Developing an argument for bundled interventions to reduce surgical site infection in colorectal surgery Seth A. Waits, MD,a Danielle Fritze, MD,a Mousumi Banerjee, PhD,a,b Wenying Zhang, MA,a James Kubus, MS,a Michael J. Englesbe, MD,a Darrell A. Campbell, Jr, MD,a and Samantha Hendren, MD, MPH,a Ann Arbor, MI
Background. Surgical site infection (SSI) remains a costly and morbid complication after colectomy. The primary objective of this study was to investigate whether a group of perioperative care measures previously shown to be associated with reduced SSI would have an additive effect in SSI reduction. If so, this would support the use of an ‘‘SSI prevention bundle’’ as a quality improvement intervention. Methods. Data from 24 hospitals participating in the Michigan Surgical Quality Collaborative were included in the study. The main outcome measure was SSI. Hierarchical logistic regression was used to account for clustering of patients within hospitals. Results. In total, 4,085 operations fulfilled inclusion criteria for the study (Current Procedural Terminology codes 44140, 44160, 44204, and 44205). A ‘‘bundle score’’ was assigned to each operation, based on the number of perioperative care measures followed (appropriate Surgical Care Improvement Project-2 antibiotics, postoperative normothermia, oral antibiotics with bowel preparation, perioperative glycemic control, minimally invasive surgery, and short operative duration). There was a strong stepwise inverse association between bundle score and incidence of SSI. Patients who received all 6 bundle elements had risk-adjusted SSI rates of 2.0% (95% confidence interval [CI], 7.9–0.5%), whereas patients who received only 1 bundle measure had SSI rates of 17.5% (95% CI, 27.1–10.8%). Conclusion. This multi-institutional study shows that patients who received all 6 perioperative care measures attained a very low, risk-adjusted SSI rate of 2.0%. These results suggest the promise of an SSI reduction intervention for quality improvement; however, prospective research are required to confirm this finding. (Surgery 2014;155:602-6.) From the Departments of Surgerya and Biostatistics,b University of Michigan, Ann Arbor, MI
THE MICHIGAN SURGICAL QUALITY COLLABORATIVE (MSQC) is a group of 52 hospitals, focused on the quality of care delivered to surgical patients in Michigan.1-3 Because colectomy is a commonly performed procedure, and because surgical site infection (SSI) is a common and costly complication after colectomy, we chose to make the reductions of SSI after colectomy a priority.4,5 Using prospectively collected data on process measures and outcomes after colectomy, we have previously identified 6 modifiable factors, each independently associated with a reduced rate of SSI after this procedure.2,6,7 Various quality improvement Accepted for publication December 6, 2013. Reprint requests: Samantha Hendren, MD, MPH, Department of Surgery, University of Michigan, 2124 Taubman Center, 1500 E. Medical Center Drive, SPC 5343, Ann Arbor, MI 48104. E-mail: [email protected]
0039-6060/$ - see front matter Ó 2014 Mosby, Inc. All rights reserved. http://dx.doi.org/10.1016/j.surg.2013.12.004
programs which target these modifiable risk factors have contributed to the nearly 3% drop in overall morbidity rates in MSQC hospitals.8 Despite the individual efficacy of these process measures, their use in an SSI prevention bundle is a matter of debate. A randomized trial by Anthony et al9 demonstrated that the rate of SSI was increased with implementation of an SSI prevention bundle. Conversely, Bull et al10 demonstrated lower rates of SSI with an SSI prevention bundle in Australian colorectal patients. Experience on a national level has shown that improved compliance with a single process measure in a complex environment is unlikely to have an appreciable effect on the desired outcome.11 To consider implementation of a statewide SSI reduction strategy within MSQC, we considered whether to focus on implementation of individual factors or to encourage implementation of all factors simultaneously, which we have called the ‘‘colectomy bundle.’’ Before embarking on the bundle implementation approach, however, we felt it important to show that implementation of
