Does decreasing the frequency of changing intravenous administration sets (>24 h) increase the incidence of sepsis in neonates receiving total parenteral nutrition?

Abstract

BACKGROUND:

The optimal timing for changing intravenous (IV) administration sets that contain total parenteral nutrition (TPN), with and without lipids, in neonates remains unknown.

OBJECTIVE:

To determine whether decreasing the frequency of changing IV administration sets (>24 h versus every 24 h) in neonates increases the incidence of sepsis within seven days of discontinuation of TPN and microbial contamination of the infusate.

METHODS:

The databases searched to identify studies that evaluated the frequency of IV administration sets on sepsis and microbial contamination of the infusate included MEDLINE, EMBASE, CINAHL, Cochrane Library, Scopus and Web of Science. The Evidence Evaluation Worksheet adapted from the American Heart Association’s International Liaison Committee on Resuscitation was used to evaluate eligible studies for quality, level of evidence and direction of support.

RESULTS:

Two studies were reviewed; however, neither of the studies reported on the outcome of sepsis. One study reported that changing IV administration sets every 48 h did not increase the rate of infusate (amino acid or lipid) contamination compared with change every 24 h, while the other study reported an increase in the lipid infusate contamination rate when IV administration sets were changed every 72 h.

CONCLUSIONS:

There is insufficient evidence to support or refute routinely changing IV administration sets every 48 h or that decreasing the frequency of set changes increases the incidence of sepsis.

Developed and established in Canadian neonatal intensive care units, EPIQ (Evidence-based Practice for Improving Quality) is a scientific method for continuous quality improvement that is evidence based; targeted on key outcomes; collaborative, involving interprofessional teams of experts; and continuous, promoting a culture of change. Based on the International Liaison Committee on Resuscitation literature review template, the EPIQ review process addresses important clinical questions by summarizing relevant studies. They are intended as guides to best practices and do not represent unique or mandatory protocols. Full versions of these reviews are available at www.epiq.ca.

Khalid Aziz MD, Assistant Editor, Paediatrics & Child Health

Hospitalized neonates require venous access either in the form of peripheral or central venous catheters for provision of total parenteral nutrition (TPN). Adverse effects of intravascular devices include sepsis, phlebitis, extravasation, line migration and thrombosis (1,2). Possible mechanisms for the occurrence of sepsis include invasion of microorganisms at the site of catheter insertion, catheter hub colonization and contamination of the infusate.

In 2010, the Centers for Disease Control (Georgia, USA) recommended that intravenous (IV) administration sets, in both adult and paediatric populations, should be replaced every 96 h if fluids that enhance microbial growth (eg, TPN or blood) are not used. For blood or blood products and TPN, with lipids or lipids alone, the administration sets should be replaced every 24 h (3). The conclusion of a Cochrane review (4) on this topic is consistent with the Centers for Disease Control recommendations. However, the authors state that the optimal period for changing IV administration sets that contain TPN, with and without lipids, is still unknown and should be investigated further. Decreasing the frequency of changing IV administration sets is associated with significant cost savings; however, this benefit needs to be balanced against the risk of sepsis, which is the focus of the present review (4,5).

The primary objective of the present review was to determine whether decreasing the frequency of changing IV administration sets (>24 h versus every 24 h) in neonates increases the incidence of sepsis within seven days of discontinuation of TPN (ie, intralipids alone or in combination with an amino acids plus dextrose solution [AADS]). The secondary objective was to compare the incidence of microbial contamination of the infusate between the two groups.

METHODS

Key words selected with synonyms were: “IV administration sets” OR “IV tubing” OR “IV infusion”; “newborn” OR “neonate” OR “infan*” OR “premature” OR “VLBW” OR “premature infant”; and “sepsis” OR “cross infection” OR “catheter-related infections”. These concepts were combined with the Boolean operator “AND”. Exclusion criteria were animal subjects, abstracts only and children/adult subjects. No language restrictions were applied.

The databases searched were MEDLINE (1946 to January 5, 2012), EMBASE (1980 to January 5, 2012), CINAHL (1982 to January 5, 2012), Scopus (1997 to January 5, 2012), Web of Science (1997 to January 5, 2012) and the Cochrane Central Register of Controlled Trials Fourth Quarter 2011. The bibliographies of all selected articles and several review articles were manually searched for additional studies.

Of the 1147 articles reviewed, no study that reported on the primary outcome was identified. Two studies reporting on the secondary outcome were eligible for inclusion and 1144 studies were not relevant. One study was excluded because it did not compare the frequency of IV tubing change, and the TPN solution was withdrawn from the original bag and transferred into a syringe for administration, with the primary outcome being contamination of the repackaged solution (6). Included studies were scored using the Evidence Evaluation Worksheet adapted from the American Heart Association’s International Liaison Committee on Resuscitation (7) by two authors (VS, NC) for level of evidence (LOE), direction of support and quality. Discrepancies were resolved using a consensus approach.

RESULTS

Two randomized clinical trials (RCTs) were included (5,8). The study characteristics and LOE are presented in Table 1.

TABLE 1.

