The Effect of Diabetes Mellitus on Surgical Site

Infections after Colorectal and Noncolorectal

From the Departments of *Surgery and ‡Internal Medicine, Albany Medical College, Albany, New York;
and †Quality Management, Albany Medical Center, Albany, New York

Patients undergoing colorectal surgery (CRS) are known to be at increased risk of surgical site in-
fection (SSI). We assessed the effect of diabetes and other risk factors on SSI in patients undergoing
CRS and patients undergoing general surgery (GS). American College of Surgeons National Surgical
Quality Improvement Program Participant Use Data File from 2005 to 2006 was used. x2 tests, t tests,
and logistic regression were used to assess the risk factors. Of the 129,909 study patients 10.1 per cent
were patients undergoing CRS. The incidence of SSI in patients undergoing CRS was 3.8 times
higher (95% CI, 3.6–4.1) than in patients undergoing GS. The incidence of SSI was higher in diabetics
than nondiabetics in patients undergoing CRS (15.4 vs 11.0%, P \ 0.001) and patients undergoing
GS (5.3 vs 3.1%, P \ 0.001). The significant univariate predictors of SSI for patients undergoing GS
and patients undergoing CRS were: males, American Society of Anesthesiologists (ASA) class, di-
abetes emergency surgery, operation time, and greater than 2 units of intraoperative red blood cell
transfusion. For patients undergoing GS, increasing age was also significant. After multivariate ad-
justment, significant predictors of SSI for patients undergoing GS and patients undergoing CRS were:
male gender, diabetes, ASA class, emergency surgery, and operation time. For patients undergoing GS,
age also remained significant. Among patients undergoing CRS, insulin-dependent diabetes mellitus
(IDDM) and noninsulin-dependent diabetes mellitus (NIDDM) were 1.32 (P \ 0.05) times more
likely than nondiabetics to develop SSI. Among patients undergoing GS, only IDDM (OR, 1.39; P \
0.001) were at increased risk. In this large hospital-based study, patients undergoing CRS were three
times more likely to get SSI than patients undergoing GS. Diabetic patients with CRS (IDDM and
NIDDM) and patients undergoing GS (IDDM) were at increased risk of SSI compared with non-
diabetics. More intense glycemic control may reduce SSI in patients undergoing CRS with diabetes.

infection (SSI) is the Patients with diabetes mellitus (DM) are known to be
P OSTOPERATIVE SURGICAL SITE
one of the most common causes of postoperative
complications in the United States. 1, 2
It is the third
at increased risk for both surgical and nosocomial in-
fections.9–11 Studies have estimated infection rates to be
most common cause of nosocomial infections and ac- to be two to five times higher in the diabetic than in the
counts for 14 to 17 per cent of all hospital-acquired nondiabetic population.9, 11 Diabetics are also more
infections.1, 3 Among surgical patients, SSI is the most than twice as likely to be hospitalized for infection and
common (38%) cause of nosocomial infections.2, 4, 5 significantly more likely to die as a result of infection.10
These infections are associated with longer intensive SSI has also been recognized as a leading cause of
care unit length of stay (average 4 days), mortality, and postoperative morbidity after colorectal surgeries.7, 8
cost ($8000 per patient).4, 6–8 The reported rates of SSI in patients undergoing co-
lorectal surgery have been inconsistent with estimates
Presented at the Annual Scientific Meeting and Postgraduate ranging from 3 to 30 per cent.12–17 The risk factors for
Course Program, Southeastern Surgical Congress, Savannah, GA,
February 20–23, 2010. SSI after colorectal surgery have not been clearly
Address correspondence and reprint requests to Ashar Ata, M.P.H., identified. The role of factors like DM or hypergly-
Department of Surgery, Albany Medical College, 47 New Scotland cemia management has not been well established for
Avenue, MC 61, Albany, NY 12208. E-mail: [email protected]. noncardiac surgeries like colorectal resection. There-
American College of Surgeons National Surgical Quality Im- fore, surgeons targeting improvement in the quality of
provement Program and the hospitals participating in the ACS
NSQIP are the source of the data used herein; they have not veri- outcomes of high-risk surgeries are unable to identify
fied and are not responsible for the statistical validity of the data specific variables that could reduce the risk of SSI.
