Clinical and Economic Consequences of Post-Operative Infections following Major Elective Surgery in U.S. Hospitals

Abstract

Background:

The incidence and consequences of post-operative infections in patients undergoing major elective surgery is not well understood.

Methods:

Using a large U.S. healthcare claims database, we identified all patients who underwent major elective surgery between January 1, 2007, and December 31, 2009. For each such patient, date of the first-noted surgery during this period was designated as the index date. Patients who developed infections within 30 d of their index date were matched to those who did not using propensity score matching. We compared hospital readmissions, mortality, and total healthcare cost during the 30-d period following index date between patients who developed post-operative infections versus those who did not.

Results:

A total of 327,618 patients met all selection criteria. At 30 d following major elective surgery, 10.9% of patients had evidence of post-operative infections, 39% of which occurred during the index admission. In propensity-matched analyses, patients with post-operative infections were about five times as likely to be readmitted to hospital (11.3% vs. 2.1%) and more than twice as likely to die (0.8% vs. 0.3%) in the 30-d period following surgery; their average total healthcare cost was $8,417 higher ($29,229 vs. $20,812) (all comparisons, p<0.01).

Conclusion:

Approximately one in 10 patients undergoing major elective surgery develop post-operative infections by day 30. Post-operative infections are associated with significantly worse clinical outcomes and higher total healthcare cost.

Many patients who undergo major surgery develop post-operative infections. Most commonly, such infections occur at the surgical site (surgical site infection [SSI]), but they may also occur at unrelated sites, especially the urinary tract, lungs, and blood stream. Post-operative infections are typically defined to include all infections that develop within 30 d of surgery (within 365 d if the infection involves a prosthetic implant) [1]. The risk of SSI is especially high in patients who undergo emergency surgery because of inadequate pre-operative preparation, contaminated wounds, and underlying comorbidity [2 –6]. Although a number of studies have addressed the incidence of specific types of post-operative infection (especially SSI) after selected elective procedures (e.g., total hip and knee replacement), there is little published information concerning the overall incidence of post-operative infection among patients who undergo major elective surgery in the United States.

Scott et al. utilized a computerized database to examine the incidence of post-operative infection among 9,016 patients who underwent surgery (i.e., cardiothoracic, gastrointestinal, neurologic, orthopedic, soft-skin tissue, and vascular) treated at a community hospital in Buffalo, NY between March 1, 1995, and December 31, 1997 [7]. They reported a 12.5% incidence of early infection (2–7 d post-operatively), and a 2.5% incidence of hospital readmission for infection between 15 and 28 d after surgery. Herwaldt et al. followed more than 3,800 patients prospectively who participated from 1995-1999 in a clinical trial of nasal mupirocin to prevent post-operative infection at two hospitals in Iowa [8]. During a mean follow up period of 30 d (range, 25 to 35 d), the incidence of post-operative infection was 11.3%. However, these results are somewhat dated and are limited to experiences at single centers.

The estimated economic impact of post-operative infections is substantial. In multivariable analyses, Herwaldt et al. reported that total hospital cost was approximately two-fold higher in general surgery patients who developed SSIs—and more than four-fold higher in patients with other types of post-operative infections (e.g., urinary tract infections, respiratory tract infections)—compared with those who did not develop post-operative infection [8]. Several other studies of the economic impact of SSI or all post-operative infections (both U.S.-based and non–U.S.-based) have been summarized in recent review articles [9,10].

In light of growing concerns about the problem of antibiotic resistant pathogens, especially methicillin-resistant Staphylococcus aureus, we undertook an examination of the incidence and clinical and economic consequences of post-operative infection using a large nationwide healthcare claims database. We chose to limit our examination to patients who underwent major elective surgery to constitute a more homogenous study population. Such a population also is of particular interest as it would be amenable potentially to pre-admission infection prophylaxis, including but not necessarily limited to home pre-operative skin sterilization and vaccination.

