Cost and Consequences of Surgical Site Infections: A Call to Arms

Abstract

Despite decades of attention and advances in medical treatments, surgical site infections (SSI) remain a significant problem to our patients. Surprisingly, the reported incidence of SSI has not changed appreciably in the past five decades, although not surprisingly the timing to diagnosis (inpatient vs. outpatient) has changed. Although the patient consequences have not escaped our notice, until recently the costs associated with SSI have been difficult to ascertain. In this review, we discuss the relevant history of SSI monitoring, the incidence as well as the costs and consequences associated with this complication.

As surgeons, we understand well that our craft entails risks of complications with associated costs and consequences. As an example, the association between surgical procedures and surgical site infection (SSI) has long been recognized, inheriting as we have the Halstedian legacy of aseptic technique. We have all witnessed the individual, sometimes devastating, effects that SSI can have on our patients. When approaching the complication of SSI, however, it is also necessary to recognize the broader scope of the problem across the surgical population.

The first “serious” attempt to look more broadly than the individual must be credited to Dr. Peter Cruse and his team in Alberta, Canada. In the late 1960s, they recognized the need to develop a method of surveillance for SSI to not only quantify the problem, but also potentially provide long-term data useful for practice-changing efforts at SSI reduction. Cruse's principal and perhaps most crucial observation was that retrospective data are insufficient to meaningfully study SSI, because hospital records were (and mostly still are) simply too inaccurate for studies of SSI.

Cruse and his team proposed a prospective wound surveillance program with the primary goal of accurately assessing hospital SSI rates, and calculated individual SSI data for each surgeon, ward, and service. These data could, they reasoned, ultimately be used to identify factors that influence SSI rates and possibly guide efforts at rate reduction. In many ways, the Foothills Hospital study and process that they developed is the precursor to the SSI component of our American College of Surgeons (ACS) National Surgical Quality Improvement Project (NSQIP).

After 10 years of prospective data collection, largely by Rosemary Foord, a surgical nurse who personally examined each wound during the inpatient period and performed 28-day follow-up on each wound with the operating surgeon, the overall SSI rate for the hospital was found to be 4.7% [1]. By recognizing the importance of wound contamination, however, they determined that it was the rate of infection of “clean” incisions that was most useful as a marker of infection control and therefore the best surrogate for aseptic practices of the institution, department, and surgeon. Wound infection rates did vary among departments, but the rate of SSI in clean wounds was consistent at approximately 1.5% [1].

This focus on clean incisions, as opposed to clean-contaminated, contaminated, or dirty incisions, also revealed the impact of endogenous contamination at operation over exogenous factors. The investigators then were able to parse out those factors that had a positive impact on SSI rates, including pre-operative showering with hexachloroprene, depilation using electric clippers, and minimizing the clipped area [1]. Advanced age (>66 y), patient co-morbidities (diabetes mellitus, obesity, and malnutrition), increased pre-operative length of stay (LOS), and increasing operative time, were all negative contributors [1].

One of the most important conclusions of Cruse work was that providing regular feedback to surgeons on their SSI rates, as well as the rates of their peers, has a strong positive impact. In their single-center effort, overall SSI rates—as well as clean incision infection rates—fell by almost 50% within six months of initiating the project [1]. While the conclusions drawn in this study may seem obvious to us now, it was the methodical surveillance of surgical incisions over time that ultimately provided these insights.

Of interest, the current landscape of SSI seems sadly similar to that reported by Dr. Cruse and his study group over 35 years ago. A recent cross-sectional study using a risk-adjusted NSQIP data set containing more than 600,000 general and vascular surgery cases confirms varying rates of SSIs by wound classification [2]. Clean wounds were associated with an overall SSI rate of 2.58%, most of which were superficial. Rates of SSI increase, as expected, with worsening class, with clean-contaminated, contaminated, and dirty wounds, respectively, associated with SSI rates of 6.67%, 8.61%, and 11.80% [2]. Although the current process to collect SSI information for NSQIP may vary slightly from the original work by the Alberta group, these NSQIP data suggest that despite advances in antibiotic agents, surgical technology, changes in clinical guidelines, evidence-based prophylaxis, and increased focus on patient safety, we continue to struggle with the same many decades old problem of SSI.

SSI and Hospital Re-Admission Rates

One factor that has had significant impact on SSI diagnosis is the overall reduction in hospital LOS. Patients are no longer routinely pre-admitted for elective surgical procedures and generally remain in the hospital for a shorter period after operation. This is in part because of changes in pay structure with the advent of Diagnosis Related Group (DRG) reimbursement, as well as the advancement of minimally invasive surgical procedures. The diagnosis of SSI, therefore, requires post-discharge surveillance even more now than ever.

