Cefazolin and an aminoglycoside compared with cefazolin alone for the antimicrobial prophylaxis of type III open orthopedic fractures.

CONTEXT: Uncertainty of antibiotic prophylaxis of type III open orthopedic fractures still exists. Controversy remains as using cefazolin as a single agent or the addition of an aminoglycoside for broader coverage to prevent infection. AIMS: The aim of the study was to determine if the combination of cefazolin and an aminoglycoside reduced infections compared with cefazolin alone. SUBJECTS AND METHODS: This was a retrospective study inclusive of patients with type III open fracture admitted between January 1, 2010, and August 31, 2014 at a level 1 trauma center, who were prophylactically treated with cefazolin alone or cefazolin and an aminoglycoside. STATISTICAL ANALYSIS USED: All analyses were performed using Microsoft Excel 2010. Chi-square or Fisher's exact tests were used for categorical data and Wilcoxon rank-sum test for skewed continuous data. Logistic regression analysis was performed on all confounding variables with P < 0.1.
RESULTS: A significantly higher percentage in the combination group developed infection (6/15 [40%] vs. 8/53 [15.1%], P = 0.035). There was a trend toward a higher odds of infection in the combination group (odds ratio: 2.99, 95% confidence interval: 0.79-11.33, P = 0.107). Infection rates due to multidrug-resistant bacteria were statistically higher with the combination group (3/15 [20%] vs. 1/53 [1.9%], P = 0.046). There were no statistically significant differences in 30-day mortality, 1-year readmission rates due to fracture complication, or length of hospital stay.
CONCLUSIONS: The results suggest that the addition of an aminoglycoside to cefazolin may not be necessary to prevent infection. Copyright:

INTRODUCTION

Open orthopedic fractures are classified according to Gustilo and Anderson, in which type III fractures have the highest infection risk.[12] Optimal antibiotic prophylaxis in type III open fractures remains uncertain. Due to the high infection rate of type III fractures, some suggest broad Gram-negative coverage with an aminoglycoside in addition to cefazolin; however, this theory has not been evaluated.[3]

The goal of this study was to assess the infection rates between cefazolin alone compared to cefazolin with an aminoglycoside for type III open fracture prophylaxis.

SUBJECTS AND METHODS

This retrospective study was conducted in the United States at a 480-bed, level I trauma center. Initially, the study was initially designed for noninferiority; however, due to calculated size requirements, it was converted to a pilot study including approximately 10% of the originally intended population. The study was approved as exempt by the local institutional review board (IRB) (SLHN-2014-98), which granted a waiver of informed consent. The study was designed entirely by the authors, and no funding was obtained. This cohort study was written in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology guideline.[4]

Patient cohort

Patients were identified for possible inclusion into the trial through the institution's trauma database for open orthopedic fracture if admitted between January 1, 2010, and August 31, 2014. After review of the electronic medical record, patients were included if they had type III open fracture of the upper extremity, lower extremity, or pelvis and had received initial prophylaxis with either cefazolin monotherapy or cefazolin plus an aminoglycoside. Only those not meeting the inclusion criteria were excluded.

The open fracture type was determined from the orthopedic and trauma surgeons' progress notes. The definition for a multidrug-resistant (MDR) organism followed those criteria previously developed by the Centers for Disease Control and Prevention (CDC).[5] Wound closure time parameters were categorized based on existing literature: immediate closure was completed during the initial surgical intervention, early closure was within 24–72 h, and delayed closure was after 72 h.[6] Fracture-related infection was determined by progress note documentation of an infection located at the site of the type III fracture that required treatment with systemic antibiotics.

Optimal antibiotic dosing in pediatric patients was obtained from Lexicomp Online: Pediatric and Neonatal Lexi-Drugs (https://online.lexi.com/lco/action/home). For adults, the optimal cefazolin dose was obtained from the Infectious Diseases Society of America and Surgical Infection Society (IDSA-SIS) trauma guidelines and the national surgical prophylaxis guidelines.[78] The optimal adult dose for gentamicin or tobramycin was identified as at least 1.5 mg/kg every 8 h or 5 mg/kg every 24 h.

