Risk factors for PICC-associated phlebitis in preterm infants using random forest algorithm

Abstract

Background

Peripherally inserted central catheter (PICC) is an important intravascular access for critically ill neonates and is widely used in neonatal intensive care unit (NICU). Phlebitis is one of the most common PICC-associated complications, with a negative impact on the PICC functionality. However, there remains insufficient research on the incidence and risk factors of PICC-associated phlebitis, particularly in preterm infants.

Objectives

To identify the incidence and explore the risk factors of PICC-associated phlebitis among preterm infants.

Methods

This is a single-center retrospective cohort study. Clinical data from all preterm infants who underwent PICC placement were collected in NICU of Nanjing Women and Children’s Healthcare Hospital in eastern China from January 01, 2019 to December 31, 2023. PICCs were divided into two groups as phlebitis group and non-phlebitis group. Based on random forest algorithm and LASSO regression, the risk factors of PICC-associated phlebitis were explored.

Results

A total of 536 PICCs inserted were included of which 161 PICCs were assigned to phlebitis group and 375 PICCs were assigned to non-phlebitis group. The incidence of PICC-associated phlebitis was 30.0% (161/536). The random forest algorithm showed that the model exhibited the least variation when the lambda (λ) value was 0.0318, the ranking according to the most contribution of PICC-associated phlebitis risk was non-central PICCs, number of attempts for successful insertions, lower limb venous catheterization, postmenstrual age at catheterization, weight at catheterization, gender, left limb venous catheterization, BPD. Multivariate Logistic regression analysis showed that non-central PICCs (OR 2.546; 95% CI 1.639–3.953), number of attempts for successful insertions ≥ 5 times (OR 2.579; 95% CI 1.012–6.571), and lower limb venous catheterization (OR 13.917; 95% CI 4.068–47.615) were risk factors for PICC-associated phlebitis among preterm infants.

Conclusions

The incidence of PICC-associated phlebitis was high among preterm infants with PICCs in NICU. Non-central PICCs, number of attempts for successful insertions ≥ 5 times, and lower limb venous catheterization served as risk factors of PICC-associated phlebitis among preterm infants.

Background

PICCs are widely used in neonatal intensive care units (NICUs) to provide intravascular access for prolonged therapy and total parenteral nutrition (TPN) to preterm infants [1]. Clinical advantages of using PICC’s include extended indwelling duration, exceptional tolerance to hyperosmolar solutions, and excellent tissue biocompatibility [2]. Although the use of PICCs offers clinical benefits, they are also associated with risks of complications, including phlebitis, infiltration, occlusion, cardiac tamponade, pericardial and pleural effusion, etc. [3, 4]. Phlebitis is one of the most common PICC-associated complications with the incidence varied from 3.2 to 34.5% [5–8]. Phlebitis leads to a negative impact on the PICC functionality and may induce non-elective removal [3, 9]. In addition, severe phlebitis causes long-term pain which may affect the development of preterm infants’ nervous system, leading to emotional cognition and behavioral disorders [10, 11]. Hence, it is of high interest to identify the risk factors of phlebitis and reduce its incidence among preterm infants.

In recent years, some researches have been conducted on identifying influencing factors for PICC-associated phlebitis in neonates. Publicly available data on the occurrence of PICC-associated phlebitis were related to lower extremity venous catheterization [12–14], while the protective factors were basilic vein [15] and axillary vein [8]. Nevertheless, these studies have primarily concentrated on neonates or preterm infants, it is unknown whether these influencing factors are applicable to very preterm infants (VPI) or very low birth weight infants (VLBWI). This uncertainty is attributable to distinct challenges in VPI/VLBWI, including but not limited to more immature and fragile vasculature, smaller vascular diameter, greater technical difficulty during catherization and prolonged catheter indwelling duration [16]. In addition, they lacked a comparative assessment of the relative importance of these factors. The random forest model not only ranks the importance of each variable but also offers greater accuracy and efficiency [17]. Meanwhile, LASSO regression can eliminate non-significant variables, significantly simplifying the relationship between dependent and independent variables while preventing model overfitting. Thus, in present study we specifically focused on VPI/VLBWI and combined the random forest algorithm with LASSO regression to identify risk factors for PICC-associated phlebitis, aiming to provide evidence-based insights for prevention.