analysis or the conclusions derived by the authors. The research literature regarding the risk of SSI in

697
698 THE AMERICAN SURGEON July 2010 Vol. 76

colorectal and noncolorectal patients undergoing general The outcome of interest was SSI, which was defined
surgery and the risk factors for SSI is lacking and often as developing either a superficial or deep SSI post-
inconsistent.1, 12–15 Most reports on the incidence and operatively. Secondarily we also reported the crude
the risk factors are based on studies with small num- effects of SSI on total hospital length of stay and on
bers of patients from a single hospital.12, 13, 15, 18 The likelihood of returning to the operating room.
availability of a large, accurate, and efficiently collected The risk factors and potential confounders assessed
surgical outcome data through the American College of for this study included age at admission (categorized
Surgeons National Surgical Quality Improvement Pro- in 15-year age intervals between 16 and 75 and an in-
gram (ACS NSQIP) allows a more efficient assessment terval comprising patients older than 75 years), gender,
of the incidence and risk factors for SSI.19, 20 race/ethnicity (white, black, Hispanic, other and un-
The purpose of our study was to assess the incidence known), DM (requiring oral medication or insulin vs
and identify risk factors of SSI in colorectal surgery none), emergency versus elective surgery, patient’s
and noncolorectal patients undergoing general surgery preoperative status as indicated by the American So-
(henceforth referred as general surgery). We were ciety of Anesthesiologists classification, number of red
specifically interested in assessing and comparing the blood cell (RBC) units given intraoperatively, and
role of diabetes and its type on risk of SSI in patients length of operation. Operation time was converted to
undergoing colorectal and general surgery. hours and number of RBC units was dichotomized to
‘‘two or less’’ and ‘‘more than two’’ categories.
Methods
Statistical Analysis
We used the ACS NSQIP Participant Use Data File
from 2005 to 2006. The data are deidentified and Health All categorical variables were compared using x2 test
Insurance Portability and Accountability Act-compliant for differences in proportions. The t test for equality of
and were originally developed as a quality improvement means was used to compare continuous variables across
initiative by the Veterans Health Administration in categories. The association between independent vari-
1994.19, 21 The ACS NSQIP program provides partici- ables and development of SSI was assessed separately
pating hospitals with data for the purposes of quality for patients undergoing colorectal surgery and patients
improvement. This ACS NSQIP data set contains in- undergoing noncolorectal general surgery. Variables
formation on 152,490 patients who underwent surgery in with P < 0.20 on bivariate analysis were further assessed
2005 and 2006. The details of the ACS NSQIP sampling in a stepwise multiple logistic regression model. The
methods, data abstraction process, and variables col- bivariate analysis is described using proportions and
lected have been described elsewhere.19, 21 The program percentages, and the results of the multivariate analysis
collects data on preoperative risk factors, intraoperative are presented and described using ORs with 95 per cent
variables, and 30-day postoperative mortality and mor- CIs. All analyses were performed using commercially
bidity outcomes for patients undergoing major surgical available software (STATA, Version 9.2; STATA Corp.,
procedures. The data are collected and uploaded elec- College Station, TX). P < 0.05 was considered statis-
tronically to the NSQIP database by each site’s surgical tically significant. All P values reported are two-sided.
clinical reviewer using a variety of methods, including This study was approved by the Institutional Review
medical chart abstraction. The sampling strategy cur- Board of our institute.
rently requires hospitals to report their first 40 consec-
utive eligible cases on an 8-day cycle. Each subsequent Results
cycle starts on a different day of the week to capture
a variety of cases and surgeons and to minimize bias in Of the 129,909 study patients, 10.1 per cent (13,089)
case selection. Patients undergoing general surgery were were patients undergoing colorectal and 116,820 were
identified using the surgical subspecialty variable. Pa- patients undergoing general surgery. The odds of SSI
tients undergoing colorectal surgery were identified in patients undergoing colorectal surgery were 3.8
using the Current Procedural Terminology codes for the times higher as compared with patients undergoing
following procedures: 44140, 44141, 44143–44147, general surgery (OR, 3.8; 95% CI, 3.61–4.10).