Endpoints

The primary endpoint was type III open fracture-related infection rates within 1 year. Secondary endpoints included 30-day mortality, 1-year readmission secondary to a complication from the fracture, length of the initial hospital stay, and type III fracture-related MDR infection within 1 year.

Statistical analysis

All analyses were performed using Microsoft Excel 2010 (Microsoft Corp., Redmond, WA). Treatment groups were compared with either Chi-square or Fisher's exact tests for categorical data and with Wilcoxon rank-sum test for skewed continuous data. P < 0.05 was considered statistically significant. Logistic regression analysis was performed on all confounding variables for the primary outcome with P < 0.1.

RESULTS

Study selection

A total of 251 open fractures were collected from the trauma database. Of these, 183 patients were excluded because they did not meet inclusion criteria: 173 did not have type III fracture and 10 patients received nonstudy antibiotics [Figure 1]. Therefore, 68 patients were included, with 53 in the cefazolin group and 15 in the combination group [Figure 1].

Figure 1

Patient population

Study characteristics

Baseline demographics are noted in Table 1, and clinical characteristics are summarized in Tables 2 and 3. In general, the two study groups were similar, with most patients being young, healthy adult males with type IIIA open fracture to the lower extremity caused by blunt trauma, who received surgery within 6 h from presentation with immediate closure and no subsequent surgical intervention. However, in the combination group, diabetes was statistically higher (2/15 [13.3%] vs. 0/52 [0%], P = 0.047) [Table 1] as was the length of antibiotic therapy after the initial surgery (median 2 days [interquartile range (IQR) 2–2.5] vs. 1.5 days [IQR 1-2], P = 0.015) [Table 3].

Table 1

Baseline demographics

Variable, unless specified, n (%)*	Cefazolin	Cefazolin + aminoglycoside	P
Age (years), median (IQR)	33 (24-51)	30 (26-37)	0.258
Male	43/53 (81.1)	11/15 (73.3)	0.510
BMI ≥30 kg/m2	16/49 (32.7)	5/14 (35.7)	1.000
Diabetes	0/52 (0)	2/15 (13.3)	0.047
Immunocompromised	1/49 (2)	1/15 (6.7)	0.417
Malnutrition	0/52 (0)	1/15 (6.7)	0.235
MDR organism history within 1 year	0/53 (0)	0/15 (0)	1.000
Smoker	11/46 (23.9)	4/15 (26.7)	1.000
Venous insufficiency	1/48 (2.1)	1/15 (6.7)	0.422

*Some variables had missing data. IQR: Interquartile range, BMI: Body mass index, MDR: Multidrug resistant

Table 2

Trauma type and fracture location

Variable, n (%)*	Classification	Cefazolin	Cefazolin + aminoglycoside	P
Fracture grade	IIIA	12/16 (75)	6/12 (50)	0.174
	IIIB	4/16 (25)	4/12 (33.3)
	IIIC	0/16 (0)	2/12 (16.7)
Injury type	Blunt	39/53 (73.6)	13/15 (86.7)	0.492
	Penetrating	14/53 (26.4)	2/15 (13.3)
Fracture location	Fibula/tibia/ankle	24/53 (45.3)	8/15 (53.3)	0.906
	Femoral	17/53 (32.1)	4/15 (26.7)
	Pelvic	1/53 (1.9)	0/15 (0)
	Upper extremity	11/53 (20.8)	3/15 (20)