Materials and Methods

Data source

Our study was based on data from the Thomson Reuters MarketScan Commercial Claims and Encounters and Medicare Supplemental and Coordination of Benefits Database. This database is comprised of paid institutional, provider, and retail pharmacy claims from a variety of health plans across the US, representing healthcare services provided to about 15 million persons annually. The database contains information on patient demographics and eligibility, inpatient and outpatient diagnoses (in International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] format), inpatient and outpatient procedures (in Physicians' Current Procedural Terminology, Fourth Edition [CPT-4] and Healthcare Common Procedure Coding System [HCPCS], and ICD-9-CM [selected plans only] formats), drugs dispensed in outpatient (e.g., retail) pharmacies (in National Drug Code format), and dates of service for all medical services and drugs.

Demographic and eligibility data also are available for plan members, including age, gender, geographic region, coverage type, and dates of eligibility for plan benefits. All data can be arrayed in chronologic order to provide a detailed longitudinal profile of all medical and pharmacy services used by each plan member. All patient-identifying information has been encrypted in the database; it is therefore compliant with the Health Insurance Portability and Accountability Act of 1996.

Study population

Our study population consisted of all persons age ≥18 y who underwent major elective surgery as inpatients between January 1, 2007, and December 31, 2009. The hospital admission for the surgery was deemed the “index admission” and the date of that admission, “the index date.” If a patient had more than one admission for major elective surgery during this 3-y period, we focused attention only on the first one. (Although there is no generally agreed-upon definition of “major surgery,” typically such procedures involve entering a body cavity or joint space via an open procedure; interested readers should consult Appendix A at www.liebertpub.com/sur for a list of all procedures that we considered to be “major.”)

To ensure that surgical procedures were elective, we excluded patients whose procedure was not performed on the day of hospital admission or the next day, as well as patients with evidence of any of the following: 1) Multiple types of surgical procedures (e.g., orthopedic and neurologic) performed on the same day; 2) source of admission other than the community (e.g., prison, skilled nursing facilities); 3) admission from an emergency department (ED); 4) non-cardiac trauma/injury (e.g., fractures), acute events (e.g., ischemic stroke), cardiac events (e.g., myocardial infarction), or infection ≤14 d before index admission; or 5) procedures performed on a Saturday or Sunday (such operations are unlikely to be elective). Evidence of any of the following also was grounds for exclusion: 1) Invalid claims information; 2) hospitalization within prior 12 mo involving implants/devices; 3) hospitalization within prior 6 mo involving any other type of surgery; 4) prior organ transplant; 5) less than 12 mo of complete data prior to the index admission: or 6) age ≥65 y and enrollment in a Medicare supplemental health plan, because the database does not include complete claims histories for such persons. (A complete listing of all ICD-9-CM, CPT-4, and HCPCS codes used to define all inclusion and exclusion criteria are available from the authors upon request.)

All remaining patients were then followed from their index admission to 30 d for post-operative infection, death, and healthcare utilization and costs. Patients were censored at the earliest of the following: 1) End of the 30-d follow up period; 2) evidence of a second major surgery (except if related to the treatment of infection [e.g., drainage/debridement]); and 3) disenrollment from the health plan.

Measures

We examined the demographic and clinical characteristics of study subjects in terms of age, gender, geographic region of residence, payer type, and the prevalence of selected comorbidities (e.g., diabetes mellitus, peripheral arterial disease, renal disease). We focused attention on comorbidities that are associated with immunocompromised status, surgical complications, or relatively high healthcare cost. Patients were considered to have evidence of these comorbidities if they had ≥2 outpatient claims ≥30 d apart, or ≥1 inpatient claim, in the 12-mo period prior to the index admission with a relevant diagnosis code/prescription (see Appendix B at www.liebertpub.com/sur for a complete listing of all relevant diagnoses). We calculated a Charlson Comorbidity Score for each patient. We also examined utilization of selected medications and healthcare services during the 12-mo period prior to index admission that have been associated with increased risk of post-operative infections (i.e., use of anti–tumor necrosis factor therapy, such as etanercept ≤2 wks; adalimumab ≤4 wks, and infliximab ≤8 wks during the period prior to surgery [11], and receipt of systemic corticosteroids, immunosuppressive agents, or chemotherapy ≤90 d prior to index admission, which were deemed to constitute evidence of immunocompromised status) or general levels of morbidity (i.e., outpatient visits including ED visits, admissions to hospital, and total healthcare costs).

