- Email: [email protected]
Risk factors for central vascular catheter-associated bloodstream infections among patients in a neonatal intensive care unit
Journal of Hospital Infection (2001) 48: 108–116 doi:10.1053/jhin.2001.0984, available online at http://www.idealibrary.com on
Risk factors for central vascular catheter-associated bloodstream infections among patients in a neonatal intensive care unit L. M. Mahieu*, A. O. De Muynck‡, M. M. Ieven†, J. J. De Dooy*, H. J. Goossens† and P. J.Van Reempts* *Departments of Paediatrics, Division of Neonatology and †Clinical Microbiology, University Hospital of Antwerp, Belgium and ‡The Department of Clinical Sciences, Institute of Tropical Medicine, Belgium
Summary: The aim of this study was to identify risk factors for catheter-associated bloodstream infection (CABSI) in neonates. We undertook a prospective investigation of the potential risk factors for CABSI (patient-related, treatment-related and catheter-related) in a neonatal intensive care unit (NICU) using univariate and multivariate techniques. We also investigated the relationship between catheter hub and catheter exit site colonization with CABSI. Thirty-five episodes of CABSI occurred in 862 central catheters over a period of 8028 catheter-days, with a cumulative incidence of 4.1/100 catheters and an incidence density of 4.4/1000 catheter days. Factors independently associated with CABSI were: catheter hub colonization (odds ratio [OR]:44.1, 95% confidence interval [CI]:14.5 to 134.4), exit site colonization (OR:14.4, CI:4.8 to 42.6), extremely low weight (1000 g) at time of catheter insertion (OR:5.13, CI:2.1 to 12.5), duration of parenteral nutrition (OR:1.04, CI:1.0 to 1.08) and catheter insertion after first week of life (OR:2.7, CI:1.1 to 6.7). In 15 (43%) out of the 35 CABSI episodes the catheter hub was colonized, in nine (26%) cases the catheter exit site was colonized and in three (9%) cases colonization was found at both sites. This prospective cohort study on CABSI in a NICU identified five risk factors of which two can be used for risk-stratified incidence density description (birthweight and time of catheter insertion). It also emphasized the importance of catheter exit site, hub colonization and exposure to parenteral nutrition in the pathogenesis of CABSI. © 2001 The Hospital Infection Society
Keywords: Surveillance; central catheter; colonization; antibiotics; neonate; risk; sepsis; low birth weight; bloodstream infection; hub; exit site.
Introduction Bloodstream infections are the most frequent nosocomial infections in neonatal intensive care unit Received 8 November 2000; revised manuscript accepted 21 February 2001 Author for correspondence: Ludo M. Mahieu, Department of Paediatrics, Division of Neonatology, University Hospital of Antwerp, Wilrijkstraat 10, B-2650 Antwerp, Belgium. E-mail: [email protected]
0195-6701/01/020108;09 $35.00
(NICU) patients. Their cumulative incidence is increasing as a result of the higher survival rate of very low birth weight infants and the increased use of central vascular catheters.1 Catheter-associated bloodstream infections (CABSI) in the NICU contribute significantly to hospital morbidity as well as to increased costs due to prolonged hospitalization.2 Both umbilical and central intravenous catheters are major risk factors for septicemia.3 Micro-organisms colonizing the catheter exit site and hub are able to
© 2001 The Hospital Infection Society
Risk for catheter sepsis in neonates
migrate along the external and internal surface to the catheter tip and cause sepsis. There are only a few reports on the epidemiology of catheter-associated infections that include both types of catheters in neonates; particularly lacking are figures on infections related to umbilical line.4,5 Moreover, in the studies reported, varying definitions are used which makes it very difficult to compare study results. A limited number of factors has been found to be associated with CABSI in neonates. We found only two studies in which adjustments were made, through multivariate analysis technique, for duration of catheterization and other potential confounding factors.6,7 To identify targets for the prevention of CABSI, by assessing the cumulative incidence and identification of independent risk factors, a prospective epidemiological study of CABSI among neonates in a NICU was undertaken. We also studied the association between catheter hub and exit-site colonization and CABSI in both umbilical as well as non-umbilical central vascular catheters. Methods Study population This prospective study was conducted from 1 November 1993 through 31 October 1995 in a 19-bed NICU at the University Hospital of Antwerp, Belgium, and included all neonates requiring at least one central vascular catheterization including umbilical (arterial and venous) and non-umbilical (subclavian and peripheral) central vascular lines. Each case of catheterization was individually evaluated. The catheter utilization ratio was calculated by dividing the number of catheter-days by the number of patient-days of the total study population. Total Parenteral Nutrition (TPN) utilization ratio was calculated in the same manner with the number of TPN days as the numerator. Protocol for catheter care Aseptic technique was used during insertion and repositioning; this included scrubbing of the nails with a brush and a two minute vigorous rubbing of the skin of the hands and arms, using 4% chlorhexidine (Hibiscrub®, Zeneca, Destelbergen, Belgium), the use of sterile gloves, drapes, gowns and facemasks. Before inserting the catheter, the patient’s
109
skin was disinfected twice by rubbing the site of insertion with sterile gauze soaked in a solution of 2% chlorhexidine in 70% isopropyl alcohol and was allowed to dry. The exit site of non-umbilical central vascular catheters was covered with sterile gauze held in place by an occlusive transparent dressing (Tegaderm®, 3M, London, Ontario, Canada). No topical antibiotics were used. This dressing was only changed if there were problems of catheter occlusion or clinical signs (e.g. temperature instability, hepatosplenomegaly, etc.) and symptoms (e.g. apnoeic spells, bradycardia, etc.) of sepsis. The exit site of umbilical lines remained uncovered, was held in place using sterile sutures (Ethilon®, Ethicon Inc., Johnson & Johnson International, Somerville, NJ, USA) and secondarily fixed with a tape to the abdominal skin. The umbilical stump was cleansed three times daily with sterile gauze soaked in a solution of 2% chlorhexidine in 70% isopropyl alcohol prior to the application of a powder containing 1.5% virginiamycin and 0.5% neomycin sulphate (Spitalen®, Bencard, Genval, Belgium). Three-way stopcocks connecting the hub with the intravenous sets were changed every 48 h, or every 24 h when used for TPN administration. The stopcocks and hubs were disinfected with a locally prepared solution of 2% chlorhexidine in 70% isopropyl alcohol using a sterile swab immediately before and after each manipulation and wrapped in sterile gauze dressing. Antibiotics were not used prophylactically, but only for treatment of suspected infections. No blood products were administered through the central vascular catheters. Data collection The following data were recorded at the time of central vascular catheter insertion: date and site of insertion (umbilical artery or umbilical vein, peripheral antecubital, saphenous or subclavian vein); venue of insertion (operating room, bedside in the NICU, other hospital); physician’s name and position (medical officer, surgeon, anaesthetist, neonatologist); whether the patient was requiring antibiotic treatment at the time of insertion; patient’s weight and postnatal age. The following data were recorded while the central vascular catheter was in place: need for repositioning as determined by X-ray; use of steroids, duration of TPN; duration of artificial ventilation; duration of catheterization, development of catheter hub and
L. M. Mahieu et al.
110
exit site colonization. In the event of CABSI, only those factors preceding the date of diagnosis were taken into consideration.
hypothermia (37⬚C), apnoea, or bradycardia] and (2) central catheter present at the time the blood culture was obtained, as previously described.9
Microbiological procedures
Statistical methods
Swabs were taken from both the catheter exit-site and hub at the time of sepsis evaluation as well as at catheter removal in those catheters not associated with infection. After disconnecting the tubing from each central vascular catheter Luer connector (hub), a sterile dry cotton swab was inserted into the hub and rotated twice through 180⬚. A culture was taken from the skin-catheter junction with another sterile cotton swab, which was rubbed along a surface area of 1 cm2, after removal of the dressing. In the microbiology laboratory, each swab was plated on a 5% horse blood agar plate. The inoculum was spread with a loop over the plate. Plates were incubated at 37⬚C for 48 h and kept at room temperature for a further two to three days. Cultures were considered positive when at least two out of the four inoculated quadrants revealed bacterial or fungal growth. Isolates were identified by standard methods to species level and their antibiotic susceptibility was determined. Coagulase-negative staphylococci were identified to species level using a rapid inhouse developed identification system.8 Paired quantitative blood cultures were taken using the DuPont Isolator 1.5® system (Unipath limited, Hampshire, UK) when clinical signs and/or symptoms suggested sepsis. If possible, one set (1 mL blood) was taken through the central vascular catheter and a second set from a peripheral vein. In cases where it was not possible to draw blood from the central vascular catheter, the non-quantitative Bactec® system (Becton-Dickinson, Cockeysville, MD, USA) was used to collect 1 mL blood from two separate peripheral veins. Only when both sets revealed the same organism (identical biotype and antibiogram), were these blood cultures considered positive.
