Enterobacter cloacae and Pseudomonas aeruginosa polymicrobial bloodstream infections traced to extrinsic contamination of a dextrose multidose vial

E

Enterobacter cloacae and Pseudomonas aeruginosa

polymicrobial bloodstream infections traced to extrinsic contamination of a dextrose multidose vial Lennox K. Archibald, MD, Maria Ramos, Matthew J. Arduino, DrPH, Sonia M. Aguero, Carmen Deseda, MD, Shailen Banerjee, PhD, and William R. Jarvis, MD

Objective: To identify risk factors for polymicrobial bloodstream infections (BSIs) in neonatal intensive care unit (NICU) patients during an outbreak of BSIs.

Design: During an outbreak of BSIs, we conducted a retrospective cohort study, assessed NICU infection control practices and patient exposure to NICU healthcare workers (HCWs), and obtained cultures of the environment and HCW hands.

contamination of electrolyte infusion solutions, such as dextrose or saline solutions or total parenteral nutrition, have been associated with nosocomial BSIs.3,4 In this report we describe the investigation of a nosocomial outbreak of polymicrobial BSIs caused by Enterobacter cloacae and Pseudomonas aeruginosa in an NICU in Puerto Rico.

Patients: During the period May 3 to 7, 1996, 5 infants contracted BSIs caused by both Enterobacter cloacae and Pseudomonas aeruginosa, and one infant contracted a BSI caused by E cloacae only. For each pathogen, all isolates were identical on DNA typing.

Results: Infants exposed to the following were more likely than nonexposed infants to have BSI: umbilical venous catheters (6/14 vs 0/7, P = .05), total parenteral nutrition given simultaneously with a dextrose/electrolyte solution (6/12 vs 0/9, P = .02), or one HCW (5/7 vs 1/13, P = .007). Neither environmental nor HCW hand cultures yielded the outbreak pathogens. Quality control cultures of intravenous solution bags were negative.

Conclusions: We speculate that a dextrose multidose vial became contaminated during manipulation or needle puncture and that successive use of this contaminated vial for multiple patients may have been responsible for BSIs. Aseptic techniques must be employed when multidose vial medications are used. Singledose vials should be used for parenteral additives whenever possible to reduce the risk of extrinsic contamination and subsequent transmission of nosocomial pathogens. (J Pediatr 1998;133:640-4)

Gram-negative organisms are important causes of primary nosocomial bloodstream infections in neonatal intensive care unit patients; these infections often result in high morbidity and mortality

rates.1,2 Moreover, BSIs in NICU patients are strongly correlated with use of umbilical or central venous catheters.2 Other factors, including unsatisfactory infusion therapy practices or extrinsic

From the Hospital Infections Program, Centers for Disease Control and Prevention, Atlanta, Georgia; and the State Epidemiology Department, Department of Health, Frio Piedras, Puerto Rico. Submitted for publication Feb 19, 1998; revision received July 15, 1998; accepted Aug 12, 1998. Reprint requests: Lennox K. Archibald, MD, Hospital Infections Program, Mailstop E-69, Centers for Disease Control and Prevention, 1600 Clifton Rd, Atlanta, GA 30333. 9/21/93784

640

BSI CDC HCW NICU PFGE PICU SNAP TPN UVC

Bloodstream infection Centers for Disease Control and Prevention Healthcare worker Neonatal intensive care unit Pulsed-field gel electrophoresis Pediatric intensive care unit Score for Neonatal Acute Physiology Total parenteral nutrition Umbilical venous catheter

Hospital A is a public hospital that admits patients to acute medical, surgical, obstetric and gynecologic, pediatric, or neonatal services; the NICU was relocated to a new building on October 4, 1995. From that date through May 2, 1996, there were 5 reported BSIs caused by E cloacae and one caused by P aeruginosa in NICU infants. During the period May 3 to 7, 1996, 6 NICU infants had BSI (5 with both E cloacae and P aeruginosa and 1 with E cloacae only). After 2 infants died, the NICU was closed to admissions on May 9, 1996; and the State Epidemiology Division, Department of Health of Puerto Rico, and the Hospital Infections Program, Centers for Disease Control and Prevention were asked to assist in an investigation. The objectives of the investigation were to identify the source of and risk factors for BSIs and to implement control measures to terminate the outbreak.

THE JOURNAL OF PEDIATRICS VOLUME 133, NUMBER 5

METHODS Case Definition and Ascertainment A case-infant was defined as any NICU infant in whom E cloacae, P aeruginosa, or both were recovered from a blood culture between April 25 and May 7, 1996 (epidemic period). Case-infants were identified by review of clinical microbiology laboratory and patient medical records from the epidemic period.

Determination of Existence of an Outbreak To determine whether the cluster of case-patients represented an outbreak, we compared the rate of E cloacae and P aeruginosa BSI in the Hospital A NICU patients during the epidemic and preepidemic (ie, October 4, 1995 to April 24, 1996) periods.