Diagnosis codes (ICD-9-CM) used to identify patients with post-operative infections are set forth in Appendix C, at www.liebertpub.com/sur, along with a frequency distribution of the total number of patients with each such code during the follow up period. Our operational definition of post-operative infection, which has been used in prior studies [7,8], included all acute infections, regardless of organ system, occurring within 30 d of hospital admission. Our list, therefore, contains diagnosis codes for some infections that are not considered to be commonly post-operative infections, but which nonetheless may occur during the post-operative period (e.g., carbuncle/furuncle, appendicitis).

Mortality during the index admission was ascertained based on discharge status. Death up to 30 d following hospital discharge was ascertained based on information on plan disenrollment, which we used as a proxy for death, following methods developed originally by Paramore et al [12]. Specifically, we assumed that all patients with evidence of disenrollment during follow up died if, during the four-week period immediately preceding disenrollment, they had evidence of a traumatic event, hospitalization (other than for reasons not typically life threatening, such as childbirth), chronic liver disease, chronic kidney disease, metastatic cancer, or infection.

We also examined the number of patients with ≥1 ED visits and hospital re-admissions, respectively, up to 30 d subsequent to the index admission. We also examined total healthcare cost during this period using total reimbursed amount (i.e., third-party payment plus patient liability) as a proxy for healthcare cost.

Analyses

We examined the incidence of post-operative infection up to 30 d following the index admission. Because demographic and clinical characteristics differed between patients with and without post-operative infections, we compared them using techniques of propensity score matching [13,14], which can reduce the effects of confounding resulting from comparing dissimilar patient populations [15]. Briefly, a propensity score was generated for each patient using a logistic regression model, which included as covariates pre-admission demographic characteristics (i.e., age, gender, geographic region, payer type), and various proxies of health status (i.e., specific comorbidities, Charlson Comorbidity Index, pre-index healthcare utilization, such as any ED visits during pre-index period, any hospital admissions during pre-index period, pre-index healthcare costs) described in the Measures section above. Once a propensity score was generated for each patient, patients who developed post-operative infections were matched to those who did not in stepwise fashion to minimize the absolute difference in propensity scores for each match. Matching was done on a 1:1 basis, and patients with evidence of post-operative infections were matched further to those without such evidence based on the type of surgery performed on the index date (e.g., orthopedic, cardiovascular). Unmatched patients were excluded from the comparisons. The statistical significance of differences between propensity-matched patients who did and did not develop post-operative infections was ascertained using paired t-tests for continuous variables that were normally distributed and Wilcoxon signed-rank tests for those that were not; McNemar and Bowker tests were used to assess the statistical significance of differences in categorical variables, as appropriate. All analyses were conducted using PC-SAS v9.2 (SAS Institute Inc, Cary, NC).

Results

We identified 1,030,968 patients who underwent major surgery between January 1, 2007, and December 31, 2009, of whom 327,618 (31.8%) met all study entry criteria (Table 1). Patients undergoing elective orthopedic procedures constituted 34.4% of the study population; gynecologic/genitourinary surgery, 29.0%; neurologic surgery, (16.5%); abdominal/bariatric/general surgery, 10.7%; cardiothoracic surgery, 5.0%; cardiovascular surgery, 2.4%; and plastic surgery, 2.1%.

Table 1.

