Analysis of Sites of Bacterial Contamination in an Enteral Feeding System 
From the * Department of Gastroenterology and
Hepatology, Correspondence: E. M. H. Mathus-Vliegen, MD, PhD, Department of Gastroenterology and Hepatology, Academic Medical Centre, University of Amsterdam, PO Box 22660, 1100 DD Amsterdam, The Netherlands. Electronic mail may be sent to e.mathus-vliegen{at}amc.uva.nl. Background: Contamination of enteral feedings is an often overlooked source for bacterial infection in the intensive care unit. A new 1-L enteral feeding system with minimal chances of touching critical areas (Nutrison Pack) was compared with routinely used 0.5-L glass bottle systems. Methods: Patients admitted to intensive care were randomized to Pack or glass bottle feeding systems. Cultures were taken from the delivery sets 5 times during the day and from feeding containers and different sites of the system after 24 hours. Results: Bacteria were present in 3 of 112 glass bottles and in 2 of 95 Pack bags. True bacterial contamination (defined as >102 colony-forming units/mL, with same bacteria also present in the delivery set) was found in none of the Packs with a 12-h (69 Packs) or a 24-h (26 Packs) hanging time and in only 1 of the glass bottles with a hanging time of 24 hours, which exceeded the advised hanging time of 8 hours. In contrast, the contamination rate of delivery sets was 48%, with increasing bacterial counts over the day and 4 subsequent days. Bacteria mainly belonged to the group of potentially pathogenic bacteria (Enterobacteriaceae and Pseudomonaceae). They likely originated from throat, lungs, and stomach and grew into and along feeding tubes upwards until they reached the delivery set. Conclusions: Prolonged hanging times of Pack bags were safe with respect to bacterial contamination. However, the bacterial safety of enteral feedings is more likely to be endangered by the endogenous route of contamination rather than exogenous contamination, as high bacterial counts were found in feeding tubes and delivery sets as a result of retrograde growth. Enteral feeding is rarely regarded as a source of infection in the highly specialized care of the intensive care patient. To ensure the microbiologic safety of enteral feedings, many improvements have been made in the last 2 decades.1–4 These include the replacement of powdered feedings by ready-to-use feedings, use of collapsible bags instead of vented feeding containers, drip chambers in delivery sets and the use of pumps that constitute a physical barrier against ascending bacteria.5,6 Feeding tubes have also improved in design and materials.7 Despite continuously improving feeding systems, many cases have been reported in which feedings became contaminated with microorganisms, very likely during handling.2–4,8–12 Several reports have described a direct link between contaminated feedings and the development of diarrhea, vomiting, fever, and formula intolerance, as well as colonization and septicemia associated with different bacterial pathogens including Enterobacter cloacae, Klebsiella pneumonia, Pseudomonas aeruginosa, Serratia marcescens, and Escherichia coli, proven by phage typing and plasmid fingerprinting.13–18 Especially in the highly vulnerable, immunocompromised intensive care patient, this might be catastrophic as infectious complications are associated with increased mortality and morbidity.19,20 Finally, changes in nutrition value and physical characteristics of the feeding may occur.21 Therefore, the chance of touching critical areas during connection of the feeding container to the feeding system should be reduced to a minimum. Moreover, prolonged hanging times for feeding containers should facilitate the administration of required amounts of nutrients without the need for additional handling. An enteral feeding system was developed that facilitates hygienic use (Nutrison Pack; Nutricia, Zoetermeer, The Netherlands). In order to assess the performance of the Pack system in an intensive care unit (ICU) setting and to compare it with routinely used systems, we investigated (a) the frequency and etiological agent of contamination of Packs compared with glass bottles; (b) the frequency and course of contamination of delivery sets during the day and over a 4-day time period; and (c) the extent to which contamination is determined by hanging times, location of feeding tube, the presence of microorganisms in throat, stomach, or lungs, or patient characteristics.