Characteristics of included studies

Author (ref), year	Study population	Intervention	Outcomes	Comment	LOE
Fox et al (5), 1999	All neonates receiving TPN and admitted to the Level III NICU during a 12-month period at the Children’s Health Centre (Edmonton, Alberta)	Control group (24 h change, n=51) Bag containing the amino acids and dextrose infusion and the lipid emulsion bottles were changed daily along with their IV administration sets Birthweight, g, mean ± SD: 2004±1057 Gestational age, weeks, mean ± SD: 32.7±5.4 Intervention group (48 h change, n=97) Bag containing the amino acids and dextrose infusion were changed daily, while the lipid emulsion bottles were changed with line sets (IV administration sets were changed every 48 h) Birthweight, g, mean ± SD: 2053±1062 Gestational age, weeks, mean ± SD: 32.9±5.2	Primary: Infusate contamination rate Secondary: Positive blood culture rate	There was no correlation between the timing of infusate contamination and positive blood culture; therefore, it is not possible to conclude that contaminated infusate caused blood infection Blood culture was only obtained at the discretion of the physician (ie, based on clinical status of the infant)	1
Matlow et al (8), 1999	Infants admitted to the NICU, The Hospital for Sick Children (Toronto, Ontario) between July 11, 1991 and October 31, 1994 for whom intravenous lipids were prescribed	Control group (24 h change, n=250) Bag containing amino acids and dextrose infusion and the lipid emulsion Contamination of bottles were changed daily along with their administration sets Birthweight, g, mean ± SD: 2318±71 Gestational age, weeks, mean ± SD: 34.2±0.35 Intervention group (72 h change, n=939) IV administration sets changed every 72 h Birthweight, g, mean ± SD: 21933±35 Gestational age, weeks, mean ± SD: 33.9±0.17	Primary: the infusate Secondary: Mortality, number of blood cultures performed, number of positive blood cultures	Cumulative risk of contamination in the 24 h versus 72 h change group was not accounted for	1

IV Intravenous; LOE Level of evidence; NICU Neonatal intensive care unit; ref Reference; TPN Total parenteral nutrition

Fox et al (5) compared the infusate contamination rates in infants receiving TPN when IV administration sets were changed every 24 h versus 48 h and were randomly assigned at a ratio of 1:2. The pharmacy prepared the bag containing the AADS and was changed daily in both groups, and the lipid emulsion bottle was changed with the line sets. Infusate for culture was obtained just before IV administration set change by withdrawing 3 mL from the sampling site, which was distal to the IV access device and below the filter. Fifty-three infants were randomly assigned to the 24 h group and 113 to the 48 h group. Two and 12 infants in the 24 h and 48 h groups, respectively, were excluded because TPN was discontinued. Three infants in the 48 h group died before sampling. One infant randomly assigned to the 48 h group had samples obtained every 24 h and was excluded. For the final analysis, data were available on 148 infants with a total of 2686 infusate samples collected. In both groups, one-third of the infusate samples were obtained from infants with central lines.

Baseline characteristics between the groups were similar. The rate of bacterial contamination of the AADS and lipid emulsion were not significantly different between the 24 h and 48 h groups (3.1% versus 2.9% and 6.0% versus 5.1%). The rate of fungal contamination was significantly greater in the 24 h versus the 48 h group (3.1% versus 0.5% [P<0.01]). This result was influenced by one patient with Candida septicaemia and Candida meningitis. Re-analysis after exclusion of this infant found no difference in the rate of contamination between groups. The number of infants with positive blood culture was not significantly different between the two groups (13 versus 20 infants). The investigators concluded that changing IV administration sets every 48 h instead of 24 h did not increase the infusate contamination rate.

In another RCT, Matlow et al (8) compared the infusate contamination rates in IV tubing of neonates receiving lipid emulsion when IV administration sets were changed every 24 h versus 72 h. Secondary outcomes included mortality, number of blood cultures performed and number of positive blood cultures. Outcomes were recorded from commencement of TPN until seven days after its cessation. The infant remained in the same group if TPN was recommenced within seven days of discontinuation; however, if more than seven days had elapsed, the participant was randomly assigned again and received a new study number (second randomization). Infants were randomly assigned at a ratio of 1:3 because the infusate was sampled three times more often in the 24 h versus the 72 h group. Infants were excluded if IV sets were used for administration of blood or blood products, if disconnected for >4 h or were without sterile gauze coverage. Infusate samples were obtained by disconnecting the IV administration set, and programming the IV pump to deliver 10 mL of infusate into a sterile container (one each from the AADS and lipid delivery set). Blood, urine and cerebrospinal fluid cultures, chest x-rays and complete blood counts were performed as clinically indicated.

