Incidence and risk factors of vancomycin-associated nephrotoxicity among neonates in the intensive care unit: a retrospective cohort study

Abstract

Background

Vancomycin-induced nephrotoxicity is associated with an increased risk of neonatal mortality and prolonged neonatal intensive care unit (NICU) stay. This retrospective cohort study aimed to assess the incidence of vancomycin-associated nephrotoxicity and its risk factors in neonates.

Methods

The study included neonates admitted to the NICU of the Maternity and Children Hospital (Makkah, Saudi Arabia) from January 2018 to December 2020 who received intravenous vancomycin for more than 48 h. The primary outcome was the incidence of newly developed acute kidney injury (AKI) after the initiation of vancomycin therapy, defined according to the neonatal modified Kidney Disease: Improving Global Outcomes criteria. Analyses included AKI incidence, comparison of clinical parameters between neonates with and without AKI, and multivariable logistic regression to identify predictors of AKI and mortality.

Results

Among 362 neonates treated with vancomycin, 11.3% developed AKI. Neonates with AKI had a lower postmenstrual age (31.3 vs. 34.1 weeks, p = 0.01) and higher rates of extreme prematurity (31.7% vs. 14.3%, p = 0.02) and extremely low birth weight (43.9% vs. 18.7%, p = 0.00). Vancomycin trough levels > 15 mg/L were more frequent in the AKI group (26.8%, 11/41) than in others (11.8%, 35/321) (p = 0.00). Only one-third of vancomycin trough measurements were obtained according to guideline-recommended timing. In multivariable analysis, each additional week of postmenstrual age was independently associated with a 10% reduction in the odds of AKI (OR 0.90; p = 0.00). Higher weight was associated with lower mortality (OR = 0.26; p < 0.001), while AKI incidence was associated with increased mortality (OR = 5.88; p < 0.001). Caffeine citrate use was associated with lower odds of mortality, but was not associated with AKI (OR = 0.21; p < 0.001).

Conclusion

Lower postmenstrual age independently predicted vancomycin-associated AKI. Lower neonatal weight and AKI were associated with increased mortality, reinforcing the importance of individualized monitoring during vancomycin therapy.

Graphical abstract

A higher-resolution version of the Graphical abstract is available as Supplementary information

Supplementary Information

The online version contains supplementary material available at 10.1007/s00467-026-07296-5.

Introduction

Vancomycin continues to be widely used in neonatal intensive care units (NICUs) worldwide. A prevalence of 26% across 29 countries demonstrates that it is one of the most frequently prescribed antibiotics among neonates receiving at least one antimicrobial therapy [1]. Despite its widespread use, nephrotoxicity remains one of the most concerning complications among patients receiving vancomycin therapy, particularly in vulnerable populations such as neonates [2]. The reported incidence rates of vancomycin-associated nephrotoxicity vary widely in neonatal cohorts, ranging from 2.7% to 20% [3], and it typically occurs within 4–8 days of vancomycin initiation [4].

In neonatal intensive care settings, vancomycin is commonly used to treat suspected or confirmed bacterial infections, particularly those caused by Staphylococcus aureus, including methicillin-resistant strains [5]. Because vancomycin has a narrow therapeutic index, determining the optimal dosing and monitoring strategy in neonates remains challenging [6]. Vancomycin use in neonates requires careful clinical consideration because of high inter-individual pharmacokinetic variability [7].

Vancomycin trough levels are routinely measured to assess efficacy, with target concentrations of 10–15 µg/mL for moderate infections and from 15–20 µg/mL for serious infections in neonates [8]. The higher target range is generally considered sufficient to achieve an area under the curve (AUC)–to–minimum inhibitory concentration ratio ≥ 400 in neonates [9, 10].

Vancomycin is primarily eliminated by the kidneys through glomerular filtration. However, neonates have a low glomerular filtration rate, which results in decreased vancomycin clearance and increased systemic exposure [11]. Vancomycin-associated nephrotoxicity resulting from accumulation of vancomycin in the proximal tubule cells leads to oxidative stress and mitochondrial dysfunction causing acute tubular injury [12]. Although vancomycin-associated acute kidney injury (AKI) is reversible, neonatal AKI increases the risk of mortality and may lead to long-term kidney consequences [13].

