An Multi-center institution-based prospective follow-up study was conducted from May 1 to August 30, 2024, at the Northwest Amhara Region Comprehensive Specialized Hospital in Northwest Ethiopia. In the Amhara region, there are seven level III neonatal intensive care units, according to WHO standards, however in northwest Amhara region five comprehensive specialized hospitals. Out of five comprehensive specialized hospitals (CSHs), three were selected using a random sampling technique: UoG CSH, Tibebe Ghion CSH, and Debre Tabor CSH. These selected hospitals admit approximately 6,912, 2,712, and 2,424 neonates annually, respectively, of whom some require specialized care. A team of pediatricians, general practitioners, neonatal nurses, and other personnel works in these institutions. They collaborate to assess, diagnose, and provide comprehensive, holistic care for these vulnerable neonates. The neonatal intensive care units offer primary services for neonates with perinatal asphyxia, including general neonatal care such as antibiotic treatment, phototherapy, blood transfusions, exchange transfusions, and ventilatory support like Continuous Positive Airway Pressure (CPAP), along with different medications.

Source population and study population

All neonates admitted to the NICU at selected comprehensive specialized hospitals in the northwest Amhara Region constituted the source population. The study population included all neonates admitted to the NICU of these hospitals from May 1 to August 30, 2024.

Inclusion and exclusion criteria

All neonates admitted to the NICU of selected comprehensive specialized hospitals in Northwest Amhara during the study period were included. Neonates with known skin infections, those admitted for less than 24 hours, and those who developed complications before arriving at the study settings were excluded.

For the first objective, the minimum required sample size is calculated using the single population proportion formula. This calculation incorporates the following statistical assumptions: P = proportion of phlebitis (50%), α/2 = the corresponding Z score for a 95% confidence interval, and (d) the margin of error (5%). n = ( za 2 ) 2 ∗ ( p ∗ ( 1 − p ) ) d 2 n = ( 1.96 ) 2 ∗ ( 0.5 ( 1 − 0.5 ) ) ( 0.05 ) 2 = 384

After adding a 10% contingency for incomplete charts, the sample size was 423.

A proportional allocation of the sample size was conducted for each selected comprehensive specialized hospital in the northwest Amhara region, based on monthly admissions derived from annual data. Specifically, UoGCSH had an average of 345 admissions per month, while TGCSH had 250. Over a four-month period, these hospitals admitted a total of 2,380 neonates. A total of 423 neonates were selected using a systematic sampling technique ( Figure 1).

Dependent variable

Incidence of phlebitis Event/censored

Independent variables

Neonatal socio-demographic variables include sex, weight, age, and gestational age. Intravenous cannula-related factors encompass antibiotics, type of fluid flushing, blood transfusion, experience in NICU, and duration of neonatal age. Additionally, neonatal comorbidities include sepsis, meningitis, birth asphyxia, respiratory distress syndrome (RDS), congenital heart disease, hypoglycemia, and necrotizing enterocolitis (NEC).

Operational definitions

Neonates who were referred for treatment against their will, experiencing disappearance, death, or discharge with a good outcome during the study period.

Follow-up time: The time admission from the NICU until either an event or censorship occurs at 28 days.

Phlebitis: According to Jackson’s visual infusion phlebitis scoring was defined as the presence of two or more signs of induration, fever, erythema, swelling, or a palpable cord with or without purulent drainage from the catheter insertion site. ¹¹

Score 1 - Slight pain or slight redness near IV insertion site. These are possible early signs of phlebitis. Score 2 - Two of the following signs are evident: pain at IV site, redness or swelling medium stage of phlebitis called early stage of phlebitis. Score 3 - All of the following signs are evident: pain along the path of the cannula, redness around the insertion site and swelling called medium stage of phlebitis. Score 4 - All of the following signs are evident and extensive: pain along the path of the cannula, redness around the insertion site, swelling, palpable venous cord so called advanced stage of phlebitis or at the start of thrombophlebitis. Score 5 - All of the following signs are evident and extensive: pain along the path of the cannula, redness around the insertion site, swelling, palpable venous cord, pyrexia called the stage of advanced thrombophlebitis. ¹¹

The data collection tool was adapted by reviewing relevant literature related to the problem under study, ensuring that all possible variables addressing the study’s objectives were incorporated. The tool consists of five parts: Part One focuses on socio-demographic characteristics, Part Two addresses neonatal comorbidities, Part Three covers cannulation, Part Four examines neonatal outcome status, and Part Five includes the Amharic version of the maternal sociodemographic sections. Data were prospectively collected by two experienced general practitioners under the supervision of two MSc neonatal nurses, who hold bachelor’s degrees and have been working in the neonatal intensive care unit.

The data collection tool was adapted by reviewing relevant literature and drawing from validated standards, such as the Ethiopian Demographic and Health Survey checklists. Initially prepared in English, it was then translated into Amharic. To ensure consistency and clarity, language experts and health professionals retranslated it back into English. For clinical variables, an English version of the chart review checklists was developed. Before data collection commenced, the tool was evaluated and critiqued by research experts to assess its validity. After obtaining written informed consent from the mothers, data were collected through interviews and by reviewing the medical records of the neonates. The data collection process involved three supervisors and data collectors. Additionally, a 5% pretest was conducted at the Gondar Comprehensive Specialized Hospital to validate the instrument or questionnaire prior to the actual data collection.

