Prolonged Perioperative Antibiotics: A Hidden Problem Free

This study was conducted at a freestanding, quaternary-care pediatric hospital with 465 total licensed beds. There are ∼97 418 patient days and 15 330 admissions each year. In 2017, there were 5701 admissions to inpatient surgical services. This facility is the only level-1 pediatric trauma center in the state and services a 7-state catchment area. This hospital also includes multiple surgical pediatric subspecialties (see below). This center has a longstanding SSI prevention committee that dictates expected duration of prophylaxis for broad categories of procedures based on national guidelines; none of the recommended durations are >24 hours. There is also a well-established stewardship program that audits all inpatient (but not outpatient) antibiotic use.^12,15,16 

All data were extracted from the local infection control and epidemiology data warehouse, which sources all encounter-level data found in the electronic medical record (EMR; Epic Systems, Verona, Wisconsin). The extraction included all procedures performed from January to December 2017, identified via the “surgery log” tables in Epic Clarity data warehouse. Extensive review and refinement of the data set were necessary to exclude duplicates, nonqualifying procedures, and to identify procedures requiring chart review (Fig 1). These data were then uploaded into the Health Insurance Portability and Accountability Act-compliant software, REDCap (Vanderbilt University, Nashville, Tennessee). Five-thousand and nine qualifying procedures were chart-reviewed for further clinical information and to confirm that they did not meet exclusion criteria. Chart review was done for any procedure that could potentially have an infectious etiology based on procedure name (eg, hardware removal), prolonged antibiotic duration, any procedure associated with an unknown duration or a discharge antibiotic prescription, or a positive C. difficile test within 30 days of the procedure. All procedures with an infectious etiology (eg, perforated appendectomy, removal of infected cardiac device) or environmentally contaminated wounds (eg, dog bite, open fracture) were excluded. All antibiotic prescriptions (inpatient and outpatient) were carefully chart-reviewed before being deemed to be surgical prophylaxis; for example, if an antibiotic was being used for an unrelated infection, this use would not be included in the prophylaxis calculations. After chart review, all surgical prophylaxis was included in the duration calculation. See Fig 1 for a detailed methods flowchart including chart review criteria and exclusions, and the Supplemental Table 2 for nonqualifying procedures and antimicrobial agents (eg, acyclovir). Data collection, chart review, and exclusion were performed by a professional research assistant and 2 surgical advance practice providers.

To calculate the actual duration of antibiotic prophylaxis, the following a priori definitions were used, after determining the antibiotic use was indeed for surgical prophylaxis via chart review. Procedures that did not receive any prophylaxis were assigned 0 hours, those with a single perioperative dose were assigned 0.5 hours, and those that received multiple doses <24 hours or >24 hours of prophylaxis were assigned their actual duration in hours from cut time to time of last administered dose. For the expected duration, procedures were assigned their actual duration up to 24 hours. To calculate potential excess, this was done with two methods. To calculate percentage of procedures with potential excess, these 2 groups were considered to meet criteria: (1) those with >1 dose, but <24 hours, and (2) those with >24 hours. The second method calculated the amount of potential excess in days of therapy (DOT). To do this, the expected duration was subtracted from the actual duration per procedure. For example, a child given cefazolin 1 hour before cut time and then another dose 7 hours later would be assigned 6 hours of actual prophylaxis (cut time to last administered dose), 6 hours of expected duration, and, thus, 0 hours of potential excess DOT, but categorized in the percentages as potential excess in the <24 hours category. If, instead, the child received 47 hours of cefazolin after cut time, the procedure would be assigned 47 hours of actual, 24 hours of expected, and, thus, 23 hours of potential excess DOT, and categorized by percentage in the potential excess >24 hours group. Potential excess DOT was calculated in hours (and converted to days by dividing by 24 hours), and in days outpatient (and converted to hours by multiplying × 24); a DOT was defined as 24 of these calculated hours. For individual procedures, we did not examine whether prophylaxis was or was not indicated (previously published);¹¹  rather, we focused on durations when prophylaxis was given. There is not a well-defined, standard duration of surgical prophylaxis for many pediatric procedures; therefore, excess was deemed “potential” rather than definitive excess.