A vancomycin course < 48 h is not typically associated with AKI, especially when established risk factors are absent. However, the literature suggests that vancomycin-associated nephrotoxicity is multifactorial. Vancomycin-related factors include elevated vancomycin serum trough levels [14], prolonged treatment duration [15] and high cumulative doses [15]. The concurrent use of nephrotoxic agents such as piperacillin/tazobactam [16] and furosemide [15] also increases the susceptibility to nephrotoxic injury. Patient-related factors such as lower gestational age [4] and low birth weight may increase the risk of vancomycin-associated nephrotoxicity [13]. Moreover, critically ill neonates with sepsis, hypotension [17], necrotizing enterocolitis [13], mechanical ventilation [18], and other severe conditions may be further predisposed to nephrotoxicity.

There is a paucity of research specifically evaluating vancomycin-associated AKI in the neonatal population in Saudi Arabia. To date, only one study has examined this association in a broader pediatric population [19], with no available data focusing exclusively on neonates. Although pharmacokinetic data have been collected from very low birth weight neonates in Saudi Arabia to optimize vancomycin dosing [20], these data have not been integrated with risk factor analyses aimed at minimizing kidney complications. This retrospective cohort study aimed to assess the incidence of vancomycin-associated nephrotoxicity and its associated risk factors in the Saudi neonatal population.

Methods

Study design, setting, and patients

This single-center retrospective cohort study was conducted at the Maternity and Children Hospital (Makkah, Saudi Arabia) from January 2018 to December 2020. Medical records were reviewed to identify neonates (< 28 days old) who were admitted to the NICU and were administered intravenous vancomycin for > 48 h. In our NICU, kidney function is assessed prior to vancomycin initiation using a standard kidney profile, and vancomycin is administered intermittently based on gestational age and postnatal age according to Neofax (Pediatrics) by IBM Micromedex [21]. Patients who had kidney impairment, including AKI or chronic kidney disease (CKD), before vancomycin administration, those who underwent hemodialysis or peritoneal dialysis before or during vancomycin administration, those with underlying congenital heart diseases and/or kidney diseases that could contribute to AKI, and those with no recorded baseline or follow-up laboratory work (serum creatinine [SCr]), were excluded. For patients who received more than one course of vancomycin, only data pertaining to the first course were extracted for the purpose of the current study to ensure independent observations and avoid confounding related to repeated exposures.

Data collection

The following information was extracted from both manual chart review and the automated electronic medical record: patient demographics (postnatal age, weight, sex, nationality); indication for vancomycin use (empiric or culture-directed); vancomycin dose regimen; duration of vancomycin therapy; vancomycin dose adjustment or dose hold (if vancomycin trough concentrations were < 10 mg/L or > 20 mg/L or if SCr increased from baseline based on the study definition of SCr); concomitant medications with special focus on potentially nephrotoxic medications such as aminoglycosides and piperacillin/tazobactam; baseline SCr level and all SCr levels recorded during vancomycin administration and after vancomycin discontinuation until patient death or discharge; length of hospital stay; and hearing assessment results (new hearing loss confirmed by audiology evaluation during vancomycin therapy or after discontinuation during NICU stay). For vancomycin trough levels, the number of the dose before which the level was obtained was recorded, with reference to the start of vancomycin therapy. The records of the patients who developed AKI were followed up for 2 years after the first date of vancomycin administration to assess long-term kidney consequences including the development of CKD. Follow-up included kidney function tests, new clinical events pertaining to kidney function, dialysis requirements, and nephrology referral, if applicable. If no follow-up visit was recorded, the patient was considered not to have developed further kidney injury.

Outcomes

The primary outcome was the occurrence of AKI, defined according to the neonatal modified Kidney Disease: Improving Global Outcomes (mKDIGO) criteria [22], with reference to changes in SCr. Patients who developed AKI were classified based on mKDIGO into stages for descriptive purposes only and were not incorporated into comparative analysis because of the small number of AKI cases. Baseline SCr was defined as the value recorded within 24 h or the most recent SCr level obtained within 5 days before vancomycin initiation. Values were classified as normal according to gestational- and postnatal-age-specific reference ranges reported in the literature [23–25]. SCr monitoring during and after vancomycin administration was based on clinical indications. Monitoring frequency varies according to patient risk category, with more frequent assessments (on a near-daily basis) in high-risk neonates, including preterm neonates, those with sepsis, hemodynamic instability, or exposure to nephrotoxic medications. The development of CKD was defined according to KDIGO [22] as functional or structural kidney damage persisting for at least 3 months. The secondary outcome was the identification of risk factors for vancomycin-associated nephrotoxicity in neonates.