Firstly, a descriptive analysis of risk factors for the first episode of CABSI was performed for each of three risk categories: (1) patient characteristics, (2) therapeutic factors and (3) catheter-related factors. Cumulative incidences (infections/100 catheters) for all risk factors were assessed by means of the Yates corrected Chi-square test or the two-tailed Fisher’s exact test. Secondly, to correct for catheterization duration as a potential confounder, the incidence density (infections/1000 catheter days) for each risk factor was determined. Variables with a P-value 0.20 were retained for inclusion in multivariate analysis. After excluding the catheters from which the results of hub or exit site culture were missing, 658 of the original 862 catheters remained for multivariate analysis. The stepwise logistic regression method was used for identification of independent predictors and for calculation of odds ratios (OR) and their 95% confidence intervals (CI). Significant interactions among variables were retained in the selected model, after ascertainment by stepwise regression. Finally, the correlation between catheter duration and duration of parenteral nutrition was determined. All analysis was performed with Stata 5.0 statistical software (State Corporation, Texas, 1996).
Definitions Catheter-associated bloodstream infection (CABSI) was defined as follows: (1) Primary bloodstream infection according to the CDC surveillance definition: a) recognized pathogen isolated from blood culture or a skin contaminant isolated from two blood cultures drawn on separate occasions, b) one of following clinical signs of infection [fever (38⬚C),
Results Study population We followed up 862 episodes of CVC catheterization with a total of 8028 catheter days among 441 neonates. The mean number of catheterization episodes per patient was 1.9 with a maximum of eight episodes. Mean birth weight of the neonates was 1796 g (range 518–5700 g); 47% weighed less than 1500 g and 18% less than 1000 g. The gestational age of the neonates varied from 24 to 43 weeks with a mean of 33 weeks. Fifty-nine percent were male. The median (quartile Q1–Q3) catheter utilization ratio did not change significantly with birth weight (1000 g:0.66 (0.35–1.07); 1001–1500 g: 0.75 (0.50–1.04); 1501–2500 g:0.73 (0.50–1.0); 2500 g:0.81 (0.53–1.0), P:0.556). In contrast, the median (quartile Q1–Q3) TPN utilization
Risk for catheter sepsis in neonates
111
ratio differed significantly between birth weight categories (1000 g:0.31 (0.09–0.52); 1001–1500 g: 0.35 (0.14–0.52); 1501–2500 g:0.21 (0.0–0.55); 2500 g:0.0 (0.0–0.0), P0.001). Characteristics of catheter insertion episodes (Table I) Umbilical lines, especially venous lines, were the most frequently used central vascular catheters. Peripheral percutaneously inserted central vascular catheters were the most frequent non-umbilical lines. A medical officer inserted by far the most of the catheters. At the time of catheter insertion, about one third of the neonates had been treated with antibiotics. Ampicillin plus netilmicin were used for early-onset infections (72 h after birth) and vancomycin with cefotaxime for hospitalacquired infections. All pathogens isolated from the blood cultures were sensitive to one of these antibiotics with exception of one patient with Pseudomonas paucimobilis sepsis. The mean catheterization
Table I
Characteristics of the episodes of catheterization at insertion
Characteristic
Number
No of catheterization episodes Insertion site of central vascular catheters, % Umbilical (UAC*, UVC†) Non-umbilical (PICVC‡, subclavical) Place of insertion, % Bedside Other hospital Operating theatre Physician responsible: medical officer, % Antibiotic treatment at insertion, % Duration of catheterization (total number of days) Overall: Mean (SD§, range) Catheter type: Mean (SD, range): Umbilical Non-umbilical Patient’s weight: Mean (SD, range): 1000 g 1000 g Use for parenteral nutrition, % Average duration of parenteral nutrition, mean (SD, range) Reposition of CVC||, % Patient ventilated during use of CVC, % Average duration of ventilation, mean (SD, range) Use of steroids, %
862 68 (24, 45) 32 (27, 4) 94.9 3.9 1.2 85 33.8 8028 9.3 (9.3, 1–110) 6.4 (4.2, 1–23) 15.7 (13.4, 1–110)¶ 12.2 (10.8, 1–110) 8.7 (8.9, 1–75) 43 11.3 (10.6, 1–70) 21 34 5.6 (1–69) 4
*UAC, umbilical arterial catheter; †UVC, umbilical venous catheter; ‡PICVC, peripheral inserted central vascular catheter; §SD, standard deviation; ¶P0.001 by Kruskal-Wallis test; ||CVC, Central vascular catheter.
duration was significantly shorter for umbilical than for non-umbilical central vascular catheters. Almost half of the central vascular catheters were used for long-term administration of TPN. Approximately one fifth of all central vascular catheters needed repositioning. There was a strong linear correlation between duration of catheter use and duration of TPN administration (r:0.83, P0.001). About one third of the patients were being artificially ventilated while the catheter was in place. During catheterization, steroids were administered in about 4% of the central vascular catheters. Catheter infections A total of 35 CABSI episodes were recorded in 862 catheters. This amounts to a cumulative incidence of 4.1/100 central vascular catheters and an incidence density of 4.4/1000 catheter days. The cumulative incidence differed according to the type of catheter: umbilical arterial catheters 2.5%; umbilical venous catheters 2.6%; peripheral inserted central venous catheters 7.2% and subclavian catheters 8.1%. The cumulative incidence was almost three times higher for non-umbilical central vascular catheters than for umbilical central vascular catheters (7.4 vs. 2.5%, P0.01). In contrast, after correction for duration of catheterisation, the incidence density was not significantly different (4.7/1000 vs. 4.3/1000 catheter-days, P0.05). The incidence densities for the different types of catheter were: umbilical arterial catheters 4.4/1000 catheter days, umbilical venous catheters 3.8/1000 catheter days, peripheral inserted central venous catheters 4.7/1000 catheter days and subclavian catheters 4.3/1000 catheter days. It was only possible to do paired quantitative blood cultures in 20% (7/35) of the CABSI episodes. In all cases the blood cultures (1 mL) taken from the central vascular catheter yielded a higher concentration of bacteria than the blood cultures (1 mL) taken from a peripheral vein (factor10). All blood cultures yielded at least 1000 cfu/ml blood. Of the pathogens responsible for CABSI, two thirds (69%) were coagulase-negative staphylococci (18 Staphylococcus epidermidis, two Staphylococcus warneri, three Staphylococcus hominis, one Staphylococcus capitis), and only 14% were Gramnegative organisms (Figure 1). No catheter-related mycotic infections were found. In 15 (43%) of the 35
L. M. Mahieu et al.
112
CNS* S. aureus E. fecalis K. pneumoniae S. marcescens P. paucimobilis 0
5
10
15
20
25
Cases Figure 1 Micro-organisms recovered from blood cultures in 35 episodes of catheter-associated blood-stream infections in neonates. Gram-positive; *Coagulase-negative styphylococcus.