44150–44153, 44155, 44156, 44160, 44204–44208,
44210–44212, 44239, 45110–45116, 45119–45121, Bivariate Analysis
45123, 45395, 45397, 45499, and 45550. Colorectal
Colorectal Surgery
surgeries comprised open or laparoscopic and partial or
complete removals of the colon or the rectum. This study The overall incidence of SSI in patients undergoing
was limited to patients 16 years of age or older at the colorectal surgery was estimated to be 11.6 per cent
time of admission. (95% CI, 11.1–12.2%) (Table 1). Among patients
No. 7 EFFECT OF DM ON SSIs ? Ata et al. 699

TABLE 1. Incidence and Unadjusted Odds Ratio for of SSI among patients that were given more than 2
Postoperative Surgical Site Infection (SSI) by Potential Risk units of RBC intraoperatively was higher as compared
Factors among Patients Undergoing Colorectal Surgery
(n 4 13,089) with those given 2 units or less (14.7 vs 11.5%, P 4
0.05). There were no differences in the SSI by age or
Variable SSI (%) OR (95% CI)
race.
Age (years)
16–30 10.8 Referent Noncolorectal General Surgery
31–45 12.9 1.23 (0.91–1.65)
46–60 11.8 1.10 (0.84–1.45) The overall incidence of SSI in patients undergoing
61–75 12.2 1.15 (0.87–1.51) general surgery was estimated to be 3.3 per cent (95%
76+ 9.9 0.91 (0.68–1.21)
Gender CI, 3.2–3.4%) (Table 2).Older patients experienced
Female 10.9 Referent significantly higher infection rates as compared with
Male 12.4 1.16y (1.04–1.29) 16 to 30 year olds and the incidence increased with
Race age. There were also some differences in incidence by
White 11.2 Referent
Black 13.2 1.20 (1.00–1.44) race/ethnic groups. Hispanics had lower (2.4%) and
Hispanic 13.5 1.24 (0.98–1.57) others higher (4.5%) rates of SSI as compared with
Other 13.6 1.24 (0.76–2.03) whites (3.5%). Males had a higher occurrence of SSI as
Unknown 12.5 1.13 (0.95–1.35)
Diabetes compared with females (3.8 vs 3.0%). The incidence of
No 11.0 Referent SSI was significantly higher in diabetics than non-
Insulin 15.5 1.48y (1.18–1.86) diabetics (5.3 vs 3.1%, P < 0.001). Similar to colorectal
Oral 15.3 1.45z (1.23–1.72) surgery, patients undergoing emergent noncolorectal
Emergency
No 11.3 Referent general surgery had significantly higher SSI rates as
Yes 13.2 1.2* (1.04–1.38) compared with electives (4.9 vs 3.0%). As compared
ASA class with ASA Class I (no known systemic disease), all
1, No 7.2 Referent other higher classes (mild to moribund systemic dis-
2, Mild 10.2 1.45* (1.02–2.05)
3, Severe 13.6 2.02z (1.43–2.86) ease) had significantly higher SSI rates. The increase
4, Life-threatening 13.2 1.94y (1.32–2.85) in the SSI incidence demonstrated a trend with con-
5, Moribund 8.1 1.12 (0.48–2.61) secutively higher ASA class. Similarly, the length of
Operation timex 1.14z (1.11–1.18) the operation was highly correlated with SSI. Every 1-
Intraoperative RBC
2 units or less 11.5 Referent hour increase in length of surgery was associated with
More than 2 units 14.7 1.33 (1.00–1.77) a 33 per cent increase in odds of being infected. (95%
* P < 0.05. CI, 31–35%). The incidence of SSI among patients
y P < 0.01. who were given more than 2 units of RBC intra-
z P < 0.001. operatively was significantly higher as compared with
x OR for per hour increase in the length of operation. those given 2 units or less (10.6 vs 3.2%, P < 0.001).
ASA, American Society of Anesthesiologists; RBC, red
blood cells.