Cohort Study To identify risk factors for BSI, we conducted a cohort study in which case-infants were compared with all other infants who were admitted to the NICU during the epidemic period but did not acquire a BSI. Factors assessed included patient demographics; duration of stay in the NICU; underlying disease; Apgar score; severity of illness as measured by the Score for Neonatal Acute Physiology5; nasogastric intubation; types and duration of intravascular catheterization, mechanical ventilation, use of intravenous antimicrobial agents, and electrolyte solutions; receipt of TPN; or exposure to NICU healthcare workers.

ARCHIBALD ET AL

nosa and E cloacae were obtained from infants who had remained in the NICU after its closure to admissions on May 9, 1996, and who were still in the NICU during the outbreak investigation. Environmental cultures of the NICU, selected on the basis of epidemiologic results, were obtained. These included cultures from work surfaces in the medication room and work surfaces surrounding the infants; tap water and respective sink areas; liquid soap; and various vials of calcium gluconate, dextrose, heparin, potassium chloride, saline, and sodium bicarbonate solutions. On the basis of a preliminary analysis of HCW exposure data, hand cultures were obtained from selected HCWs who were more likely to have had contact with case-infants. Hand cultures were obtained with the Handiwipe method.6 All specimens, cultures, and isolates of the outbreak pathogens that had been saved by Hospital A were sent to CDC for appropriate culture, species confirmation, and antimicrobial susceptibility testing. DNA characterization of P aeruginosa and E cloacae isolates was achieved by pulsedfield gel electrophoresis.7

Statistical Analysis All data were collected on standardized forms and analyzed by using Epi Info version 6.02 software (CDC, Atlanta, Ga). Categorical variables were compared by using the χ2 test or, when appropriate, Fisher’s exact test. Medians of continuous variables were compared by using the Wilcoxon 2-sample test. Relative risk and 95% CIs were calculated.

Procedure Review We reviewed infection control policies and procedures and conducted observational studies to assess infection control, housekeeping, and NICU HCW handwashing practices. A questionnaire was administered to NICU HCWs to assess handwashing practices, the prevalence of hand dermatitis caused by handwashing, and use of vials of dextrose or saline solution for the preparation of electrolyte or antimicrobial solutions during the epidemic period.

Microbiology Point prevalence nasopharyngeal and rectal surveillance cultures for P aerugi-

RESULTS Patients When we compared the number of NICU infants with P aeruginosa and E cloacae BSIs in the pre-epidemic and epidemic periods, we found a significant increase in the number of infants with BSIs during the epidemic period (6/153 vs 6/27, P < .01). Data on infant clinical condition; intravascular devices; type of infusates; blood tests; and start and stop dates of invasive procedures, devices, and infusates were often incomplete or missing from infant medical records. In

addition, progress notes on NICU infants were often incomplete or cursory. One week before the cluster of infections occurred, an infant who had been admitted to the NICU during April, 1996, had sepsis and died. At delivery, cultures of this infant’s placenta yielded P aeruginosa and E cloacae. However, cultures of blood drawn from this infant during his NICU stay remained negative. Although the antimicrobial susceptibility testing pattern of the P aeruginosa isolate from the placenta of this infant was similar to that of the case-infant P aeruginosa isolate, the susceptibility pattern of the placental E cloacae isolate was different from that of the case-infant E cloacae isolate. The P aeruginosa and E cloacae isolates from this dead infant had not been saved for further analyses.

Case Characteristics Six infants met the case definition. Cases occurred daily over 5 days from May 3 to 7; there were 15 noncase-infants. Case-infants had a median gestational age of 32 weeks (range, 28 to 32 weeks), a median birth weight of 1349 g (range, 1023 to 1847 g), and a median NICU stay of 4.5 days before acquiring BSI. All 6 case-infants had respiratory distress syndrome, and 2 of 6 died (Table I). The cause of death of these 2 infants was recorded in their medical records as sepsis caused by P aeruginosa and E cloacae.

Cohort Study Case- and noncase-infants were similar in sex, gestational age, birth weight, Apgar scores, SNAP scores on admission, underlying disease, and type and place of delivery. In contrast, case-infants were significantly more likely than noncase-infants to have received TPN (prepared only in the pharmacy and delivered to the NICU on a daily basis) given simultaneously with an intravenous electrolyte/dextrose infusion, known as “mezclas” (prepared in the NICU), or an umbilical venous catheter (Table II). Exposure data to 86 NICU HCWs (55 nurses, 15 respiratory therapists, 14 doctors, and 2 medical students) were obtained. Case-infants were more likely than noncase-infants to have had contact with one particular HCW (relative risk, 641

ARCHIBALD ET AL

THE JOURNAL OF PEDIATRICS NOVEMBER 1998

Table I. Characteristics of case-infants in the NICU at Hospital A from April 25 to May 7, 1996