CABSI episodes the catheter hub was colonized, in nine (26%) cases the catheter exit site was colonized and in three (9%) cases colonization was found at both sites. Thus catheter colonization at the exit site and/or hub was found in 27 out of 35 (77%) CABSI. In this study, exit site colonization had 32% sensitivity (11/34) and 94% specificity (610/648) for CABSI. Its negative predictive value (NPV) was 96% (610/633) and its positive predictive value (PPV) was 23% (11/49). For hub colonization, we found 46% sensitivity (15/33) and 97% specificity (625/646) for CABSI. The NPV for hub colonization was 97% (625/643) with a PPV of 42% (15/36). Univariate analysis (Table II) allowed the retention of 10 out of 13 potential predicting factors (P 0.20) for inclusion in the multivariate model: weight1000 g at time of insertion; catheter colonization of the hub; exit site colonization; duration of total parenteral nutrition; insertion by consultant in charge of the neonate; insertion in the operating room; non-umbilical insertion; use of corticosteroids; postnatal age of more than 1 week at insertion and catheterization duration. Artificial ventilation, repositioning of the central vascular catheter and antibiotic treatment at time of insertion were not associated with CABSI. The independent risk factors predictive for CABSI are shown in Table III. After adjustment by logistic regression, besides catheter hub and exit site colonization, only TPN duration during catheter use, extremely low weight at time of CVC insertion (1000 g) and postnatal age of more than 1 week at insertion were significantly associated with CABSI. The risk of CABSI increased with increasing duration of TPN exposure: adjusted OR:1.0 for 0 days; OR:4.36 for 1–7 days (95%
CI:1.59–11.90, P:0.004); OR:7.08 for 8–14 days (95% CI:2.77–18.11, P0.001); OR:6.24 for 14 days (95% CI:2.04–19.13; P:0.001) (Figure 2). The infection risk for patients with a weight of1000 g at CVC insertion was approximately three times greater than that of patients with a weight 1000 g (%CABSI, 9.2 for those 1000 g; %CABSI, 3.0 for those 1000 g). The cumulative incidence of CABSI for patients with a postnatal age of more than 1 week at insertion was approximately three times higher than those less than 1 week at insertion (8.7 vs. 3.2%). The catheterization duration itself was not an independent risk factor for CABSI. Discussion Although indispensable for the treatment of sick neonates in the NICU, central vascular catheters present a major risk for nosocomial bacteraemia as well as the increased cost of hospitalization.2,3 Surveillance of CABSI and identification of risk factors are essential if effective preventive strategies are to be developed. However, due to differences in definition and lack of adjustment for duration of catheterization, the results of published studies are difficult to compare and the strength of the identified risk factors should be viewed with caution, as most analyses have not been adjusted for potential confounding factors. The laboratory confirmation of catheter-related sepsis in neonates is difficult for several reasons: (1) paired quantitative cultures of blood obtained from peripheral vein and through the central line is frequently impossible; (2) in practice, central lines are not removed to perform cultures of the tip at sepsis evaluation because of the drawbacks and uncertain value of reinsertion of a central line in small neonates; and (3) the use of molecular epidemiological techniques, to show identity of organisms isolated from the blood and other sites is not routine in most hospitals. Therefore, for surveillance purposes, we defined CABSI as a primary bloodstream infection in association with a central line as suggested by others.9 In our neonates, the CABSI cumulative incidence was found to be almost three times higher when non-umbilical lines were used (7.4 vs. 2.5% in umbilical lines). This association was, however, not significant when the incidence density was compared (4.7/1000 vs. 4.3/1000 catheter days in umbilical lines). Although the daily attributable
Risk for catheter sepsis in neonates
Table II
113
Risk factors for catheter-associated bloodstream infection: univariate analysis (n:862)
Risk Factor
Modality
Cumulative incidence/ 100 CVC*
Incidence density/1000 catheter days
Catheterization duration§
0–7 8–14 14 Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No 0 1–7 8–14 14 Yes No
2.0 5.9¶ 7.4¶ 22.4|| 3.6 41.7|| 2.8 7.1¶ 2.7 5.2 3.5 4.4 3.4 9.2|| 3.0 8.0** 3.4 8.7¶ 3.2 3.9 4.1 15.2¶ 3.6 1.4 5.9¶ 9.2|| 8.2¶ 2.5¶ 7.4
5.0 5.6 3.0 23.1|| 3.8 26.2|| 3.0 4.7 4.3 4.3 4.7 4.9 3.8 8.2** 3.5 7.0 4.0 6.2 4.0 4.5 4.5 7.6 4.2 2.6 7.2** 7.2** 3.0 4.3 4.7
Exit site colonization§ Hub colonization§ Insertion at bedside§ On antibiotics at insertion§ Patient ventilated§ Patient’s weight 1000 g§ Physician: Consultant§ Postnatal age7 days§ Reposition§ Steroid use§ TPN†† duration (days)§
Umbilical insertion§
OR†
1 3.06 3.95 7.7
95% CI ‡
P
1.26, 7.53 1.4, 11.12 3.5, 16.91
0.01 0.006 0.001
24.8
11, 55.8
0.001
1.6
1.2, 2.3
0.002
1.49
0.75, 2.96
0.25
1.3
0.64, 2.7
0.461
3.3
1.6, 6.6
0.001
2.4
1.14, 5.21
0.021
2.9
1.4, 5.98
0.004
0.9
0.4, 2.2
0.869
4.8
1.7, 13.2
0.003
1 4.35 7.08 6.2 2.7
1.59, 11.9 0.004 2.77, 18.11 0.001 2.04, 19.13 0.001 5.26, 1.43 0.004
*CVC, central vascular catheter; †OR, Odds Ratio; ‡CI, Confidence Interval; §Retained for inclusion into multivariate analysis, P0.20; ¶Yates-corrected Chi-square P0.01; ||Yates-corrected Chi-square P0.001; ** Yates-corrected Chi-square or Pearson’s Chi-square test P0.05; †† TPN, total parenteral nutrition. 8
Odds ratio for CABSI
7 6 5
Table III infection
4
Risk Factor
OR*
95% CI†
SE‡
Hub colonization Exit site colonization Weight1000 g Postnatal age 7 days TPN|| days (until CABSI**) Hub and exit site colonization
44.1§ 14.4§ 5.13§ 2.74¶ 1.04¶ 0.06¶
14.5, 134.4 4.8, 42.6 2.1, 12.5 1.1, 6.7 1.0, 1.08 0.01, 0.5
25.1 7.97 2.3 1.25 0.02 0.07
3 2 1 0
1–7 8–14 TPN exposure (days)
> 14
Figure 2 Effect of exposure to parenteral nutrition (TPN) on the risk for catheter-associated bloodstream infection (CABSI). Odds ratio for CABSI; Lower Limit 95% CI; Odds Ratio.
Independent risk factors for catheter-associated bloodstream
Observation 658, Chi-square:92.12, P0.0001, Log Likelihood:984.9, area under receiver operating characteristic curve:0.91 *OR, Odds Ratio; † CI, Confidence Interval; ‡SE, Standard Error; §P0.001; ¶P0.05; ||TPN, total parenteral nutrition; **CABSI, catheter-associated bloodstream infection.
114
risk for infection in umbilical and non-umbilical lines does not differ significantly, the cumulative incidence of the latter is higher because the duration of catheterization is more than twice as long (Table I). In one study, umbilical artery catheters were found to be associated more frequently with sepsis than umbilical venous catheters (5 vs. 3%).5 These findings were not confirmed in our study (2.5 vs. 2.6%). In our study, a CABSI rate of 8.1% was found in subclavian lines which is at the lower end of the range of 6–46% in infants with Broviac catheters.10,11 With peripherally inserted central vascular catheters, a CABSI cumulative incidence of 7.2% was found, which is consistent with the published figures of 1.3–12.9%.12–16 Only one study of neonates revealed a CABSI incidence density of 1/1000 catheter-days which is much lower than the 4.7/1000 found in this cohort study.14 In that study, the infection rate may have been underestimated as only catheter hub-related infections were reported. The incidence densities of CABSI of the present study, stratified by birth weight categories are about half of those in the National Nosocomial Infections Surveillance study (NNIS) of the CDC.4 The finding that the incidence density increases with decreasing birth weight corroborates the NNIS data. Little is known about the factors that predispose to CABSI in NICU patients. In arterial umbilical lines, very low birth weight and antimicrobial therapy, and in venous umbilical lines infusion of TPN and high birth weight have been identified as risk factors.5 This study revealed one independent patient-related risk factor, besides catheter colonization: extremely low weight (<1000 g) at the time of catheter insertion. Different factors may be responsible for the higher frequency of CABSI in low weight infants also found in other studies. Firstly, longer exposure to TPN including intravenous lipid emulsions as shown by the TPN utilization ratio in the study population. This is supported by data for very low birth weight infants for whom intravenous lipid emulsions were a major determinant of coagulase-negative staphylococcal bacteraemia, even after adjustment for severity of illness.17 Secondly, more severe clinical conditions leading to increased catheter manipulations. The Day 1 Score for Neonatal Acute Physiology (SNAP) and the Clinical Risk Index for Babies (CRIB) are useful in predicting the risk of bacteremia;18,19 the immature immune system may also be a factor.20 Some risk factors for CABSI reported by others, such as antimicrobial therapy, were not
