The Barrow Randomized Operating Room Traffic (BRITE) Trial

Abstract

Surgical site infections (SSIs) account for 14% to 20% of all nosocomial infections.1 They have been shown to increase a patient's risk of death 2- to 11-fold,2 and they cost our nation an estimated $3.5 to $10 billion annually.3 Neurosurgical SSIs are of particular concern, because the potential consequences of SSIs in the central nervous system are dire. SSIs are known to be multifactorial, with a large number of both fixed and modifiable risk factors present in the preoperative, intraoperative, and postoperative periods. Among the modifiable intraoperative risk factors, cleanliness of the operating room (OR) environment is often targeted in SSI prevention strategies. Observational studies have shown that OR bacterial air counts are directly related to OR activity.4 As such, several national organizations, including the Centers for Disease Control and Prevention (CDC), the Joint Commission, and the Association of Operative Registered Nurses, have published recommendations for reducing SSIs that include calls to restrict OR traffic to essential personnel only.5-7 In the 15 years since publication of these recommendations, numerous studies have been published that attempt to correlate OR traffic with SSI risk. The level of evidence in these studies is low, with major methodological flaws and heterogeneous methods making it difficult to interpret what effect, if any, reducing OR traffic has on SSI rate.8 The purpose of this study was to provide high-quality evidence regarding the effect of OR traffic on SSI risk. METHODS We conducted a 2-phase study to determine whether increased OR traffic is associated with increased SSI rates. OR traffic recording was automated via infrared personnel counters. Data were collected on patient and procedure characteristics, length of surgery, and OR traffic. Each patient had ≥8 weeks of follow-up to identify the incidence of SSI. SSIs were defined according to current CDC criteria.5 Any patient undergoing a procedure in 1 of the 11 neurosurgical ORs at our single institution was included in the study. Minors (age <18 years), pregnant patients, and prisoners were excluded. This study was approved by our institutional review board (#12BN124) and was registered at clinicaltrials.gov (NCT01783769). Phase 1 Phase 1 was a prospective, observational study comprising 21 weeks of data collection and analysis. Data were collected on case OR number (ORs 1-11), attending neurosurgeon, case type, case length, and traffic rate (events/min) through each OR door from opening incision to skin closure. An independent, professional statistician calculated both descriptive and comparison statistics. Comparison statistics included evaluations of the above variables on SSI rate. The χ2 test and independent samples t test were used. Phase 2 Phase 2 was a randomized, single-blinded, controlled clinical trial comprising 12 weeks of data collection and analysis. The 11 neurosurgical ORs were divided into groups A (ORs 1-5) and B (ORs 6-11, see Figure 1). For each of the 12 weeks of data collection, group A or B was randomly assigned to a low-traffic (LT) or regular-traffic (RT) protocol. LT protocol ORs had stop signs placed on all OR doors. Signs included written instructions (in multiple languages) that all nonessential traffic was prohibited. No observers were allowed in the OR, medical student access was restricted to those who were committed to staying for the case entirety, and essential personnel were asked to avoid using the main door.FIGURE 1: Phase 2 OR randomization groups. AC, autoclave door; OR, operating room; SS, scrub sink door. Used with permission from Barrow Neurological Institute, Phoenix, Arizona.The reasoning behind restricting OR traffic through the main door was based on the underlying theory of association between OR traffic and SSI rate: ORs are equipped to maintain positive air pressure and laminar air flow, which minimize the amount of contaminant-laden air near the patient. Increased personnel traffic increases the shedding of airborne contaminants from skin and clothing, and increased door openings disrupt OR positive pressure and laminar air flow, allowing for dirtier corridor air to enter the OR and contaminate an open incision. The main door is the largest of the 4 OR doors, and it opens to the OR transport corridors, where air bacterial counts have been shown to be much higher.4 Randomization of OR groups A and B to the LT or RT protocols was accomplished via 2 levels of randomization by an independent observer (www.random.org). Descriptive and comparison statistics were calculated as in phase 1 with the addition of the American Society of Anesthesiologists classification and the surgical wound classification as variables. RESULTS Phase 1 Overall, 