The data were cleaned for inconsistencies and missing values, with necessary amendments considered before analysis. The cleaned data were then entered into Epi-data 4.6.0.2 software and subsequently exported to STATA 14 statistical software. Descriptive statistics, including mean with standard deviation, median with interquartile range, frequency, and percentages, were computed based on the nature of the variables. Each participant’s outcome was dichotomized into censored and event categories. The incidence density rate was calculated for the entire study period. Kaplan-Meier methods were employed to estimate the median failure time and cumulative probability of failure, with the KM plot and log-rank tests used to compare failure function curves between groups. Before performing Cox proportional hazard regression, model goodness-of-fit was assessed using the Schoenfeld residual test and Cox-Snell residuals, while multicollinearity was evaluated using variance inflation factors. Bivariable Cox proportional hazard regression was conducted for each predictor variable, including those with p-values <0.2 in the multivariable analysis. Adjusted hazard ratios with 95% confidence intervals and p-values <0.05 were utilized to determine statistical significance. Finally, the results were presented through text, graphs, figures, and tables.

In this study, the vast majority of neonates (88.18%) were admitted to the hospital for less than three days. More than half of the neonates (53.66%) were born at term, with a median weight of 2900 g (IQR 850). Additionally, over two-thirds of the newborns (69.27%) were appropriate for their gestational age. Notably, almost half of the neonates (78.01%) developed sepsis, while 26.95% developed meningitis, and 28.13% experienced birth asphyxia ( Table 2).

Most of the peripheral intravenous cannulas were inserted in the anterior cubital fossa of the neonates’ hands. This accounted for 197 (46.57%) of the cannulas and represented approximately two-thirds of the neonates, totaling 241 (52.16%) who received maintenance D10 and N/S. During the study period, 121 (28.61%) patients were flushed before receiving medication. Half of the neonates, 227 (53.66%), had dwelling times of less than four days. Among the neonates, one-fourth, or 110 (26.00%), had their peripheral intravenous cannula inserted by a nurse with less than six months of experience ( Table 3).

The PHA was assessed using the Schoenfeld residuals test. The results indicated that the p value for each covariate, as well as for all covariates together, was above 0.05 (p = 0.2322).

In this study, we followed the neonates for a maximum of 17 days from the date of admission. Throughout the follow-up period, we made a total of 4,590 observations of neonate days. This number was obtained by adding the amount of time each neonate was followed during the study, measured in days. The overall incidence of phlebitis was 20.9 per 1,000 neonate-day observations (95% CI: 17.12-25.54) during the entire follow-up period. Of the 423 neonates admitted to the NICU, 96 (22.69%) (95% CI: 18.93-26.94) developed the event of interest, phlebitis.

Out of a total of 96 neonates, 22.69% showed early signs of phlebitis. The distribution of scores is as follows: early signs of phlebitis - 15 (15.62%), early stage of phlebitis - 24 (25%), medium stage of phlebitis - 23 (23.96%), advanced stage of phlebitis or the start of thrombophlebitis - 21 (21.88%), and advanced thrombophlebitis - 13 (13.54%) ( Figure 2).

The Kaplan-Meier failure function indicated that the likelihood of developing phlebitis increased with longer follow-up times ( Figure 3).

The goodness of fit for the fitted model was checked using the Cox-Snell residual test. As shown in the figure, the Cox-Snell regression model was adequate ( Figure 4).

In the bivariable Cox regression analysis, several variables were found to be associated with phlebitis (p < 0.25), including birth asphyxia, flushing of the cannula before medication, dwelling, trial of cannulation, birth weight, educational status, congenital heart disease, and sepsis. However, in the multivariable Cox regression analysis, only five variables remained as significant predictors of phlebitis (p < 0.05).

Specifically, neonates with birth asphyxia had a 2.1 times higher hazard of developing phlebitis compared to those without birth asphyxia (Adjusted Hazard Ratio [AHR] 2.1; 95% Confidence Interval [CI] 1.38–3.19). Neonates who underwent three trials of cannulation had a 2.5 times higher hazard of developing phlebitis compared to those who underwent one or two trials (AHR 2.5; 95% CI 1.01–6.36). Additionally, neonates with birth weights of 4000 g or greater had a three times higher hazard of developing phlebitis compared to those with low and normal birth weights (AHR 3.05; 95% CI 1.17–7.96).

On the other hand, neonates who had their cannula flushed were 54% less likely to develop phlebitis compared to their counterparts (Adjusted Hazard Ratio [AHR] 0.54; 95% Confidence Interval [CI] 0.31–0.97). Similarly, neonates who had cannulas in place for less than 4 days had a 47% lower likelihood of developing phlebitis compared to those who dwelled longer (AHR 0.47; 95% CI 0.29–0.765) ( Table 4).

This study is among the first to specifically examine phlebitis in neonates, addressing a significant gap in the existing literature.

It does not explore a multidisciplinary approach to care that includes pharmacists and infection control specialists for optimizing IV therapy management. Additionally, it only assesses the vein viewer. However, it does not consider its availability in both hospitals or the newer catheter designs and materials that are less likely to irritate veins, such as those with smoother surfaces or biocompatible coatings.

Ethical clearance was obtained from the University of Gondar collage of medicine and health science Specialized hospital institutional research ethics review committee ( IRERC)/35/17/2024 of the Collage of Medicine and Health science Specialized Referral Hospital ( CMHSSH). The letter of permission was obtained from the North West Gondar Public Hospital medical director’s office. Finally, both written and verbal informed consent was obtained from the mothers after explaining all the purposes, benefits, confidentiality of the information, and the voluntary nature of participation in the study before the data collection. All methods were carried out in accordance with the declarations of Helsinki and relevant guidelines and regulations. The names and/or identification numbers of the study participants were not recorded on the data collection tool. All data were kept strictly confidential and used only for this study’s purposes.