*Some variables had missing data

Table 3

Clinical characteristics

Variable, unless specified, n (%)*	Cefazolin	Cefazolin + aminoglycoside	P
Optimal initial antibiotic dosing	17/53 (32.1)	8/15 (53.3)	0.132
Time from ED to surgery (h), median (IQR)	3 (2-9)	4 (2.5-7)	0.807
Glucose ≥180 mg/dL±1 day from first surgery	8/49 (16.3)	4/15 (26.7)	0.453
O2 saturation <90% ± 1 day from first surgery	1/53 (1.9)	0/15 (0)	1.000
Temperature <96.8°F±1 day from first surgery	17/52 (32.7)	5/15 (33.3)	1.000
Antiseptic skin preparation done for first surgery	52/53 (98.1)	15/15 (100)	1.000
Antibiotic duration (days), median (IQR)	1.5 (1-2)	2 (2-2.5)	0.015
Additional surgeries (n), median (IQR)	0 (0-1)	0 (0-2.5)	0.261
Optimal antibiotic prophylaxis for subsequent surgeries	8/19 (42.1)	3/7 (42.9)	1.000
Wound closure time
Immediate	37/53 (69.8)	9/15 (60)	0.365
Early	3/53 (5.7)	0/15 (0)
Delayed	13/53 (24.5)	6/15 (40)

*Some variables had missing data. IQR: Interquartile range, ED: Emergency department

Outcomes

For the primary outcome, there were statistically fewer patients in the cefazolin group who developed a type III fracture-related infection within 1 year compared to the combination group (8/53 [15.1%] vs. 6/15 [40%], P = 0.035) [Table 4]. This resulted in the monotherapy group having a relative risk (RR) of 0.38 or an absolute risk (AR) reduction of 24.9% (number-needed-to-treat of 5). Rates of infection due to an MDR organism were significantly higher in the combination group (3/15 [20%] vs. 1/53 [1.9%], P = 0.046) Table 4, which resulted in a RR of 10.6 or an AR increase of 18.1% (number-needed-to-harm of 6). The only MDR organism that occurred in the monotherapy group was Pseudomonas fluorescens (intrinsically resistant to cefazolin; susceptible to aminoglycosides); whereas, those that occurred in the combination therapy group included two methicillin-resistant Staphylococcus aureus (gentamicin susceptible) isolates and one Stenotrophomonas maltophilia (intrinsically resistant to cefazolin and aminoglycosides). There were no statistically significant differences in 30-day mortality, 1-year readmission rates, or length of hospital stay [Table 4].

Table 4

Primary and secondary endpoints

Variable, unless specified, n (%)*	Cefazolin	Cefazolin + aminoglycoside	P
Primary outcome
Fracture-related infection within 1 year	8/53 (15.1)	6/15 (40)	0.035
Secondary outcomes
MDR organism fracture-related infection within 1 year	1/53 (1.9)	3/15 (20)	0.046
30-day mortality	4/53 (7.5)	0/15 (0)	0.569
Fracture-related readmission within 1 year	13/50 (26)	7/15 (46.7)	0.128
Initial length of stay (days), median (IQR)	5 (3-9)	9 (5-11.5)	0.053

*Some variables had missing data. IQR: Interquartile range, MDR: Multidrug resistant

Logistic regression was performed on all independent variables with P < 0.1 (prophylaxis type, antibiotic duration, length of stay, and diabetes) to control for confounding effects on the primary outcome. However, because no diabetic patient developed an infection, the model failed due to perfect separation. Therefore, the analysis was rerun with all of the other variables included [Table 5], which resulted in a low but acceptable Chi-square overall model fit (P = 0.09). With regression, combination treatment trended toward an increased risk of infection (odds ratio 2.99, 95% confidence interval 0.79–11.33, P = 0.107), but this was not significant.

Table 5

Logistic regression analysis of primary outcome*

Variables	OR (95% CI)	P
Combination therapy given	2.99 (0.79-11.33)	0.107
Initial length of hospital stay	1.04 (0.94-1.15)	0.452
Days of antibiotic therapy postoperatively	1.17 (0.85-1.61)	0.323

*Chi-square overall model fit, P=0.09. OR: Odds ratio, CI: Confidence interval

DISCUSSION

Based on existing practice guidelines and clinical trials, the optimal antibiotic prophylaxis for type III open orthopedic fractures is largely unknown. To the best of our knowledge, this is the first study that has attempted to compare the two prophylaxis recommendations set forth from the three national guidelines.[389] Our study aimed to assess clinically meaningful outcomes while capturing important confounding factors that may affect rates of postoperative infection.