Patients Patients admitted to the ICU who were expected to receive enteral feeding for >1 day via a PVC nasogastric tube (Argyle Salem Sump Tube; Sherwood Medical, Tullamore, Ireland) or an endoscopically placed polyurethane jejunal tube (Flocare; Nutricia Healthcare, Chatel-St. Denis, Switzerland) were eligible. Patients were randomized by coded envelopes to a system routinely used or a Pack system during the first 4 feeding days. Patients who participated in the selective decontamination study were not eligible.22 Patients were excluded if they had tubes placed outside the ICU, were fed >24 hours before the present study, or had their tubes replaced or their feeding program interrupted during these 4 days. Patients who left the ICU before the completion of the 4 study days were excluded, as were patients who had to be fed by special feedings that are not commercially available in ready-to-use containers. The routinely used system consisted of a 0.5-L glass bottle (Nutricia) which was connected to the delivery set (Kangaroo; Sherwood Services, Mansfield, MA) by piercing its sterile elastomer cap by a spike. A Y-port for medication and flushing of the tube was present 20 cm from its distal end. According to the ICU protocol, the maximum hanging time of the glass bottle was 8 hours. The Pack system (Nutrison Pack; Nutricia) consisted of a 1-L triple-foil-laminated collapsible bag without the need of an air inlet. The delivery set (Flocare 800; Nutricia) was connected to the Pack by screwing, herewith spiking the sterile seal. The Y-port was at 8 cm from the distal end. The maximum recommended hanging time of the Pack as indicated by the manufacturer was 24 hours. Patients were divided into those who needed 1 or 2 Packs per 24 hours. The flow rate of the feeding was regulated by a Kendall Kangaroo Control 324 Sherwood pump for the glass bottle and a Flocare 800 Nutricia pump for the Pack system. Patients began receiving 20 mL/hour, and feeding was increased by steps of 20 mL/hour according to tolerance on the following days. The study was approved by the Medical Ethical Committee. Age, gender, ICU referral diagnosis, acute physiology and chronic health evaluation (APACHE) score at admission, daily Therapeutic Intervention Scoring System (TISS) score, and medication (antibiotics and acid-suppressing drugs) were recorded.
Methods Samples of food were taken aseptically via the Y-port at the start and 9.5, 12, 16, and 20 hours later after clamping the part delivery access to the distal tube in order to obtain food from the proximal part of the system (Figure 1). Packs that were exchanged after 12 hours were closed with a sterile screw cap. Disconnected bottles were also capped. Both were stored at 4°C until bacterial analysis. At the end of a 24-hour feeding period, the complete delivery set, including bottle or Pack, was replaced and stored at 4°C. The next day, various sites of the feeding system were sampled for bacteriologic analysis.
For bacterial analysis of food and gastric contents, decimal dilutions in physiologic saline were plated on tryptone soy agar. Colony-forming units per mL (cfu/mL) were determined after 2 days of aerobic incubation at 37°C. The detection limit for microorganisms was 1 cfu/mL. Food was regarded as contaminated when >100 cfu/mL were present. To confirm that food in the feeding container was contaminated during use, the same microorganism had to be present in the delivery set. Contaminated food was spread on a blood agar plate for bacterial identification. For throat and bronchial aspirates, a semiquantitative method was applied, using blood agar plates. Positive cultures of throat, stomach, and bronchial aspirate were defined as harboring >103 cfu/mL. For the identification of colonies, standard biochemical and microbiologic techniques were used. The identified microorganisms were categorized into 5 groups according to pathogenicity: group 1, normal throat flora; group 2, low pathogenic bacteria; group 3, potentially pathogenic for a normal population; group 4, potentially pathogenic for the hospital population; and group 5, highly pathogenic bacteria.23–25 Yeasts were a separate group.23,26
Statistical Analysis
Thirty-seven patients (19 men, 18 women) were randomized to Pack or glass bottle systems (Table I). Twenty-seven patients received feeding via a nasogastric tube and 10 via a nasojejunal tube. Patients' characteristics were comparable and differed only with respect to the APACHE score at admission, which was higher in the glass bottle group.