L. M. Mahieu et al.
found significant in this particular study.14,15 One study revealed that a low dose of vancomycin in TPN solutions is effective in the prevention of nosocomial sepsis in extremely low birth weight infants.21 The relationship found by others between duration of catheterization and risk for CABSI14 was not confirmed. There was a strong correlation between duration of catheterization and duration of TPN infusion therapy, the latter being associated with CABSI cumulative incidence (Table II). Therefore catheterization duration must be considered as a confounding factor for CABSI and so infection rates should be corrected by using incidence density measures (CABSI per 1000 catheter-days).13 When crude nosocomial sepsis densities are compared between NICUs, they should be adjusted for line utilization ratio in the units, as is done in the NNIS.4 We believe this is the first published report on the importance of both hub and catheter exit sites as risk factors for CABSI in neonates. A previous study, found that catheter exit site and especially catheter hub colonization at the time of clinical symptoms of infection correlated with hospitalacquired sepsis in neonates. Addition of catheter exit site and hub cultures to a diagnostic scoring system, composed of laboratory and clinical parameters of infection, significantly improves the discriminatory power of the latter.22 We found a high negative predictive value for CABSI of exit site and hub colonization (96 and 97% respectively) but a rather low positive predictive value if the exit site or hub was colonized (23 and 42% respectively). The predictive values depend very strongly on the prevalence of the disease studied. As a consequence, higher positive predictive values would be expected when the prevalence of CABSI is higher in the patient population than in our cohort (overall 4.1%). In postoperative adult patients with central lines, one recent study has shown a 100% negative predictive value for catheter tip colonization if neither the skin nor the hub was colonized, and a positive predictive value of 68% if the skin and 52% if the hub was colonized.23 Similar observations were made in adults receiving TPN24 where catheter exit site and especially hub colonization were strongly associated with CABSI (OR:6 and 19 respectively). This supports the hypothesis that both catheter hub and exit site colonization is important in the pathogenesis of CABSI. Although several studies in adults have proven the catheter hub to be an important portal of entry for organisms causing CABSI,23–25 only one study supports this in neonates.26 A recent
Risk for catheter sepsis in neonates
study of neonates using genotyping techniques, revealed that 66% of coagulase-negative staphylococci on central vascular catheter-tips, originated from stopcocks.27 We believe this is the first study suggesting that prolonged hospitalization prior to catheter insertion is associated with increased infectious catheter complications. A previous study has shown that length of stay is an independent risk factor for hospitalacquired bacteraemia with coagulase-negative staphylococcal which are major pathogens in central vascular catheters.28 Duration of catheterization may be another contributing factor to higher infection rates in catheters placed after the first week of life. Indeed, in the first week the catheters inserted are usually short-term umbilical lines. This explains why the incidence densities are equal but not the cumulative incidences between catheters placed before and after the first postnatal week (Table II). Both umbilical and percutaneous central vascular lines are associated with severe infectious complications in neonates. As colonization of the exit site and hub are important risk factors for CABSI, more studies are needed to explain the pathogenesis of catheter hub and exit site colonization. The adjusted infection rates described here provide specific surveillance data for interhospital comparison and also to assess the implementation of preventable measures to control infections in neonates.
Acknowledgements The authors thank Dr Guillaume Van Melckebeke for his help in data collection and data entry into the computer and Mr Rizwan Afzal for his biostatistical assistance. They are most grateful to the clinicians that recorded the data at their patients’ bedside.
References 1. Gladstone IM, Ehrenkranz RA, Edberg SC, Baltimore RS. A ten-year review of neonatal sepsis and comparison with the previous fifty-year experience. Pediatr Infect Dis J 1990; 9: 819–825. 2. Leroyer A, Bedu A, Lombrail P et al. Prolongation of hospital stay and extra costs due to hospital-acquired infection in a neonatal unit. J Hosp Infect 1997; 35: 37–45. 3. Gaynes RP, Edwards JR, Jarvis WR, Culver DH, Tolson JS, Martone WJ. Nosocomial infections among neonates in high-risk nurseries in the United States. Pediatrics 1996; 98: 357–361.
115
04. CDC NNIS System. National Nosocomial Infections Surveillance (NNIS) Report, May 1997. Am J Infect Control 1997; 25: 477–487. 05. Landers S, Moise AA, Fraley JK, Smith EO, Baker CJ. Factors associated with umbilical catheter related sepsis in neonates. Am J Dis Child 1991; 145: 675–680. 06. Maas A, Flament P, Pardou A, Deplano A, Dramaix M, Struelens MJ. Central venous catheter-related bacteraemia in critically ill neonates: risk factors and impact of a prevention programme. J Hosp Infect 1998; 40: 211–224. 07. Pandit PB, Pandit FA, Govan J, O’Brien K. Complications associated with surgically placed central venous catheters in low birth weight neonates. J Perinatol 1999; 19: 106–109. 08. Ieven M, Verhoeven J, Pattyn SR, Goossens H. Rapid and economical method for species identification of clinically significant coagulase-negative staphylococci. J Clin Microbiol 1995; 33: 1060–1063. 09. Sherertz RJ. Surveillance for infections associated with vascular catheters. Infect Control Hosp Epidemiol 1996; 17: 746–752. 10. Grisoni ER, Mehta SK, Connors AF. Thrombosis and infection complicating central venous catheterisation in neonates. J Pediatr Surg 1986; 21: 772–776. 11. Sadiq HF, Devaskar S, Keenan WJ, Weber TR. Broviac catheterisation in low birth weight infants: incidence and treatment of associated complications. Crit Care Med 1987; 15: 47–50. 12. Durand M, Ramanathan R, Martinelli B, Tolentino M. Prospective evaluation of percutaneous central venous silastic catheters in newborn infants with birth weight of 510 to 3920 grams. Pediatrics 1986; 78: 245–249. 13. Loeff DS, Matlak ME, Black RE, Overall JC, Dolcourt JL, Johnson DG. Insertion of a small central venous catheter in neonates and young infants. J Pediatr 1982; 17: 944–949. 14. Chathas MK, Paton JB, Fisher DE. Percutaneous central venous catheterisation. Am J Dis Child 1990; 144: 1246–1250. 15. Neubauer AP. Percutaneous central iv access in the neonate: experience with 535 silastic catheters. Acta paediatr 1995; 84: 756–760. 16. Cairns PA, Wilson DC, McClure BG, Halliday HL, McReid M. Percutaneous central venous catheter use in the very low birth weight neonate. Eur J Pediatr 1995; 154: 145–147. 17. Avila-Figueroa C, Goldmann DA, Richardson DK, Gray JE, Ferrari A, Freeman J. Intravenous lipid emulsions are the major determinant of coagulasenegative staphylococcal bacteremia in very low birth weight newborns. Pediatr Infect Dis J 1998; 17: 10–17. 18. Gray JE, Richardson DK, McCormick MC, Goldmann DA. Coagulase-negative staphylococcal bacteremia among very low birth weight infants: relation to admission illness severity, resource use and outcome. Pediatrics 1995; 95: 225–230. 19. International Neonatal Network. The CRIB (clinical risk index for babies) score: a tool for assessing initial neonatal risk and comparing performance of neonatal intensive care units. Lancet 1993; 342: 193–198.
116
20. Bellanti JA, Bower AL. Immunology of the fetus and the newborn. In: Avery GB, Ed. Neonatology: Pathophysiology and Management of the Newborn, Philadelphia $$ Lippincott, 1981: 701–702. 21. Kacica MA, Horgan MJ, Ochoa L, Sandler R, Lepow ML, Venezia RA. Prevention of gram-positive sepsis in neonates weighing less than 1500 grams. J Pediatr 1994; 125: 253–258. 22. Mahieu LM, De Muynck AO, De Dooy JJ, Laroche SM, Van Acker KJ. Prediction of nosocomial sepsis in neonates by means of a computer weighted bedside scoring system (NOSEP-score). Crit Care Med 2000; 28: 2026–2033. 23. Bach A, Eberhardt H, Frick A, Schmidt H, Bottiger BW, Martin E. Efficacy of silver-coating central venous catheters in reducing bacterial colonization. Crit Care Med 1999; 27: 515–516. 24. De Cicco M, Panarello G, Chiardia V. Source and of microbial colonization of parenteral nutrition catheters. Lancet 1989; 2: 1258–1261.
L. M. Mahieu et al.
25. Moro ML, Vigano EF, Lepri AC. The central venous catheter-related infections study group. Risk factors for central venous catheter-related infections in surgical and intensive care units. Infect Control Hosp Epidemiol 1994; 15: 253–264. 26. Salzman MB, Isenberg HD, Shapiro JF, Lipsitz PJ, Rubin LG. A prospective study of the catheter hub as the portal for entry of microorganisms causing catheter-related sepsis in neonates. J Infect Dis 1993; 167: 487–490. 27. Mueller-Premru M, Gubina M, Kaufmann ME, Primozic J, Cookson BD. Use of semi-quantitative and quantitative culture methods and typing for studying the epidemiology of central venous catheter-related infections in neonates on parenteral nutrition. J Med Microbiol 1999; 48: 451–460. 28. Goldmann DA. Coagulase-negative staphylococci: interplay of epidemiology and bench research. Am J Infect Control 1990; 18: 211–221.
