Safety of a 2-Day Antibiotic Regimen After Delayed Chest Closure Post Pediatric Cardiac Surgery

Abstract

Background:

There is no consensus for the length of prophylactic antibiotics after delayed chest closure (DCC) postcardiac surgery in pediatrics. In September 2014, our institution’s pediatric cardiac intensive care unit changed the policy on length of prophylactic antibiotics after DCC from 5 days (control) to 2 days (study group). The objective of the study was to determine whether a 2-day course of antibiotics is as effective as a 5-day course in preventing blood stream and sternal wound infections in pediatric DCC.

Methods:

Retrospective and prospective study. Primary end points included incidence of sternal wound infections and positive sternal imaging for infection. Surrogate markers of infection were collected at 4 time points.

Results:

During the study period, 139 patients had DCC postcardiac surgery of which 110 patients were included for analysis, 54 patients in the control and 56 in the study group. There was no difference in total number of positive wound cultures/chest computed tomography (CT) findings (4/54 [7.5%] control vs 5/56 [8.9%] study group, P = .3), positive blood cultures (P = .586), median postsurgical length of stay (P = .4), or readmissions within 30 days postsurgery (P = .6). All secondary end points were similar in both groups except peak heart rate between weeks 2 and 4 (P = .041).

Conclusion:

Two days of prophylactic antibiotics is not inferior to 5 days of prophylactic antibiotics after DCC following pediatric cardiac surgery.

Introduction

Delayed chest closure (DCC) is a common practice in pediatric congenital heart surgery, as it has been shown to decrease mortality and minimize complications from postsurgical edema related to open heart surgery and cardiopulmonary bypass.^1,2 However, the practice increases the risk of infection by removing the body’s primary defense mechanism.^1,2 To reduce the risk of infection, these infants have a Gore-Tex patch in place to provide a physical barrier against infection and are treated prophylactically with broad-spectrum antibiotics, while the chest remains open.^1,2 DCC usually occurs within 2 to 4 days after the primary surgery. Although multiple studies have shown DCC significantly increases the risk of infection, increasing the need for prophylactic antibiotics, only a few studies address the length of antibiotics for prophylaxis.^3–5

Based on adult studies, the society of thoracic surgeons (STS) have recommended a prophylactic antibiotics course for 48 hours after cardiac surgery.^6,7 Due to the lack of clinical evidence, practice across pediatric cardiothoracic teams varies nationally,^8,9 with very limited data on the duration of antibiotics post-DCC. An informal survey by our team at the University of Florida of 10 different institutions ranked highly in the US News and World Report showed no consensus on the length of antibiotics after DCC in the pediatric population. The common practices varied between stopping antibiotics at time of chest closure to continuing prophylactic antibiotic coverage for 7 days following chest closure. In the past, it has been the practice in the pediatric cardiac intensive care unit (PCICU) at the University of Florida to continue broad-spectrum antibiotics for 5 days after delayed chest closure to minimize the risk of infection.

In September 2014, based on a review of practices at other institutions, the decision was made to decrease the length of antibiotic prophylaxis after DCC to 2 days. We decided to monitor the rate of sternal infections (SSI) and inflammatory markers of inflammation over the next 2 years to assess for increased incidence of infection. Specifically, our study aimed to determine whether 2 days of antibiotic prophylaxis after DCC provided the same protection as 5 days of prophylactic antibiotic coverage. We hypothesized that 2 days of prophylactic antibiotics after DCC is not inferior in preventing infections as 5 days of antibiotics.

Patients and Methods

Research Plan

This was a retrospective and prospective observational study. The study was approved by the University of Florida Institutional Review Board (IRB). The study included pediatric patients undergoing cardiac surgery and returning to the PCICU unit without primary closure of the chest from July 1, 2012 to March 30, 2016. In September 2014, the antibiotic protocol in the PCICU was changed to decrease the length of prophylactic antibiotics after DCC from 5 to 2 days. Our study evaluated the incidence of postsurgical infection in both the groups. No other changes were made to the standard of care for this patient population. For patients who had cardiac surgery and were discharged prior to IRB approval, a waiver of consent was granted for their inclusion in the study. For patients having cardiac surgery after IRB approval, verbal consent was obtained prior to inclusion in the study. Local IRB waived written consent for our study, as all data were collected as deidentified set, in accordance with our institutional guidelines.

