Inappropriate Use of Peripherally Inserted Central Catheters in Pediatrics: A Multisite Study Free

This multisite, retrospective cohort study was a collaborative effort between 4 pediatric medical centers representing the Midwest and Southeast regions of the United States and has been reported in alignment with the Strengthening the Reporting of Observational Studies in Epidemiology Guidelines.⁸  Two of the medical centers are medium size (200–400 beds), 1 small (<200 beds), and 1 large (>400 beds). All 4 sites are standalone children’s hospitals affiliated with a medical school. The study cohort included all PICCs placed in hospitalized children aged <18 years between January 2019 and December 2021. PICCs placed in patients admitted to the NICU were excluded, as per the scope of miniMAGIC guidance.

The population of interest was identified using electronic medical record (EMR)-based databases at each participating institution. Encounters with PICCs were identified using the line/drain/airway section of the EMR. Measures unavailable in the database were collected manually via retrospective chart review. To ensure consistency and accuracy in data collection and recording, a detailed protocol with clear definitions and directions was developed. Each collected variable had predetermined answer choices. Because only deidentified data were harmonized by the coordinating center, the institutional review board at each site deemed the study as exempt.

PICC utilization trends were assessed over time, pre and post-miniMAGIC guideline, and between sites using the proportion of total PICCs placed over the total inpatient encounters that satisfied the inclusion/exclusion criteria. All PICCs from the identified cohort were classified as appropriate, inappropriate, or uncertain appropriateness on the basis of dwell time, type of infusate (peripheral compatible or not), and continuation at discharge. Definitions are presented in Fig 1.

A more detailed analysis was performed for the patients with PICCs in the inappropriate and uncertain appropriateness groups. Variables were selected on the basis of established best practices.^9,10  The following characteristics of the patient and clinical encounter were collected: demographic data, existence of chronic complex conditions, service requesting the PICC, primary diagnostic category (medical versus surgical, infectious versus noninfectious), and patient location (eg, intensive care, acute care). PICC data collected included main reason for PICC placement, use for blood draws, number of lumens, inserter type (eg, physician, nurse), insertion vessel, number of attempts, dwell time, reason for removal, and catheter-associated complications (VTE, CLABSI, concern for VTE, concern for CLABSI, infiltration, dislodgement, phlebitis, occlusion, line dysfunction/fracture).

Definitions for catheter-associated complications were based on national and international benchmarks that could be pragmatically applied to a retrospective chart audit. A catheter-associated VTE was defined as ultrasound finding of thrombus at the catheter site.¹¹  Concern for VTE was defined as obtaining an ultrasound to evaluate for catheter-related thrombus that was negative. CLABSI was defined using the National Healthcare Safety Network surveillance definition of a laboratory-confirmed infection where a PICC was in place for >2 calendar days before a positive blood culture, and was also in place the day of or the day before the positive culture.¹²  Concern for CLABSI was defined as a negative blood culture obtained while the catheter was in place. Additional catheter-associated complications such as infiltrations, phlebitis, and line dysfunction could not be objectively defined because of the limitation of the retrospective nature of the study and the reliance on existing EMR documentation.

Descriptive statistics were used to report patient and catheter characteristics, with PICCs being the primary unit of analysis. Continuous variables were summarized using median and interquartile range, whereas categorical variables were summarized as n (%). The Cochran-Mantel-Haenszel test was used to compare differences in appropriateness over time and between institutions. Interrupted time series analysis was used to compare the rates of inappropriate PICCs by month pre and post-miniMAGIC guideline. A P value < .05 was considered significant. SPSS version 28.0 (IBM Corp., Armonk, New York), SAS version 9.4 (SAS Institute Inc., Cary, North Carolina), and R¹³  were used for statistical analysis.

A total of 6051 PICCs were placed during the study period at the participating sites. The rate of PICC use decreased overtime; however, 16% to 20% of the PICCs placed each year were uncertain or inappropriate (Fig 2A). Over the 3-year study period, 9% (n = 550) of total PICCs were inappropriate and 9% (n = 550) were of uncertain appropriateness. Significant difference in PICC appropriateness was noted between years, P = .042, and between sites, P < .0001 (Fig 2B). The difference in appropriateness between years and sites remained significant (P = .025 and < .0001, respectively) after removing the PICCs in the inappropriate and uncertain groups that were placed in children that had:

1. >1 PICC during the same encounter (n = 21); or

2. those in which the PICC was removed early because of a complication (n = 70), assuming those cases may have been appropriate if a catheter-related complication never occurred.

When examining differences in inappropriate PICC use by month pre and post-miniMAGIC guideline (Fig 2C), we found a 22% decrease in use; however, the decrease was not statistically significant (relative risk 0.78, 95% confidence interval [0.55–1.11], P = .17).