Multivariate Analysis
undergoing colorectal surgery, there were no differ- Among patients undergoing colorectal surgery (Ta-
ences in the SSI by age or race. However, males had ble 3), diabetics were estimated to be 1.32 times more
a higher occurrence of SSI as compared with females likely than nondiabetics to develop SSI after adjust-
(12.4 vs 10.9%, P < 0.01). The incidence of SSI was ment for the effect of other variables. The association
higher in diabetics as compared with nondiabetics for diabetics treated with oral medication (95% CI,
(15.4 vs 11.0%, P < 0.001). Patients who had un- 1.11–1.57) was slightly stronger than for insulin-
dergone emergency surgery had significantly higher dependent diabetics (95% CI, 1.05–1.68). After ad-
rates as compared with those who had elective surgery justment, patients with ASA Class III was 1.84 times
(13.2 vs 11.3%, P < 0.05). As compared with Ameri- (95% CI, 1.30–2.62) and those with ASA Class IV was
can Society of Anesthesiologists (ASA) Class I (no 1.68 times (95% CI, 1.13–2.48) more likely to develop
known systemic disease), Classes II, III, and IV(mild SSI as compared with those with ASA Class I. Patients
to life-threatening systemic disease) had significantly undergoing emergency colorectal surgery were 1.29
higher SSI infection rates (7.2 vs 10.2, 13.6, and times more likely to be infected (SSI) as compared with
13.2%, respectively; P < 0.001). The length of the those undergoing elective surgery (95% CI, 1.10–1.50).
operation (in hours) was highly correlated with SSI. Every 1-hour increase in operation time was associated
Every 1-hour increase in length of surgery was asso- with 1.15 times increase in the odds for SSI (95% CI,
ciated with a 14 per cent increase in odds of being 1.11–1.18). Male gender was also borderline signifi-
infected. (95% CI, 11–18%; P < 0.001). The incidence cant and was adjusted for in the multivariate model.
700 THE AMERICAN SURGEON July 2010 Vol. 76

TABLE 2. Incidence and Unadjusted Odds Ratio for TABLE 3. Multivariate Adjusted Odds Ratios for Postoperative
Postoperative Surgical Site Infection (SSI) by Potential Risk Surgical Site Infection among Patients Undergoing
Factors among Patients Undergoing Noncolorectal General Colorectal Surgery
Surgery (n 4 116,820)
Variables OR (95% CI)
Variable SSI (%) OR (95% CI)
Diabetes
Age (years) No Referent
16–30 1.8 Referent Insulin 1.32* (1.05–1.68)
31–45 2.9 1.59z (1.38–1.83) Oral 1.32y (1.11–1.57)
46–60 3.6 2.00z (1.73–2.27) ASA class
61–75 3.8 2.12z (1.84–2.43) 1, No Referent
76+ 4.1 2.30z (1.97–2.68) 2, Mild 1.40 (0.99–1.99)
Gender 3, Severe 1.84y (1.3–2.62)
Female 3.0 Referent 4, Life-threatening 1.68* (1.13–2.48)
Male 3.8 1.29z (1.21–1.38) 5, Moribund 0.88 (0.38–2.08)
Race Emergency
White 3.5 Referent No Referent
Black 3.7 1.07 (0.96–1.19) Yes 1.29y (1.10–1.5)
Hispanic 2.4 0.70y (0.61–0.79) Operation timex 1.15z (1.11–1.18)
Other 4.5 1.32* (1.03–1.69) Gender
Unknown 3.0 0.84y (0.76–0.95) Female Referent
Diabetes Male 1.11 (1.00–1.24)
No 3.1 Referent
Insulin 6.5 2.22z (1.97–2.5) * P < 0.05.
Oral 4.6 1.51z (1.36–1.69) y P < 0.01.
Emergency z P < 0.001.
No 3.0 Referent x OR for per hour increase in the length of operation.
Yes 4.9 1.64z (1.52–1.78) ASA, American Society of Anesthesiologists.