Selection of Patients Undergoing Major Elective Surgery, by Index Procedure ^*

Criteria	Patients
Admitted to hospital with ≥1 diagnostic/procedure codes indicative of major elective surgery during study period (see Appendix A), and	1,030,968
Age between 18 and 64 y as of admission date or age ≥65 y as of admission date and not enrolled in Medicare supplemental plans, and	1,012,613
≥12 mo enrollment prior to admission, and	522,904
No invalid/missing information on claims during pre-index or follow up period, and	478,209
No evidence of multiple procedures from different cohorts on the index date, and	470,670
No evidence of hospitalization that involved surgeries with implants/devices during the pre-index period, and	462,118
No evidence of transplant during the entire pre-index period, and	459,947
No evidence of hospitalization that involved surgeries without implants/devices during the 6-mo period immediately prior to the index date, and	459,110
No evidence of care rendered ≤14 d prior to and including the index date in prison or correctional facilities, group homes, inpatient long-term care, other inpatient care (including acute-care facilities), skilled nursing facilities, nursing facilities, custodial care facilities, hospices, adult living care facilities, and/or comprehensive in-patient rehabilitation facilities, and	445,511
First code for major elective surgery on admission date or day immediately subsequent, and	406,913
No evidence of admission to hospital for the index procedure from the ED, and	406,592
No evidence of cardiac events or immune deficiency within 14 d of index admission and ^**	403,754
No evidence of infection within 14 d of index admission, and ^**	384,187
No evidence of non-cardiac acute event/injury/trauma within 14 d of index admission, and ^**	335,493
First code for major elective surgery not on a weekend	327,618
Total in study	327,618

All values are number of patients.

Including principal diagnosis on index admission.

ED=emergency department.

A total of 10.9% of patients had evidence of infections within 30 d of their index admission, approximately 39% of which occurred during the index admission. Rates of post-operative infections were lower among men than women (33.2% vs. 37.1%, respectively; p<0.01). Patients with evidence of post-operative infections also were more likely to have various comorbidities than those without these infections, including (but not limited to) immuncompromising conditions (9.6% vs. 6.1%, respectively), diabetes mellitus (5.8% vs. 3.8%), evidence of prior infection (14.6% vs. 9.1%), and coronary artery disease (6.0% vs. 4.2%). Patients with post-operative infections also had higher mean total healthcare cost during the 6-mo pre-index period ($6416 vs. $3697 for those without these infections) (all comparisons p<0.01; Table 2).

Table 2.

Clinical and Demographic Characteristics of Patients Undergoing Major Elective Surgery, by Post-Operative Infection Status at Day 30

Characteristic	Post-operative infections (N=35,703)	No post-operative infections (N=291,915)	p
Mean (SD) age, y	56.9 (16.5)	56.2 (15.8)	<0.01
No. (%) males	11,854 (33.2)	108,435 (37.1)	<0.01
Geographic residence
Northeast	3,122 (8.7)	26,845 (9.2)	<0.01
North central	11,481 (32.2)	91,573 (31.4)	<0.01
South	15,115 (42.3)	124,225 (42.6)	<0.01
West	5,985 (16.8)	49,272 (16.9)	<0.01
Payer type
HMO	4,971 (13.9)	39,433 (13.5)	<0.01
POS	3,019 (8.5)	27,115 (9.3)	<0.01
PPO	20,527 (57.5)	167,756 (57.5)	<0.01
Other	7,186 (20.1)	57,611 (19.7)	<0.01
Comorbidities/health history
Immunocompromising conditions	3,410 (9.6)	17,827 (6.1)	<0.01
Open wounds/wound care	509 (1.4)	2,139 (0.7)	<0.01
Diabetes mellitus	2,065 (5.8)	11,133 (3.8)	<0.01
Liver disease	453 (1.3)	2,210 (0.8)	<0.01
Peripheral arterial disease	856 (2.4)	4,609 (1.6)	<0.01
Renal disease	879 (2.5)	3,821 (1.3)	<0.01
Alcohol/drug abuse	517 (1.4)	2,479 (0.8)	<0.01
Heart failure	839 (2.3)	3,834 (1.3)	<0.01
Malnutrition	121 (0.3)	367 (0.1)	<0.01
Prior infection	5,207 (14.6)	26,553 (9.1)	<0.01
Hypertension	4,462 (12.5)	25,771 (8.8)	<0.01
Dyslipidemia	2,335 (6.5)	14,866 (5.1)	<0.01
Coronary artery disease	2,145 (6.0)	12,284 (4.2)	<0.01
Cardiogenic shock	4 (0)	26 (0)	<0.01
Cardiomyopathy	295 (0.8)	1,561 (0.5)	<0.01
Cardiac dysrhythmias
Atrial fibrillation	833 (2.3)	4,403 (1.5)	<0.01
All other	1,071 (3.0)	6,031 (2.1)	<0.01
Valvular heart disease	962 (2.7)	5,344 (1.8)	<0.01
Prior stroke/TIA	955 (2.7)	4,981 (1.7)	<0.01
Dialysis
Peritoneal	11 (0)	27 (0)	<0.01
Hemodialysis	87 (0.2)	344 (0.1)	<0.01
Chronic respiratory conditions	1,859 (5.2)	7,941 (2.7)	<0.01
Blood dyscrasias	220 (0)	936 (0.3)	<0.01
Sepsis/septic shock	298 (0.8)	819 (0.3)	<0.01
Overweight	516 (1.4)	2,369 (0)	<0.01
Underweight	16 (0)	37 (0.0)	<0.01
Leukemia, lymphoma, or metastasis	444 (1.2)	1,923 (0.7)	<0.01
Autoimmune diseases	489 (1.4)	2,201 (0.8)	<0.01
Neutropenia	83 (0.2)	434 (0.1)	<0.01
Pulmonary tuberculosis	5 (0)	28 (0)	0.43
Mean (SD) Charlson Comorbidity Index	0.4 (1.3)	0.3 (1.0)	<0.01
Pre-index healthcare utilization
No. (%) with anti-TNF use	70 (0.2)	421 (0.1)	0.05
Mean (SD) outpatient visits	5.5 (13.4)	3.7 (10.5)	<0.01
No. (%) with ≥1 ED visits	2,046 (5.7)	10,517 (3.6)	<0.01
No. (%) with ≥1 hospitalizations	4,706 (13.2)	25,266 (8.7)	<0.01
Mean (SD) pre-index healthcare cost, $	6,416 (23,094)	3,697 (15,011)	<0.01