Procedures were attributed to a subspecialty based on the surgeon of record’s subspecialty. Combination cases were assigned to the first surgeon on the EMR panel. The subspecialties with enough volume to analyze independently were: general surgery (including liver and kidney transplants and colorectal care), neurosurgery, otolaryngology (ENT), cardiothoracic surgery (including cardiac transplants), orthopedic surgery, and urology. Those grouped for analysis (“other”) included plastic surgery, gynecology, gastroenterology, ophthalmology, and oral surgery. The 1028 procedure types were grouped into procedure-similar categories (233) for ease of analysis. This was done first on the basis of surgical subspecialty and second on the basis of procedure similarity (method, body area, underlying diagnosis). For example, shoulder, knee, elbow, and ankle arthroscopies performed by orthopedic surgery were all categorized as “arthroscopy” and attributed to that subspecialty. See Supplemental Table 1 for a list of procedures and category designation.

Analysis was conducted at multiple levels: hospitalwide, by surgical subspecialty, and by procedure category. For hospitalwide and surgical subspecialty analysis, prophylaxis duration was attributed to one of the following groups: no surgical prophylaxis, a single perioperative dose, multiple doses encompassed in 24 hours, and multiple doses extending beyond 24 hours. Selected procedure categories were examined by duration of surgical prophylaxis in hours to assess variability.

C. difficile testing in 2017 was done by the BioFire FilmArray Gastrointestinal Panel. Local policy requires an unformed stool for the sample to be run, and results in children aged <1 year were routinely suppressed. We extracted reported positive tests for C. difficile within 30 days of the procedure, and charts were reviewed to confirm it was a new positive (no positive test in the previous 30 days).

Statistical analysis was performed in Minitab (V.19). Descriptive statistics were used throughout and interval plots showing the mean and 95% confidence intervals were used to illustrate duration of prophylaxis for select surgical procedures (Fig 2). These procedures were selected in the following manner: for each of the high-volume surgical subspecialties, we identified the top procedures that contributed to >24 hours of prophylaxis with the intention of selecting procedures with the most excess and therefore displaying duration variability. Four to 5 procedure categories were selected for each surgical subspecialty on the basis of the percentage of those procedure categories that contributed to >24 hours of prophylaxis (ie, only procedure categories that contributed >5% to that subspecialty’s >24-hours prophylaxis group were included). For a full list of the procedure categories that received >24 hours of prophylaxis within each subspecialty, see Supplemental Table 1. An χ² analysis was preformed to compare C. difficile occurrence between the patients who received prophylaxis (single dose, multiple <24 hours, multiple >24 hours) and those who received no prophylaxis.

This project was approved by the state’s multiple institutional review board (submission #19-2032) and considered exempt.

A total of 15 651 surgical procedures were initially included for analysis, 5009 of which required chart review on the basis of the criteria described in the methods. Of these, 2756 were excluded on the basis of the criteria in Methods and Fig 1. Thus, a total of 12 895 procedures were included in the final analysis. Of these procedures, 5778 received no perioperative prophylaxis (45% of all procedures), 3853 received a single dose (30% of all procedures), 1784 received multiple doses within 24 hours (14% of all procedures), and 1480 received >24 hours (11% of all procedures) (Fig 3).

For the 1480 procedures that extended beyond 24 hours of prophylaxis, the potential excess antibiotic DOT was 5733. This was composed of 11% of ENT procedures (3092 potential excess DOT), 12% of all orthopedic procedures (218 DOT), 9% of all general surgery procedures (483 potential excess DOT), 33% of all neurosurgery procedures (149 potential excess DOT), 12% of all cardiac procedures (353 potential excess DOT) and 11% of all urology procedures (882 potential excess DOT). The combined group of surgical subspecialties included in the other group had a combined procedure count of 1306, of which 98 were >24 hours (8%), contributing 556 potential excess DOT. See Table 1 and Supplemental Table 1 information for more data on the <24 hours, single dose, and no prophylaxis groups.