Methods

Study population and design

This study was a single-center retrospective cohort study conducted in NICU at a tertiary hospital in eastern China from January, 2019 to December, 2023. Most preterm infants in NICU were transferred from delivery room or operating room in this hospital. The study was approved by the Medical Ethics Committee of Nanjing Women and Children’s Healthcare Hospital (Approval No: 2020KY068).

Inclusion criteria included: (1) gestational age < 32 weeks or birth weight < 1500 g, (2) born in our hospital or transferred to our hospital within 24 h after birth, (3) PICCs inserted within the first 30 days of life and retention time ≥ 72 h.

Exclusion criteria included: (1) failure to puncture or retention time of PICCs < 72 h, (2) preterm infants with PICC inserted in other NICUs or hospitals, (3) preterm infants had already received one or more PICCs previously, (4) congenital malformations, chromosomal/genetic abnormalities, or immunodeficiency diseases, (5) incomplete medical records.

Sample size

Based on the events per variable (EPV) principle, the sample size was calculated as

$$n=\frac{\text{number of independent variables}\times \text{required events per variable}}{\text{expected incidence of phlebitis in preterm infants}}$$

According to literature reports [18], the incidence of phlebitis in preterm infants was 0.22. With 5 planned independent variables and an EPV set at 20, the calculated sample size was 454.

Primary objectives: To assess the risk factors of PICC-associated phlebitis in preterm infants.

Secondary objectives: To assess the incidence and timing of PICC-associated phlebitis and rate of non-elective removal due to phlebitis in preterm infants.

Data collection

Collected data included (1) general information of preterm infants, including gender, gestational age, birth weight, 5-min Apgar scores, etc., (2) clinical data on complications in preterm infants, including respiratory distress syndrome (RDS), hemodynamically significant patent ductus arteriosus (hsPDA), early onset sepsis (EOS), late-onset sepsis(LOS), stage Ⅱ or higher necrotizing enterocolitis (NEC), brain injury (including intraventricular hemorrhage and periventricular white matter softening), moderate and severe bronchopulmonary dysplasia (BPD), extrauterine growth retardation (EUGR), (3) data on PICC catheterization, including postmenstrual age at catheterization, weight at catheterization, postnatal age at catheterization, vein selected for insertion, number of attempts for successful insertions, catheter tip position, whether the catheter tip site was adjusted post first X-ray, time of phlebitis occurrence, clinical treatment for phlebitis, duration of phlebitis, and PICC outcomes after phlebitis.

Study definition and diagnostic criteria

Diagnostic criteria

1. hsPDA was diagnosed clinically and by ultrasound as previously described [19].

2. Sepsis was diagnosed following the Expert Consensus on Diagnosis and Treatment of Neonatal Sepsis (2019 Edition) [20].

3. BPD was diagnosed according to the 2001 criteria established by the U.S. National Institute of Child Health and Human Development [21].

4. EUGR was defined as the weight below the 10th percentile of the same gestational age and sex at postmenstrual age (PMA) 36 weeks or at discharge according to the 2013 Fenton curve for children [22].

5. The diagnostic criteria for RDS, NEC, and brain injury were based on the Practice of Neonatology (5th edition) [23].

Relevant definitions

(1) Phlebitis was an inflammatory reaction of the venous intima layer. The diagnostic criteria for phlebitis were referred to the American Infusion Nurses Society (INS) [24], including 5 grades, (A) grade 0: no symptoms; (B) grade I: redness with or without pain at the infusion site; (C) grade II: pain at the infusion site accompanied by redness and/or edema; (D) grade III: pain at the infusion site accompanied by redness and/or edema, cord-like formation, and palpable cord-like veins; and (E) grade IV: pain at the infusion site accompanied by redness and/or edema, cord-like formation, palpable vein cords longer than 1 inch, and pus discharge. The phlebitis in this study was PICC-associated phlebitis and included grades I to IV. (2) Elective removal was defined as removal for the end of the therapy and non-elective removal was defined as removal due to complications associated with the line. (3) Catheter tip position adjustment referred to pull or insert the catheter if it was not a central line after X-ray.