SD=standard deviation; HMO=health maintenance organization; POS=point of service; PPO=preferred provider organization; TIA=transient ischemic attack; TNF=tumor necrosis factor; ED=emergency department. Numbers in parentheses represent percentages unless noted otherwise.

A total of 26,593 patients with post-operative infections (74.5% of all such patients) were propensity matched to an equal number of patients without evidence of post-operative infections. The demographic and clinical characteristics of propensity-matched patients with and without post-operative infections were similar (data not shown). In propensity-matched analyses, during the 30-d follow up period, post-operative infections were associated with an increased likelihood of ED visits (10.9% vs. 3.3% for those who did not develop such infections), and readmission to hospital (11.3% vs. 2.1%); risk of death was more than two-fold higher among those with evidence of post-operative infections (0.8% vs. 0.3%; p<0.01 for all comparisons). Healthcare cost during the 30 d also was $8,417 higher among patients with post-operative infections (mean [standard deviation]=$29,229 [$40,715] vs. $20,812 [$20,266] patients without such infections; p<0.01).

Discussion

We report the incidence of post-operative infection among a large nationwide sample of patients who underwent major elective surgery between 2007 and 2009. Approximately one in 10 patients who underwent major elective surgery developed post-operative infections within 30 d. In propensity matched analyses, post-operative infections were associated with an increased likelihood of readmission to hospital or ED visits, a greater likelihood of death, and higher total healthcare costs. Our findings indicate that the clinical and economic consequences of post-operative infections are substantial.

Our incidence rates were similar to those reported in the two prior studies described in the Introduction [7,8]. However, there are several differences between the methods used in our study and those used in prior research. The study by Scott et al. was conducted at a single hospital in Buffalo, NY between 1995 and 1997. Early infection was defined as a new antibiotic regimen started between 2 and 7 d post-operatively, and the authors note that their definition does not distinguish between clinically suspected (but unproved) and documented infection. Whereas this definition suggests that the 12.6% incidence rate of early post-operative infection they reported was inflated, the only post-discharge infections reported were those that resulted in readmission between 15 d and 28 d post-operatively and thus the incidence rate for post-discharge infections (2.5%) was clearly an underestimate of the total (both inpatient and outpatient) post-discharge infection rate. In the study by Herwaldt et al., all post-operative infections were counted if they were deemed nosocomial but many infections acquired post-discharge may have not qualified, whereas our definition was not designed to distinguish between post-discharge infections that did and did not involve a hospital-acquired pathogen. However, we emphasize that despite differences in methodology, our findings of marked increases in healthcare utilization and costs for patients with post-operative infections are consistent with these studies.