Patient Care

The population included patients with acquired or congenital heart disease from newborn to 21 years of age admitted to the PCICU after cardiac surgery with an open chest and placed on broad-spectrum antibiotic prophylaxis. The historical control group included patients who had cardiac surgery before the change in prophylactic antibiotic course (from July 1, 2012 to August 30, 2014). The study group included patients undergoing cardiac surgery after the change in antibiotic course length (September 1, 2014 to March 30, 2016). All patients received a dose of cefazolin prior to incision for both the initial surgery and the DCC. The prophylactic antibiotics of choice were vancomycin and cefepime. Antibiotics were changed in patients with positive cultures based on organism sensitivities to antibiotics. Data collection began at 24 hours prior to the initial surgery and continued for 30 days following surgery. Data were collected until April 30, 2016 (30 days after the final delayed chest closure in March 2016).

Primary and Secondary End Points

The primary end points included the incidence of sternal wound infections as documented by total number of positive sternal wound cultures and/or positive CT imaging in the clinical setting suspicious for sternal wound infection. Secondary end points include positive blood cultures, length of stay, peak leucocyte counts, C-reactive protein, lactate, temperature, heart rate, lowest pH, and systolic, diastolic, and mean blood pressures collected at 4 time points. The 4 time points were presurgery, 24 hours postsurgery, 1 week after chest closure, and peak values between week 2 and week 4 or week 2 and discharge, whichever occurred first. Patients who underwent heart transplant, extracorporeal membrane oxygenation (ECMO), or ventricular assist device (VAD) were excluded from the study due to their increased risk of infections over the rest of the population.

Data Collection

Laboratory data collection included arterial blood gases, venous blood gases, C-reactive protein, erythrocyte sedimentation rate, complete blood count and differential, basic metabolic panel, complete blood count, liver function panel, and urinalysis. Vital signs included core and peripheral temperatures, heart rate, oxygen saturation, systolic and diastolic blood pressures, and mean arterial pressures were collected. Data on infections or concern for infections included culture(s) date, source, reason, organism, and sensitivities. Antibiotic profile included date, type, length, reason for initiating, and changes in antibiotics with type and reason for changing. Admission data included date, admission diagnosis, date and time of surgery, date and time of chest closure, discharge date, discharge diagnoses, readmission dates, and readmission diagnoses. Laboratory data and vital signs were collected at 4 time points: presurgery, 24 hours postsurgery, 1 week after chest closure, and peak values between week 2 and week 4 or week 2 and discharge if discharged prior to 4 weeks. Congenital heart disease was classified based on the STS classification, which places each of the lesion into 5 “The Society of Thoracic Surgeons-European Association for Cardio-Thoracic Surgery (STAT)” categories. Category 1 includes surgeries with the lowest risk of death, and category 5 includes the surgeries with the highest risk of death.

Statistical Analysis

We use the noninferiority method to determine whether the 2-day regimen was comparable to the 5-day regimen. After assessment of the incidence of the surgical site infections prior to the change in protocol change was assessed, the noninferiority confidence limits were then calculated. Satterthwaite t test was used on both length of stay and age. The log of length of stay was also used to help minimize differences in the 2 groups. Two different statistical methods were used to measure the strength of association between the covariate (study group) and the outcome variables. The first method was logistic regression and the second method was the 2-sided, 2-sample t test. Logistic regression was used for outcome variables that are binary. Two-sided t tests were used for outcome variables that are continuous. No correction for multiple comparisons was applied.

