Full title: peripheral venous catheter complications in children: predisposing factors in a multicenter prospective cohort study

We conducted a prospective, multicenter study over a period of 2 months, including April and May of 2015.Two university hospitals in Tunis participated: The Children’s Hospital Béchir Hamza of Tunis (CHBH), the pediatric Hospital in Tunisia, and La Rabta Hospital (LRH),a predominantly adult hospital hosting a pediatric department specialized in inherited metabolic diseases. In total, four pediatric departments and one pediatric surgery department participated in the study. The overall hospitalization capacity of these departments is 260 beds. Only the pediatric department of LRH is a mother-child ward. In the other wards, children are hospitalized by themselves and mothers are allowed to visit for a few hours during the day.

All patients admitted to the hospital and requiring at least one PIVC were included. Children referred from another department with a pre-existing PIVC were excluded. Any children who underwent PIVC placement later during hospitalization were included. Children transferred out with a PIVC in place were also noted, but they were not followed up in the wards. These PVCs were labeled as non-complication cases.

We calculated that a sample of at least 200 PVCs was clinically practical and sufficient to provide a heterogenous sample containing all important factors that contribute to complications and would be large enough to overcome potential biases [27].Therefore, we used a convenient study period of 2 months based on feasibility and the number of patients treated in the participating wards.

Study participation was initiated by a member of the research team during routine morning visits who collected baseline data and attached the data collection form to the patient’s medical file.

Each patient was given a case number, and each catheter was numbered in the order of the time of insertion.

The primary outcomes were (i) the lifespan of the PIVC (calculated from the time of insertion to the time of removal) and (ii) the occurrence of a complication (occlusion, infiltration, phlebitis, necrosis, displacement or infection). Both parameters were defined by clinicians per their standard clinical judgment. No formal assessment tool was used. The secondary outcomes were variables with potential significant associations with PVC complications, including patient characteristics (age, gender, medical diagnosis), indication for PVC, PIVC gauge, the site of insertion, medications and fluids, and administration method. These specific variables were chosen based on previous literature reporting risk factors of PVC complications in children with consideration for the feasibility of their measurement given the limited human resources dedicated to this study.

Details related to the primary and secondary outcome variables were collected on a paper form designed for this study. Two research nurses dedicated to the study completed the forms during business hours based on information from monitoring sheets used in each ward. Every day, each nurse went into one of the two hospitals on an alternating basis to ensure a daily presence in the study sites.

The participating hospitals do not have special intravenous insertion teams, so all PIVCs were inserted by nurses who had completed PIVC insertion training. Nurses were not aware of the study. Alcohol 70%was used to decontaminate the skin and devices involved in catheter manipulation. Both 24- and 22-gauge Teflon cannulae were used (B-CAT I.V. cannula, BIÇAKCILAR®, Istanbul, Turkey and PRIMAFLON® I.V cannula, Med Devices Lifesciences Ltd., London, UK).Ultrasound and vein visualization devices were not available and were not used for PIVC insertion during the study. Tape dressings (Neoplast ® universal tape and universal tape perforated with zinc oxide, Adhe-elss.a. Sousse Tunisia) were used for the insertion site and the catheter hub. When necessary, additional nonsterile tape, bandages and splints were used. No extension set was used between the PIVC and devices used for infusions or syringes for the administration of boluses. The infusions and medications were given by either the gravitational method (with or without a volume-controlled burette (VCB)) (Alaris® GW volumetric pump, CareFusion, Hampshire, UK and Romsons ® vented infusion set, Romsons Junior, Agra India and Infusion burette SO.F.A.P TN, Tunis, Tunisia), infusion pumps or electronic syringes (depending on their availability) (Lifepum ® Syringe pump, Well Kang Ltd., Dublin, Ireland, Perfusor ® compact, Braun and Phoenix® syringe pump, Foures, GRADIGNAN - France), or bolus injection. PIVCs were flushed with normal saline solution before use. The insertion site was assessed during nursing rounds or if a sign (from the patient himself or his mother, the medical staff or the infusion device) of potential dysfunction emerged. Dressings were replaced if they were loose or if visible ooze was noticed. In all cases, the catheters were discontinued or replaced based on clinical indications (no routine changes to the catheters).