A total of 1189 infants receiving TPN were enrolled and accounted for 1278 randomizations. There was no statistically significant difference in the AADS infusate contamination rate between the 72 h versus the 24 h change group (1.1% versus 0.36% [P=0.76]). There was a statistically significant difference in the lipid emulsion infusate contamination rate between the 72 h versus the 24 h change group (3.54% versus 1.35% [P=0.001]). Regression analysis revealed that the risk of infusate contamination was significantly greater in the 72 h compared with the 24 h group (OR 2.69 [95%CI 1.40 to 5.13]; P=0.003). The mortality rate was not different between the two groups. The rate of blood culture sampling was significantly higher in the the 72 h versus the 24 h change group (6.11 versus 4.99 per 100 patient days on TPN [P=0.017]); however, the rate of positive blood cultures was not different between the two groups (2.01 versus 1.76 per 100 patient days on TPN [P=0.36]). Despite the microbial contamination rate being higher in the 72 h versus 24 h set change group, the authors did not consider it to be clinically significant and opted to change their tubing change unit policy for neonates receiving lipid infusion to 48 h.

Quality of the studies

The RCT by Fox et al (5) was of good quality, while the study by Matlow et al (8) was of fair quality. In both studies, the primary outcome was infusate contamination rate rather than sepsis, which is more clinically relevant. Despite the fact that several line cultures were obtained from each infant, each infusate culture was treated as an ‘independent sample’ in both studies. Statistical analyses did not account for repeated sampling. In addition, information was not provided on the number of TPN administration set manipulations (ie, disconnection of sets from the IV device or injection into the administration set).

Major limitations of the Matlow et al (8) study included: randomization of each infant on more than one occasion (ie, more likely to be sick and need TPN); inability to obtain infusate samples in approximately 50% of randomizations (691 of 1278 randomizations) on the weekend; inability to determine whether individuals obtaining infusate samples were blinded to the tubing change groups; cumulative risk of contamination in the 24 h specimens was one day versus three days in the 72 h group; and whether the contamination rate varied depending on the type of vascular access the infant had (ie, peripheral versus central lines).

DISCUSSION

Based on the present review, no studies involving the neonatal population have evaluated the incidence of sepsis when the frequency of IV administration set change was decreased. Regarding microbial infusate contamination rate, in the study by Fox et al (5), the investigators concluded that changing administration sets every 48 h instead of 24 h did not increase the infusate contamination rate, while Matlow et al (8) concluded that despite the microbial contamination rate being higher in the 72 h versus 24 h set change group, they did not consider it to be clinically significant and opted to change their tubing change unit policy to 48 h for neonates receiving lipid infusion.

The practice of changing IV administration sets every 24 h to mitigate the effects of contaminated solutions was based on the reports by Maki et al (9) and Fleer et al (10). Maki et al (9) reported a nationwide epidemic of Gram-negative sepsis in adults that was attributed to contamination of IV products and was linked to a change in the manufacturing process of the cap. Similarly, Fleer et al (10) reported an outbreak of Staphylococcus epidermidis that was linked to contaminated TPN solution when it was transferred from bottles into syringes, which was then infused into the infants.

Subsequent reports (11–13) have suggested that the origin of most IV catheter sepsis is the hub and, therefore, the practice of manipulating the hub during daily changes of IV administration sets was questioned. In the adult literature, there were several studies that documented that increasing the frequency of IV administration set changes from 24 h to 48 h (14) and from 48 h to 72 h (15,16) did not influence the infusate contamination rate and occurrence of sepsis. Lai (17) reported that the phlebitis rate in patients with 72 h versus 96 h IV administration set changes was not significantly different. Comparable studies in the neonatal population are lacking. Of the two eligible studies included in the present review, one was opposed to the clinical question while the other supported it.

Several other factors have been attributed to the occurence of sepsis including the type of IV catheters. Silicone catheters have been associated with reduction in thrombus formation compared with other materials (18) but are associated with increased risk of bacterial adherence compared with the use of polyurethane catheters (19). Recently, catheters impregnated with minocycline and rifampin or bonded with chlorhexidine/silver sulfadiazine have been shown to decrease nosocomial bloodstream infections in the adult population (20,21). However, there are no such data in the paediatric population. Both of the included studies in the present review did not mention of the type of catheters used. Other factors include the number of punctures required to insert the catheters, colonization of the catheter either from the inner lumen or the outside wall of the catheter with organisms originating from the infant’s skin at the time of insertion, handling of the catheter hub and from infusions that are contaminated (6,22,23). Both of the included studies in the present review did not mention the length of time the IV catheters were in place during the study period, number of punctures required to insert the catheters and number of times IV administration sets were accessed.

Consensus on science

There is no evidence that decreasing the frequency of IV administration set changes increases the risk of neonatal sepsis. One RCT found that changing TPN administration sets every 48 h did not increase the rate of bacterial or fungal contamination of amino acids or lipid infusions when compared with sets changed every 24 h and was deemed to be safe (LOE 1) (5). While another RCT found that there was an increased lipid infusate contamination rate in the 72 h change group, however, the rate of positive blood cultures with the same organisms found in the infusate was not significant (LOE 1) (8). None of these studies were powered to detect a difference in the rate of sepsis.

RECOMMENDATION

There is insufficient evidence to support or refute that decreasing the frequency of IV administration set changes increases the incidence of sepsis. There is evidence from one good-quality RCT supporting the safety of changing TPN and lipid IV administration sets every 48 h compared with every 24 h. Based on this single study, there is insufficient evidence to support or refute routinely changing IV administration sets every 48 h.