The high risk of post-operative infection, especially SSI, after emergency major surgery has been well documented [2 –6]. Our study confirms that the risk of post-operative infection after elective major surgery is also substantial. Of particular note, we found that 61% of post-operative infections occurred subsequent to discharge from the index admission. Guidelines for post-operative infection prophylaxis focus attention on in-hospital, peri-operative measures (e.g., pre-operative antiseptic showering, antimicrobial prophylaxis, aseptic surgical procedures, prompt removal of urinary catheters) [16 -18]. Our finding of a high rate of post-operative infection well beyond the peri-operative period suggests that future research concerning prophylaxis of post-operative infections should be directed to measures that can reduce their incidence subsequent to hospital discharge.

We note that observed differences in utilization and cost between patients who developed post-operative infections and those who did not may not be wholly attributable to such infections. Healthcare claims databases such as the one we used have only limited clinical information, and do not contain information on a number of important risk factors for post-operative infection, such as obesity, disease severity, use of antibiotic prophylaxis prior to surgery, sterility of the surgical field and operating equipment, the degree to which operating room staff follow proper sterile procedure(s), and contamination of the injury/surgical site [19]. It is therefore possible that, despite our use of propensity score matching, patients who developed post-operative infections differed in other unmeasured important respects from patients who did not. The degree to which our results may suffer from such residual confounding is unknowable.

On a related matter, whereas we were able to match approximately 75% of patients who experienced post-operative infections (using methods of propensity scoring) to an equal number of patients who did not experience such infections, we could not match the remaining 25% (i.e., the propensity scores of one-quarter of patients who experienced post-operative infection did not overlap with those who did not experience such infections). It is possible that patients with post-operative infections that could not be matched differ in important ways from those that could not (e.g., they may have had higher levels of comorbidities or higher pre-index healthcare cost). The impact of the “omitted” patients on the generalizability of our findings to the entire population of patients with post-operative infections is unknown.

There are a number of limitations of our study. First, our findings are based on analyses of healthcare claims, and therefore are subject to the inherent limitations of such data, including possible errors of omission and commission in coding and limited clinical information. Second, and related to the first, our operational definitions of post-operative infections were based on ICD-9-CM diagnosis codes. To the extent that patients were treated for infections either without being seen by a physician (e.g., an antibiotic prescription prescribed over the phone), or following a physician visit that did not result in the recording of an infection-related ICD-9-CM diagnosis code on a healthcare claim submitted for reimbursement, they would have been misclassified as not having experienced a post-operative infection. Without access to patients' medical records, the degree to which such misclassification occurred must remain conjectural. However, we note that unrecorded infections are likely to have been minor (e.g., strep throat, ear infection). Moreover, to the degree that such misclassification occurred, it would render our findings somewhat conservative by minimizing the differences between those who did versus did not develop post-operative infections. Third, persons in our database all were commercially insured and comparatively few were age ≥65 y. Because persons of advanced age may be more likely to develop post-operative infections [7,8,20 -22], our estimates may be conservative.

In conclusion, our study suggests that approximately one in 10 patients undergoing major elective surgery develop post-operative infections within 30 d of their procedure, most commonly following hospital discharge. Patients who develop these infections have significantly higher risk of death, higher levels of healthcare utilization, and higher healthcare costs. We believe that our findings highlight the potential value of improved infection prophylaxis in this large patient population.

Footnotes

Acknowledgments

This work was supported by Pfizer Inc. The authors thank Douglas Giganti, M.D., an employee of Pfizer, who provided comments and feedback on study methodology. The authors also thank Larry Gertzog, M.S., an employee of Policy Analysis Inc., who assisted with the statistical programming required for the conduct of this study.

Author Disclosure Statement

Ariel Berger, MPH, John Edelsberg, MD, MPH, and Gerry Oster, PhD, are employed by Policy Analysis Inc., Brookline, Massachusetts, which received financial support from Pfizer in connection with the development of the study and the manuscript. Holly Yu, MSPH is employed by Pfizer. Pfizer reviewed the study research plan and the study manuscript; data management, processing, and analyses were conducted by the authors, who made all final decisions with respect to the analyses and manuscript preparation.

