Search results

Our initial search yielded 1,464 articles. After removing duplicates, we screened the titles and/or abstracts of 1,044 unique references. Of these, we reviewed the full texts of 57 articles, and we ultimately included 40 studies in our qualitative review, as summarized in the PRISMA flow diagram (Fig. 1).

Fig. 1.

PRISMA flow diagram of literature search and study selection.

Study characteristics

Among these 40 selected studies, 3 (8%) were randomized-controlled trials,^{Reference Anthony, Murray and Sum-Ping12–Reference Ruiz-Tovar, Llavero, Morales and Gamallo14} 29 (72%) were prospective studies,^{Reference Albert, Bataller and Masroor15–Reference Ohman, Wan and Guthrie43} and 8 (20%)^{Reference Bell44–Reference Keenan, Speicher, Thacker, Walter, Kuchibhatla and Mantyh51} were retrospective trials. Most studies (35, or 88%) were single-center studies, and 5 (12%) were multicenter trials. Study locations included the United States (n = 28), Spain (n = 3), Canada (n = 2), Japan (n = 2), England (n = 1), the Netherlands (n = 1), Italy (n = 1), Singapore (n = 1), and Australia (n = 1). Most studies (n = 33) reported overall SSI incidence (Table 1). Several studies did not report overall SSI incidence but did report superficial SSI incidence (n = 4),^{Reference Ruiz-Tovar, Llavero, Morales and Gamallo14,Reference Berenguer, Ochsner, Lord and Senkowski38–Reference Yamamoto, Morimoto and Kita40} and/or organ-space SSI incidence (n = 3)^{Reference Forbes, Stephen and Harper41,Reference Mammo, Peeples, Grodsky, Honaker and Wasvary42,Reference Keenan, Speicher, Thacker, Walter, Kuchibhatla and Mantyh51} (Table 1).

Table 1.

Studies of Colorectal Infection Prevention Bundles and Their Effect on Surgical Site Infection (SSI) Rates

Note. aOR, adjusted odds ratio; O/E, observed/expected; NR, not reported.

^a SSI outcomes are reported as superficial SSI for references 19, 41, 42, and 43; organ-space SSI for references 11, 44, and 45; and overall SSI for the remainder of the studies.

^b Assigned based on Downs and Black score as follows: high (score ≤19); medium (score 20–24); low (score 25–31); indeterminate, if total score could not be calculated (studies did not report sufficient raw data).

Study quality and bias assessment

The risk of bias was low in 5 studies (12%), medium in 14 studies (35%), high in 17 studies (43%), and indeterminate in 4 studies (10%) (Table 1). The average Downs and Black individual study bias assessment score was 20.82 (range, 15–26). Of the 36 studies that provided enough raw data to allow power calculations, 18 (50%) were powered at <80%.

Composition of colorectal infection prevention bundles

There was significant heterogeneity in bundle composition across studies. The number of interventions recommended by the American College of Surgeons and Surgical Infection Society (ACS/SIS) clinical guidelines for SSI prevention^{Reference Ban, Minei and Laronga5} varied between 2^{Reference Yamamoto, Morimoto and Kita40} and 13^{Reference Connolly, Foppa, Kazi, Denoya and Bergamaschi19,Reference Hewitt, Tannouri and Burkhart26} per bundle. The most common ACS/SIS bundle interventions were (1) intravenous antibiotic prophylaxis, present in 40 studies (100%), as standard of care, or with special emphasis on antibiotic choice and dosing as part of the intervention bundle; (2) normothermia maintenance in 32 studies (80%); (3) perioperative glucose control in 22 studies (55%); (4) appropriate hair removal in 20 studies (50%); (5) mechanical and oral bowel preparation in 18 studies (45%); (6) preoperative bathing with chlorhexidine in 16 studies (40%); (7) standardized postoperative dressing removal and wound care in 16 studies (40%); (8) wound closure protocol including glove with or without gown change and separate instrument tray in 15 studies (38%); (9) use of wound protectors in 11 studies (28%); (10) use of supplemental oxygen postoperatively in 7 studies (18%); (11) cutaneous antisepsis with a solution containing chlorhexidine gluconate and isopropyl alcohol in 6 studies (15%); (12) preoperative smoking cessation in 4 studies (10%); (13) wearing appropriate surgical attire, restricted to the operating room (OR) in 2 studies (5%). Other interventions included minimizing OR traffic,^{Reference Connolly, Foppa, Kazi, Denoya and Bergamaschi19,Reference Crolla, van der Laan, Veen, Hendriks, van Schendel and Kluytmans20,Reference Elia-Guedea, Cordoba-Diaz de Laspra, Echazarreta-Gallego, Valero-Lazaro, Ramirez-Rodriguez and Aguilella-Diago22,Reference Keenan, Speicher and Nussbaum46,Reference Keenan, Speicher, Thacker, Walter, Kuchibhatla and Mantyh51} peritoneal lavage with antibiotics,^{Reference Ruiz-Tovar, Llavero, Morales and Gamallo14,Reference Ohman, Wan and Guthrie43} wound irrigation with antibiotic solution^{Reference Albert, Bataller and Masroor15,Reference Hewitt, Tannouri and Burkhart26} or with saline before closure,^{Reference Benlice16,Reference Gorgun, Rencuzogullari and Ozben23,Reference Lutfiyya, Parsons and Breen29,Reference Perez-Blanco, Garcia-Olmo, Maseda-Garrido, Najera-Santos and Garcia-Caballero31,Reference Yamamoto, Morimoto and Kita40} postoperative daily chlorhexidine showers,^{Reference Cima, Dankbar and Lovely18,Reference DeHaas, Aufderheide, Gano, Weigandt, Ries and Faust21,Reference Russell, Chung and Riad32} and daily wound cleaning with chlorhexidine.^{Reference Russell, Chung and Riad32,Reference Keenan, Speicher and Nussbaum46,Reference Keenan, Speicher, Thacker, Walter, Kuchibhatla and Mantyh51}