According to the CDC, many factors may potentially contribute to the development of surgical site infections: age, malnutrition, diabetes, smoking, obesity, immunosuppression, inadequate preoperative skin preparation, and suboptimal antibiotic prophylaxis.[710] Perioperative hyperglycemia, hypoxemia, and hypothermia may also be possible risk factors, according to the CDC and other literature.[1112] In our study, both cefazolin and combination groups had similar rates of these confounders, except for the higher incidence of diabetes in the combination group. However, neither of these diabetic patients went on to develop an infection.

Studies in the trauma literature reveal that fracture subcategory, type of trauma, location, initial antibiotic characteristics and duration, time to initial surgery, and wound closure time can influence infection rates.[3891314] The risk for infection quickly rises from type IIIA to IIIC.[12] In our study, the subcategories reported were similar between groups, with most being type IIIA. A limiting factor was that only a small number (16/53 [30.2%]) of patients in the monotherapy group had their type III fracture subcategory noted.

Many recommendations in the national guidelines are based on discrepant data and expert opinion. Although the recommended time until surgery is within 6–8 h from injury, one study found that this did not significantly affect infection rates.[1315] The time from emergency department presentation to the operating room in both of our study groups was comparable at well below 6 h. The preferred type of wound closure has also varied over time. Historically, late closure was favored.[14] However, recent data show most infections to be nosocomial; so, primary closure may be beneficial, which is how most of our patients were managed.[16]

Based upon data showing increased infection rates with delays beyond 3 hours, antibiotics should be started as soon as possible after the injury.[17] While we were unable to directly compare antibiotic dosing times between our groups, we were able to deduce that most patients received antibiotics within 3 h, as the median time from presentation until surgery was 3–4 h.

Antibiotic duration recommendations in these patients are conflicting. The SIS guidelines recommend 24 h of prophylaxis, whereas, the EAST guidelines recommend 3 days from initial surgery or 1 day from wound closure.[39] Both of our patient groups were managed with the EAST guidelines, and despite a statistically longer duration in our combination group, this did not impact infection rates.

This observational pilot study was done in an attempt to begin to test the hypothesis that cefazolin monotherapy may have similar outcomes to aminoglycoside combination therapy in patients with type III open fractures. The analysis revealed that patients treated with cefazolin monotherapy had significantly lower rates of type III fracture-related infection than those receiving cefazolin plus an aminoglycoside. After controlling for confounding variables, a trend toward higher odds of infection in the combination group persisted although this was no longer significant. The higher incidence of infection with the combination may have been driven by the statistically significant increased risk of developing an MDR infection. Although our study did not account for known resistance risk factors, such as recent antibiotic use or hospitalization, most of our patients were young, healthy males admitted from the community, and did not have a history of MDR infections or colonization.

The limitations to our study should be noted. It was not adequately powered to assess for noninferiority of the cefazolin monotherapy regimen. The retrospective nature of the study increases the likelihood of potential reporting bias, and the lack of randomization allows for possible selection bias. However, these potential biases were at least partially mitigated by controlling for potential confounders.

CONCLUSIONS

This retrospective study suggests that the addition of an aminoglycoside to cefazolin may not decrease risk of infection compared to cefazolin alone for prophylaxis of type III open fractures. The addition of the aminoglycoside may increase the risk of developing subsequent MDR infections and adverse effects. Randomized controlled trials or large, multicentered, observational studies are necessary to confirm our study's findings, as our results are only hypothesis generating.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.

Ethical conduct of research

This study was approved by the St. Luke's University Health Network Institutional Review Board. The authors followed applicable EQUATOR Network (http://www.equator-network.org/) guidelines during the conduct of this research project.