa-tients with a BMI greater than 40 had a nearly 2-foldincrease in wound complication compared with normal-weight patients. Finally, patients with a BMI of greater than40 were 1.5 times more likely to have a respiratory com-plication compared with normal-weight patients. Neverthe-less, after adjusting for patient risk factors and hospital-levelvariation, obesity is not an independent risk factor for in-creased costs in intestinal surgery. It should be noted thatobesity may impact costs and the likelihood of complicationdirectly as well as indirectly through comorbid conditionsthat are related to obesity (eg, diabetes). Thus, our resultsmay be conservative to the extent that our risk-adjustmentmodels control for comorbid conditions that are related toobesity and reflect the direct impact of obesity on costs andcomplications controlling for indirect effects that occurthrough other comorbid conditions.

Obesity is associated with sleep apnea, atelectasis, andhypoventilation syndrome. It is also associated with de-creased vital capacity, functional residual capacity, forcedexpiratory volume in 1 second, and arterial oxygen tension.In our patient population, we found that patients with a BMIof 30 to 40 did not have greater respiratory complications.However, patients with BMI greater than 40 were 1.5 timesas likely to develop a respiratory complication as normal-weight patients. Previous studies have also shown a corre-lation with obesity and respiratory complications. In a ret-rospective analysis by Merkow et al11 with data from 3,202colectomy patients, a BMI greater than 30 was found to beassociated with pulmonary embolus but not with pneumo-nia. In a retrospective analysis by Kuduvalli et al12 with datafrom 4,713 coronary artery bypass surgery patients, a BMIgreater than 35 was associated with prolonged mechanicalventilation. We also found a correlation between obesityand wound complications. Obesity is associated with met-abolic disturbances. Adipose tissue was once thought to be

just a lipid storage organ, but we now know that it is

metabolically active with endocrine, paracrine, and immu-nologic properties. Obesity also leads to the metabolic syn-drome, which is a proinflammatory, prothrombotic state.13

Patients in obesity class I to II were 1.44 times more likelyto develop a wound complication compared with normal-weight patients. Patients in class III were 1.84 times aslikely to develop a wound complication. In the retrospectiveanalysis by Kuduvalli et al,12 a BMI greater than 30 wasssociated with sternal wound infections, and a BMI greaterhan 35 was associated with sternal wound infections andarvest site infections. In another study by Lu et al,14 4,228atients undergoing coronary artery bypass graft proceduresere analyzed, and a BMI greater than 30 was found to be

n independent risk factor for sternal wound infection. In aetrospective analysis by Merkow et al11 with data from

3,202 colectomy patients, a BMI greater than 35 was asso-ciated with surgical site infection and wound dehiscence.Morbidly obese patients were 2.6 times more likely to incura surgical site infection, superficial or deep, and 3.5 timesmore likely to experience wound dehiscence. In a retrospec-tive analysis by Myles et al15 with data from 574 electivecesarean section patients, a BMI greater than 30 was asso-ciated with an increased risk of postoperative infectiouscomplications, including endomyometritis, wound infec-tions, urinary tract infections, septic pelvis thrombophlebi-tis, and pneumonia.

We used the World Health Organization classification ofBMI to stratify patient information into 4 categories. Otherstudies have looked at alternative measures of obesity asidefrom BMI such as abdominal wall thickness, hip girdle fatthickness, and retrorenal visceral fat. In a study by House etal16 examining complication in patients after undergoingpancreaticoduodenectomy, it was found that a BMI greaterthan 30 was associated with a higher risk of wound infec-tion. However, visceral fat �2.5 cm was associated with anearly 3 times greater risk of postoperative complicationsand a nearly 2.5 times greater risk of developing pancreaticfistula. Another study by Tanaka et al17 compared BMI withvisceral fat area in patients undergoing gastrectomy. BMIsgreater than 25 were only associated with increased bloodloss. However, a high visceral fat area (�100 cm2) at thelevel of the umbilicus estimated from computed tomogra-phy scans was associated with a 6.5-fold increased risk ofpancreatic fistula.17 Aside from BMI and alternative mea-ures of obesity, preoperative serum albumin has been an-ther parameter that has been investigated in relation tourgical complications. Preoperative albumin correlated in-ersely with complications, length of stay, postoperativetay, intensive care unit stay, mortality, and resumption ofral intake.18 Albumin levels correlated with complications

in patients undergoing esophageal or pancreatic procedureswho sustained significantly higher complication rates. How-ever, in colonic surgery, complication rates did not correlatewith albumin levels. Colonic procedures used the fewestresources at comparable albumin levels. This study corre-

lates with our group of patients undergoing intestinal sur-

440 The American Journal of Surgery, Vol 204, No 4, October 2012

gery showing that malnourished patients (BMI �18) werenot at an increased risk of complications and that preoper-ative serum albumin was not an independent risk factor forcomplications.