Risk factors for central vascular catheter-associated bloodstream infections among patients in a neonatal intensive care unit L. M. Mahieu*, A. O. De Muynck‡, M. M. Ieven†, J. J. De Dooy*, H. J. Goossens† and P. J.Van Reempts* *Departments of Paediatrics, Division of Neonatology and †Clinical Microbiology, University Hospital of Antwerp, Belgium and ‡The Department of Clinical Sciences, Institute of Tropical Medicine, Belgium
Summary: The aim of this study was to identify risk factors for catheter-associated bloodstream infection (CABSI) in neonates. We undertook a prospective investigation of the potential risk factors for CABSI (patient-related, treatment-related and catheter-related) in a neonatal intensive care unit (NICU) using univariate and multivariate techniques. We also investigated the relationship between catheter hub and catheter exit site colonization with CABSI. Thirty-five episodes of CABSI occurred in 862 central catheters over a period of 8028 catheter-days, with a cumulative incidence of 4.1/100 catheters and an incidence density of 4.4/1000 catheter days. Factors independently associated with CABSI were: catheter hub colonization (odds ratio [OR]:44.1, 95% confidence interval [CI]:14.5 to 134.4), exit site colonization (OR:14.4, CI:4.8 to 42.6), extremely low weight (1000 g) at time of catheter insertion (OR:5.13, CI:2.1 to 12.5), duration of parenteral nutrition (OR:1.04, CI:1.0 to 1.08) and catheter insertion after first week of life (OR:2.7, CI:1.1 to 6.7). In 15 (43%) out of the 35 CABSI episodes the catheter hub was colonized, in nine (26%) cases the catheter exit site was colonized and in three (9%) cases colonization was found at both sites. This prospective cohort study on CABSI in a NICU identified five risk factors of which two can be used for risk-stratified incidence density description (birthweight and time of catheter insertion). It also emphasized the importance of catheter exit site, hub colonization and exposure to parenteral nutrition in the pathogenesis of CABSI. © 2001 The Hospital Infection Society
Keywords: Surveillance; central catheter; colonization; antibiotics; neonate; risk; sepsis; low birth weight; bloodstream infection; hub; exit site.
Introduction Bloodstream infections are the most frequent nosocomial infections in neonatal intensive care unit Received 8 November 2000; revised manuscript accepted 21 February 2001 Author for correspondence: Ludo M. Mahieu, Department of Paediatrics, Division of Neonatology, University Hospital of Antwerp, Wilrijkstraat 10, B-2650 Antwerp, Belgium. E-mail: [email protected]
0195-6701/01/020108;09 $35.00
(NICU) patients. Their cumulative incidence is increasing as a result of the higher survival rate of very low birth weight infants and the increased use of central vascular catheters.1 Catheter-associated bloodstream infections (CABSI) in the NICU contribute significantly to hospital morbidity as well as to increased costs due to prolonged hospitalization.2 Both umbilical and central intravenous catheters are major risk factors for septicemia.3 Micro-organisms colonizing the catheter exit site and hub are able to
© 2001 The Hospital Infection Society
Risk for catheter sepsis in neonates
migrate along the external and internal surface to the catheter tip and cause sepsis. There are only a few reports on the epidemiology of catheter-associated infections that include both types of catheters in neonates; particularly lacking are figures on infections related to umbilical line.4,5 Moreover, in the studies reported, varying definitions are used which makes it very difficult to compare study results. A limited number of factors has been found to be associated with CABSI in neonates. We found only two studies in which adjustments were made, through multivariate analysis technique, for duration of catheterization and other potential confounding factors.6,7 To identify targets for the prevention of CABSI, by assessing the cumulative incidence and identification of independent risk factors, a prospective epidemiological study of CABSI among neonates in a NICU was undertaken. We also studied the association between catheter hub and exit-site colonization and CABSI in both umbilical as well as non-umbilical central vascular catheters. Methods Study population This prospective study was conducted from 1 November 1993 through 31 October 1995 in a 19-bed NICU at the University Hospital of Antwerp, Belgium, and included all neonates requiring at least one central vascular catheterization including umbilical (arterial and venous) and non-umbilical (subclavian and peripheral) central vascular lines. Each case of catheterization was individually evaluated. The catheter utilization ratio was calculated by dividing the number of catheter-days by the number of patient-days of the total study population. Total Parenteral Nutrition (TPN) utilization ratio was calculated in the same manner with the number of TPN days as the numerator. Protocol for catheter care Aseptic technique was used during insertion and repositioning; this included scrubbing of the nails with a brush and a two minute vigorous rubbing of the skin of the hands and arms, using 4% chlorhexidine (Hibiscrub®, Zeneca, Destelbergen, Belgium), the use of sterile gloves, drapes, gowns and facemasks. Before inserting the catheter, the patient’s
109
skin was disinfected twice by rubbing the site of insertion with sterile gauze soaked in a solution of 2% chlorhexidine in 70% isopropyl alcohol and was allowed to dry. The exit site of non-umbilical central vascular catheters was covered with sterile gauze held in place by an occlusive transparent dressing (Tegaderm®, 3M, London, Ontario, Canada). No topical antibiotics were used. This dressing was only changed if there were problems of catheter occlusion or clinical signs (e.g. temperature instability, hepatosplenomegaly, etc.) and symptoms (e.g. apnoeic spells, bradycardia, etc.) of sepsis. The exit site of umbilical lines remained uncovered, was held in place using sterile sutures (Ethilon®, Ethicon Inc., Johnson & Johnson International, Somerville, NJ, USA) and secondarily fixed with a tape to the abdominal skin. The umbilical stump was cleansed three times daily with sterile gauze soaked in a solution of 2% chlorhexidine in 70% isopropyl alcohol prior to the application of a powder containing 1.5% virginiamycin and 0.5% neomycin sulphate (Spitalen®, Bencard, Genval, Belgium). Three-way stopcocks connecting the hub with the intravenous sets were changed every 48 h, or every 24 h when used for TPN administration. The stopcocks and hubs were disinfected with a locally prepared solution of 2% chlorhexidine in 70% isopropyl alcohol using a sterile swab immediately before and after each manipulation and wrapped in sterile gauze dressing. Antibiotics were not used prophylactically, but only for treatment of suspected infections. No blood products were administered through the central vascular catheters. Data collection The following data were recorded at the time of central vascular catheter insertion: date and site of insertion (umbilical artery or umbilical vein, peripheral antecubital, saphenous or subclavian vein); venue of insertion (operating room, bedside in the NICU, other hospital); physician’s name and position (medical officer, surgeon, anaesthetist, neonatologist); whether the patient was requiring antibiotic treatment at the time of insertion; patient’s weight and postnatal age. The following data were recorded while the central vascular catheter was in place: need for repositioning as determined by X-ray; use of steroids, duration of TPN; duration of artificial ventilation; duration of catheterization, development of catheter hub and
L. M. Mahieu et al.
110
exit site colonization. In the event of CABSI, only those factors preceding the date of diagnosis were taken into consideration.
hypothermia (37⬚C), apnoea, or bradycardia] and (2) central catheter present at the time the blood culture was obtained, as previously described.9
Microbiological procedures
Statistical methods
Swabs were taken from both the catheter exit-site and hub at the time of sepsis evaluation as well as at catheter removal in those catheters not associated with infection. After disconnecting the tubing from each central vascular catheter Luer connector (hub), a sterile dry cotton swab was inserted into the hub and rotated twice through 180⬚. A culture was taken from the skin-catheter junction with another sterile cotton swab, which was rubbed along a surface area of 1 cm2, after removal of the dressing. In the microbiology laboratory, each swab was plated on a 5% horse blood agar plate. The inoculum was spread with a loop over the plate. Plates were incubated at 37⬚C for 48 h and kept at room temperature for a further two to three days. Cultures were considered positive when at least two out of the four inoculated quadrants revealed bacterial or fungal growth. Isolates were identified by standard methods to species level and their antibiotic susceptibility was determined. Coagulase-negative staphylococci were identified to species level using a rapid inhouse developed identification system.8 Paired quantitative blood cultures were taken using the DuPont Isolator 1.5® system (Unipath limited, Hampshire, UK) when clinical signs and/or symptoms suggested sepsis. If possible, one set (1 mL blood) was taken through the central vascular catheter and a second set from a peripheral vein. In cases where it was not possible to draw blood from the central vascular catheter, the non-quantitative Bactec® system (Becton-Dickinson, Cockeysville, MD, USA) was used to collect 1 mL blood from two separate peripheral veins. Only when both sets revealed the same organism (identical biotype and antibiogram), were these blood cultures considered positive.