1944 cases were evaluated in phase 1. The average case length was 177 minutes and the total number of door openings was 136, giving a total-door traffic rate of 0.77 events/min. The main-door traffic rate was 0.21 events/min. Thirty-four SSIs were identified during follow-up. Three of these patients had >1 procedure before wound washout, and so it was impossible to know which case resulted in the SSI. These 3 cases were therefore excluded from the traffic analyses. The overall SSI rate for phase 1 was 1.75%, which is consistent with previously published rates for neurosurgical procedures. Risk factor assessment for SSI showed that case length and case type were strongly associated with increased SSI risk (P = .003 and P < .001, respectively). Case types that had higher SSI rates were cranioplasties, open craniotomies, and posterior spinal fusions (Figure 2). There was no significant difference in total-door traffic rate between the SSI and non-SSI groups (P = .78). There was, however, a significant difference in main-door traffic rate between SSI and non-SSI groups (P < .001), with the non-SSI cases having a paradoxically higher main-door traffic rate (SSI main-door rate = 0.154 events/min, non-SSI main-door rate = 0.213 events/min). There were no significant differences in traffic rates in any of the other 3 OR doors (data not shown). Table 1 summarizes phase 1 results.FIGURE 2: Percentage of surgical site infections by case type. Asterisk (*) indicates that cranioplasties, open craniotomies, and posterior spinal fusion procedures have a higher risk for SSI than other procedures noted. Ant/lat, anterior/lateral; crani, craniotomy; decomp, decompression; endo/bx, endoscopy/biopsy; IPGs, implantable pulse generators; periph, peripheral; post, posterior; rev, revision; SSI, surgical site infection; vv/nn, vasculature/nerve. Used with permission from Barrow Neurological Institute, Phoenix, Arizona.TABLE 1: Phase 1 Comparison of SSI and Non-SSI Groupsa,bPhase 2 Overall, 1116 cases were evaluated in phase 2. Thirty-six cases were identified as take-backs for infection, 4 of which had numerous procedures before take-back, prohibiting traffic data analysis. Six of the remaining 32 cases did not meet CDC criteria for SSI, leaving 26 SSIs meeting CDC criteria. Analyses were completed for both the 26 CDC-defined take-backs and all 32 take-backs combined. Possible confounders included American Society of Anesthesiologists class, wound class, case type, case length, OR number, attending surgeon, number of emergent cases, and number of night/weekend cases. No significant differences were found between the LT and RT protocol arms for any of these variables. See Tables 2-7 for phase 2 randomization data.TABLE 2: Phase 2 Data by OR Traffic GroupaTABLE 3: Randomization by Case TypeaTABLE 4: Randomization by ASA ClassaTABLE 5: Randomization by Wound ClassaTABLE 6: Randomization by OR NumberaTABLE 7: Randomization by Attending NeurosurgeonaWith all 12 weeks of traffic data combined, no significant differences in total-door or main-door traffic rates were found, although the LT group trended toward a reduced main-door traffic rate (P = .08). To determine if the LT protocol itself had an effect on SSI rate, we compared SSI rates in the LT and RT groups and found a paradoxical trend toward higher SSI risk in the LT protocol for CDC-defined SSIs only (3.2% vs 1.5%, respectively, P = .06) and for all take-backs (3.8% vs 2.1%, respectively, P = .09) (Table 8).TABLE 8: Phase 2 Outcomes: Effect of Traffic Rates on SSIsaTo evaluate if actual reduced traffic has an effect on SSI rate, we evaluated each week of the study individually and found that during 4 weeks of the study, there were significant differences in the total-door traffic rate between study arms, and in 10 weeks of the study, the main-door traffic rates were significantly different. SSI rates were then calculated for these weeks alone, and we found no effect of reduced total-door or main-door traffic rates for both CDC-defined SSIs (P = .75 and P = .83, respectively) and all take-backs (P = .58 and P > .99, respectively). See Table 8 for a summary of phase 2 results. Several other analyses were performed. LT and RT groups were reassigned based on those cases that had total-door and main-door traffic rates in the lowest quartile for the entire study population. For this analysis, we found no significant difference in infection rate between high-traffic and low-traffic groups for both the 26 CDC-defined SSIs (high traffic = 2.3%, low traffic 1.8%; P = .63) and all 32 take-backs (high traffic = 2.8%, low traffic = 2.5%; P = .82). We also selected attending physicians who performed ≥30 cases and who had a reduced total-door traffic rate when operating in the LT protocol. This analysis also failed to show any significant