The study was approved by local Clinical Research Ethics Committee. In our study there was no experimentation or intervention on patients. The manuscript did not contain any recognizable patient data, so a written informed consent was not required. Verbal parental permission was obtained and a subject’s consent was also obtained when appropriate.

The data were processed using SPSS Version 21 software. For categorical variables, we calculated percentages. For quantitative variables, we calculated the means, standard deviations, medians and interquartile ranges (IQRs). For intergroup comparisons, we used the Chi2 test or the Fisher exact test for categorical variables. Hazard ratios (HR) were analyzed with 95% confidence intervals (CI_95%). For quantitative variables, we used Student’s t-test or the non-parametric Mann-Whitney Utest as indicated. The Spearman test was used to determine correlations between two quantitative variables (such as the lifespan of a PIVC and the age of a patient or the total duration of intravenous (IV) therapy). The Kruskal Wallis test was used to test for differences in the median lifespan within 3 or more subgroups of a variable (such as administration method and study site). For the multivariate analysis, we included variables with a p-value < 0.2 in the univariate analysis and any other variable predicted to be a potential risk factor for the outcome variables (the occurrence of complications and PIVC lifespan) based on previous literature. The forced entry method was used to introduce all predictor variables in one block. For the variable occurrence of a complication, we used logistic binary regression and for the variable PIVC lifespan, we used linear regression. A p-value <0.05 was considered significant.

In this study, 98 children were included. The mean age was 5.2 ± 4.7 years (range 0.1–18). The medical diagnoses groups were respiratory diseases (24.5%), infectious diseases (15.3%), hematological diseases (18.3%), metabolic diseases (20.4%), dehydration (2%), surgery (6.1%), neurological diseases (3.1%) and other diseases (11.2%). The details of the participants’ characteristics are given in Table 1.

Two hundred fifteen PIVCs were inserted with a mean of 2.28 ± 1.69 PIVCs/patient (range 1–9). Twenty-two-gauge catheters were used in 53.7% of the cases. The insertion site was in the hand in 82.2% of the cases. The main indications for PVCs were drug administration in 67.8% of the cases, infusion in 24.8% of the cases and blood product transfusion in 7.5% of the cases. The details of the device characteristics are given in Table 2. The gravitational method (with or without VCB) was predominantly used, accounting for 77.7% of the cases. The total duration of IV therapy was significantly longer when a VCB was used (14.91 ± 4.01 days) compared to the gravitational method (7.62 ± 4.01 days) or an infusion pump (6.23 ± 4.54 days) (p < 0.001).

Complications were observed in 111 PIVCs (51.9%)in 46 children (46%), including infiltration in 80 cases (37.2%), accidental removal in 19 cases (8.8%), phlebitis in 11 cases (5.1%) and skin necrosis in one case. No local infection or device-related bloodstream infections were observed. Among the PIVCs with complications, 67 (60%) were replaced with a new device.

Among the patient- and device-related variables considered in the univariate analysis, the following variables were significantly associated with the occurrence of local complications: a small catheter gauge (24-gauge) (p = 0.023), the use of the gravitational method, especially with a VCB (p = 0.036), a shorter indwelling time (p < 0.001), a longer duration of IV therapy (p < 0.001),medical diagnosis, especially respiratory and infectious diseases (p = 0.047), the type of IV drug used, especially antibiotics (p = 0.031), including cefotaxime(p = 0.024) and vancomycin (p = 0.031), and the use of proton pump inhibitors (PPIs) (p = 0.004). All of these variables were considered in the multivariate model and the use of PPIs (p = 0.011), a longer duration of IV therapy (p < 0.001) and a shorter indwelling time (p = 0.011) were found to be statistically significant.