Results

During the study period, 139 patients had DCC after cardiac surgery. Twenty-nine patients underwent a heart transplant, ECMO, or VAD implant and were excluded from the study. One hundred and ten patients were included in our analysis, 54 patients in the control group (5-day regimen) and 56 in the study group (2-day regimen). Median age for the whole cohort was below 1 month (interquartile range [0-1]; Table 1), with no difference between the groups P = .17). Similarly, the median number of days with an open chest was not significantly different between the groups (Table 2; P = .12). During the time period (between control and study group), our severity of case volume increased with an increase in the number of patients ≥STAT 4 mortality score (Table 1; P = .059).

Table 1.

Demographics.

Variable	Control (n = 54)	Study Group (n = 56)	Incidence (n = 110)	P Value
Males, n (%)	31 (57%)	32 (57%)	63 (57%)	.89
Age				.17
Mean, months	5 ± 19	1 ± 2	3 ± 14
Median (IQR)	0 (0-2)	0 (0-1)	0 (0-1)
Range	0-129	0-10	0-129
CPB time, minutes	127 ± 57	138 ± 44	132 ± 51	.24
STAT category 4, n (%)	45 (83%)	39 (69%)	84 (76%)	.16
STAT category 5, n (%)	8 (15%)	15 (27%)	23 (21%)
STAT score ≥ 4, n (%)	7 (13%)	15 (27%)	22 (20%)	.059

Abbreviations: IQR, interquartile range; CPB, cardiopulmonary bypass; STAT, The Society of Thoracic Surgeons-European Association for Cardio-Thoracic Surgery.

Table 2.

Markers of Infection.

Variable	Control (n = 54)	Study Group (n = 56)	P Value
Days with open chest			.12
Mean	3.5 ± 1.2	3.2 ± 1.5
Median (IQR)	3 (3-4)	3 (2-4)
Range	1-9	1-8
Positive sternal wound infection	3 (5.5%)	5 (8.9%)	.21
Positive CT infection	1 (1.8%)	0	.32
Positive blood culture	1 (1.8%)	2 (3.6%)	.58
Length of antibiotics	5.8 ± 2.8	2.7 ± 1.7	<.01
Length of stay			.40
Mean (days)	25 ± 25	31 ± 46
Median (IQR)	18 (12-38)	16 (9-31)
Range	6-175	6-280

Abbreviations: CT, computed tomography; IQR, interquartile range.

Bold face is statistically significant value (p < 0.05).

Imaging and Cultures in All Patients

A total of 8 positive sternal wound cultures (bacterial and fungal) were identified between the 2 groups, 3 were noted in the preintervention group and 5 were noted in the postintervention group (P = .21). Of the 3 isolated positive wound cultures in the 5-day group, one was methicillin-sensitive Staphylococcus aureus (MSSA), one was methicillin-resistant Staphylococcus aureus (MRSA) and other was nonspeciated gram-positive cocci in pairs. Of the 5 cultures positive in the 2 day group included 2 positive for MRSA, 1 for MSSA, and 1 for Candida species and Peptostreptococcus, respectively. Only 1 patient in the 5-day group had a positive chest CT scan showing a sternal abscess and none in the 2-day group. There were no cases of mediastinitis in our entire cohort. The total number of wound infections as defined by positive wound cultures and/or positive chest CTs was 4 (7.6%) of 54 in the control group and 5 (8.9%) of 56 in the study group (P = .3). Three positive blood cultures were noted in both the groups, 2 in the study group and 1 in the control group (P = .586).

Length of Stay

Median presurgical length of stay was 12 days and ranged from 0 to 224 days, with no difference between the groups (Table 2; P = .33). The median postsurgical length of stay was 17 (11-31) days with no difference between the groups (Table 2; P = .4).

Secondary Outcomes

Almost all the secondary end points were similar in both the groups. The only exception was peak heart rate that was significantly higher in the intervention group between week 2 and week 4 (P = .041; Supplemental Table 1).