Firstly, a descriptive analysis of risk factors for the first episode of CABSI was performed for each of three risk categories: (1) patient characteristics, (2) therapeutic factors and (3) catheter-related factors. Cumulative incidences (infections/100 catheters) for all risk factors were assessed by means of the Yates corrected Chi-square test or the two-tailed Fisher’s exact test. Secondly, to correct for catheterization duration as a potential confounder, the incidence density (infections/1000 catheter days) for each risk factor was determined. Variables with a P-value 0.20 were retained for inclusion in multivariate analysis. After excluding the catheters from which the results of hub or exit site culture were missing, 658 of the original 862 catheters remained for multivariate analysis. The stepwise logistic regression method was used for identification of independent predictors and for calculation of odds ratios (OR) and their 95% confidence intervals (CI). Significant interactions among variables were retained in the selected model, after ascertainment by stepwise regression. Finally, the correlation between catheter duration and duration of parenteral nutrition was determined. All analysis was performed with Stata 5.0 statistical software (State Corporation, Texas, 1996).
Definitions Catheter-associated bloodstream infection (CABSI) was defined as follows: (1) Primary bloodstream infection according to the CDC surveillance definition: a) recognized pathogen isolated from blood culture or a skin contaminant isolated from two blood cultures drawn on separate occasions, b) one of following clinical signs of infection [fever (38⬚C),
Results Study population We followed up 862 episodes of CVC catheterization with a total of 8028 catheter days among 441 neonates. The mean number of catheterization episodes per patient was 1.9 with a maximum of eight episodes. Mean birth weight of the neonates was 1796 g (range 518–5700 g); 47% weighed less than 1500 g and 18% less than 1000 g. The gestational age of the neonates varied from 24 to 43 weeks with a mean of 33 weeks. Fifty-nine percent were male. The median (quartile Q1–Q3) catheter utilization ratio did not change significantly with birth weight (1000 g:0.66 (0.35–1.07); 1001–1500 g: 0.75 (0.50–1.04); 1501–2500 g:0.73 (0.50–1.0); 2500 g:0.81 (0.53–1.0), P:0.556). In contrast, the median (quartile Q1–Q3) TPN utilization
Risk for catheter sepsis in neonates
111
ratio differed significantly between birth weight categories (1000 g:0.31 (0.09–0.52); 1001–1500 g: 0.35 (0.14–0.52); 1501–2500 g:0.21 (0.0–0.55); 2500 g:0.0 (0.0–0.0), P0.001). Characteristics of catheter insertion episodes (Table I) Umbilical lines, especially venous lines, were the most frequently used central vascular catheters. Peripheral percutaneously inserted central vascular catheters were the most frequent non-umbilical lines. A medical officer inserted by far the most of the catheters. At the time of catheter insertion, about one third of the neonates had been treated with antibiotics. Ampicillin plus netilmicin were used for early-onset infections (72 h after birth) and vancomycin with cefotaxime for hospitalacquired infections. All pathogens isolated from the blood cultures were sensitive to one of these antibiotics with exception of one patient with Pseudomonas paucimobilis sepsis. The mean catheterization
Table I
Characteristics of the episodes of catheterization at insertion
Characteristic
Number
No of catheterization episodes Insertion site of central vascular catheters, % Umbilical (UAC*, UVC†) Non-umbilical (PICVC‡, subclavical) Place of insertion, % Bedside Other hospital Operating theatre Physician responsible: medical officer, % Antibiotic treatment at insertion, % Duration of catheterization (total number of days) Overall: Mean (SD§, range) Catheter type: Mean (SD, range): Umbilical Non-umbilical Patient’s weight: Mean (SD, range): 1000 g 1000 g Use for parenteral nutrition, % Average duration of parenteral nutrition, mean (SD, range) Reposition of CVC||, % Patient ventilated during use of CVC, % Average duration of ventilation, mean (SD, range) Use of steroids, %
862 68 (24, 45) 32 (27, 4) 94.9 3.9 1.2 85 33.8 8028 9.3 (9.3, 1–110) 6.4 (4.2, 1–23) 15.7 (13.4, 1–110)¶ 12.2 (10.8, 1–110) 8.7 (8.9, 1–75) 43 11.3 (10.6, 1–70) 21 34 5.6 (1–69) 4
*UAC, umbilical arterial catheter; †UVC, umbilical venous catheter; ‡PICVC, peripheral inserted central vascular catheter; §SD, standard deviation; ¶P0.001 by Kruskal-Wallis test; ||CVC, Central vascular catheter.
duration was significantly shorter for umbilical than for non-umbilical central vascular catheters. Almost half of the central vascular catheters were used for long-term administration of TPN. Approximately one fifth of all central vascular catheters needed repositioning. There was a strong linear correlation between duration of catheter use and duration of TPN administration (r:0.83, P0.001). About one third of the patients were being artificially ventilated while the catheter was in place. During catheterization, steroids were administered in about 4% of the central vascular catheters. Catheter infections A total of 35 CABSI episodes were recorded in 862 catheters. This amounts to a cumulative incidence of 4.1/100 central vascular catheters and an incidence density of 4.4/1000 catheter days. The cumulative incidence differed according to the type of catheter: umbilical arterial catheters 2.5%; umbilical venous catheters 2.6%; peripheral inserted central venous catheters 7.2% and subclavian catheters 8.1%. The cumulative incidence was almost three times higher for non-umbilical central vascular catheters than for umbilical central vascular catheters (7.4 vs. 2.5%, P0.01). In contrast, after correction for duration of catheterisation, the incidence density was not significantly different (4.7/1000 vs. 4.3/1000 catheter-days, P0.05). The incidence densities for the different types of catheter were: umbilical arterial catheters 4.4/1000 catheter days, umbilical venous catheters 3.8/1000 catheter days, peripheral inserted central venous catheters 4.7/1000 catheter days and subclavian catheters 4.3/1000 catheter days. It was only possible to do paired quantitative blood cultures in 20% (7/35) of the CABSI episodes. In all cases the blood cultures (1 mL) taken from the central vascular catheter yielded a higher concentration of bacteria than the blood cultures (1 mL) taken from a peripheral vein (factor10). All blood cultures yielded at least 1000 cfu/ml blood. Of the pathogens responsible for CABSI, two thirds (69%) were coagulase-negative staphylococci (18 Staphylococcus epidermidis, two Staphylococcus warneri, three Staphylococcus hominis, one Staphylococcus capitis), and only 14% were Gramnegative organisms (Figure 1). No catheter-related mycotic infections were found. In 15 (43%) of the 35
L. M. Mahieu et al.
112
CNS* S. aureus E. fecalis K. pneumoniae S. marcescens P. paucimobilis 0
5
10
15
20
25
Cases Figure 1 Micro-organisms recovered from blood cultures in 35 episodes of catheter-associated blood-stream infections in neonates. Gram-positive; *Coagulase-negative styphylococcus.