ASA class
1, No 1.3 Referent
2, Mild 2.5 2.01z (1.73–2.33) TABLE 4. Multivariate Adjusted Odds Ratios for Postoperative
3, Severe 5.1 4.23z (3.65–4.90) Surgical Site Infection among Patients Undergoing
4, Life-threatening 7.0 5.91z (4.93–7.10) Noncolorectal General Surgery
5, Moribund 6.2 5.19z (3.02–8.92) Variables Odds Ratio (95% CI)
Operation timex 1.33z (1.31–1.35)
Intraoperative RBC Diabetes
2 units or less 3.2 Referent No Referent
More than 2 units 10.6 3.57z (2.97–4.28) Insulin 1.39z (1.23–1.58)
Oral 1.07 (0.96–1.20)
* P < 0.05. ASA class
y P < 0.01. 1, No
z P < 0.001. 2, Mild 1.83z (1.57–2.14)
x OR for per hour increase in the length of operation. 3, Severe 3.12z (2.67–3.66)
ASA, American Society of Anesthesiologists; RBC, red 4, Life-threatening 3.54z (2.92–4.31)
blood cells. 5, Moribund 2.67y (1.54–4.62)
Emergency
No Referent
Yes 2.04z (1.87–2.22)
Among patients undergoing noncolorectal general Operation timex 1.30z (1.28–1.33)
surgery, insulin-treated diabetics were estimated to Gender
Female Referent
be 1.39 times more likely to develop SSI as com- Male 1.16z (1.09–1.24)
pared with nondiabetics after adjustment for all other Age category (years)
significant variables (Table 4). Oral hypoglycemic- 16–30 Referent
treated diabetics (noninsulin-dependent DM) were not 31–45 1.34z (1.16–1.55)
46–60 1.38z (1.2–1.6)
at increased risk of SSI. As compared with those with 61–75 1.26y (1.09–1.46)
ASA Class I, patients in all ASA classes were signif- 75+ 1.27* (1.08–1.5)
icantly more likely to be infected. The odds for SSI * P < 0.05.
increased from 1.83 times (95% CI, 1.57–2.14) for y P < 0.01.
Class II to 3.54 times (95% CI, 2.92–3.41) for Class IV z P < 0.001.
and dropped to 2.67 times (95% CI, 1.54–4.62) for x OR for per hour increase in the length of operation.
ASA, American Society of Anesthesiologists.
Class V. Patients undergoing emergency general sur-
geries were twice as likely to develop SSI as compared
with those undergoing elective surgeries (95% CI, 1.28–1.33). The effect of increasing age and male
1.87–2.22). The odds for SSI increased 1.3 times for gender were not explained by other variables and were
every hour increase in the length of surgery (95% CI, also adjusted for in the multivariate model.
No. 7 EFFECT OF DM ON SSIs ? Ata et al. 701

Effect of Surgical Site Infection on Length of Stay and the well as data collection methods and tools have sig-
Likelihood of Returning to the Operating Room nificantly varied. Studies have pointed out that the
The length of stay for patients undergoing colorectal NNIS data are voluntary and self-reported, include in-
surgery was on an average 3.21 days longer (95% CI, fection diagnosed after discharge, and it does not in-
1.95–4.47) for infected patients as compared with clude all procedures.16, 22, 23 The results of our study
noninfected patients. Similarly, for patients undergo- are based on ACS NSQIP data from years 2005 to 2006
ing general surgery, the total hospital length of stay for from 121 hospitals (59% academic and 41% commu-
infected patients was 7.04 days (95% CI, 6.64–7.45) nity) hospitals. The data are collected, validated, and
longer as compared with noninfected patients. Infected submitted by a trained surgical clinical reviewer at
patients undergoing colorectal surgery were 2.13 times each site to a web-based system that ensures complet-
(95% CI, 1.81–2.50) more likely to return to the op- eness, uniformity, and validity of the data. The ACS
erating room as compared with noninfected patients. NSQIP data address the potential source variation in
Likewise, infected patients undergoing general surgery the results from previous studies and the data are also
were 4.24 times (95% CI, 3.88–4.64) more likely to relatively current as compared with NNIS reports.