Statistical analyses

Data were analyzed using SPSS version 21.0 (IBM Corp., Armonk, NY, USA). Normality was assessed via histograms and the Shapiro–Wilk test. Categorical variables are presented as counts (percentages) and compared using the chi-square or Fisher’s exact test (when ≥ 25% of expected counts < 5). Numeric data were checked for normality; normally distributed data are presented as the mean ± standard deviation, and non-normally distributed data are presented as the median and quartiles. The Student t-test was used to compare normally distributed numeric data between AKI and non-AKI groups, and the Mann–Whitney test was used for non-normally distributed numeric data. All comparisons were two-sided, with an alpha of 0.05. Spearman’s correlation examined associations between vancomycin trough and maximum creatinine levels. Logistic regression identified predictors of AKI and mortality. AKI analysis was performed twice: for the entire cohort and for patients with trough levels drawn 30 min before the fourth dose. Univariable regression tested potential factors including: vancomycin trough level, concurrent nephrotoxic drugs, therapy duration, weight, gestational week, postmenstrual age, and delivery mode. Significant variables (p < 0.1) were included in a multivariable model with backward elimination (Wald method). Both initial and final models are reported.

Results

Study population

A total of 1093 neonates were admitted to the NICU during the period covered by the current study. Among those admitted to the NICU, 471 neonates received vancomycin, with a prevalence of 43%. Among neonates who received vancomycin, 109 were excluded based on eligibility criteria (Fig. 1). A total of 362 patients were included in the analysis, of whom 41 developed AKI during vancomycin treatment (stage 1, n = 21; stage 2, n = 16; stage 3, n = 4 patients), resulting in an overall incidence of 11.3%. The distribution of AKI severity stages, expressed as the percentage of affected patients, is presented in Fig. 2.

Fig. 1.

Flowchart of neonate inclusion and exclusion. SCr, serum creatinine

Fig. 2.

Distribution (%) of AKI severity stages according to neonatal modified Kidney Disease: Improving Global Outcomes criteria. AKI, acute kidney injury; SCr, serum creatinine

Patient characteristics

Baseline demographic and clinical characteristics by AKI status are presented in Table 1. Patients with AKI were younger (median, 3.0 vs. 5.0 days; p = 0.05) and had lower weight at vancomycin initiation (1.1 vs. 1.7 kg; p = 0.01). Moreover, postmenstrual age was significantly lower in neonates who developed AKI than in those without AKI (31.3 vs. 34.1 weeks, p = 0.01). Extremely low birth weight (< 1 kg) and gestational age < 28 weeks were also significantly associated with AKI (p = 0.00 and p = 0.02, respectively). Septic shock (14.6% [6/4] vs. 4.4% [14/321]; p = 0.01) and neonatal seizures (9.8% [4/41] vs. 2.5% [8/321]; p = 0.01) were more common in the AKI group. No significant differences were found for sex, nationality, delivery mode, or feeding type (all p > 0.05).

Table 1.