PICC insertion criteria

(1) Umbilical vein catheterization failed and there was no contraindication for PICC catheterization. PICC contraindications, including: history of deep vein thrombosis, severe coagulation disorders, vascular malformations, active severe hemorrhage, systemic infections (e.g., sepsis and bacteremia), severe local skin infections at proposed insertion site, or burn injuries [25]. (2) Need more than 6 days of parenteral nutrition or long-term use of irritating drugs after umbilical vein was removed [26]. (3) The need for parenteral nutrition with a dextrose concentration of > 10%. (4) Difficult to insert peripheral venous catheterization and the expected fasting time more than one week.

Data management and quality control

Two data entry clerks underwent uniform training before the start of the research. EpiData software was applied to establish the database. The technical staff checked the case records, verified the accuracy and consistency of the data, and solved associated problems in time.

Statistical methods

SPSS 26.0 and R Studio software were used for statistical analysis. For univariate analysis, categorical data were expressed as frequencies and percentages (%) and comparisons between groups were made using the χ² test. Using the occurrence of phlebitis as the dependent variable, statistically significant independent variables identified in the univariate analysis were incorporated into a random forest algorithm constructed with the Random Forest package in R 4.5.0 to obtain variable importance rankings. The glmnet function was then employed to perform LASSO regression analysis on variables showing statistical significance in the univariate analysis, thereby screening for key variables. Finally, the selected variables were included in a binary logistic regression for multivariate analysis. The significance level was set at α = 0.05.

Results

A total of 536 PICCs inserted in 536 preterm infants were included, 161 PICCs in phlebitis group and 375 PICCs in non-phlebitis group. We excluded 192 PICCs, 3 PICCs inserted in other hospitals, 117 PICCs with infants gestational age or birth weight not match, 34 PICCs with failure placement or retention time less than 3 days, 17 PICCs transfer or death during PICC indwelling, and 21 PICCs with repeated catheterization or incomplete case data. (see Fig. 1).

Fig. 1.

Flowchart for inclusion in the study

The incidence and timing of phlebitis in preterm infant

The incidence of PICC-associated phlebitis among preterm infants was 30.0% (161/536). Among 161 PICC-associated phlebitis, the number of grade I, grade II, grade III and grade IV were 45 cases, 71 cases, 39 cases and 6 cases, respectively. Grade I to grade III PICC-associated phlebitis were better by setting the puncture body limb higher appropriately and using Hirudoid massage, and the PICCs could continue for intravenous therapy. However, all PICCs of grade IV phlebitis were removed due to the severity of the condition. The incidence of non-elective removal caused by PICC-associated phlebitis was 3.7% (6/161) and1.1% (6/536) of all PICCs. The median occurrence time was 3.0 (2.0, 4.0) days after successful placement, and the median duration of phlebitis was 4.0 (3.0, 6.0) days.

Demographics characteristics between phlebitis group and non-phlebitis group

There were significant difference in terms of gender, gestational age at birth and birth weight between two groups (P < 0.05 for all). There was no significant difference in the incidence of mode of delivery, 5-min Apgar scores ≤ 7, premature rupture membranes > 18 h, etc. between two groups (P > 0.05 for all), see Table 1.

Table 1.