return to the operating room as compared with non- Diabetes has been previously identified as a risk
infected patients. factor for SSI in patients undergoing general and
vascular surgery in a study by Neumayer et al. based
on the same data.14 The authors, however, did not as-
Discussion
sess colorectal surgeries separately and also did not
The purpose of our study was to estimate the in- compare the effect of insulin-dependent and non-
cidence of SSI and the effect of diabetes and other risk insulin-dependent diabetics. Other reports based on
factors on SSI in patients undergoing colorectal and data from single institutes that assessed the risk factors
noncolorectal general surgical procedures. In this for SSI in colorectal surgeries did not find diabetes
study, we found that incidence of SSI among patients to be an independent risk factor.12, 16 In our study, di-
undergoing colorectal surgery was 11.6 per cent and it abetes was found to be an important independent risk
was more than three times higher than patients un- factor for both colorectal and general surgeries. Di-
dergoing general surgery (3.3%). Risk factors for SSI abetics had the highest infection rates when compared
were similar for both surgery types, but the strength of with any other risk factors. As compared with general
associations was stronger for patients undergoing co- surgery, diabetes was a stronger risk factor for patients
lorectal surgery. The rate of incisional SSI for patients undergoing colorectal surgery and both insulin-de-
undergoing general surgery in our study (3.3%) is pendent and noninsulin-dependent diabetics were at
similar to that reported by recent studies.1, 14 For co- increased risk of SSI. For patients undergoing general
lorectal surgeries, the reported incidence of SSI varies surgery, only insulin-dependent diabetics were at in-
from 3 to 30 per cent12–17, 22 and the average rates creased risk of SSI. Age was not an independent risk
for such infections is estimated to be roughly 10 per factor for SSI in colorectal surgery but was for general
cent.16 Most of these reports are based on small num- surgery. Crude analysis of the association of SSI with
bers of patients from a single hospital. The most hospital length of stay and likelihood of returning to
widely used reporting system used currently is the the operating room after the surgery showed that SSI
National Nosocomial Infections Surveillance (NNIS) was associated with longer length of stay and higher
system. It is based on a large sample from multiple ‘‘return to operating room’’ rates in both patients un-
institutes throughout the United States. A comparison dergoing colorectal surgery and those undergoing
of the SSI rates reported in the present study with the general surgery.
rates reported nationally by the NNIS System showed In this study, we decided to exclude organ space
that our rates for colorectal surgery are similar to the infections because we believe that the mechanism and
those reported for the highest risk category in the NNIS the risk factors for organ space SSIs are different from
report, which reports rates of 3.98 per cent for cases in incisional SSIs as has been reported by other studies
the M,0 risk index category to 11.25 per cent for cases as well.16 The results of this study are therefore gen-
in the risk index category of ‘‘3.’’1 The variation in the eralizable to incisional SSIs only. Another limitation of
reported rates of colorectal surgeries is the result of our study was that we did not have information on
varying definitions of SSI, methods of performing perioperative glucose levels in the diabetic as well as
the assessments for infection, different patient pop- the nondiabetic patients. The role of hyperglycemia
ulations, the inclusion of infections identified after and its management therefore could not be assessed.
discharge, and, most importantly, the small sample The increased costs, increased length of stay, and in-
sizes. The NNIS report also is based on data from 1992 creased morbidity and mortality associated with SSI
to 2004, a time period over which surgical practices as have helped to broaden the scope of research surrounding
702 THE AMERICAN SURGEON July 2010 Vol. 76

SSI. National initiatives such as the Surgical Care 9. Hruska LA, Smith JM, Hendy MP, et al. Continuous insulin
Improvement Project have developed process mea- infusion reduces infectious complications in diabetics following
sures based on evidence-based practice recommen- coronary surgery. J Card Surg 2005;20:403–7.
dations to reduce SSI. Despite adoption of these 10. Shah BR, Hux JE. Quantifying the risk of infectious dis-
eases for people with diabetes. Diabetes Care 2003;26:510–3.
measures, the rate of SSI in patients undergoing co-
11. Shilling AM, Raphael J. Diabetes, hyperglycemia, and in-
lorectal surgery remains relatively high. Our research fections. Best Pract Res Clin Anaesthesiol 2008;22:519–35.
demonstrates increased risk of SSI in both patients 12. Smith RL, Bohl JK, McElearney ST, et al. Wound infection
undergoing colorectal surgery and those undergoing after elective colorectal resection. Ann Surg 2004;239:599–605;
general surgery if they have insulin-dependent DM. discussion 605–7.
Intense glycemic control has become standard in pa- 13. Tang R, Chen HH, Wang YL, et al. Risk factors for surgical
tients undergoing cardiac surgery as a result of re- site infection after elective resection of the colon and rectum:
search demonstrating lower sternal wound infections a single-center prospective study of 2,809 consecutive patients.
and lower mortality in critically ill patients not only Ann Surg 2001;234:181–9.
with diabetes, but also with insulin resistance.24 To 14. Neumayer L, Hosokawa P, Itani K, et al. Multivariable
date, there has been no definitive study in colorectal predictors of postoperative surgical site infection after general and
vascular surgery: results from the patient safety in surgery study.
patients to confirm that tight glycemic control reduces
J Am Coll Surg 2007;204:1178–87.