References

Garner

, Jarvis

, Emori

, et al. CDC definitions for nosocomial infections. In: Olmsted

, ed. APIC Infection Control and Applied Epidemiology: Principles and Practice. St. Louis. Mosby, 1996: A1–A20.

Sorensen

, Hemmingsen

, Kallehave

, et al. Risk factors for tissue and wound complications in gastrointestinal surgery. Ann Surg, 2005; 241:654–658.

Kaya

, Yetim

, Dervisoglu

, et al. Risk factors for and effect of a one-year surveillance program on surgical site infection at a university hospital in Turkey. Surg Infect, 2006; 7:519–526.

Neumayer

, Hosokawa

, Itani

, et al. Multivariable predictors of postoperative surgical site infection after general and vascular surgery: Results from the Patient Safety in Surgery Study. J Am Coll Surg, 2007; 204:1178–1187.

Di Leo

, Piffer

, Ricci

, et al. Surgical site infections in an Italian surgical ward: A prospective study. Surg Infect, 2009; 10:533–538.

Lee

, Feng

, We

, et al. Outcome and risk factors associated with surgical site infections after cardiac surgery in a Taiwan medical center. J Microbiol Immunol Infect, 2010; 43:378–385.

Scott

, Forrest

, Feuerstein

, et al. Factors associated with postoperative infection. Infect Control Hosp Epidemiol, 2001; 22:347–351.

Herwaldt

, Cullen

, Scholz

, et al. A prospective study of outcomes, healthcare resource utilization, and costs associated with postoperative nosocomial infections. Infect Cont Hosp Epidemiol, 2006; 27:1291–1298.

Broex

, van Asselt

, Bruggeman

, van Tiel

. Surgical site infections: How high are the costs?. J Hosp Infect, 2009; 72:193–201.

10.

Graf

, Ott

, Vonberg

, et al. Surgical site infections—economic consequences for the health care system. Langenbecks Arch Surg, 2011; 396:453–459.

11.

Giles

, Bartlett

, Gelber

, et al. Tumor necrosis factor inhibitor therapy and risk of serious postoperative orthopedic infection in rheumatoid arthritis. Arthritis Rheum, 2006; 55:333–337.

12.

Paramore

, Thomas

, Knopf

, et al. Estimating costs of care for patients with newly diagnosed metastatic cancer. Clin Colorect Cancer, 2006; 6:52–58.

13.

Rubin

. Matching using propensity scores: Relating theory to practice. Biometrics, 1996; 52:249–264.

14.

Rubin

. Estimating causal effects from large data sets using propensity scores. Ann Intern Med, 1997; 127:757–763.

15.

Heckman

, Ichimura

, Todd

. Matching as an econometric evaluation estimator: Evidence from evaluating a job training programme. Rev Econ Studies, 1997; 64:605–654.

16.

Mangram

, Horan

, Pearson

, et al. Guideline for prevention of surgical site infection, 1999. Hospital Infection Control Practices Advisory Committee. Infect Cont Hosp Epidemiol, 1999; 20:250–278.

17.

Woods

, Dellinger

. Current guidelines for antibiotic prophylaxis of surgical wounds. Am Fam Physician, 1998; 57:2731–2740.

18.

Surgical Care Improvement Project. Available at: www.jointcommission.org/surgical_care_improvement_project/. Accessed September 21, 2012.

19.

Torpy

. Postoperative infections. JAMA, 2010; 303:2544.

20.

Pessaux

, Msika

, Atalia

, et al. Risk factors for postoperative infectious complications in noncolorectal abdominal surgery: A multivariate analysis based on a prospective study of 4718 patients. Arch Surg, 2003; 138:314–324.

21.

Gross

, Rapuano

, Adrignolo

, et al. Nosocomial infections: Decade-specific risk. Infect Control, 1983; 4:145–147.

22.

Nicolle

, Huchcroft

, Cruse

. Risk factors for surgical wound infection among the elderly. J Clin Epidemiol, 1992; 45:357–364.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.05 MB

0.06 MB

0.08 MB

0.00 MB