JSTOR

Use of Peripherally Inserted Central Catheters to Prevent Catheter‐Associated Bloodstream Infection in Children

Received January 14, 2009; accepted May 4, 2009; electronically published August 31, 2009.

Bloodstream infection (BSI) is the most common healthcare‐associated infection in pediatric intensive care units (ICUs).^1,2 Risk factors for central venous catheter (CVC)–associated BSI are poorly understood in middle‐income developing countries. We used a prospective cohort study design to evaluate the infection rate and risk factors for BSI associated with short‐term use (duration, less than 30 days) of a CVC in a pediatric hospital in Rio de Janeiro.

The Instituto de Puericultura e Pediatria Martagão Gesteira (IPPMG) is a tertiary care pediatric hospital with 60 beds in 6 wards. This is a reference center for patients requiring specialists in multiple diseases and admits patients aged 0–18 years. Because there is no pediatric ICU, patients with critical conditions are admitted to the wards.

All patients admitted to the wards from March 2003 through March 2006 who had a short‐term CVC inserted at the IPPMG were monitored daily by the infection control team for development of CVC‐associated BSI. A structured questionnaire based on National Nosocomial Infection Surveillance System criteria was used for active surveillance.³ If a patient had multiple CVCs in place simultaneously, 1 catheter‐day was assigned for each day of multiple CVC use, and the first catheter placed was used for surveillance purposes. CVCs inserted at another hospital, CVCs in place before admission to the wards, and CVCs placed for less than 1 day were excluded from analysis in this study.

Peripherally inserted central catheters (PICCs) were inserted by a nurse, and other CVCs for short‐term use (ie, those placed in the subclavian, intrajugular, and femoral veins) were inserted by surgeons. Before catheter insertion, 4% chlorhexidine‐germicide solution and 0.5% chlorhexidine‐alcohol solution were used to prepare skin. After insertion, 5% chlorhexidine‐alcohol solution was applied, and dressings were monitored by nursing staff. Conventional gauze dressings were applied to CVCs and were replaced with new dressings every 48 hours. Transparent dressings were applied to PICCs and were changed if bleeding and soiling occurred or if they did not stay in place.

CVCs (PICCs or conventional catheters) were removed immediately if patients showed signs of local infection or, for febrile patients, if their fever had no explanation other than local infection. The 5‐cm portion of the tip of each CVC removed because of suspected infection was sent to the microbiology laboratory for culture; 2 samples of peripheral blood from the corresponding patient were also sent for culture.⁴ Patients were observed for up to 48 hours after CVC removal.

Data analyses were performed using Stata, version 9.0 (Stata). Bivariate analysis was performed using a Wilcoxon 2‐sample test (also known as the Mann‐Whitney U test). The Fisher exact test was used for analysis of categorical variables. Variables with a P value of .20 or less were included in the multivariate analysis. A main‐effects logistic regression model was fitted using the stepwise maximum likelihood estimation technique. The level of significance for removal of a variable in backward regression was .10. Interactions were assessed using the −2 log likelihood ratio test. The Pearson χ² goodness‐of‐fit test and the Hosmer‐Lemeshow test were used to evaluate the fitness of the model. This study was reviewed and approved by the IPPMG ethical research committee.

During the study, 166 patients required 313 CVCs. Eighty‐two CVCs did not satisfy inclusion criteria, and 27 CVCs were not evaluated because they were associated with infection at the catheter‐exit site rather than with BSI. Thus, 204 CVCs were analyzed. Among the 166 patients in the study, the hospitalization duration was 12,370 patient‐days, and the duration of CVC use was 2,197 CVC‐days. The patients were aged 24 days to 14 years (median, 31.89 months), and 106 (64%) were male. Of the 204 CVCs studied, 38 (16%) were from patients with BSI, yielding an incidence of 17.30 infections per 1,000 CVC‐days. Ten BSIs were classified as CVC‐related BSI, and 28 were classified as CVC‐associated BSI. Eighteen CVC‐associated BSIs were laboratory‐confirmed infections, and 10 were classified as clinical sepsis. Among 204 CVCs followed up, 53% were inserted by the Seldinger technique, 28% required surgical cut down, and 19% were placed percutaneously. Thirty‐six percent of CVCs were inserted in brachiocephalic veins, 30% in internal jugular veins, 17% in femoral veins, 15% in subclavian veins, and 2% at another anatomic site. Ninety percent of CVCs were inserted using full‐barrier precautions.

Findings of bivariate analysis are shown in the Table. Multivariate analysis revealed that total parenteral nutrition use (odds ratio [OR], 2.68 [95% confidence interval {CI}, 1.13–6.37]) and the number of CVC‐days (OR, 1.06 [95% CI, 1.01–1.12]) were independently associated with CVC‐associated BSI. PICC use was protective against CVC‐associated BSI (OR, 0.17 [95% CI, 0.03–0.96]).

Table.  Bivariate Analysis of Risk Factors for Central Venous Catheter (CVC)–Associated Bloodstream Infection (BSI) among Patients at a Brazilian Pediatric Hospital, 2003–2006

Open New Window

In this study, the incidence of CVC‐associated BSI was 17.3 infections per 1,000 catheter‐days, which is greater than the incidence observed in most pediatric ICUs in industrialized countries.^1,2 It is important to note that this study involved patients admitted to hospital wards, some of whom were in critical condition, and that 90% of CVCs were inserted using full‐barrier precautions. An important risk factor for CVC‐associated BSI is the duration of CVC use.^1,2 By use of multivariate analysis, we demonstrated that the odds of CVC‐associated BSI increased a mean of 6% per CVC‐day, although this rate is not linear and probably increases with each CVC‐day.

We compared the risk of CVC‐associated BSI among patients who received PICCs with that among patients who received other CVCs. Of interest, PICC use was independently protective against CVC‐associated BSI. Several advantages of PICC use have been described elsewhere,^5‐7 but the protection against CVC‐associated BSI conferred by PICC use has not been well established among children.¹ To our knowledge, our study is the first to report such protection.

Of final note, our hospital does not have a pediatric ICU. Therefore, factors associated with patients in critical condition, such as mechanical ventilation, arterial catheter use, and urinary catheter use, were not studied because of sample size limitations.