Readmissions

Of 11 readmissions within 30 days of surgery, 5 were in the control and 6 in the intervention group (P = .6). There were no readmissions concerning sternal wound infection. All readmissions were related to complications of the cardiac surgery, fluid overload (eg, respiratory distress), or feeding intolerance.

Cost of Antibiotics

An estimated cost of antibiotic prophylactic treatment was calculated for 2-day and 5-day antibiotic courses to estimate savings in antibiotic cost associated with the decrease in length of antibiotic treatment. For neonates, we estimate a decrease in medication cost of USD$3.24 per patient. For infants, the decrease in medication cost is estimated at USD$8.10 per patient and for adolescents USD$168.30 per patient. For the patients receiving a 5-day antibiotic regimen, the cost of vancomycin (including trough, medication dispensing, supplies and delivery to the patient room) at our institution is estimated at USD$987 for neonates, USD$995 for infants and younger children, and USD$1262 for adolescents. For the younger patients having a 2-day antibiotic regimen, the estimated cost of vancomycin trough, medication dispensing, supplies, and delivery to the patient room is estimated at USD$412 for neonates, infants, and younger children, while it is USD$465 for older patients and adolescents. Thus, a 2-day antibiotic prophylactic treatment can result in an estimated total cost saving of USD$575 to USD$583 for younger patient and USD$797 for the older patient and adolescent for each patient for the entire antibiotic course. This estimate does not include nurse charges, as data were not available (Table 3).

Table 3.

Cost of Antibiotics.

Variable	Dose	Cost/Dose	Cost/Day	2 Day Cost	5 Day Cost	Difference
Cefepime
Neonates (<30 days)	30 mg/kg q12	$0.37	$0.74	$1.48	$3.70	−$2.22
Infants (>30 days)	50 mg/kg q8	$0.60	$1.80	$3.60	$9.00	−$5.40
Adolescent	2 gm q8	$10	$30	$60	$150.00	−$90.00
Vancomycin
Neonates (<30 days)	10 mg/kg q12	$0.17	$0.34	$0.68	$1.70	−$1.02
Infants (>30 days)	15 mg/kg q8	$0.30	$0.90	$1.80	$4.50	−$2.70
Adolescent	1 gm q8	$8.70	$26.10	$52.20	$130.50	−$78.30
Age-group	Dips Fee	Dispensing	Vanc Trough Cost	Labs	Total Cost
Vanc dispensing/monitoring fees
2 Days	$40 dose	$160	$91	$91	$251
5 Days		$400		$182	$582
Age-Group	Dips Fee	Dispensing	Total Cost
Cefepime dispensing
2 Days	$40 dose	$160	$160
5 Days		$400	$400

Abbreviations: q, every; Vanc, vancomycin.

$: USD.

Comment

Prophylactic antibiotics have been in use since the 1950s. The duration of prophylactic antibiotic regimens postcardiac surgery has been explored in adults, yet minimal literature is available in pediatrics.^3,4 Murray et al were one of the first to describe the use of a particular regimen of antibiotics for prophylaxis following DCC after neonatal cardiac surgery.⁵ In their cohort, they evaluated surgical site infection before and after a standard antibiotic protocol. They evaluated the dosages as well as time of antibiotic administration. For closed sternotomy and DCC, the antibiotic duration was 48 hours and 24 hours postsurgery, respectively. They had a 6.21% to 5.80% rate of SSI in their cohort before and after implementation of the protocol, respectively (P = .88). The median duration to an SSI infection was 20 days, with MSSA being the most common offender. In the results, they do not specifically document the number of patients who underwent DCC. In this perspective, we are the first center to look at an antibiotic protocol focused specifically on DCC, with median age across the study being below 1 month. Our current regimen uses antibiotics for 24 hours after a primary closed sternotomy and for 48 hours after DCC.