In conclusion, obesity leads to increased wound andrespiratory complications after intestinal surgery and anincreased risk of developing any complication. Patients witha BMI of 30 to 40 were 1.21 times more likely to have anycomplication compared with normal-weight patients andpatients with a BMI greater than 40 were 1.41 times morelikely to have any complication. Obese patients were morelikely to have respiratory and wound complications. Pa-tients with a BMI of 30 to 40 had a nearly a 1.5-foldincrease in wound complications, and patients with a BMIgreater than 40 had a nearly 2-fold increase in woundcomplication compared with normal-weight patients. Fi-nally, patients with a BMI of greater than 40 were 1.5 timesmore likely to have a respiratory complication comparedwith normal-weight patients. Nevertheless, after adjustingfor patient risk factors and hospital-level variation, obesityalone is not an independent risk factor for increased costs inintestinal surgery. Our study looked specifically at patientsundergoing intestinal surgery. Our patient population ex-cluded bariatric surgery patients, and only 3% of our pa-tients had a BMI greater than 40. Further studies couldanalyze a larger group of patients undergoing any generalsurgery procedure.

Acknowledgments

The authors would also like to acknowledge the VASurgical Quality Data Use Group for its role as scientificadvisers and for the critical review of data use and analysispresented in this manuscript.

References

1. Wang Y, Beydoun MA. The obesity epidemic in the United States-gender, age, socioeconomic, racial/ethnic, and geographic character-istics: a systematic review and Meta-regression analysis. Epidemiol

Rev 2007;29:6–28.

2. Khuri SF, Daley J, Henderson WG, et al. The Department of Veteran’sAffairs NSQIP: the first national, validated, outcome-based, risk-ad-justed, and peer-controlled program for the measurement and enhance-ment of the quality of surgical care. Ann Surg 1998;228:491–507.

3. Itani KMF. Fifteen years of the national surgical quality ImprovementProgram in review. Am J Surg 2009;198(suppl):S9–18.

4. Azoulay A, Doris NM, Filion KB, et al. The use of the transition costaccounting system in health services research. Cost Eff Resour Alloc2007;5:11.

5. Daley J, Khuri SF, Henderson W, et al. Risk adjustment of thepostoperative morbidity rate for the comparative assessment of thequality of surgical are: results of the national Veterans Affairs surgicalrisk study. J Am Coll Surg 1997;185:328–40.

6. WHO. Obesity: Preventing and Managing the Global Epidemic. Re-port of a WHO consultation. WHO Technical Report Series 894.Geneva: World Health Organization; 2000.

7. Raghunathan TE. What do we do with missing data? Some options foranalysis of incomplete data. Annu Rev Public Health 2004;25:99–117.

8. Khan S, Sclabas G, Reid-Lombardo K, et al. Does body mass index/morbid obesity influence outcome in patients who undergo pacrea-tuduodenectomy for pancreatic adenocarcinoma? J Gastrointest Surg2010;14:1820–5.

9. Efron JE, Uriburu JP, Wexner SD, et al. Restorative proctocolectomywith ileal pouch anal anastomosis in obese patients. Obes Surg 2001;11:246–51.

10. Hawn MT, Bian J, Leeth RR, et al. Impact of obesity on resourceutilization for general surgical procedures. Ann Surg 2005;241:821–6.

11. Merkow RP, Bilimoria KY, McCarter MD, et al. Effect of body massindex on short-term outcomes afer colectomy for cancer. J Am CollSurg 2009;208:53–61.

12. Kuduvalli M, Grayson AD, Oo AY, et al. The effect of obesity onmid-term survival following coronary artery bypass surgery. EurJ Cardiothrac Surg 2003;23:368–73.

13. Doyle SL, Lysaght J, Reynolds JV. Diagnostic in obesity and compli-cations: obesity and post-operative complications in patients undergo-ing non-bariatric surgery. Obes Rev 2009;11:875–6.

14. Lu JC, Grayson AD, Jha P, et al. Risk factors for sternal woundinfection and mid-term survival following coronary artery bypasssurgery. Eur J Cardiothorac Surg 2003;12:943–9.

15. Myles TD, Gooch J, Santolaya J. Obesity as an independent risk factorfor infectious morbidity in patients who undergo cesarean delivery.Obstet Gynecol 2002;100:959–64.

16. House MG, Fong Y, Arnaoutakis DJ, et al. Preoperative predictors forcomplication after pancreaticoduodenectomy: impact of BMI andbody fat distribution. J Gastrointest Surg 2008;12:270–8.

17. Tanaka K, Miyashiro I, Yano M, et al. Accumulation of excess visceralfat is a risk factor for pancreatic fistula formation after total gastrec-tomy. Ann Surg Oncol 2009;16:1520–5.

18. Kudsk KA, Tolley EA, DeWitt RC, et al. Preoperative albumin andsurgical site identify surgical risk for major postoperative complica-

tions. JPEN J Parenter Enteral Nutr 2003;27:1–9.