Adherence to colorectal SSI prevention bundles

Adherence to colorectal SSI prevention bundles was reported in 28 studies (70%). Composite adherence rates to the entire bundle, reported in 7 studies (25%), ranged from 19% in a 9-intervention bundle^{Reference Tanner, Kiernan and Hilliam33} to 92% in a 3-intervention bundle.^{Reference Berenguer, Ochsner, Lord and Senkowski38} In terms of adherence to individual bundle interventions, several studies reported adherence rates ≥90% to 1 or more Surgical Care Improvement Project (SCIP) measures: correct preoperative antibiotic prophylaxis, hair removal by clippers, normothermia maintenance, with or without preoperative glycemic control.^{Reference Anthony, Murray and Sum-Ping12,Reference Kwaan, Weight and Carda13,Reference Cima, Dankbar and Lovely18,Reference Crolla, van der Laan, Veen, Hendriks, van Schendel and Kluytmans20,Reference Hedrick, Heckman, Smith, Sawyer, Friel and Foley24,Reference Hewitt, Tannouri and Burkhart26,Reference Pastor, Artinyan, Varma, Kim, Gibbs and Garcia-Aguilar30,Reference Weiser, Gonen and Usiak35,Reference Wick, Galante and Hobson36,Reference Berenguer, Ochsner, Lord and Senkowski38,Reference Forbes, Stephen and Harper41,Reference Keenan, Speicher and Nussbaum46,Reference Tillman, Wehbe-Janek, Hodges, Smythe and Papaconstantinou49,Reference Keenan, Speicher, Thacker, Walter, Kuchibhatla and Mantyh51} In contrast, other studies reported lower adherence rates to these interventions, with antibiotic prophylaxis adherence ranging from 45% to 85%,^{Reference Albert, Bataller and Masroor15,Reference Benlice16,Reference Perez-Blanco, Garcia-Olmo, Maseda-Garrido, Najera-Santos and Garcia-Caballero31,Reference Tanner, Kiernan and Hilliam33,Reference Mammo, Peeples, Grodsky, Honaker and Wasvary42,Reference Bert, Giacomelli and Amprino52} and normothermia ranging from 33% to 77%.^{Reference Albert, Bataller and Masroor15,Reference Hedrick, Turrentine and Smith25,Reference Liau, Aung and Chua28,Reference Perez-Blanco, Garcia-Olmo, Maseda-Garrido, Najera-Santos and Garcia-Caballero31,Reference Tanner, Kiernan and Hilliam33} Interventions related to surgical wound closure (ie, glove with or without gown change, separate instrument tray, redraping, and/or wound lavage) had adherence rates ≥85% in several studies.^{Reference Gorgun, Rencuzogullari and Ozben23,Reference Hewitt, Tannouri and Burkhart26,Reference Ghuman, Chan, Karimuddin, Brown, Raval and Phang45} Adherence to preoperative bathing with chlorhexidine and/or use of chlorhexidine-impregnated wipes ranged between 89% and 92%.^{Reference Benlice16,Reference Bert, Giacomelli and Amprino52}