Case no. 1 2 3 4 5 6

Sex

Gestational age (wk)

Birth weight (g)

No. days in NICU before first positive blood culture

Organisms isolated from blood culture

Outcome

F M M F M M

32 28 30 32 32 32

1278 1023 1145 1509 1420 1847

4 4 4 5 3 7

E cloacae, P aeruginosa E cloacae, P aeruginosa E cloacae E cloacae, P aeruginosa E cloacae, P aeruginosa E cloacae, P aeruginosa

Survived Died Died Survived Survived Survived

Table II. Potential risk factors for BSIs in the NICU at Hospital A from April 25 to May 7, 1996

Categorical exposure variables TPN given simultaneously with mezclas UVC Single HCW

9.3; 95% CI 1.3 - 64) (Table II). Also, case-infants were more likely than noncase-infants to have leukopenia (4,300/mm3 vs 15,100/mm3, P = .06) or thrombocytopenia (18,000/mm3 vs 243,000/mm3, P < .01). Moreover, the median SNAP score of case-infants at the time of positive blood culture was significantly higher than that on admission to the NICU (9 vs 2, P < .05). When we considered only those infants who had an in situ UVC, infants who received TPN simultaneously with mezclas were more likely than infants who did not receive these infusates to have BSIs (6/12 vs 0/2); this association, however, was not statistically significant. Because of the small numbers of case- and noncase-patients, multivariate analysis could not be done.

Procedure Review The NICU has an open design with 2 isolation rooms and can accommodate 18 infants. It is separated from the pediatric intensive care unit by 2 short passageways containing doors. Off one of these passageways are the rooms for the preparation of intravenous medications. During our observational study, the work surfaces of the medication room were cluttered with used utensils, vials, and filled syringes amidst pools of stag642

Attack rate among exposed infants 6/12 6/14 5/7

nant tap water. The NICU had 7 sinks with hot and cold water taps; one had a water faucet controlled with a knee lever; the remainder were manual. Soap was obtained from wall dispensers fitted with foot-operated pumps. At the start of a morning shift, nurses prepared all the intravenous medications that their patients would require for the rest of the day. The filled syringes and vials were left at room temperature on a paper mat on the work surface of the medication room. Before withdrawal of fluid, the tops of the vials were cleaned with povidone-iodine. At times, needles were introduced into the vial for fluid removal before the povidone-iodine on the vial top had dried properly. Intravenous medications and electrolyte solutions other than TPN were prepared by HCWs in the NICU. Antimicrobial agents were reconstituted for intravenous use in the medication room by addition of sterile water. Reconstituted vials of antimicrobial agents were used for multiple patients and then discarded after 1 hour. The intravenous electrolyte solution known as mezclas consists of calcium gluconate, potassium chloride, 50% dextrose solution, and normal saline. The 50% dextrose solution was contained in 50-mL multidose

Attack rate among unexposed infants 0/9 0/7 1/13

P value .02 .05 .007

vials, which were left at room temperature on the work surface in the medication room and reused to prepare mezclas for other infants in the NICU or PICU. The volume of dextrose solution removed from a 50-mL vial at any one time ranged from 5 to 20 mL. Hence, depending on the volumes removed, a 50-mL vial might last from a few hours to 1 week. The vials were discarded when empty. Thus it is plausible that several NICU infants could have received mezclas prepared from a single vial of 50% dextrose. There was no documentation as to which infants received diluents from particular vials, nor were batch numbers recorded. All but one vial used for preparation of mezclas during the outbreak had been discarded by the time our investigation had started; the one available vial was sent to CDC for culture of its contents. All TPN preparations (including lipids and amino acids) were prepared in the pharmacy with aseptic procedures under laminar flow hoods. Quality control cultures of the work surfaces in the pharmacy TPN preparation areas and of the contents of randomly chosen TPN bags, routinely obtained by pharmacy and microbiology personnel, were negative for gram-negative rods over the 12 months

THE JOURNAL OF PEDIATRICS VOLUME 133, NUMBER 5

ARCHIBALD ET AL

before the outbreak. Electrolytes or other medications were not added to TPN bags after they arrived in the NICU. Lipids and amino acid preparations were placed in separate bags, which were connected to the patient by means of a pump; the preparations entered the pump through separate inlets and were then infused into the infant through a common line attached to the UVC. Manipulations of the UVC were performed by HCWs. TPN bags, tubing, and connectors were changed every 24 hours. TPN bags were brought up daily from the pharmacy to the NICU; according to hospital protocol, TPN bags should have been attached immediately to the infants’ venous access devices. In practice, however, TPN bags were often left standing at room temperature, for various periods, on a work surface in the NICU before being attached to the UVC. During our observational study, various HCWs were noted to have unsatisfactory handwashing practices before and between attending to patients. Most of the observed initial handwashings were cursory in nature (ie,