Of the 233 procedure categories, 162 (70%) had at least 1 procedure that extended beyond 24 hours of antibiotic prophylaxis. For each high-volume subspecialty, the procedure categories that contributed the most to extended prophylaxis (>24 hours) by subspecialty were as follows: for the 152 cardiac surgery cases: cardiac catherization (42; 28%) and hypoplastic repair (23; 15%); for the 381 total ENT cases: tympanostomy/-plasty (44; 12%) and acoustic implants (43; 11%); for the 235 general surgery cases: colorectal procedures (57; 24%) and laparoscopy/-otomy (41; 17%); for the 198 neurosurgery cases: resection and excision (70; 34%) and foreign body placement (43; 22%); for the 238 orthopedic surgery cases: osteotomy (62; 22%) and posterior and anterior spinal fusion (47; 17%); and for the 133 urology cases: hypospadias and epispadias repair (46; 35%) and urethrocutaneous fistulas (15; 11%). See Supplemental Table 1 for all procedures that extended beyond 24 hours of prophylaxis.

Specific subspecialty procedures revealed great variability in duration of surgical prophylaxis within 1 procedure. For example, patients who underwent a septoplasty received anywhere between no prophylaxis and 378 hours of prophylaxis (median 187.5 hours). Similarly, hypospadias procedures ranged from no prophylaxis to 429 hours (median 192 hours). Orthopedic arthroscopies ranged from no prophylaxis to 168 hours (median 0.5 hours) and abdominal laparoscopies and laparotomies without entry into bowel ranged from no prophylaxis to 211 hours (median 0.5 hours). See Fig 2 and Supplemental Table 1 for more examples.

There were 67 positive C. difficile tests in the 30 days after the 12 895 procedures (0.5%). When comparing those who did receive antibiotic surgical prophylaxis (54 of 7117; 0.76%) to those who did not (13 of 5778; 0.22%), this was statistically significant (P = <.01) and was more prevalent in the beyond–24-hours prophylaxis group (18 of 1480; 1.2%), though we were unable to adjust for differences in patient factors among groups. See Table 1 for breakdown of C. difficile results.

Surgical prophylaxis that extends beyond 24 hours is considered unnecessary by most current guidelines, and many recent publications advocate for a single dose.^1–5  Our study demonstrated that, even in a facility with an active ASP, there was still gross overextension of surgical prophylaxis beyond 24 hours across all surgical subspecialties. At this regional children’s hospital, a total of 11% of all procedures in 2017 received >24 hours of surgical prophylaxis, resulting in 5733 potential excess DOT. An additional 14% of procedures received multiple doses within 24 hours, also potentially unnecessary. To our knowledge, this is the first report of a hospitalwide assessment of surgical prophylaxis durations. Our results point to a need for continued quality improvement and antimicrobial stewardship to effect change at our center, a need that is reasonably extrapolated to other hospitals, adult and pediatric alike. This goal is broadly supported.¹⁴  Framing the problem as potential excess DOT has created the platform needed to spark interest around quality improvement in this area. Potential excess DOT can be used as a scale to identify subspecialties where quality improvement efforts will have the greatest impact.

Notably, in the literature, extended durations of prophylaxis are associated with increases in ADEs, such as C. difficile-positive testing and use of allergy medications on the day of surgery.^6,11  We did observe an association with antibiotic surgical prophylaxis and C. difficile; however, our results were limited by lack of ability to adjust by patient complexity.