Three studies reported SSI incidence by bundle adherence: Jaffe et al^{Reference Jaffe, Meka and Semaan27} observed an overall SSI rate of 0.9% versus 16% when there was adherence to all 6 bundle elements versus only 0–2 bundle elements, respectively. Vu et al^{Reference Vu, Collins and Seese34} reported an overall SSI rate of 4.5% versus 6.5% in the highest versus lowest adherence quartile, respectively (P = .049). Waits et al^{Reference Waits, Fritze and Banerjee50} reported an overall SSI rate of 2.5% for adherence to all 6 bundle interventions versus 17.5% for adherence to only 1 bundle intervention.^{Reference Waits, Fritze and Banerjee50}

Implementation strategies for colorectal SSI prevention bundles

The following implementation strategies were most commonly used: (1) multidisciplinary collaborative team or steering committee, led by a colorectal surgery champion, with members from surgery, anesthesia, quality improvement, infection control, infectious disease, pharmacy, hospital administrators^{Reference Albert, Bataller and Masroor15–Reference Connolly, Foppa, Kazi, Denoya and Bergamaschi19,Reference DeHaas, Aufderheide, Gano, Weigandt, Ries and Faust21,Reference Hedrick, Heckman, Smith, Sawyer, Friel and Foley24,Reference Pastor, Artinyan, Varma, Kim, Gibbs and Garcia-Aguilar30,Reference Vu, Collins and Seese34,Reference Weiser, Gonen and Usiak35,Reference Forbes, Stephen and Harper41,Reference Keenan, Speicher and Nussbaum46,Reference Tillman, Wehbe-Janek, Hodges, Smythe and Papaconstantinou49} ; (2) surgical leadership spearheading bundle implementation^{Reference Benlice16,Reference Cima, Dankbar and Lovely18,Reference Connolly, Foppa, Kazi, Denoya and Bergamaschi19,Reference Hewitt, Tannouri and Burkhart26,Reference Berenguer, Ochsner, Lord and Senkowski38,Reference Keenan, Speicher and Nussbaum46} ; (3) hospital administration leadership support for process enforcement and resource allocation^{Reference Crolla, van der Laan, Veen, Hendriks, van Schendel and Kluytmans20,Reference Wick, Hobson and Bennett37} ; (4) educational meetings with relevant frontline clinicians^{Reference Connolly, Foppa, Kazi, Denoya and Bergamaschi19,Reference Hedrick, Heckman, Smith, Sawyer, Friel and Foley24,Reference Hewitt, Tannouri and Burkhart26,Reference Wick, Hobson and Bennett37,Reference Forbes, Stephen and Harper41,Reference Tillman, Wehbe-Janek, Hodges, Smythe and Papaconstantinou49} ; (5) assignment of responsibility and accountability for key bundle interventions to most relevant clinicians^{Reference Hedrick, Heckman, Smith, Sawyer, Friel and Foley24,Reference Hedrick, Turrentine and Smith25,Reference Lutfiyya, Parsons and Breen29} ; (6) use of checklist to record process measures and improve adherence with prevention bundle^{Reference Bull, Wilson and Worth17,Reference Connolly, Foppa, Kazi, Denoya and Bergamaschi19,Reference Hedrick, Turrentine and Smith25,Reference Hewitt, Tannouri and Burkhart26,Reference Wick, Hobson and Bennett37,Reference Reames, Krell, Campbell and Dimick39} ; (7) use of electronic order sets and automatic reminders to facilitate bundle related workflows^{Reference Cima, Dankbar and Lovely18,Reference Hedrick, Heckman, Smith, Sawyer, Friel and Foley24,Reference Russell, Chung and Riad32,Reference Wick, Galante and Hobson36,Reference Keenan, Speicher and Nussbaum46} ; (8) standardization of clinical practices and protocols across all operating units^{Reference Hedrick, Heckman, Smith, Sawyer, Friel and Foley24,Reference Hedrick, Turrentine and Smith25} ; (9) performance feedback to staff and clinicians on a routine basis^{Reference Cima, Dankbar and Lovely18,Reference Crolla, van der Laan, Veen, Hendriks, van Schendel and Kluytmans20,Reference Hedrick, Heckman, Smith, Sawyer, Friel and Foley24,Reference Hedrick, Turrentine and Smith25,Reference Lutfiyya, Parsons and Breen29,Reference Russell, Chung and Riad32,Reference Vu, Collins and Seese34,Reference Wick, Galante and Hobson36,Reference Forbes, Stephen and Harper41,Reference Keenan, Speicher and Nussbaum46} ; and (10) overall promotion of culture of safety and openness to change.^{Reference Connolly, Foppa, Kazi, Denoya and Bergamaschi19,Reference Crolla, van der Laan, Veen, Hendriks, van Schendel and Kluytmans20,Reference Wick, Galante and Hobson36,Reference Reames, Krell, Campbell and Dimick39}

Effect of colorectal bundles on incidence of overall SSI (meta-analysis)