Surgery Volume 155, Number 4
the 6 bundle elements, on a statewide basis, and on an individual hospital basis, was actually associated with a lower rate of SSI. Because SSI is a modifiable complication after colectomy, pursuing best practice is a priority within our collaborative. With this study, we have investigated whether there is a dose–response relationship between SSI and infection prevention interventions, considering whether the sum is better than the individual components. We hypothesize that increased compliance with the ‘‘colectomy bundle’’ will yield decreased rates of SSI. METHODS Setting. This is a retrospective cohort study from the MSQC, a voluntary network of 52 hospitals that collect data on surgical patients for the purpose of quality improvement. Twenty-four MSQC hospitals participated in the Colectomy Project, a focused effort to improve SSI in colectomy operations across the collaborative from 2008 to 2011. The MSQC is funded by Blue Cross Blue Shield of Michigan, a private, not-for-profit insurance company. Although Blue Cross Blue Shield provides financial support for the project, they are not involved in the policy recommendations that are developed within the collaborative. MSQC hospitals are predominantly community hospitals. Each participating hospital employs $1 person(s) to prospectively collect data on general and vascular surgery patients. Patient selection uses an algorithm designed to minimize selection bias. In addition to patient and 30-day outcome variables, a subset of hospitals (24 sites) also collected additional data about the process of care on colectomy patients (Current Procedural Terminology codes 44140, 44160, 44204, and 44205). MSQC data collection is Institutional Review Board exempt, and the current study was performed with University of Michigan Institutional Review Board review, from a limited data set derived from the MSQC database. Patients with missing data for the outcome6 or the key explanatory variables (perioperative care processes) were excluded. Primary outcomes. The primary outcome variable was development of a SSI within 30 days of operation. SSIs were defined as superficial, deep and/or organ space as previously described (Fig 1). The primary exposure variable was degree of compliance with a group of 6 perioperative measures. All bundle measures were previously investigated and found to be independently associated with increased risk of SSI.2 Bundle measures were:
Waits et al 603
Fig 1. Michigan Surgical Quality Collaborative (MSQC) surgical site infection6 prevention measure checklist.
1. Appropriate (Surgical Care Improvement Project [SCIP]-2) selection of intravenous prophylactic antibiotics; 2. Postoperative normothermia (temperature of >98.68F); 3. Oral antibiotics with mechanical bowel preparation, if used12,13 (Nichols preparation14); 4. Postoperative day 1 glucose #140 mg/dL; 5. Minimally invasive surgery; and 6. Short operative duration as defined by <100 or >100 minutes as a dichotomous outcome.
SCIP-2–compliant antibiotic choice was categorized into 3 compliance levels as has been previously described: Compliant, choices with inadequate spectrum (‘‘undertreatment group’’), and choices that were not compliant because an approved choice had an additional antibiotic added (‘‘overtreatment group’’). For the purposes of the analysis, the ‘‘overtreatment group’’ was considered to be SCIP-2 compliant. Compliance with weight-based antibiotic dosing antibiotic redosing was not considered for this analysis. Participating hospitals with <10 cases were excluded from the analysis. Of 4,085 colectomy operations performed from 2008 to 2010, 2 patients had zero bundle measures followed and were excluded from the analysis. Additional patient-specific covariates. In addition to the patient variables under study, additional patient demographic and hospital stay–specific data were analyzed and adjusted to control for individual patient risk. These variables included body mass index of >30 kg/m2, history of alcohol abuse, history of corticosteroid therapy, urgent operation, age >70 years, appropriate antibiotic timing (SCIP-1), wound class, functional status, black race, wound left open, albumin, and International Classification of Diseases, 9th edition, diagnosis code associated with the colectomy operation. Analysis. The primary analysis tested the association between each patient’s receipt of the care processes in the SSI-prevention bundle and SSI