None of the other variables, such as the ward, a patient’s age and gender, insertion site (in the hand, foot or head), the type of infusion, the administration of blood products, or the method of drug administration, contributed to the occurrence of local PVC complications. Table 3 shows the comparison of children and device characteristics in the two groups with and without complications.

When we considered each type of complication individually, some variables were associated with the occurrence of a specific type of complication in the multivariate analysis. Infiltration (or extravasation) was associated with a shorter indwelling time (p < 0.001), a longer duration of IV therapy (p = 0.007), the use of antibiotics, including ceftriaxone (p = 0.035) and vancomycin (p = 0.032), and PPI use (p = 0.005). Phlebitis was associated with PPI use (p = 0.031) and ceftriaxone (p = 0.017) and vancomycin (p = 0.011) use. Accidental PIVC removal was associated with a shorter indwelling time (p = 0.011).

The cumulative lifespan of a PIVC was up to 15,011 h. The mean lifespan was 68.82 ± 35.71 h (range 1–168). The factors associated with PIVC lifespan were catheter gauge, the administration method, the rank of the IV device in each patient, the nature of the infusion and the study site. The mean lifespan was shorter for 24-gauge catheters (61.32 ± 33.25 h) compared to 22-gauge catheters (77.21 ± 36.26 h) (p < 0.001). When the gravitational method was used to administer infusions or drugs, the lifespan of the PIVC was longer (72.56 ± 35.85 h) when an electronic pump or syringe was used (60.03 ± 33.88 h) (p < 0.001). The lifespan of a PIVC decreased significantly with the rank of the device in the same patient (Pearson coefficient = −0.216, p = 0.001). The lifespan was shorter with the use of 10% dextrose (49.95 ± 28.70 h) compared to 5% dextrose (74 ± 39.39 h) and normal saline solutions (98.67 ± 47 h) (p = 0.005). Participants from the pediatric department of LRH had the shortest PIVC lifespan (57.14 ± 33.34 h) compared to the other departments (lifespan ranging from 65 ± 28.73 to 96 ± 33.94 h)(p = 0.005). When we considered the subgroup of PIVCs with complications, we found that the difference in lifespan between the different study sites was non-significant (p = 0.080). In the subgroup of devices without complications, the lifespan was significantly lower in the pediatric department of LRH(64.7 ± 32.56 h) than in the other wards (lifespan ranging from 86 ± 34 to 96 ± 33 h)(p = 0.001).

In the multivariate analysis, the use of 10% dextrose was the only factor associated with a reduced PIVC lifespan (p = 0.003). Table 4 shows the details of the comparison of the means and standard deviations PIVC lifespan between subgroups of patients and devices characteristics.

Our study is the first multicenter cohort study on PVC complications in a tertiary pediatric population in a developing country. Our results could be regarded as preliminary data for the implementation of practices to reduce the incidence of PVC complications in children in Tunisia and in other developing countries.

However, a few limitations exist in this study. The lack of funding limited the number of people dedicated to data collection. The two nurses who collected the data were volunteers, as this study was part of their end-of-study project, and they could only be present during business hours. This could be a source of bias as some patients, especially those who were admitted and released or transferred during night shifts, and rapidly removed devices could have been missed. To some extent, this could explain the longer mean lifespan of PIVCs in our study compared to other studies in children.

The variables studied were numerous, but because of limited feasibility due to limited resources for our study, we could not integrate all potential factors that could affect the occurrence of PVC complications, including details related to PIVC fixation and dressing, maintenance practices, and details regarding PIVC use (the sizes of the syringes used to administer each medication, the type of pump used, the infusion rate, and adherence to recommendations for each medication or device used) [15, 25, 39, 42‐44]. These variables should be considered in future studies to evaluate actions that can be taken in our hospitals to reduce PVC-related complications.