A population-based retrospective cohort study in Ontario, Canada, including more than 600,000 surgical cases from multiple specialties, reported an overall 30-day SSI rate of 13.5% confirms this [3]. Interestingly, more than half of SSIs (58%) were diagnosed after hospital discharge (median 10 days) [3]. This study also suggests several implications for the healthcare system at large. Post-discharge SSIs were associated with increased re-operation rates, emergency department visits, hospital re-admissions, and death. Previously validated risk indices for health-care–associated infection (HAI), however, based on patient characteristics and procedure length, failed to accurately predict patients at risk for the development of post-discharge SSI [3]. Independent predictors included rural residence, alcoholism, diabetes mellitus, obesity, and, curiously, shorter procedure duration. These predictors contributed only weakly to post-discharge SSI, however, and patients with a post-discharge SSI were more comparable to patients with no post-operative infection than they were to those with an in-hospital SSI [3].

Similar results were found in a prospective cohort study of 1,506 general surgery patients in Spain. This single center study, conducted at a tertiary center, showed an overall SSI rate of 14.8% [4]. The authors compared patients with a diagnosis of SSI during the index hospitalization (7.7%) with those with an SSI diagnosis after discharge (7.1%). While their identification of risk factors for development of in-hospital SSI was consistent with known models (American Society of Anesthesiologists [ASA] physical status classification, severity of illness, National Nosocomial Infections Surveillance Index—now the National Health Safety Network of the Center for Disease Control [CDC], and Study on the Effectiveness of Nosocomial Infection Control [SENIC] index), they could not identify independent risk factors for SSI after discharge [4].

Patients in whom post-discharge SSI developed were, again, more comparable to those who never had SSI than to those in whom SSI developed before discharge [4]. Notably, in this study, most post-discharge SSI occurred in clean cases (e.g., breast surgical procedure, vascular operation, and herniorrhaphy). Body mass index was a significant factor for all SSI, but notably, a surgeon's SSI rate was most strongly associated with development of in-hospital SSI, not post-discharge SSI [4].

Diagnosis of SSI after hospital discharge and the resultant increase in hospital re-admission rates because of this complication were the focus of a recent retrospective analysis of a single center's NSQIP database [5]. Of 3,663 patients undergoing inpatient general surgical procedures, SSI developed in 9.9%; 48% of these patients received an SSI diagnosis after hospital discharge and 54% of this patient subset required re-admission [5]. Again, these patients showed a trend toward fewer co-morbidities, more favorable ASA classification, and higher likelihood of clean or clean-contaminated wound classification than patients with a diagnosis of SSI in hospital [5]. Notably, re-admission rates varied according to method and timing of evaluation. Of patients presenting to the emergency department, 50% were re-admitted, compared with 6% presenting to clinic [5]. Re-admission was also strongly associated with time of day, with 54.5% patients presenting after 5 pm being re-admitted versus 7.5% of patients evaluated between 8 am and 5 pm [5]. Some of these findings may be because of local practice patterns; however, it is clear from these data that SSIs are real contributors to the economic burden of healthcare and impact on the patient.

Using a standard 30-day surveillance measure, Gibson et al. [5] found that delayed superficial and deep incisional SSIs were most likely to be discovered in the first week after discharge. This is consistent with previous studies and, given the paucity of proven risk models for delayed SSIs, offers a potential avenue for improvement in our clinical outcomes. These data suggest that careful surveillance in the early post-discharge period might replace the “in-hospital surveillance” that was such a part of operations before the advent of DRGs and minimally invasive surgical procedures.

Nevertheless, recent work by Merkow et al. [6] did not identify a peak in the timing of re-admission for SSI, and early and late re-admissions did not differ in cause. Because re-admissions occurred steadily over a 30-day period in their analysis, there was no inflection point above which a longer initial hospitalization would not lead to a reduction in unplanned re-admission rate [6]. Whatever the time course, it seems prudent to include careful monitoring post-discharge, with the premise that early recognition of post-discharge SSI might allow early intervention and decrease re-admission rates, morbidity, and thus overall costs. Post-discharge wound surveillance programs, using available electronic medical records and telemedicine technologies, as described by Shah et al. in this issue, could offer a medium to facilitate such future efforts.

Patient-Level Cost of SSI

No discussion of the costs of SSI would be complete without examining the human costs paid by affected patients. For all adults with SSI, we can imagine a series of potential human costs suffered because of SSI. While hospital re-admission rate has been suggested as an appropriate surrogate for patient-absorbed morbidity, this fails to account for a wide range of consequences. Unfortunately, there are few useful data on the impact of SSI on patients beyond re-admission rate and death. Pain, loss of income, and reduced quality of life are not accounted for in current models. In addition, certainly the economic and non-economic burden to patients' families is even more difficult to quantify.