Demographics characteristics between phlebitis group and non-phlebitis group

Demographics	Phlebitis group (n = 161)	Non-phlebitis group (n = 375)	χ2	P
Gender			4.765	0.029
Male	78(48.4)	220 (58.7)
Female	83 (51.6)	155 (41.3)
Gestational age at birth			10.727	0.005
< 28w	29 (18.0)	119 (31.7)
28 ~ 316/7w	115 (71.4)	226 (60.3)
≥ 32w	17 (10.6)	30 (8.0)
Birth weight			9.987	0.007
< 1000 g	30 (18.6)	120 (32.0)
1000 ~ 1499 g	104 (64.6)	202 (53.9)
≥ 1500 g	27 (16.8)	53 (14.1)
Mode of delivery			0.161	0.689
Normal delivery	73 (45.3)	163 (43.5)
Cesarean section	88 (54.7)	212 (56.5)
Five-minute Apgar scores ≤ 7			0.790	0.374
Yes	12 (7.5)	37 (9.9)
No	149 (92.5)	338 (90.1)
Premature rupture membranes > 18 h			0.190	0.663
Yes	35 (21.7)	88 (23.5)
No	126 (78.3)	287 (76.5)
Amniotic fluid abnormalities			1.792	0.181
Yes	22 (13.7)	69 (18.4)
No	139 (86.3)	306 (81.6)
Maternal gestational diabetes			0.023	0.878
Yes	34 (21.1)	77 (20.5)
No	127 (78.9)	298 (79.5)
Maternal hypertension in pregnancy			2.319	0.128
Yes	18 (11.2)	61 (16.3)
No	143 (88.8)	314 (83.7)
Small for gestational age infants			0.093	0.760
Yes	17 (10.6)	43 (11.5)
No	144 (89.4)	332 (88.5)

Comparison of preterm complications of infants between two groups

The incidence of moderate/severe BPD was higher in non-phlebitis compared to the phlebitis group (P < 0.05). There were no significant difference in the incidence of RDS, hsPDA, EOS, LOS, NEC ≥ stage Ⅱ, Brain injury, EUGR between two groups (P > 0.05 for all), see Table 2.

Table 2.

Comparison of preterm complications of infants between two groups

	Phlebitis group (n = 161)	Non-phlebitis group (n = 375)	χ2	P
RDS (n, %)	158(98.1)	356(94.9)	2.937	0.087
hsPDA (n, %)	98(60.9)	253(67.5)	2.169	0.141
EOS (n, %)	30(18.6)	92(24.5)	2.230	0.135
LOS (n, %)	26(16.1)	69(18.4)	0.391	0.532
NEC ≥ stage Ⅱ (n, %)	16(9.9)	32(8.5)	0.273	0.602
Brain injury (n, %)	15(9.3)	55(14.7)	2.839	0.092
Moderate/severe BPD (n, %)	18(11.2)	91(24.3)	11.907	0.001
EUGR (n, %)	81(50.3)	188(50.1)	0.001	0.970

^*RDS, respiratory distress syndrome; hsPDA, hemodynamically significant patent ductus arteriosus; EOS, early inset sepsis; LOS, late-onset sepsis; NEC, necrotizing enterocolitis; BPD, bronchopulmonary dysplasia; EUGR, extrauterine growth retardation

Catheter characteristics between phlebitis group and non-phlebitis group

The variables that reached significance included postmenstrual age at catheterization, weight at catheterization, number of attempts for successful insertions (P < 0.05 for all). The proportion of non-central PICCs, lower limb venous catheterization, and left limb venous catheterization in phlebitis group were significantly higher than those in non-phlebitis group (P < 0.05 for all), see Table 3.

Table 3.

Catheter characteristics between phlebitis group and non-phlebitis group

	Phlebitis group (n = 161)	No phlebitis group(n = 375)	χ2	P
Non-central PICCs (n, %)	66 (41.0)	77 (20.5)	24.106	< 0.001
Catheter tip position adjustment (n, %)	36 (22.4)	92 (24.5)	0.293	0.589
Lower limb venous catheterization (n, %)	158 (98.1)	315 (84.0)	21.703	< 0.001
Left limb venous catheterization (n, %)	68(42.2)	115(30.7)	6.705	0.010
Postmenstrual age at catheterization (n, weeks)			12.886	0.002
< 28	12(7.5)	67(17.9)
28 ~ 316/7	96(59.6)	223(59.5)
≥ 32	53(32.9)	85(22.7)
Weight at catheterization (n, grams)			11.856	0.003
< 1000	28(17.4)	119(31.7)
1000 ~ 1499	105(65.2)	207(55.2)
≥ 1500	28(17.4)	49(13.1)
Postnatal age at catheterization (n, days)			4.768	0.190
< 3	6(3.7)	11(2.9)
3 ~ 6	68(42.2)	185(49.3)
7 ~ 13	65(40.4)	148(39.5)
≥ 14	22(13.7)	31(8.3)
Number of attempts for successful insertions (n, times)			7.309	0.026
< 3	129(80.1)	331(88.3)
3 ~ 4	19(11.8)	31(8.3)
≥ 5	13(8.1)	13(3.5)
Central line-associated bloodstream infection (n, %)	2(1.2)	9(2.4)	0.751	0.386
Malposition (n, %)	19(11.8)	31(8.3)	1.664	0.197