CABSI episodes the catheter hub was colonized, in nine (26%) cases the catheter exit site was colonized and in three (9%) cases colonization was found at both sites. Thus catheter colonization at the exit site and/or hub was found in 27 out of 35 (77%) CABSI. In this study, exit site colonization had 32% sensitivity (11/34) and 94% specificity (610/648) for CABSI. Its negative predictive value (NPV) was 96% (610/633) and its positive predictive value (PPV) was 23% (11/49). For hub colonization, we found 46% sensitivity (15/33) and 97% specificity (625/646) for CABSI. The NPV for hub colonization was 97% (625/643) with a PPV of 42% (15/36). Univariate analysis (Table II) allowed the retention of 10 out of 13 potential predicting factors (P 0.20) for inclusion in the multivariate model: weight1000 g at time of insertion; catheter colonization of the hub; exit site colonization; duration of total parenteral nutrition; insertion by consultant in charge of the neonate; insertion in the operating room; non-umbilical insertion; use of corticosteroids; postnatal age of more than 1 week at insertion and catheterization duration. Artificial ventilation, repositioning of the central vascular catheter and antibiotic treatment at time of insertion were not associated with CABSI. The independent risk factors predictive for CABSI are shown in Table III. After adjustment by logistic regression, besides catheter hub and exit site colonization, only TPN duration during catheter use, extremely low weight at time of CVC insertion (1000 g) and postnatal age of more than 1 week at insertion were significantly associated with CABSI. The risk of CABSI increased with increasing duration of TPN exposure: adjusted OR:1.0 for 0 days; OR:4.36 for 1–7 days (95%
CI:1.59–11.90, P:0.004); OR:7.08 for 8–14 days (95% CI:2.77–18.11, P0.001); OR:6.24 for 14 days (95% CI:2.04–19.13; P:0.001) (Figure 2). The infection risk for patients with a weight of1000 g at CVC insertion was approximately three times greater than that of patients with a weight 1000 g (%CABSI, 9.2 for those 1000 g; %CABSI, 3.0 for those 1000 g). The cumulative incidence of CABSI for patients with a postnatal age of more than 1 week at insertion was approximately three times higher than those less than 1 week at insertion (8.7 vs. 3.2%). The catheterization duration itself was not an independent risk factor for CABSI. Discussion Although indispensable for the treatment of sick neonates in the NICU, central vascular catheters present a major risk for nosocomial bacteraemia as well as the increased cost of hospitalization.2,3 Surveillance of CABSI and identification of risk factors are essential if effective preventive strategies are to be developed. However, due to differences in definition and lack of adjustment for duration of catheterization, the results of published studies are difficult to compare and the strength of the identified risk factors should be viewed with caution, as most analyses have not been adjusted for potential confounding factors. The laboratory confirmation of catheter-related sepsis in neonates is difficult for several reasons: (1) paired quantitative cultures of blood obtained from peripheral vein and through the central line is frequently impossible; (2) in practice, central lines are not removed to perform cultures of the tip at sepsis evaluation because of the drawbacks and uncertain value of reinsertion of a central line in small neonates; and (3) the use of molecular epidemiological techniques, to show identity of organisms isolated from the blood and other sites is not routine in most hospitals. Therefore, for surveillance purposes, we defined CABSI as a primary bloodstream infection in association with a central line as suggested by others.9 In our neonates, the CABSI cumulative incidence was found to be almost three times higher when non-umbilical lines were used (7.4 vs. 2.5% in umbilical lines). This association was, however, not significant when the incidence density was compared (4.7/1000 vs. 4.3/1000 catheter days in umbilical lines). Although the daily attributable
Risk for catheter sepsis in neonates
Table II
113
Risk factors for catheter-associated bloodstream infection: univariate analysis (n:862)
Risk Factor
Modality
Cumulative incidence/ 100 CVC*
Incidence density/1000 catheter days
Catheterization duration§
0–7 8–14 14 Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No 0 1–7 8–14 14 Yes No
2.0 5.9¶ 7.4¶ 22.4|| 3.6 41.7|| 2.8 7.1¶ 2.7 5.2 3.5 4.4 3.4 9.2|| 3.0 8.0** 3.4 8.7¶ 3.2 3.9 4.1 15.2¶ 3.6 1.4 5.9¶ 9.2|| 8.2¶ 2.5¶ 7.4
5.0 5.6 3.0 23.1|| 3.8 26.2|| 3.0 4.7 4.3 4.3 4.7 4.9 3.8 8.2** 3.5 7.0 4.0 6.2 4.0 4.5 4.5 7.6 4.2 2.6 7.2** 7.2** 3.0 4.3 4.7
Exit site colonization§ Hub colonization§ Insertion at bedside§ On antibiotics at insertion§ Patient ventilated§ Patient’s weight 1000 g§ Physician: Consultant§ Postnatal age7 days§ Reposition§ Steroid use§ TPN†† duration (days)§
Umbilical insertion§
OR†
1 3.06 3.95 7.7
95% CI ‡
P
1.26, 7.53 1.4, 11.12 3.5, 16.91
0.01 0.006 0.001
24.8
11, 55.8
0.001
1.6
1.2, 2.3
0.002
1.49
0.75, 2.96
0.25
1.3
0.64, 2.7
0.461
3.3
1.6, 6.6
0.001
2.4
1.14, 5.21
0.021
2.9
1.4, 5.98
0.004
0.9
0.4, 2.2
0.869
4.8
1.7, 13.2
0.003
1 4.35 7.08 6.2 2.7
1.59, 11.9 0.004 2.77, 18.11 0.001 2.04, 19.13 0.001 5.26, 1.43 0.004
*CVC, central vascular catheter; †OR, Odds Ratio; ‡CI, Confidence Interval; §Retained for inclusion into multivariate analysis, P0.20; ¶Yates-corrected Chi-square P0.01; ||Yates-corrected Chi-square P0.001; ** Yates-corrected Chi-square or Pearson’s Chi-square test P0.05; †† TPN, total parenteral nutrition. 8
Odds ratio for CABSI
7 6 5
Table III infection
4
Risk Factor
OR*
95% CI†
SE‡
Hub colonization Exit site colonization Weight1000 g Postnatal age 7 days TPN|| days (until CABSI**) Hub and exit site colonization
44.1§ 14.4§ 5.13§ 2.74¶ 1.04¶ 0.06¶
14.5, 134.4 4.8, 42.6 2.1, 12.5 1.1, 6.7 1.0, 1.08 0.01, 0.5
25.1 7.97 2.3 1.25 0.02 0.07
3 2 1 0
1–7 8–14 TPN exposure (days)
> 14
Figure 2 Effect of exposure to parenteral nutrition (TPN) on the risk for catheter-associated bloodstream infection (CABSI). Odds ratio for CABSI; Lower Limit 95% CI; Odds Ratio.
Independent risk factors for catheter-associated bloodstream
Observation 658, Chi-square:92.12, P0.0001, Log Likelihood:984.9, area under receiver operating characteristic curve:0.91 *OR, Odds Ratio; † CI, Confidence Interval; ‡SE, Standard Error; §P0.001; ¶P0.05; ||TPN, total parenteral nutrition; **CABSI, catheter-associated bloodstream infection.
114
risk for infection in umbilical and non-umbilical lines does not differ significantly, the cumulative incidence of the latter is higher because the duration of catheterization is more than twice as long (Table I). In one study, umbilical artery catheters were found to be associated more frequently with sepsis than umbilical venous catheters (5 vs. 3%).5 These findings were not confirmed in our study (2.5 vs. 2.6%). In our study, a CABSI rate of 8.1% was found in subclavian lines which is at the lower end of the range of 6–46% in infants with Broviac catheters.10,11 With peripherally inserted central vascular catheters, a CABSI cumulative incidence of 7.2% was found, which is consistent with the published figures of 1.3–12.9%.12–16 Only one study of neonates revealed a CABSI incidence density of 1/1000 catheter-days which is much lower than the 4.7/1000 found in this cohort study.14 In that study, the infection rate may have been underestimated as only catheter hub-related infections were reported. The incidence densities of CABSI of the present study, stratified by birth weight categories are about half of those in the National Nosocomial Infections Surveillance study (NNIS) of the CDC.4 The finding that the incidence density increases with decreasing birth weight corroborates the NNIS data. Little is known about the factors that predispose to CABSI in NICU patients. In arterial umbilical lines, very low birth weight and antimicrobial therapy, and in venous umbilical lines infusion of TPN and high birth weight have been identified as risk factors.5 This study revealed one independent patient-related risk factor, besides catheter colonization: extremely low weight (<1000 g) at the time of catheter insertion. Different factors may be responsible for the higher frequency of CABSI in low weight infants also found in other studies. Firstly, longer exposure to TPN including intravenous lipid emulsions as shown by the TPN utilization ratio in the study population. This is supported by data for very low birth weight infants for whom intravenous lipid emulsions were a major determinant of coagulase-negative staphylococcal bacteraemia, even after adjustment for severity of illness.17 Secondly, more severe clinical conditions leading to increased catheter manipulations. The Day 1 Score for Neonatal Acute Physiology (SNAP) and the Clinical Risk Index for Babies (CRIB) are useful in predicting the risk of bacteremia;18,19 the immature immune system may also be a factor.20 Some risk factors for CABSI reported by others, such as antimicrobial therapy, were not