Gheorghe et al. [7] performed a systematic review of the current literature on SSI health-related quality of life and found no consistent methodology employed for collecting this type of data. They found only a handful of articles that adhered to evidence-based practice identifying utility values, and most incorporated values that were irrelevant to post-operative SSI [7]. All studies reviewed were consistent in the message that SSI posed a significant burden to patients, but there was no reliable SSI disutility estimate, and the heterogeneous burden of SSI contributes greatly to this problem. They suggest that future studies focus on specific types of SSI, use more consistent methods of data collection, and widen the time frame in which we look at patient-level costs [7].

One group disproportionately affected by SSI is adults older than age 65. They bear a higher burden of the human cost, as well as the financial costs, of this common post-operative complication, and they represent an increasingly larger percentage of the US population. In a retrospective matched outcomes study, Kaye et al. [8] examined the impact of SSI on patients age 65 and older. At one tertiary hospital and seven community hospitals in North Carolina, 561 patients with a diagnosis of deep incisional or organ space SSI were compared with 576 patients without any SSI. Of elderly patients with SSI, 8.6% died of the complication within 90 days, representing an odds ratio of 3.51. This is higher than the two-fold estimate of death from SSI for non-elderly adults.

Poor pre-operative functional status and age over 80 years were significant predictors of death in this study [8]. In addition, there was an average increase in LOS of greater than 14 days, including re-admission, compared with seven to 10 days for all-comers. Rather than an increase in total costs of $3,000 to $29,000 per SSI, because estimates vary for all adult surgical patients, elderly patients had an average increase of greater than $40,000 [8].

The ultimate cost of any surgical complication is, of course, death, and there are several studies that address this. First is a single-center, case matched follow-up study by Kirkland et al. [9] that shows roughly double the deaths in patients with a diagnosis of SSI, compared with SSI-free patients (7.8% vs. 3.5%). In another study of case-fatality of SSI, Astagneau et al. [10] used data obtained from a French SSI surveillance network, and of 38,973 patients, SSI developed in 3.4%. Overall, 1.5% of patients in the study died, but a significantly higher percentage of patients with SSI died than of patients without (5.8% vs. 1.3%). Of all deaths, 38% were thought to be directly attributable to SSI, and these were typically deep incisional and organ space infections [10]. Similar to results from Kaye et al. [8], patients over the age of 65 in that study were observed to have a higher death rate.

Economic Burden of SSI

To put the issue of SSI and re-admission into a larger perspective, Merkow et al. [6] recently evaluated reasons for unplanned post-operative re-admission using ACS NSQIP data from almost 500,00 patients from 346 hospitals. They showed re-admission rates after surgical discharge of 5.7%, and the number one cause was SSI (19.5%), followed by ileus or obstruction (10.3%)⁶. Other significant causes included bleeding/anemia, pulmonary complications, venous thromboembolism, dehydration or malnutrition, and sepsis [6]. Importantly, only 2.3% of unplanned readmissions could be attributed to an exacerbation of previously identified problems. Given that one in five of the roughly 6% of patients who have surgical re-admission are re-admitted as a consequence of SSI, SSI has been identified as a huge potential economic burden.

Zimlichman et al. [11] recently estimated costs of the five most significant HAIs: SSI, central line associated blood stream infection (CLABSI), catheter associated urinary tract infection (CAUTI), ventilator associated pneumonia (VAP), and Clostridium difficile infection by meta-analysis of 26 studies conducted between 1998 and 2013. They estimate that SSI is the third most costly HAI on an individual case basis, at $20,785, mostly because of the profound impact SSI has on LOS. The annual burden of these five HAIs they estimated to be $9.8 billion, with SSI accounting for the largest overall contribution at 33.7% of that cost. It is interesting that there has been a great deal of energy expended to stamp out CAUTI in our hospitals, which in comparison are numerous but relatively cheap events, while comparably little attention is being paid to outpatient SSI. As payers increasingly deny reimbursement for preventable complications such as SSI, it is imperative that we continue to drive initiatives aimed at reducing SSI rates.

The primary mechanism by which HAI impacts cost is from increased LOS requirements. Hassan et al. [12], in a retrospective analysis from the New Jersey Universal Billing database, sought to provide an analysis of the impact of HAI in 6.4% of admitted patients with a diagnosis of HAI. They estimate that the increased LOS directly attributable to HAI is 3.3 days. Each hospital day for patients with HAI had an associated cost of $3,144, a daily cost 24% higher than an infection-free hospital day. This amounts to a total additional cost of at least $10,375, with a projected annual cost burden of $16.6 billion.

