The development of a risk score for unplanned removal of peripherally inserted central catheter in newborns1

OBJECTIVE: to develop a risk score for unplanned removal of peripherally inserted central catheter in newborns. METHOD: prospective cohort study conducted in a neonatal intensive care unit with newborn babies who underwent 524 catheter insertions. The clinical characteristics of the newborn, catheter insertion and intravenous therapy were tested as risk factors for the unplanned removal of catheters using bivariate analysis. The risk score was developed using logistic regression. Accuracy was internally validated based on the area under the Receiver Operating Characteristic curve. RESULTS: the risk score was made up of the following risk factors: transient metabolic disorders; previous insertion of catheter; use of a polyurethane double-lumen catheter; infusion of multiple intravenous solutions through a single-lumen catheter; and tip in a noncentral position. Newborns were classified into three categories of risk of unplanned removal: low (0 to 3 points), moderate (4 to 8 points), and high (≥ 9 points). Accuracy was 0.76. CONCLUSION: the adoption of evidence-based preventative strategies based on the classification and risk factors faced by the newborn is recommended to minimize the occurrence of unplanned removals.


Introduction
Obtaining venous access in newborn babies admitted to neonatal intensive care units (NICUs) to infuse hyperosmolar, vesicant or irritating solutions is a challenge for nursing professionals. A peripherally inserted central catheter (PICC) is a central vascular access device placed at the bedside by a professionally qualified doctor or nurse, the tip of which is positioned close to the heart, preferably in the vena cava (1)(2) .
Although the use of this device is increasingly frequent in NICUs, due to high insertion success rates and lower infection rates compared to surgically inserted central catheters (3) , studies carried out in Brazil show that rate of catheter-related complications range between 41 (4) and 50.8% (5) , while international studies reveal lower rates, between 2.9 (6) and 31.7% (7) . Mechanical, thrombotic and infectious complications limit the effectiveness of PICCs and may lead to its removal earlier than planned.
Studies with newborn babies which aimed to contribute towards preventing these complications and consequently reduce the occurrence of unscheduled PICC removal have identified a number of risk factors, including the insertion of the catheter through femoral veins (8) , spending more than sixty minutes on catheter insertion (9) , and non-central tip position (10) . However, the role of other potential risk factors among newborns, such as the clinical and anthropometric characteristics of the newborn, the type of catheter used and number of catheter lumens, the type of intravenous infusion in question, and previous PICC insertion history also merit investigation.
Since the majority of catheter-related complications are preventable, the development of a risk score for unplanned PICC removal which considers the prognostic value of various risk factors is an innovative initiative for the advancement of nursing knowledge. Risk scores are potentially valuable tools for informing the decisions made by nurses, since they aid these professionals to estimate the likelihood of unplanned removal of bedside catheters prior to insertion, enabling case-by-case planning of care to attenuate risk. Based on a previous study (11) conducted in the same NICU which observed that the prevalence of unscheduled PICC removal was 37.7%, the minimum odds ratios which could be detected for a binary stratification variable with the sample size used in this study (524 PICC insertions), at the 5% level of significance, with The management of PICCs in this institution follows the guidelines set out in an institutional protocol designed by nurses from the venous catheters study group, based on the literature (12)(13) and recommendations given by institutions accredited to provide capacity building on the insertion, maintenance and removal of PICCs by the Regional (2) or Federal Council of Nursing (14) . PICC insertion must be recommended by a doctor after an assessment of the newborn's clinical condition and venous network and is an aseptic procedure which is conducted at the bedside by a qualified nurse. The medical and nursing teams use  The removal of a catheter was defined as unplanned when it was due to complications such as obstruction, rupture, tip migration, phlebitis, thrombosis, catheterrelated bloodstream infection, swelling, infiltration, leakage, and accidental catheter removal.
Data was stored in an excel spreadsheet using double entry and analysed in the R 3.01 environment.
After applying the eligibility criteria, the data from 80% of the cases of PICC insertion was used to develop the risk score, while the data from the remaining 20% of cases was used for the internal validation of the risk score. First, the quantitative variables were analysed using averages and standard deviation. Qualitative variables were also analysed to ascertain the absolute and relative frequency distribution. Bivariate analysis was conducted to ascertain whether there was an association between variables and the outcome (unscheduled removal of PICC) using the Student t-test for continuous variables, the Chi-squared test or Fisher's exact test for categorical variables, and the estimation of relevant risk and 95% confidence interval.
The significance level was set at 5%. The risk score was developed by conducting stepwise logistic regression using forward selection with the variables which were shown to have a significant association under the bivariate analysis.
Only statistically significant and noncollinear variables were retained. The risk score was constructed based on the magnitude of correlation of the coefficients of each variable in the logistic equation. The predictive capacity of the score was evaluated based upon the area below the Receiver Operating Characteristic curve (ROC curve). The points of the ROC curve were used to construct three risk categories for unscheduled PICC removal: low, medium and high risk. For internal validation of the tool, the risk score was applied to the data reserved for validation to evaluate its predictive capacity in relation to the outcome based on the absolute and relative frequency distribution of the three risk categories.

Results
A total of 17,341 infants were born during the study period, of which 1,482 were admitted to the NICU. Of this total, 460 underwent intravenous therapy, resulting in a total of 563 PICC insertions. After exclusion based on study eligibility criteria, the sample was reduced to 436 newborns who underwent a total 524 PICCs which was divided into two data sets: data used to develop the risk score (80% of the PICCs = 419); and data used for the initial validation of the risk score (20% the PICCs =105).
Other mechanical complications such as obstruction, therapy (23) . However, another study which evaluated 188 PICCs inserted in newborn babies revealed that the complication rate was higher in infants who received continuous infusion of heparin than in those who did not (23.7/1,000 catheter days versus 17.2/1,000 catheter days) (24) . Given the fact that complications in newborn babies such as haemorrhaging, thrombocytopenia, and bleeding disorders may be related to continuous heparin infusion (23) , conclusive evidence to the contrary is required to support this practice. the insertion site is exposed (21) .  (10) . A preventative strategy for this risk factor is the accurate measurement of the length of the catheter, i.e., the distance between the puncture site to the vena cava along the vein (1) , together with close monitoring of the newborn for initial signs of complications such as infiltration (10) .
Another risk factor was the type of catheter used: findings suggest that risk of unscheduled removal was four times greater with polyurethane double-lumen catheters than with single-lumen silicone catheters.
However evidence showing which material is best is not conclusive. A study compared the silicone catheter with an anti-reflux valve and the polyurethane PICC without a valve in 26 adults and concluded that the prevalence of complications between the two groups was similar (26) .
However, the occurrence of complications is influenced not only by the material, but also the number of lumens.
A study which analysed 4,000 PICC placements in adults

Conclusion
The risk score for unplanned PICC removal developed by this study is a potentially useful tool for the