604 Waits et al
Surgery April 2014
Table. Demographic variables for bundle analysis Variable N Age > 70 y (%) Race (% black) Functional status (%)y ASA (average group) History of alcoholism (%) Smoking history (within last year; %) BMI (kg/m2) Preoperative corticosteroids (%) Urgent procedure (% elective) Wound class1 Wound left open (%) Albumin Preoperative open wound (%) Diagnosis (%) Colon cancer Colon adenoma or polyp Other infectious or inflammatory Obstruction or ileus or volvulus or hernia Vascular insufficiency Other cancer Diverticular disease or fistula Operative technique (%) Partial colectomy with anastomosis Partial colectomy with removal of terminal ileum and anastomosis Laparoscopic partial colectomy with anastomosis Laparoscopic partial colectomy with removal of terminal ileum and anastomosis
P 1 Measure 2 Measures 3 Measures 4 Measures 5 Measures 6 Measures value* 99 35.35 18.18 16.16 0.80 3.03 18.18 34.34 5.05 80 2.47 3.03 1.08 4.04
552 45.11 14.67 8.70 0.69 3.26 17.93 39.71 4.89 87 2.32 2.18 1.24 1.63
1,179 42.15 12.30 7.46 0.61 3.82 18.91 35.24 3.90 91 2.21 0.85 1.15 1.70
1,438 37.34 9.74 4.17 0.50 3.20 19.96 32.61 4.24 95 2.17 0.21 1.16 0.76
730 33.84 7.67 2.19 0.37 3.01 19.18 29.49 2.74 99 2.10 0.00 1.10 0.27
87 31.03 5.75 2.30 0.38 2.30 24.14 24.14 1.15 100 2.06 0.00 0.98 0.00
— <.01 <.01 <.01 <.01 NS NS <.01 <.01 <.01 <.01 <.01 <.01 <.01
30.30 9.09 2.02 10.10 9.09 11.11 15.15
40.94 9.78 3.08 8.15 1.99 9.78 18.48
46.40 13.23 1.61 5.68 0.85 6.96 17.22
41.72 17.87 0.83 4.45 0.28 5.70 22.53
39.32 17.95 0.27 1.37 0.14 5.07 31.78
36.78 31.03 0.00 0.00 0.00 4.60 20.69
<.01 <.01 <.01 <.01 <.01 <.01 NS
*P < .05. yEqual to % dependent, 1 average of American College of Surgeons-National Surgical Quality Improvement Program Standard Wound Class (clean = 1, clean contaminated = 2, contaminated = 3, dirty = 4). ASA, American Society of Anesthesiologists; BMI, body mass index.
risk. To do this, each subject was assigned a score (1–6) based on how many of the process measures were administered. To test for an independent association of this variable with SSI, we recognized the need to adjust for patient risk factors and for the clustering of patients within hospitals (hierarchical generalized linear model). The rationale for this approach is that unmeasured differences between hospitals might affect SSI risk and also be associated with patterns of perioperative care. As such, a hierarchical generalized linear model was created with SSI as the outcome. Our model included the primary exposure variable and all covariates specified above. A random, hospitalspecific intercept was added to capture the heterogeneity across hospitals. Model estimates were obtained using likelihood based approach in STATA v.11.2 (College Station, TX).
RESULTS Within the 24 Michigan hospitals participating in the colectomy project, 4,085 colectomy operations were performed during the study period. The Table depicts demographic characteristics between the compliance groups. Before patientand hospital-level risk adjustment, body mass index, age >70 years, wound class, functional status, race, presence of open wound, American Society of Anesthesiologists class, albumin, and history of colon cancer were found to be significantly different when compared across the middle 3 compliance bundles (2/3/4). There was a strong association between increasing bundle compliance and lower SSI; patients receiving the entire bundle had riskadjusted SSI rate of 2.5% (95% confidence interval [CI], 1.4–4.5%). Furthermore, within each
Surgery Volume 155, Number 4
Waits et al 605
Fig 2. Risk-stratified surgical site infection (SSI) rate as a function of the number of SSI prevention measures followed (appropriate Surgical Care Improvement Project-2 antibiotics, postoperative normothermia, oral antibiotics with bowel preparation, perioperative glycemic control, minimally invasive surgery, and short operative duration).