Contamination of Pack and Glass Bottle
Contamination of Delivery Sets
Also, an increase in contamination over the 4-day period was evident. The contamination frequency on day 4 was significantly higher than after similar hanging times on day 1 (p < .05). In addition, there was a correlation between contamination of the delivery set after 20 and 24 hours and the succeeding set (p < .05 and < .01, respectively). The relative risk of a subsequently infected delivery set after a previously contaminated set was 4.1 (95% confidence interval 2.26–7.36). To determine the source of contamination and to study the possibility of retrograde growth, other sites of the delivery set were examined after a 24-hour feeding period (Figure 3). On all 4 days, the contamination frequency in the drip chamber and feeding container was significantly lower than at the distal end and the Y-port (p < .001). The contamination frequency at the Y-port (p = .032) increased significantly over the days, indicating other sources of contamination due to handling by administration of drugs through the Y-port or retrograde growth from the distal end upward into the delivery set.
Determinants of Contamination of Delivery Sets
Relationship Between Species of Microorganisms at Different Places To investigate whether microorganisms in the delivery set were originating from bacteria already present in the patient, positive cultures of the delivery set had to be preceded by positive cultures from the throat, stomach, or lungs on the previous day or earlier the same day containing the same microorganism. The concordance of bacteria is presented in Table III. The bacteria that were found in the contaminated delivery set were also the predominant bacteria in the cultures of throat, lungs, and stomach. These concordant bacteria mainly concerned bacteria belonging to group 4, followed by group 2 bacteria. Microorganisms found at the Y-port were already present in the previous set.
In this study, the safety of the Pack feeding system with regard to microbial contamination was evaluated in daily practice in the ICU and compared with the routine practice of glass bottles. A contamination criterion of 102 cfu/mL was used, according to studies on colonization and infection in immunocompromised patients. International recommendations of acceptable levels of contamination during feeding are lacking, and definitions vary from any bacteria present to >105 cfu/mL.8,12,27–33 In only 1 of 112 bottle-fed cases, levels of >100 cfu/mL were found throughout the whole feeding system at the end of 24 hours, a hanging time that far exceeded the advised hanging time of 8 hours. For another 2 bottles and 2 Packs, similar contamination levels were found in the feeding containers only, at the end of the hanging time without any contamination of the drip chamber of the connected delivery set. It is unclear what caused the contamination in these 4 feeding containers. Possibly, outgrowth of bacteria might have occurred at the rim between cap and bottle, which contaminated the feeding when the feeding container was disconnected to obtain samples for culture.
Although the average hanging time for the Pack (15.1 hours) in this study
was almost twice as long as for the bottles (8.7 hours), no difference was
found in contamination rates. Even Packs that hung for 24 hours did not show
an increased contamination risk, thus demonstrating the safety of the Pack.
The absent contamination of Packs in our study is notably better than the 12%
contamination rate found in surgical and medical
wards.34 The low
level of contamination of glass bottles was an unexpected finding and far
below a 59%–68% contamination in simulated
conditions3 and
4%–15% in clinical
use.10,12,16,27,28
It is unclear whether the low contamination rate for both Pack and bottle
might be ascribed to the higher level of nursing care on the ICU. We
previously reported a 4% contamination rate in ICU patients when using a limit
of 102
cfu/mL,35 whereas
Wagner et al36
found a significant contamination, defined as While handling of the system upon connection and a subsequent risk of contamination of the feeding container have received most attention in the past,5–12 our study demonstrates that the safety of the whole system is not endangered by the feeding container but by the contamination risk of the delivery set. Some indication of this has already been reported previously.9,10,35–39 The numbers of bacteria present in the Pack and bottle were far outnumbered by those present in the delivery set. These bacteria closely resembled the bacteria present in the patient's throat, lungs, and stomach. During the first day and over the successive feeding days, the percentage of contaminated delivery sets increased. As the bacteria did not originate from the Pack or bottle, other sources of contamination should be considered, either due to handling of the Y-port by the administration of drugs or due to bacterial retrograde growth from the distal end of the feeding tube upward into the delivery set. We have been able to confirm that retrograde growth from the feeding tube of the patient is an important source of contamination by comparing the bacteria grown from the tubes to those grown from the patient. Microorganisms present in the patient can enter the feeding tube. The