L. M. Mahieu et al.
found significant in this particular study.14,15 One study revealed that a low dose of vancomycin in TPN solutions is effective in the prevention of nosocomial sepsis in extremely low birth weight infants.21 The relationship found by others between duration of catheterization and risk for CABSI14 was not confirmed. There was a strong correlation between duration of catheterization and duration of TPN infusion therapy, the latter being associated with CABSI cumulative incidence (Table II). Therefore catheterization duration must be considered as a confounding factor for CABSI and so infection rates should be corrected by using incidence density measures (CABSI per 1000 catheter-days).13 When crude nosocomial sepsis densities are compared between NICUs, they should be adjusted for line utilization ratio in the units, as is done in the NNIS.4 We believe this is the first published report on the importance of both hub and catheter exit sites as risk factors for CABSI in neonates. A previous study, found that catheter exit site and especially catheter hub colonization at the time of clinical symptoms of infection correlated with hospitalacquired sepsis in neonates. Addition of catheter exit site and hub cultures to a diagnostic scoring system, composed of laboratory and clinical parameters of infection, significantly improves the discriminatory power of the latter.22 We found a high negative predictive value for CABSI of exit site and hub colonization (96 and 97% respectively) but a rather low positive predictive value if the exit site or hub was colonized (23 and 42% respectively). The predictive values depend very strongly on the prevalence of the disease studied. As a consequence, higher positive predictive values would be expected when the prevalence of CABSI is higher in the patient population than in our cohort (overall 4.1%). In postoperative adult patients with central lines, one recent study has shown a 100% negative predictive value for catheter tip colonization if neither the skin nor the hub was colonized, and a positive predictive value of 68% if the skin and 52% if the hub was colonized.23 Similar observations were made in adults receiving TPN24 where catheter exit site and especially hub colonization were strongly associated with CABSI (OR:6 and 19 respectively). This supports the hypothesis that both catheter hub and exit site colonization is important in the pathogenesis of CABSI. Although several studies in adults have proven the catheter hub to be an important portal of entry for organisms causing CABSI,23–25 only one study supports this in neonates.26 A recent
Risk for catheter sepsis in neonates
study of neonates using genotyping techniques, revealed that 66% of coagulase-negative staphylococci on central vascular catheter-tips, originated from stopcocks.27 We believe this is the first study suggesting that prolonged hospitalization prior to catheter insertion is associated with increased infectious catheter complications. A previous study has shown that length of stay is an independent risk factor for hospitalacquired bacteraemia with coagulase-negative staphylococcal which are major pathogens in central vascular catheters.28 Duration of catheterization may be another contributing factor to higher infection rates in catheters placed after the first week of life. Indeed, in the first week the catheters inserted are usually short-term umbilical lines. This explains why the incidence densities are equal but not the cumulative incidences between catheters placed before and after the first postnatal week (Table II). Both umbilical and percutaneous central vascular lines are associated with severe infectious complications in neonates. As colonization of the exit site and hub are important risk factors for CABSI, more studies are needed to explain the pathogenesis of catheter hub and exit site colonization. The adjusted infection rates described here provide specific surveillance data for interhospital comparison and also to assess the implementation of preventable measures to control infections in neonates.
Acknowledgements The authors thank Dr Guillaume Van Melckebeke for his help in data collection and data entry into the computer and Mr Rizwan Afzal for his biostatistical assistance. They are most grateful to the clinicians that recorded the data at their patients’ bedside.
References 1. Gladstone IM, Ehrenkranz RA, Edberg SC, Baltimore RS. A ten-year review of neonatal sepsis and comparison with the previous fifty-year experience. Pediatr Infect Dis J 1990; 9: 819–825. 2. Leroyer A, Bedu A, Lombrail P et al. Prolongation of hospital stay and extra costs due to hospital-acquired infection in a neonatal unit. J Hosp Infect 1997; 35: 37–45. 3. Gaynes RP, Edwards JR, Jarvis WR, Culver DH, Tolson JS, Martone WJ. Nosocomial infections among neonates in high-risk nurseries in the United States. Pediatrics 1996; 98: 357–361.
115
04. CDC NNIS System. National Nosocomial Infections Surveillance (NNIS) Report, May 1997. Am J Infect Control 1997; 25: 477–487. 05. Landers S, Moise AA, Fraley JK, Smith EO, Baker CJ. Factors associated with umbilical catheter related sepsis in neonates. Am J Dis Child 1991; 145: 675–680. 06. Maas A, Flament P, Pardou A, Deplano A, Dramaix M, Struelens MJ. Central venous catheter-related bacteraemia in critically ill neonates: risk factors and impact of a prevention programme. J Hosp Infect 1998; 40: 211–224. 07. Pandit PB, Pandit FA, Govan J, O’Brien K. Complications associated with surgically placed central venous catheters in low birth weight neonates. J Perinatol 1999; 19: 106–109. 08. Ieven M, Verhoeven J, Pattyn SR, Goossens H. Rapid and economical method for species identification of clinically significant coagulase-negative staphylococci. J Clin Microbiol 1995; 33: 1060–1063. 09. Sherertz RJ. Surveillance for infections associated with vascular catheters. Infect Control Hosp Epidemiol 1996; 17: 746–752. 10. Grisoni ER, Mehta SK, Connors AF. Thrombosis and infection complicating central venous catheterisation in neonates. J Pediatr Surg 1986; 21: 772–776. 11. Sadiq HF, Devaskar S, Keenan WJ, Weber TR. Broviac catheterisation in low birth weight infants: incidence and treatment of associated complications. Crit Care Med 1987; 15: 47–50. 12. Durand M, Ramanathan R, Martinelli B, Tolentino M. Prospective evaluation of percutaneous central venous silastic catheters in newborn infants with birth weight of 510 to 3920 grams. Pediatrics 1986; 78: 245–249. 13. Loeff DS, Matlak ME, Black RE, Overall JC, Dolcourt JL, Johnson DG. Insertion of a small central venous catheter in neonates and young infants. J Pediatr 1982; 17: 944–949. 14. Chathas MK, Paton JB, Fisher DE. Percutaneous central venous catheterisation. Am J Dis Child 1990; 144: 1246–1250. 15. Neubauer AP. Percutaneous central iv access in the neonate: experience with 535 silastic catheters. Acta paediatr 1995; 84: 756–760. 16. Cairns PA, Wilson DC, McClure BG, Halliday HL, McReid M. Percutaneous central venous catheter use in the very low birth weight neonate. Eur J Pediatr 1995; 154: 145–147. 17. Avila-Figueroa C, Goldmann DA, Richardson DK, Gray JE, Ferrari A, Freeman J. Intravenous lipid emulsions are the major determinant of coagulasenegative staphylococcal bacteremia in very low birth weight newborns. Pediatr Infect Dis J 1998; 17: 10–17. 18. Gray JE, Richardson DK, McCormick MC, Goldmann DA. Coagulase-negative staphylococcal bacteremia among very low birth weight infants: relation to admission illness severity, resource use and outcome. Pediatrics 1995; 95: 225–230. 19. International Neonatal Network. The CRIB (clinical risk index for babies) score: a tool for assessing initial neonatal risk and comparing performance of neonatal intensive care units. Lancet 1993; 342: 193–198.
116
20. Bellanti JA, Bower AL. Immunology of the fetus and the newborn. In: Avery GB, Ed. Neonatology: Pathophysiology and Management of the Newborn, Philadelphia $$ Lippincott, 1981: 701–702. 21. Kacica MA, Horgan MJ, Ochoa L, Sandler R, Lepow ML, Venezia RA. Prevention of gram-positive sepsis in neonates weighing less than 1500 grams. J Pediatr 1994; 125: 253–258. 22. Mahieu LM, De Muynck AO, De Dooy JJ, Laroche SM, Van Acker KJ. Prediction of nosocomial sepsis in neonates by means of a computer weighted bedside scoring system (NOSEP-score). Crit Care Med 2000; 28: 2026–2033. 23. Bach A, Eberhardt H, Frick A, Schmidt H, Bottiger BW, Martin E. Efficacy of silver-coating central venous catheters in reducing bacterial colonization. Crit Care Med 1999; 27: 515–516. 24. De Cicco M, Panarello G, Chiardia V. Source and of microbial colonization of parenteral nutrition catheters. Lancet 1989; 2: 1258–1261.
L. M. Mahieu et al.
25. Moro ML, Vigano EF, Lepri AC. The central venous catheter-related infections study group. Risk factors for central venous catheter-related infections in surgical and intensive care units. Infect Control Hosp Epidemiol 1994; 15: 253–264. 26. Salzman MB, Isenberg HD, Shapiro JF, Lipsitz PJ, Rubin LG. A prospective study of the catheter hub as the portal for entry of microorganisms causing catheter-related sepsis in neonates. J Infect Dis 1993; 167: 487–490. 27. Mueller-Premru M, Gubina M, Kaufmann ME, Primozic J, Cookson BD. Use of semi-quantitative and quantitative culture methods and typing for studying the epidemiology of central venous catheter-related infections in neonates on parenteral nutrition. J Med Microbiol 1999; 48: 451–460. 28. Goldmann DA. Coagulase-negative staphylococci: interplay of epidemiology and bench research. Am J Infect Control 1990; 18: 211–221.