Our data indicate that a 48-hour antibiotic regimen following DCC is noninferior to a 5-day antibiotic regimen. The rate of wound infections is comparable to preintervention SSI rate of 7.6% and postoperative SSI rate of 8.8%. Nelson-McMillan et al specially evaluated DCC in infants following cardiac surgery via the STS database and its associated infectious issues.¹⁰ There were 6127 index operations spanning from 2007 to 2013. In their category of infectious complications, endocarditis, pneumonia, blood stream infections, wound dehiscence, and mediastinitis were included. At least 1 infection complication occurred in 18.7%, compared to 6.6% among neonates and infants without their sternums left open (SLO). Our infectious complications with addition of isolated blood stream infections is 9.6%(5/54) versus 12.5%(7/56). The median duration of SLO was 3 days in our cohort, which is similar to the cohort from the STS database review.

In addition, we evaluated for surrogate markers of infection including white cell counts, the percentages of bandemia and neutrophils, markers of poor perfusion, and relevant vitals at 4 time points. These surrogate markers of infection showed no significant changes between the groups. Of all the centers that have explored the use of antibiotic prophylaxis, we are likely the first to include an exhaustive list of clinical surrogate markers. The recommendation for antibiotic duration for the pediatric population following cardiac surgery has yet to be defined. The recommendations are even more obscure when it comes to DCC. The STS recommendations focus predominantly on adults with no mention about DCC. In this context, the work by Nelson-McMillan et al reveals a significant infection rate as high as 18.7% in infants with DCC after cardiopulmonary bypass.¹⁰ With varied antibiotic practices in the pediatric world and the increased risk of infection from DCC, our cohort indicates that transition from a 5-day antibiotic protocol to 2-day protocol can be done safely.

During the time period, our severity of case volume increased with an increase in the number of patients in the STAT 4-5 category, with a difference that neared significance for patient with STAT score≥4 (Table 1). Our case mix mortality decreased to below 1% across all cases in 2014 to 2016. The clinical practices in the care of the patients continued to evolve slowly with no drastic changes other than the length of antibiotic evaluated in this study. The total number of incidence of DCC is low compared to other cardiac surgeries. Attempting to show superiority between the 2 regimens would require a multicenter study; hence, we used a noninferiority model to analyze our results. The same surgeon performed the surgeries during the entire time frame of the study. The average daily cost of antibiotics at wholesale cost at our institution is shown in Table 3. This is the noted cost for our particular institution, which has a contract to purchase the antibiotics in large quantities at much lower prices. This in addition to the data on the cost of vancomycin trough, medication dispensing, supplies, and delivery to the patient room indicates a saving of an estimated USD$600 for the younger patient and USD$800 for the adolescent patient for the complete antibiotic course. This does not include cost of nursing to dispense and the possible medical errors from 3 more days of antibiotics. This intervention has obvious antibiotic stewardship, as it can help decrease antibiotic resistance in the neonatal and pediatric population by limiting the duration of exposure.

There are many limitations to our study, with the foremost being a single-center study with a small sample size. A noninferiority model helps to overcome some of the limitations of having a small sample size, but there needs to be larger sample size to adequately power the study. As this study was brought upon by a change in standard of care, we used a historical control. Furthermore, since the study was started after the change, part of our intervention patients was included as retrospective, and the other part was enrolled prospectively. During the study period, we transitioned from a mixed pediatric ICU to a dedicated pediatric Cardiac ICU in 2014, but this should not limit the generalization of the results, as there was a focused cardiac team before and after the transition. The incidence of readmissions due to surgical infection in both groups may be underestimated, as we could have missed a minority of patients who presented at other hospitals for treatment postdischarge. However, most patients remained in our medical system and followed up with our pediatric cardiologist after discharge.

Conclusion

Two days of prophylactic antibiotics after DCC following pediatric cardiac surgery did not provide inferior protection from sternal wound infection and sepsis in the pediatric population Understanding the limitations of a single-center data, a multicenter collaboration among the STS centers participating in pediatric cardiac surgeries could help delineate the true duration of prophylactic antibiotics post DCC.