SSI. Findings of our study warrant that tighter glyce- 15. Walz JM, Paterson CA, Seligowski JM, Heard SO. Surgical
mic control needs to be further evaluated as a tool to site infection following bowel surgery: a retrospective analysis of
decrease SSI in diabetics and in patients who are 1446 patients. Arch Surg 2006;141:1014–8; discussion 1018.
highly susceptible to such infections after surgical 16. Blumetti J, Luu M, Sarosi G, et al. Surgical site infections
procedures like colorectal resection. after colorectal surgery: do risk factors vary depending on the type
of infection considered? Surgery 2007;142:704–11.
17. Gaynes RP, Culver DH, Horan TC, et al. Surgical site in-
REFERENCES fection (SSI) rates in the United States, 1992–1998: the National
1. National Nosocomial Infections Surveillance (NNIS) System Nosocomial Infections Surveillance System basic SSI risk index.
Report. Data summary from January 1992 through June 2004, Clin Infect Dis 2001;33(suppl 2):S69–77.
issued October 2004. Am J Infect Control 2004;32:470–85. 18. McConnell YJ, Johnson PM, Porter GA. Surgical site in-
2. National Nosocomial Infections Surveillance (NNIS) System fections following colorectal surgery in patients with diabetes:
Report. Data summary from January 1992–June 2001, issued association with postoperative hyperglycemia. J Gastrointest Surg
August 2001. Am J Infect Control 2001;29:404–21. 2009;13:508–15.
3. Smyth ET, Emmerson AM. Surgical site infection surveil- 19. Khuri SF. The NSQIP: a new frontier in surgery. Surgery
lance. J Hosp Infect 2000;45:173–84. 2005;138:837–43.
4. Mangram AJ, Horan TC, Pearson ML, et al. Guideline for 20. Bentrem DJ, Cohen ME, Hynes DM, et al. Identification of
prevention of surgical site infection, 1999. Hospital Infection specific quality improvement opportunities for the elderly un-
Control Practices Advisory Committee. Infect Control Hosp Epi- dergoing gastrointestinal surgery. Arch Surg 2009;144:1013–20.
demiol 1999;20:250–78; quiz 279–80. 21. Khuri SF, Daley J, Henderson W, et al. The Department of
5. Weiss CA III, Statz CL, Dahms RA, et al. Six years of sur- Veterans Affairs’ NSQIP: the first national, validated, outcome-
gical wound infection surveillance at a tertiary care center: review based, risk-adjusted, and peer-controlled program for the mea-
of the microbiologic and epidemiological aspects of 20,007 surement and enhancement of the quality of surgical care. National
wounds. Arch Surg 1999;134:1041–8. VA Surgical Quality Improvement Program. Ann Surg 1998;228:
6. Dimick JB, Chen SL, Taheri PA, et al. Hospital costs asso- 491–507.
ciated with surgical complications: a report from the private-sector 22. de Oliveira AC, Ciosak SI, Ferraz EM, Grinbaum RS.
National Surgical Quality Improvement Program. J Am Coll Surg Surgical site infection in patients submitted to digestive surgery:
2004;199:531–7. risk prediction and the NNIS risk index. Am J Infect Control 2006;
7. Astagneau P, Rioux C, Golliot F, Brucker G. Morbidity and 34:201–7.
mortality associated with surgical site infections: results from the 23. Rioux C, Grandbastien B, Astagneau P. The standardized
1997–1999 INCISO surveillance. J Hosp Infect 2001;48:267–74. incidence ratio as a reliable tool for surgical site infection sur-
8. Kirkland KB, Briggs JP, Trivette SL, et al. The impact of veillance. Infect Control Hosp Epidemiol 2006;27:817–24.
surgical-site infections in the 1990s: attributable mortality, excess 24. van den Berghe G, Wouters P, Weekers F, et al. Intensive
length of hospitalization, and extra costs. Infect Control Hosp insulin therapy in the critically ill patients. N Engl J Med 2001;345:
Epidemiol 1999;20:725–30. 1359–67.