hospital increasing bundle compliance was associated with lower SSI risk, suggesting an association not confounded by site. Also, hospitals with higher compliance with the bundle had lower SSI rates; hospitals with median 3, 4, or 5 practices had mean SSI rates of 10.0% (95% CI, 5.8– 16.9%), 8.3% (95% CI, 4.9–13.7%), and 4.1% (95% CI, 2.3–7.2%), respectively (Fig 2). DISCUSSION Postoperative SSI is a common and morbid complication for patients after colectomy. Although the MSQC had initially focused on quality improvement through individual interventions, we wanted to build a case for bundling interventions to improve compliance and more efficiently reduce SSI. Previous studies considering bundled measures for prevention of SSI have not conclusively demonstrated the efficacy of these interventions.9,10 In this study, we sought to demonstrate the potential of a bundle of process measures to bring SSI rates down to previously unobtainable levels. We found a progressive increase in the implementation of the bundle elements was associated with a stepwise decrease in SSI. The medical literature has demonstrated efficacy of bundling anti-infectious interventions. Bundled measures aimed at reducing the incidence of intensive care unit catheter related blood
stream infections were demonstrated to nearly eliminate this risk.15,16 Similar dramatic improvements have also been seen in the prevention of ventilator associated pneumonia.17 The surgical literature concerning bundling has been less concrete. A recently published, randomized, clinical trial of transabdominal colorectal patients examined the use of bundled interventions based on evidence-supported interventions.9 The study concluded that bundling did not reduce SSI and formal testing of bundled interventions should occur before formal implementation. In contrast, our study makes a strong case for the efficacy of such a strategy utilizing evidence-based treatment recommendations from our own collaborative. Limitations of this work deserve consideration. Many patients who undergo urgent operations may not be candidates for all process measures. For example, surgical candidates with complex surgical histories may present difficulties with adherence to bundle elements such as laparoscopic approach and shorter operative duration. Despite this, the rapid rise in utilization of laparoscopic surgery has demonstrated that many contraindications remain relative. The inherent challenges of urgent colectomy operations are also difficult to control. Although priority (elective versus urgent) was used as a covariate in our final model, may have introduced some selection bias into our analysis.
606 Waits et al
Additionally, surgeon identifiers were not available for this analysis. Although our primary outcome was controlled for patient- and hospital-level variation, surgeon-specific practice patterns may also introduce additional selection bias. Finally, as demonstrated in the Table, it is clear that compliance with process measures is more common in a select group of ‘‘low-risk’’ patients. This highlights the observation that patients who received all bundle measures were more likely to undergo elective operations in a controlled setting. Although these data are limited by the inherent bias of retrospectively obtained data, it does provide an example of the potential efficacy within the framework of a randomized, controlled trial, providing hope that SSI rates could be reduced to previously unobtainable levels using a bundled approach. There are important implications of this study in our collaborative. Providing an association between increasing bundle compliance and decreasing SSI provides a framework for implementation of quality improvement measures. Our findings may provide direction for marketing and education on the topic and a rationale for hospitals to develop a bundle implementation plan. Development of quality improvement tools such as a checklist will also enhance cooperation and participation in the initiative (Fig 1).18 Innovative approaches may be necessary to improve institutional compliance, including involving stakeholders outside of the surgical staff. An example of this approach includes involving anesthesia staff, which has recently been shown to improve antibiotic administration compliance by 15%.19 In conclusion, our analysis provides a case for implementing a bundle of best practices for minimizing SSI in patients undergoing colon surgery in Michigan. We have shown that increasing compliance to this bundle is associated with rates of infection nearly 80% lower than baseline values. Our work builds a scaffold for compliance and provides a reachable target for a significant number of operative cases across the state. As we move forward within the state of Michigan, we will now focus on education, implementation support, and close follow-up of outcomes.