tube then acts as a reservoir, and the delivery set can subsequently become contaminated through retrograde growth. In a nutritious environment, bacteria will multiply. This was demonstrated by a more heavily contaminated distal end of the delivery set where it connects to the feeding tube compared with the Y-port situated 8–20 cm above. This also explains why a subsequent delivery set had a very high risk of becoming contaminated when a previous one was infected. Although not borne out by our study that was based on the results of cultures and not on the outcome of patients, a reduction of the bacterial load in the immunocompromised ICU patient would require a more frequent replacement of feeding tubes, which is rather cumbersome for jejunal feeding tubes, and a more frequent exchange of delivery sets. Cost containment is a major issue in health care, and delivery sets are expensive.40 In this context, the use of delivery sets for 48 hours in the UK2 and recent advice41 to extend the use of delivery sets to 72–96 hours should urgently be reconsidered. Our data support further understanding of the possible causes of infections in critically ill patients. Potential pathogenic hospital flora colonize the digestive tract and subsequently cause nosocomial pneumonia or contaminate the feeding tube and delivery set. The finding that microorganisms found in the feeding at the Y-port did not reflect the normal microbiologic flora and that potential pathogenic hospital-acquired species were found relatively more frequently substantiates this notion. Obviously, a selection takes place that probably depends on microbiologic characteristics such as motility and adherence capacity.1,4,21,23–25,42 The most pathogenic species found in this study (ie, bacteria of group 4) are motile and can therefore more easily grow upstream via the tube toward its connection to the delivery set. Due to this supposed selective retrograde growth, the highest correlation between Y-port and patients' sites was found for Enterobacteriaceae and Pseudomonas species.
Prolonged hanging times of up to 24 hours are safe with the Pack, a system that reduces the chance of touching critical areas to a minimum upon connecting the delivery set to the feeding container. Cheaper standard bottles also performed well for a shorter hanging time of 8 hours. Hitherto, the safety of prolonged use of enteral food was analyzed by a risk analysis of the contamination of the container, which underestimates the risk for the patient as shown by our study. We demonstrated a high risk of contamination of feeding systems near their connection to the feeding tube. It is alarming that Enterobacteriaceae and Pseudomonaceae, which were most frequently found to contaminate the delivery set, are the most pathogenic group of microorganisms isolated in this study. They apparently reside in the patient and grow upwards until they reach the delivery set. Here, in a nutritious environment, they may multiply to high numbers and reenter the patient with the feeding. Therefore, delivery systems should be manufactured without Y-ports or other niches where bacteria can proliferate, tube feedings should not be interrupted to prevent bacterial proliferation during periods of standstill in the delivery set, and delivery sets should be exchanged more frequently. Other options would be to impregnate feeding tubes with antibacterial compounds or by using selective gut decontamination. As selective gut decontamination is used routinely in our ICU our current studies are related to bacterial contamination of feeding systems and bacterial overgrowth. We are much indebted to Dr L. R. Verdooren, retired associate professor of design and analysis of experiments, Wageningen Agricultural University, Wageningen, The Netherlands, and to Dr A. E. Zwinderman, professor of clinical epidemiology and biostatistics, University of Amsterdam, Amsterdam, The Netherlands, for their help with statistics. The study was in part financially supported by a grant from Numico Research B.V., Wageningen, The Netherlands, who had no influence or voice as to the conduct and outcome of the study. Received for publication August 5, 2005. Accepted for publication July 26, 2006.
Journal of Parenteral and Enteral Nutrition, Vol. 30, No. 6,
519-525 (2006) This article has been cited by other articles:
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
|
|
 |
|
Top
MATERIALS AND METHODS
RESULTS
= .05) with 80% power (β = .2), 90 bottles and 90 Packs were
required. Colony counts were log transformed. A value of 0.9 cfu/mL was used
when <1 cfu/mL was found in order to enable log transformation. Descriptive
statistics were used to describe group characteristics. The nonparametric
Wilcoxon test and t-test were used to compare groups. Associations
were analyzed with Yates' corrected 
2 test (Fisher's exact
test, where appropriate) and correlation statistics (Pearson's r or
Spearman's 
). To compare the course of the contamination over the
different times of the day and over the 4 different days, regression lines per
patient were drawn and regression coefficients were used for trend analysis.
Usually, a 2-sided p value < .05 was used. A 1-sided p
value < .05 was chosen when an expected stabilizing or worsening was
investigated. 
100 cfu/mL (cfu, colony-forming units;