Monetary costs of HAI and specifically SSI notoriously have been hard to evaluate because of lack of consensus on appropriate methodology. By simply comparing costs of treatment of patients with and without HAI, even with appropriate case matching, we fail to fully account for patient heterogeneity. There is also the issue of endogeneity bias, because unmeasured factors may simultaneously impact multiple measured factors. For example, LOS and rate of HAI can confound each other: Infection increases the likelihood of increased LOS, just as LOS increases likelihood of infection [13]. Such looped causality between dependent and independent variables must be addressed in future studies. In a systematic review of available literature on economic analyses of HAI between 2001 and 2004, Stone et al. [14] noted a trend toward increased interest in the subject.

Nevertheless, there is little standardization of methodology in determining the incremental costs solely attributable to HAI, which likely explains the observed variation in cost estimates. The SENIC data from 1985 cited a $5–$10 billion annual impact of HAI, while the CDC has reported costs of $5 billion, but estimates elsewhere in the literature have reached as high as $29 billion per year [12]. Future studies will require the continued careful scrutiny and ingenuity of our statistician colleagues to draw meaningful conclusions about the costs of HAI and SSI.

Conclusions

This review should serve as a surgical “call to arms” on the cost and consequences of SSI. The SSI occur frequently after discharge, and current data clearly show that the models that have been used historically to predict patients at risk for HAI are not predictive for patients in whom post-discharge SSI develops. Post-discharge SSIs are now the most common cause for re-admission for our surgical patients, and these infections are among the most expensive HAI. This is a significant burden on our patients and our healthcare system.

As surgeons, we have been included in significant efforts to prevent VAP, CLABSI, and CAUTI in our patients, but when it comes to SSI, it has been more or less “business as usual.” Our well-intended efforts have been focused on meeting our NSQIP metrics—timeliness of antibiotic administration, correct antibiotic agents for the surgical indication, discontinuation of antibiotic agents within 24 hours, and so on. Lately, there are even those who have taken great fascination with what kind of hats and shoe covers we should be wearing and who launders our scrub clothing. The simple fact remains: Despite methodologic variations in reported incidence of SSI in the literature, these incidence rates have not declined appreciably during the roughly five decades since the pioneering work of Dr. Cruse and his team began [1]. We must pull our collective heads out of the sand and aggressively study and understand who will be most at risk for post-discharge SSI, how to make the diagnosis early, and implement therapy to obviate both the morbidity and the associated costs of such infections.

Footnotes

Author Disclosure Statement

No competing financial interests exist.

References

Cruse

. Wound infection surveillance. Rev Infect Dis, 1981; 3:734–737.

Ortega

, Rhee

, Papandria

, et al. An evaluation of surgical site infections by wound classification system using the ACS-NSQIP. J Surg Res, 2012; 174:33–38.

Daneman

, Lu

, Redelmeier

. Discharge after discharge: Predicting surgical site infections after patients leave hospital. J Hosp Infect, 2010; 75:188–194.

Delgado-Rodriguez

, Gomez-Ortega

, Sillero-Arenas

, Llorca

. Epidemiology of surgical-site infections diagnosed after hospital discharge: A prospective cohort study. Infect Control Hosp Epidemiol, 2001; 22:24–30.

Gibson

, Tevis

, Kennedy

. Readmission after delayed diagnosis of surgical site infection: A focus on prevention using the American College of Surgeons National Surgical Quality Improvement Program. Am J Surg, 2014; 207:832–839.

Merkow

, Ju

, Chung

, et al. Underlying reasons associated with hospital readmission following surgery in the United States. JAMA, 2015; 313:483–495.

Gheorghe

, Moran

, Duffy

, et al. Health utility values associated with surgical site infection: A systematic review. Value Health, 2015; 18:1126–1137.

Kaye

, Anderson

, Sloane

, et al. The effect of surgical site infection on older operative patients. J Am Geriatr Soc, 2009; 57:46–54.

Kirkland

, Briggs

, Trivette

, et al. The impact of surgical-site infections in the 1990s: Attributable mortality, excess length of hospitalization, and extra costs. Infect Control Hosp Epidemiol, 1999; 20:725–730.

10.

Astagneau

, Rioux

, Golliot

, et al. Morbidity and mortality associated with surgical site infections: Results from the 1997–1999 INCISO surveillance. J Hosp Infect, 2001; 48:267–274.

11.

Zimlichman

, Henderson

, Tamir

, et al. Health care-associated infections: A meta-analysis of costs and financial impact on the US health care system. JAMA Intern Med, 2013; 173:2039–2046.

12.

Hassan

, Tuckman

, Patrick

, et al. Cost of hospital-acquired infection. Hosp Top, 2010; 88:82–89.

13.

Kilgore

, Ghosh

, Beavers

, et al. The costs of nosocomial infections. Med Care, 2008; 46:101–104.

14.

Stone

, Braccia

, Larson

. Systematic review of economic analyses of health care-associated infections. Am J Infect Control, 2005; 33:501–509.