We included 30 studies that reported overall SSI incidence outcomes in our meta-analysis (Fig. 2 and Supplemental Fig. S4 online). We analyzed SSI outcomes separately for the single study that reported SSI outcomes by 2 different surveillance methodologies (NHSN and NSQIP).^{Reference Rumberger, Vittetoe, Cathey, Bennett, Heidel and Daley48} Cumulatively, there were 894 SSI among the 10,627 patients in the study arm versus 1,561 SSI among the 10,074 patients in the control arm. We found a statistically significant pooled SSI RR of 0.56 (95% CI, 0.48–0.65), with 71.2% heterogeneity in our random effects model. There was no significant publication bias according to the Egger test (−0.48; P = .60) and Begg’s Kendall score (Q = −23; P = .69). Given the high statistical heterogeneity, we analyzed the bundle effect on SSI incidence over 2 periods: 2000–2015 and 2016–2018. We found a pooled SSI RR of 0.60 (95% CI, 0.49–0.70) with 73.7% heterogeneity for studies published during 2000–2015, and an RR of 0.48 (95% CI, 0.41–0.57) with 0.0% heterogeneity for 2016–2018. Cumulative effects analysis showed that all studies included showed a protective SSI effect, except for the study by Anthony et al,^{Reference Anthony, Murray and Sum-Ping12} which showed an SSI RR of 2.49 (Supplemental Fig. S5 online).

Fig. 2.

Forest plot showing pooled effect of colorectal bundles on the incidence of overall surgical site infection. Note. RCT, randomized controlled trial; PCS, prospective cohort, single center; PCM, prospective cohort, multicenter; RCS, retrospective cohort, single center; RCM, retrospective cohort, multicenter.

Effect of colorectal bundles on incidence of superficial SSI (meta-analysis)

We included 21 studies that reported superficial SSI incidence outcomes in our meta-analysis (Supplemental Fig. S3 online). There were 808 SSIs among the 19,984 patients in the study arm, versus 1,094 SSIs among the 19,479 patients in the control arm. We found a statistically significant pooled SSI RR of 0.56 (95% CI, 0.42–0.75), with 83.4% heterogeneity in our random-effects model. Publication bias was present according to the Egger test (−2.1; P = .009), but not according to Begg’s Kendall score (Q = −32; P = .34).

Effect of colorectal bundles on incidence of deep SSI (meta-analysis)

We included 11 studies that reported deep SSI incidence outcomes in our meta-analysis. (Supplemental Fig. S2 online). There were 36 SSIs among the 3,365 patients in the study arm, versus 78 SSIs among the 3,787 patients in the control arm. In our fixed-effects model, we calculated a statistically significant pooled SSI RR of 0.67 (95% CI, 0.45–0.98), with 0.0% heterogeneity. There was no significant publication bias according to the Egger test (−0.332.1; P = .059) and Begg’s Kendall score (Q = −4; P = .71)

Effect of colorectal bundles on incidence of organ-space SSI (meta-analysis)

We included 17 studies that reported organ-space SSI incidence outcomes into the meta-analysis. There were 203 SSIs among the 6,115 patients in the study arm, versus 344 SSIs among the 5,980 patients in the control arm. In our random-effects model, we found a statistically significant pooled SSI RR of 0.63 (95% CI, 0.49–0.81), with 44.5% heterogeneity (Supplemental Fig. S1). There was no significant publication bias according to the Egger test (2.1; P = .059) and Begg’s Kendall score (Q = 30; P = .23).

Effect of bundle size on SSI (meta-regression)

In our meta-regression analysis, we found that bundle size was statistically significantly related to reduction of overall SSI in our meta-regression model (Table 2). The highest SSI reduction (59%) was associated with the largest bundle size, corresponding to colorectal SSI prevention bundles whose composition included ≥11 ACS/SIS guideline-recommended interventions.

Table 2.

Meta Regression Models Examining Bundle Size Categories and Reduction of Surgical Site Infections

Note. Het I², test of heterogeneity; RR, relative risk; CI, confidence interval. Bundle size represents number guideline-recommended interventions included in each bundle.

Effect of individual bundle interventions on SSI

The following individual colorectal bundle interventions significantly correlated with SSI incidence: (1) mechanical bowel preparation combined with oral antibiotics (R = −0.68; P = .003) and (2) preoperative chlorhexidine showers (R = −0.49; P = .04) for organ-space SSI.

The following bundle interventions correlated with highest superficial SSI reductions: (1) wound closure protocols, including separate instrument trays and glove with or without gown change prior to surgical wound closure (R = −0.55; P = .009), and (2) standardized postoperative wound dressing change at 48 hours (R = −0.59; P = .005)