Selection of risk factors for PICC-associated phlebitis in preterm infants

Variable importance ranking

Using the occurrence of phlebitis as the dependent variable, variables with statistically significant differences identified in the univariate analysis were incorporated into a random forest algorithm. The Random Forest package in R Studio was utilized to generate the output results. Based on the mean decrease in accuracy, the variables were ranked in descending order of importance as follows: non-central PICCs, number of attempts for successful insertions, lower limb venous catheterization, postmenstrual age at insertion, birth weight, weight at catheterization, gestational age at birth, gender, left limb venous catheterization, BPD, see Fig. 2.

Fig. 2.

Prioritization of risk factors for PICC-associated phlebitis in preterm infants

Variable selection

The results of the LASSO regression analysis are presented in Fig. 3. As shown, the penalty intensity increased with higher λ values. When the lambda (λ) value reached 0.0318, the model achieved optimal stability. Ultimately, eight risk factors were selected, ranked in descending order of importance as follows: non-central PICCs, number of attempts for successful insertions, lower limb venous catheterization, postmenstrual age at catheterization, weight at catheterization, gender, left limb venous catheterization, BPD.

Fig. 3.

Feature variable selection based on LASSO regression analysis

Multivariate logistic regression analysis of PICC-associated phlebitis in preterm infants

With the occurrence of PICC-associated phlebitis as the dependent variable, multivariate binary logistic regression analysis was performed. Based on the LASSO regression results, the top 8 most significant independent variables identified by the random forest algorithm were incorporated into a binary logistic regression analysis. The catheter tip position was a binary variable, with central PICCs value of 0 and non-central PICCs value of 1. The catheterization vein was also a binary variable, with upper limb vein value of 0 and lower limb vein value of 1. Number of attempts for successful insertions was three-category variable, < 3 times value of 0, 3 ~ 4 times value of 1, ≥ 5 times value of 2. Multivariate logistic regression analysis showed that non-central PICCs (OR 2.546; 95% CI 1.639–3.953), the number of attempts for successful insertions ≥ 5 times (OR 2.579; 95% CI 1.012–6.571), and lower limb venous catheterization (OR 13.917; 95% CI 4.068–47.615) were independent risk factors for PICC-associated phlebitis in preterm infants, see Table 4.

Table 4.

Multivariate logistic regression analysis of PICC-associated phlebitis in preterm infants

Multivariate analysis	B	SE	OR	95% CI	P
Number of attempts for successful insertions
3 ~ 4 times	0.415	0.357	1.515	0.753 ~ 3.048	0.245
≥ 5 times	0.947	0.477	2.579	1.012 ~ 6.571	0.047
Non-central PICCs	0.934	0.225	2.546	1.639 ~ 3.953	< 0.001
Lower limb venous catheterization	2.633	0.628	13.917	4.068 ~ 47.615	< 0.001
Constant	− 3.496	0.760

Discussion

Preterm infants with PICCs are at high risk of PICC-associated phlebitis in NICU. The present study showed that the incidence of PICC-associated phlebitis in preterm infants was 30.0% (161/536), which was higher than the recommended rate of no more than 5% by the Association of Intravenous Nurses [27], and higher compared with the studies by Bashir (3.6%, 30/827) [5] and Liu (22.1%, 23/104) [18]. This may be attributed to the gestational age and weight restrictions for preterm infants in present study, as well as to the uniform use of 1.9 Fr catheters without adopting finer gauges tailored to the delicate venous characteristics of this population. Phlebitis is fundamentally an inflammatory response of the vascular endothelium [1]. When veins are subjected to mechanical or chemical stimuli, the endothelial cells lining the venous walls become damaged [28], triggering the release of inflammatory mediators, cytokines, and chemokines. Concurrently with this mediator release, selectins, integrins, and adhesion molecules facilitate the adhesion of leukocytes (particularly neutrophils) to the endothelial surface. Subsequently, endothelial cells release additional chemokines that attract nearby leukocytes to migrate toward the inflammatory site, where they transmigrate through the vascular endothelial gaps into the venous wall tissue, ultimately leading to the development of phlebitis. Preterm birth may disrupt or prematurely terminate vascular development, leading to endothelial dysfunction [29]. This may explain why preterm infants’ blood vessels exhibit heightened inflammatory responses to mechanical stimuli compared to term infants.