The inappropriate and uncertain appropriateness PICC groups included a total of 1100 PICCs placed in 1079 children. Median age of the children in the groups was 4 years (interquartile range 0–11); 54% were male (Table 1). Fifty-five percent had a complex medical condition, 79% had a primary medical diagnosis, and 64% were treated for an infection during their hospitalization. The most common services requesting the PICC were: 24% critical care, 22% general pediatrics, and 20% pulmonary (Table 2). The most common reason for placing the PICC was prolonged antibiotic course (56%), followed by durable access (30%). Of the children in both the inappropriate and uncertain appropriateness groups admitted to the general pediatrics or pediatric hospitalist service, 75% (185 of 247) used the PICC for prolonged antibiotic courses. The most common infections treated being meningitis/encephalitis (25%, 47 of 185), pneumonia (13%, 24 of 185), and joint and bone infections (12%, 23 of 185). The most common reason for placing a PICC by the critical care service was the need for durable access (53%, 139 of 264), followed by prolonged antibiotic courses (33%, 88 of 264), with pneumonia and meningitis/encephalitis remaining the 2 most treated infections (29% and 25%, respectively). Within the group of PICCs placed by pulmonary service, 92% (203 of 220) were placed for prolonged antibiotic courses and, within this group, 86% PICCs were for treatment of pulmonary exacerbation of cystic fibrosis. Within the group of patients who had a primary surgical diagnosis (21%, 227 of 1100), the most common service requesting a PICC was neurosurgery (28%, 63 of 227), and the most 2 common primary diagnoses for this group were ventriculoperitoneal shunt malfunction/infection and epilepsy.

The majority of PICCs were placed by nurses (61%) and advance practice providers (35%). Eighty-eight percent of PICCs had a single lumen (Table 2). Ninety-two percent of PICCs stayed in place until completion of therapy; 6% were removed because of a catheter-associated complication. Rate of CLABSI in this cohort was 1% (n = 10) and rate of VTE was 1% (n = 13) (Table 3). The most common complication was catheter dislodgement (3%, n = 29), with 26 of 29 being removed by the patient. Concern for CLABSI was common in both the inappropriate and uncertain appropriateness groups, with a total of 184 catheters (17%) having at least 1 blood culture drawn while the catheter was in place.

This multisite, retrospective study confirms a common concern regarding the inappropriate use of PICCs in hospitalized children. The study quantifies the rate of appropriate, uncertain appropriateness, and inappropriate use of PICCs at 4 children’s hospitals in the United States, according to widely accepted international guidelines published in 2020.^5,6  Our findings suggest that the overall rate of PICC use from 2019 to 2021 is trending down (5.9% to 5.2%), which is likely multifactorial in nature (eg, evidence supporting shorter IV antibiotic courses for pediatric infections, local initiatives to reduce catheter-associated complications). However, up to 20% of PICCs placed each year do not meet criteria for appropriateness per the miniMAGIC recommendations. We did not find a statistically significant decrease in use of inappropriate PICC after the miniMAGIC guideline publication in June 2020. Per the guideline, peripheral catheters such as midline catheters or long peripheral catheters are more appropriate choices if anticipated length of therapy is ≤14 days for peripherally compatible infusates.⁶  During the US hospitalization peak of the corona disease 2019 pandemic (2020), the rate of PICC use was the highest of the 3 years included because of the lower number of total hospital encounters, the denominator (Fig 2A). The reason for the higher rate is unclear, but it’s likely multifactorial; it may be related to potentially higher acuity in the patients hospitalized during 2020 and other system-level changes (eg, staffing models, workforce configuration, infection prevention).

Similar to adult studies,¹⁴  the most common reason for placing a PICC of uncertain or inappropriate indications was prolonged antibiotic courses. Therefore, local or national quality initiatives should target this population specifically. For infections that require ≤14 days of IV antibiotics, a midline catheter or long peripheral catheter could be considered as an alternative to a PICC, even if the initial catheters may need to be replaced to complete therapy.^6,15,16  Midline catheters were found to be safe and reliable alternatives to PICCs for prolonged IV antibiotic therapy in children with cystic fibrosis.¹⁶  However, future effectiveness and implementation studies should further evaluate the introduction of midline catheters as an alternative to PICCs in general hospitalized pediatric patients. In addition, using a PICC for prolonged IV antibiotic therapy in pediatric infections such as complicated pneumonia, complicated appendicitis, and bone/joint infections is considered low-value care by clinicians and researchers in the field.^17–20  Yet, 2 of the most common reasons for PICC placement for prolonged antibiotic courses in the general pediatrics population were treatment of pneumonia and bone/joint infections. The decision to place a PICC in children with complicated infections is likely compounded by need for frequent laboratory draws, reducing the number of pokes for the patient. In our study, 80% of PICCs were used for blood draws at least once.