Extended durations of antibiotics for surgical prophylaxis were spread across all subspecialties and all procedure categories, indicating inherent variability in care. Notably, potential excess durations appeared to fall within 2 distinct categories: purposeful, as demonstrated by long durations of therapy (eg, 7–14 days with an outpatient prescription), or inadvertent, where an active order was not discontinued in the intended time frame (eg, 26 hours given). Our attempts to address this variability point to various causes. First, some attending physicians were not aware of updated prophylaxis recommendations; second, some attending physicians were not aware they did things differently than their colleagues, exacerbated by the fact that recommendations from pediatric surgical governing bodies is often unclear or nonexistent; third, often orders were placed by learners, who simply did not include durations. In terms of processes, there are various approaches. For inadvertent overextension, tools in the EMR (such as automatic stops) and improved order sets may be the easiest path. Inadvertent variability may be approached with education about national and subspecialty guidelines and with consensus-building where clear guidance does not yet exist. Liaisons in each surgical subspecialty are currently using these data to develop detailed local surgical prophylaxis recommendations.

To this end, we are working with each surgical subspecialty to develop guidance with plans to repeat the current analysis for specific procedure categories after implementation. For example, at this facility, hypospadias repair (urology) and tympanostomy tube placement (ENT) appeared to be receiving purposeful extended durations of antibiotics. Though hypospadias repair with catheter placement and tympanostomy tube placement are not explicitly mentioned by national guidelines, there exists subspecialty-specific studies endorsing < 24 hours of surgical antibiotic prophylaxis for both procedures; thus, these are areas for local standardization.^17–25  Second, where no specific data exist, some logic in extrapolating data from similar risk procedures might inform care. For example, data on shorter prophylaxis (<24 hours) for spinal fusion operations with hardware placement may be applicable to cases that are categorized as clean and involve both an implant and prolonged operative time. This kind of improvement will require collaboration between the ASP and stakeholders within each subspecialty, a process that has already begun at this center with meetings involving subspecialty and quality heads. Through this process, we encounter that both a lag in education about available adult guidelines and the absence of data-driven pediatric guidelines have contributed to this systemic overuse.

National societies need to be aware of the need for more detailed guidance, as well. Use of data in databases, such as National Surgical Quality Improvement Program Pediatric (NSQIP-P) and The Pediatric Health Information System database, might be harnessed to inform care. Prospective, grant-funded initiatives can help better define appropriate durations. Just recently, NSQIP-P released the data collected during a pilot program for antibiotic stewardship. The NSQIP-P report for our institution mirrored these data; thus, the NSQIP-P report will be powerful in tracking change after quality improvement projects are implemented. This will be particularly impactful if surgical antibiotic prophylaxis can be correlated with SSI incidence for particular pediatric surgical categories. Although we will evaluate some of our more common procedures with variability in prophylaxis for SSIs, it is unlikely we will be statistically powered to reach definitive conclusions.

There are multiple limitations to this study. First, as described above, we were not judging if a procedure qualified or not for prophylaxis; thus, our excess DOT is likely an underestimate. However, we were inversely unable to determine if those that did not receive prophylaxis should have and whether there were cases where antibiotics prophylaxis was insufficient. Second, our data are from 2017, the year new guidelines for surgical prophylaxis were published, which might influence results. Third, we calculated excess based on durations >24 hours of antibiotics; however, as many procedures require none or only a single dose of antibiotics, this excess is likely an underestimate. Note that, though there may be concern some of the prolonged antibiotics were given for reasons other than prophylaxis, all cases with prophylaxis beyond 24 hours were chart-reviewed (5009 cases) to determine the antibiotic indication, thus allaying this concern. Fourth, surgical site infection data were not included in this study, so we were unable to give the incidence of SSI within each of our prophylaxis groups. Lastly, as these data were collected from 2017, it may not reflect current practices and overestimate the current quantity of antibiotic overuse.

Our data demonstrate that, at a large, quaternary-care pediatric hospital with an active ASP, potentially unnecessary antibiotic prophylaxis was still prevalent and is a worthy area for improvement in pediatric centers. Based on these data, we have taken steps to address duration of prophylaxis by surgical subspecialty to improve standardization and compliance within our facility. Most importantly, this work reveals the need for a national effort to harness current databases and fund new research to define and implement appropriate durations of surgical prophylaxis in pediatric patients.

We thank Drs Jennifer Bruny and Shannon Acker for their support in reviewing and critiquing our paper from a surgeon’s perspective, and Kelly Pearce for data extraction.