The result of multivariate logistic regression analysis showed that non-central PICCs was an independent risk factor for PICC-associated phlebitis in preterm infants (OR 2.546; 95% CI 1.639–3.953) (P < 0.05). The results of random forest algorithm showed that it was ranked in number one according to the most contribution of PICC-associated phlebitis risk. A retrospective cohort study by Jain et al. [30] analyzed complications associated with PICC tip position in neonates and showed an increase risk of mechanical complications with peripheral catheter tip location versus central catheter tip location. Similarly, van et al. [16] emphasized that a central tip position was crucial to secure successful catheter retention for the duration of therapy. Racadio et al. [31] also compared the incidence of PICC-associated complications between central lines and non-central lines in pediatrics. They found that non-central PICCs were associated with higher rate of phlebitis compared with central PICCs. In present study, we found the similar trends, although the study population was different. Central lines tip position is in the superior vena cava or inferior vena cava with large lumen diameter and fast blood flow velocity which can dilute highly irritating or hypertonic liquids quickly. Moreover, the tip of central line is parallel to the blood vessel wall which also decreases irritation to the blood vessel. This may explain the lower incidence of PICC-associated phlebitis with central lines compared to non-central lines. Intracavitary electrocardiography (IC-ECG) could increase accuracy of optimal tip position and reduce phlebitis incidence compared with traditional X-ray [32]. It could be recommended for clinical providers as the primary method for guiding PICC tip placement.

The present study showed that the number of attempts for successful insertions ≥ 5 times was an independent risk factor for PICC-associated phlebitis in preterm infants (OR 2.579; 95% CI 1.012–6.571) (P < 0.05). Preterm infants, especially VPI/VLBWI, have immature peripheral blood vessels with narrow diameters, thin tubes, and few elastic fibers and smooth muscles. Furthermore, procedural pain and irritability may stimulate cortisol and catecholamines release, consequently inducing vasospasm [11]. PICC insertion and catheter advancement present inherent technical challenges. If resistance is felt while advancing the catheter, it is advisable to avoid repeated attempts, as this can lead to increased mechanical irritation of the vascular endothelium, significantly raising the risk of mechanical phlebitis. Fluoroscopic or contrast imaging should be promptly performed to evaluate the cause. A Japanese case report [33] indicates that when catheter advancement was initially impeded, venography could be employed to identify venous stenosis or obstruction, thereby facilitating successful catheter placement. A prospective observational cohort study [34] demonstrated that a dedicated nursing team could increase the first attempt for successful central line placement and reduce the incidence of phlebitis. Topical anesthetics such as EMLA cream and lidocaine gel are commonly used to provide analgesia for needle-related procedures [35]. Nevertheless, the safety of drug analgesia for preterm infants needs further verification to reach a consensus in medical staff.

This study showed that lower limb venous catheterization (OR 13.917; 95% CI 4.068–47.615) (P < 0.05) was an independent risk factor for PICC-associated phlebitis in preterm infants. Ma et al. [14] reported significantly higher risks of infiltration and phlebitis associated with lower extremity (LE) PICCs in neonates with gastroschisis, particularly when catheters were placed during silo reduction or within 5 day post-abdominal closure. A retrospective study [13] of neonates on comparing outcomes of PICCs placed in upper extremity (UE) versus LE (including 138 cases in UE group and 227 cases in LE group) also found that LE PICCs were more likely to be removed for phlebitis. These results are consistent with the present study. However, a meta-analysis [9] showed comparable result in the incidence of phlebitis between UE group and LE group. These contrasting results may be due to the wide ranges of gestational age and birth weight among the included infants. Even though LE PICCs were associated with higher incidence of phlebitis, UE PICCs demonstrated increased risks of malposition and life-threatening pleural or pericardial effusions in neonates [4, 13]. Currently, PICC insertion sites are chosen based on the inserter’s preference and availability of suitable peripheral vein for cannulation rather than which extremity offers less complication [5]. Our finding alerts clinicians to necessity of additional considerations regarding PICC-associated complications. For LE PICCs, clinicians should increase awareness of high incidence of phlebitis and appropriately elevate the puncture limb or apply pharmacological agents (e.g., Hirudoid^®) post-procedure as soon as possible to prevent phlebitis or mitigate its severity.