REFERENCES 1. Henke PK, Kubus J, Englesbe MJ, Harbaugh C, Campbell DA. A statewide consortium of surgical care: a longitudinal
Surgery April 2014
investigation of vascular operative procedures at 16 hospitals. Surgery 2010;148:883-9. Hendren S, Fritze D, Banerjee M, et al. Antibiotic choice is independently associated with risk of surgical site infection after colectomy: a population-based cohort study. Ann Surg 2013;257:469-75. Campbell DA Jr, Kubus JJ, Henke PK, Hutton M, Englesbe MJ. The Michigan Surgical Quality Collaborative: a legacy of Shukri Khuri. Am J Surg 2009;198(5 Suppl.):S49-55. Wick EC, Hirose K, Shore AD, et al. Surgical site infections and cost in obese patients undergoing colorectal surgery. Arch Surg 2011;146:1068-72. Mahmoud NN, Turpin RS, Yang G, Saunders WB. Impact of surgical site infections on length of stay and costs in selected colorectal procedures. Surg Infect 2009;10:539-44. Edmiston CE, Spencer M, Lewis BD, et al. Reducing the risk of surgical site infections: did we really think SCIP was going to lead us to the promised land? Surg Infect 2011;12:169-77. Krapohl GL, Phillips LR, Campbell DA Jr, et al. Bowel preparation for colectomy and risk of Clostridium difficile infection. Dis Colon Rectum 2011;54:810-7. Share DA, Campbell DA, Birkmeyer N, et al. How a regional collaborative of hospitals and physicians in Michigan cut costs and improved the quality of care. Health Aff 2011; 30:636-45. Anthony T, Murray BW, Sum-Ping JT, et al. Evaluating an evidence-based bundle for preventing surgical site infection: a randomized trial. Arch Surg 2011;146:263-9. Bull A, Wilson J, Worth LJ, et al. A bundle of care to reduce colorectal surgical infections: an Australian experience. J Hosp Infect Aug 2011;78:297-301. Meeks DW, Lally KP, Carrick MM, et al. Compliance with guidelines to prevent surgical site infections: as simple as 1-2-3? Am J Surg 2011;201:76-83. Toneva GD, Deierhoi RJ, Morris M, et al. Oral antibiotic bowel preparation reduces length of stay and readmissions after colorectal surgery. J Am Coll Surg 2013;216:756-62. Englesbe MJ, Brooks L, Kubus J, et al. A statewide assessment of surgical site infection following colectomy: the role of oral antibiotics. Ann Surg 2010;252:514-9. Nichols RL, Broido P, Condon RE, Gorbach SL, Nyhus LM. Effect of preoperative neomycin-erythromycin intestinal preparation on the incidence of infectious complications following colon surgery. Ann Surg 1973;178:453-62. Pronovost P. Interventions to decrease catheter-related bloodstream infections in the ICU: the Keystone Intensive Care Unit Project. Am J Infect Control 2008;36: S171.e1-5. Berenholtz SM, Pronovost PJ, Lipsett PA, et al. Eliminating catheter-related bloodstream infections in the intensive care unit. Critical Care Med 2004;32:2014-20. Resar R, Pronovost P, Haraden C, Simmonds T, Rainey T, Nolan T. Using a bundle approach to improve ventilator care processes and reduce ventilator-associated pneumonia. Jt Comm J Qual Patient Saf 2005;31:243-8. Haynes AB, Weiser TG, Berry WR, et al. A surgical safety checklist to reduce morbidity and mortality in a global population. N Engl J Med 2009;360:491-9. Fahy BG, Bowe EA, Conigliaro J. Perioperative antibiotic process improvement reaps rewards. Am J Med Qual 2011;26:185-92.