difference in SSI rate between LT and RT groups for either the 26 CDC-defined SSIs (LT = 2.5%, RT = 0.9%; P = .27) or all 32 take-backs (LT = 2.9%, RT = 1.4%; P = .33). As previously described, 4 SSIs were excluded from phase 2 analyses because the patients had multiple procedures before wound washout. In a final set of analyses, these 4 cases were all assigned to the RT group. Analyses were completed for all 12 weeks of data comparing infection rates between the LT and RT protocols (P = .30), weeks of reduced total-door rate only (CDC SSIs, P = .55; all take-backs, P = .75), and weeks of reduced main-door rate only (CDC SSIs, P = .29; all take-backs, P = .41). Adding the 4 previously eliminated SSIs to the RT group did not result in significant differences in SSI rates between the 2 traffic protocol groups. Study Power Post hoc power analyses for each phase of the study were performed by an independent, professional statistician. Phase 1 had >0.99 power to detect a medium effect size (d = 0.5) in traffic rate between the SSI and non-SSI groups. The power of phase 1 to detect a small effect size was >0.95. Power analysis for phase 2 showed that with all weeks combined, the power to detect a small effect size on traffic between the 2 protocols was 0.93. The power to detect a large effect size on SSI rate was 0.80. DISCUSSION Study Limitations Although our study was overpowered to detect a difference in traffic rate between both phase 1 and phase 2 SSI and non-SSI groups, phase 2 was powered only to detect a large effect size of traffic rate on SSI risk. As such, if OR traffic has a medium or small effect on SSI risk, this study may not identify that relationship. However, given the phase 1 results and the data trend showing, if anything, an increased SSI risk among the patients randomly assigned to the LT protocol, we are confident in concluding that increased OR traffic has no meaningful effect on SSI risk. This study is furthermore limited by its reliance on infrared personnel counters for monitoring traffic. Although the automation of OR traffic monitoring can be considered a strength of this study, this method also opens the study to potential sources of error such as equipment malfunction and OR personnel tampering. This risk was minimized by performing overnight test traffic of a known quantity (10 door openings for each OR door). These tests demonstrated that the technology was working correctly. Reported and witnessed incidences of OR personnel tampering with automated personnel counters were fortunately very infrequent. Interpretation These data provide a strong argument against a significant association between OR traffic and SSI risk. Finding a significantly reduced main-door traffic rate among SSIs in phase 1 was unexpected. Given that no significant difference was found in traffic rate between SSI and non-SSI groups for other OR doors or all doors combined, we must consider that increased main-door traffic is protective against SSI, possibly secondary to changes in OR air flow dynamics. It is also possible that by attempting to limit all OR traffic, we are unintentionally limiting traffic that is somehow beneficial. Further consideration of this finding is needed. CONCLUSION SSIs represent a significant source of patient morbidity and mortality, and are a major economic burden carried by the national health care economy. Efforts directed at minimizing SSIs are therefore necessary. As with any medical decision, the potential benefits of a national health policy must be weighed against its risks. The main risk of policies that propose to restrict OR access is that the policies might limit the quality of surgical education. Limited OR access for observers, nursing/medical students, or even residents threatens the education of future essential OR personnel, and so if surgical teaching institutions are to adopt such national recommendations, rigorous data are needed to show that the potential benefit of OR restriction is worth the risk. The above data provide strong evidence that the potential benefit of OR traffic restriction in reducing SSI rates is, at best, trivial and is possibly nonexistent. These data should better enable teaching institutions to weigh the benefits of opening ORs to interested learners against the questionable risk of increased OR traffic. Disclosures Dr Justin C. Clark and Dr Michael A. Bohl received a research grant from Medtronic, Inc. The authors have no personal, financial, or institutional interest in any of the drugs, materials, or devices described in this article. Acknowledgments The authors would like to acknowledge all the OR nursing, anesthesia, and support staff who helped make this study possible, as well as the observers, medical students, and residents who worked hard to ensure the study protocol was followed.