Strengths and limitations

The results of present study supplemented the current knowledge about the risk factor and rank for the PICC-associated phlebitis in preterm infants (gestational age < 32 weeks or birth weight < 1500 g) in NICU. Our finding highlighted critical prevention for PICC-associated phlebitis management and provided a scientific basis for optimizing PICC clinical practice among preterm infants. This study had several limitations that should be considered when interpreting the results. First, this study was a retrospective study, we obtained the catheter tip location from record in the database, and it was impossible to review all the X-rays. The tips of some described central lines maybe located at non-central locations. Second, this study was single-center retrospective cohort study. A multi-center study would be more meaningful and convincing. Third, the population included in this study was preterm infants with gestational age < 32 weeks or birth weight < 1500 g, which may limit the generalizability of the results in neonatal patients of other backgrounds. Furthermore, due to the insufficient sample size of Grade Ⅳ phlebitis, we did not perform a sub-analysis to specifically identify the risk factors. Further studies are required to more fully explore the risk of Grade Ⅳ phlebitis among preterm infants, aiming to reduce non-elective removal.

Conclusion

In conclusion, PICC-associated phlebitis is common among preterm infants in NICU. Non-central PICCs, number of attempts for successful insertions ≥ 5 times, and lower limb venous catheterization sever as risk factors of PICC-associated phlebitis in preterm infants. Non-central PICCs should be discouraged in preterm infants as these catheters are associated with higher incidence of phlebitis. Strategies to reduce number of attempts for successful insertions should be considered in clinical practice, and further research is warranted to explore preventive measures. Moreover, for the site selection of PICC placement, clinicians should consider many factors, including infants’ clinical conditions, availability of suitable peripheral vein, and extremity-specific PICC complications additionally before making a final decision.

Abbreviations

PICC

Peripherally Inserted Central Catheter

NICU

Neonatal Intensive Care Unit

TPN

Total parenteral nutrition

VPI

Very preterm infants

VLBWI

Very low birth weight infants

EPV

Events per variable

RDS

Respiratory distress syndrome

hsPDA

Hemodynamically significant patent ductus arteriosus

EOS

Early inset sepsis

LOS

Late-onset sepsis

NEC

Necrotizing enterocolitis

BPD

Bronchopulmonary dysplasia

EUGR

Extrauterine growth retardation

PMA

Postmenstrual age

INS

The American Infusion Nurses Society

OR

Odds Ratio

CI

Confidence Interval

NANN

The Internation Association of Neonatal Nurses

IC-ECG

Intracavitary electrocardiography

LE

Lower extremity

UE

Upper extremity

Author contributions

X.S. and W.X. conceptualized and designed the study, drafted the initial manuscript, and approved the final manuscript as submitted; P.D. and R.G. collected and analyzed the data, reviewed and revised the manuscript; Q.W. and J.L. reviewed and revised the manuscript; B.L. and A.Z. made supportive contributions and contributed to the critical revision of the manuscript. All the authors have read and approved the manuscript.

Funding

This study was supported by the Nanjing Medical University Science and Technology Development Fund Project (NMUB20230073).

Data availability

The datasets generated or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The study followed the Declaration of Helsinki and was conducted in accordance with the relevant local guidelines and regulations. The study was approved by the Medical Ethics Committee of Nanjing Women and Children’s Healthcare Hospital (Approval No: 2020KY068). In addition, waiver of informed consent was granted by the Medical Ethics Committee of Nanjing Women and Children’s Healthcare Hospital.

Consent for publication

Not applicable.

Competing interest

The authors declare no competing interests.