Our study also identified significant variation in appropriateness between the 4 participating sites. The driver of this variation is not clear. Potential contributors to appropriate use of PICCs may be related to available vascular access resources, as noted in previous studies.^21,22  For example, the site with the highest percentage of appropriate PICCs (site 1) has resources such as a vascular access team, ability to place midline catheters or long peripheral catheters, and the use of a catheter selection guide. In contrast, 2 of the other sites (including site 4), lack a catheter selection guide but they do have the other 2 resources available. None of the sites had implemented the miniMAGIC guideline at the time of the study. More research is needed to identify the barriers and facilitators that drive appropriate use of PICCs in pediatrics. In adult studies, the use of long peripheral catheters or midline catheters and a catheter selection guide decreased the use of inappropriate PICCs and, therefore, central catheter-associated complications.^23–27 

Overall, our data show that clinicians are appropriately selecting mostly single lumen PICCs, considering rate of CLABSI increases with each increase in number of lumens²⁸ ; however, more opportunities to decrease multilumen PICC use exist. In our study, 12% (n = 130) of PICCs had 2 or more lumens even when the majority of patients were outside of the ICU (72%) and all received medications that were peripherally compatible. Regarding PICC placement, we found that 19% of PICCs (n = 207) required >1 attempt for successful placement, with 7% requiring 3 or more attempts. Rates of CLABSI and VTE were lower in our population compared with other populations or pooled data from systematic reviews and meta-analysis.⁴  This is an important finding to consider when balancing the risks and benefits of selecting a PICC. Our population was primarily hospitalized on the acute care floor, so patients likely had fewer additional risk factors for developing CLABSI or VTE (eg, severity of illness).^29,30  We were unable to reliably look at trends of CLABSI and VTE over the study period because of the small number of complications overall; this is similar to published adult studies led by Chopra et al.^31,32  In addition to the harm to patients, the cost of 1 CLABSI and 1 VTE episode is estimated to be $55 646 and $27 686, respectively.^33,34  The use of resources needed to monitor for these complications add additional cost. In our study, blood cultures were obtained in 17% of the cases, whereas catheter in place (184 catheters) and ultrasound of extremity were obtained in 3% of the cases (33 catheters). The ultrasounds were always completed to assess for PICC-associated VTE development; however, it was not clear if the blood cultures were obtained because of concern for CLABSI or because of concern for bacteremia secondary to presenting illness.

Our study has a few limitations related to the retrospective nature of the study design and reliance on EMR documentation. When dividing the PICCs into the 3 groups (appropriate, inappropriate, uncertain), assumptions were made that the PICCs were nonurgently placed and were not used for central venous monitoring, both being difficult to gauge from the EMR documentation, but both situations would have made those PICCs appropriate. Need for frequent laboratory draws or difficulty obtaining IV access are sometimes reasons for placing an appropriate PICC. Similarly, a patient who is stable versus unstable in the ICU settings may have different vascular access needs, and the miniMAGIC has specific recommendations for this specialized population. Because this information is not consistently documented in the EMR, we chose not to incorporate it in our definition of appropriateness. The primary criteria devising the 3 appropriateness groups was dwell time of ≤7 days, 8 to 14 days, and >14 days. We recognize that, in some cases, the PICCs were removed before completion of therapy because of a complication, having shorter dwell time than originally anticipated. Those cases may have been otherwise appropriate. To work with this limitation, we examined trends and variation between sites after removing the complication cases. To maintain consistency in measures collected from the EMR and to allow for data harmonization, we used predetermined definitions; however, this strategy may have missed infections less commonly encountered or cases of multiple infections (eg, pneumonia and bacteremia). In addition, we only performed chart review to confirm inclusion/exclusion criteria and to collect more detailed data for the inappropriate and uncertain appropriateness groups, so we assumed that the remaining of the PICCs would, by the process of elimination, be appropriate. Because data from all 4 sites were harmonized after they were deidentified, we were not able to determine the number of patients with >1 encounter with a PICC during the study period. Additional catheter-associated complications (eg, infiltration, occlusion, dislodgment) can only be identified on the basis of EMR, so the rates described in our study may be an underestimate because documentation is sometimes lacking or incomplete. Finally, all participating sites are standalone teaching hospitals and the pattern of PICC use may not accurately describe that of a community-based hospital.

The use of PICCs has decreased over the past few years, but up to 20% of catheters each year are still not appropriately placed per the miniMAGIC. PICCs that are inappropriate or of uncertain appropriateness are more likely to be used for prolonged IV antibiotics and safe alternatives are available to support this administration. Thus, this population may be a target for future improvement efforts.