Characteristics of neonatal patients with and without AKI

Variable	Total (n = 362)	Without AKI (n = 321)	With AKI (n = 41)	p Value
Postnatal age (days)	5.0 [3.0, 9.3]	5.0 [3.0, 9.5]	3.0 [2.0, 9.0]	0.05*
Postmenstrual age (weeks)	33.9 [30.0, 37.8]	34.1 [30.4, 37.9]	31.3 [27.5, 36.8]	0.01*
Weight (kg)	1.6 [1.1, 2.6]	1.7 [1.1, 2.7]	1.1 [0.9, 2.3]	0.01*
Sex				0.74
Male	203 (56.1)	142 (44.2)	17 (41.5)
Female	159 (43.9)	179 (55.8)	24 (58.5)
Nationality				0.53
Saudi	222 (61.3)	195 (60.7)	27 (65.9)
Non-Saudi	140 (38.7)	126 (39.3)	14 (34.1)
Birth weight classification				0.00*
NBW	106 (29.3)	98 (30.5)	8 (19.5)
LBW	83 (22.9)	77 (24.0)	6 (14.6)
VLBW	95 (26.2)	86 (26.8)	9 (22.0)
ELBW	78 (21.5)	60 (18.7)	18 (43.9)
Gestational week				0.02*
FT	116 (32)	106 (33.0)	10 (24.4)
LPT	83 (22.9)	78 (24.3)	5 (12.2)
VPT	104 (28.7)	91 (28.3)	13 (31.7)
EPT	59 (16.3)	46 (14.3)	13 (31.7)
Mode of birth				0.98
Spontaneous vaginal delivery	195 (53.9)	173 (53.9)	22 (53.7)
Cesarean section	167 (46.1)	148 (46.1)	19 (46.3)
Feeding at admission				0.08
Breast milk	9 (2.5)	9 (2.8)	0 (0.0)
Formula	140 (38.7)	118 (36.8)	22 (53.7)
Not reported	213 (58.8)	194 (60.4)	19 (46.3)
Reason for admissiona
Respiratory distress syndrome	101 (27.9)	90 (28.0)	11 (26.8)	0.87
Sepsis	72 (19.9)	62 (19.3)	10 (24.4)	0.44
Neonatal seizure	12 (3.3)	8 (2.5)	4 (9.8)	0.01*
Neonatal jaundice	8 (2.2)	7 (2.2)	1 (2.4)	0.91
LBW	23 (6.4)	20 (6.2)	3 (7.3)	0.79
Preterm	31 (8.6)	28 (8.7)	3 (7.3)	0.76
Other reason for admission	85 (23.5)	79 (24.6)	6 (14.6)	0.16
Diagnosisa
Respiratory distress syndrome	267 (73.8)	232 (72.3)	35 (85.4)	0.07
Sepsis	164 (45.3)	141 (43.9)	23 (56.1)	0.14
Neonatal seizures	52 (14.4)	45 (14.0)	7 (17.1)	0.60
Neonatal jaundice	46 (12.7)	44 (13.7)	2 (4.9)	0.11
ELBW	25 (6.9)	19 (5.9)	6 (14.6)	0.04*
Preterm	34 (9.4)	30 (9.3)	4 (9.8)	0.93
Meningitis	45 (12.4)	43 (13.4)	2 (4.9)	0.12
Septic shock	20 (5.5)	14 (4.4)	6 (14.6)	0.01*
Asphyxia	29 (8.0)	26 (8.1)	3 (7.3)	0.86
Retinopathy	15 (4.1)	12 (3.7)	3 (7.3)	0.28

Values are presented as median [quartiles] or n (%)

*Statistically significant

^aVariable is not mutually exclusive

AKI, acute kidney injury; NBW, normal birth weight (> 2.5 kg); LBW, low birth weight (1.5–2.5 kg); VLBW, very low birth weight (1–1.5 kg); ELBW, extremely low birth weight (< 1 kg); FT, full term (> 37 weeks); LPT, late preterm (32–37 weeks); VPT, very preterm (28–31 weeks); EPT, extremely preterm (< 28 weeks)

Among neonates receiving vancomycin, 80.1% (290/362) were treated empirically and 19.9% (72/362) were treated with culture-confirmed therapy, with no significant difference between the two groups (p = 0.443) (Table 2). Staphylococcus epidermidis was the most common isolate in initial cultures and was recognized as both a pathogen in NICU neonates and a possible contaminant (51.4% [37/60]), mostly from blood samples (66.7% [48/62]) (Online Resources 1 and 2).

Table 2.

Comparison of clinical vancomycin administration and concomitant medication use during therapy in neonates with and without AKI

Variable	Total (n = 362)	Without AKI (n = 321)	With AKI (n = 41)	p Value
Vancomycin use				0.44
Definite (culture confirmed)	72 (19.9)	62 (19.3)	10 (24.4)
Empirical	290 (80.1)	259 (80.7)	31 (75.6)
Vancomycin dose (mg/kg)				0.68
Mean ± SD	10.3 ± 2.0	10.3 ± 2.1	10.1 ± 0.5
Median [quartiles]	10.0 [10.0, 10.0]	10.0 [10.0, 10.0]	10.0 [10.0, 10.0]
Vancomycin frequency				0.05*
Every 8 h	104 (28.7)	95 (29.6)	9 (22.0)
Every 12 h	187 (51.7)	169 (52.6)	18 (43.9)
Every 18 h	60 (16.6)	49 (15.3)	11 (26.8)
Every 24 h	9 (2.5)	7 (2.2)	2 (4.9)
Every 36 h	1 (0.3)	1 (0.3)	0 (0.0)
Every 48 h	1 (0.3)	0 (0.0)	1 (2.4)
Duration of therapy (days)	4.0 [3.0, 6.0]	4.0 [3.0, 6.0]	4.0 [3.0, 6.0]	0.85
Cumulative vancomycin dose per kilogram	90.0 [60.0, 135.0]	90.0 [60.0, 140.0]	80.0 [54.4, 118.5]	0.15
Trough level category
< 5 mg/L	98 (28.3)	87 (28.5)	11 (26.8)	0.07
5–10 mg/L	129 (37.3)	117 (38.4)	12 (29.3)
10–15 mg/L	72 (20.8)	65 (21.3)	7 (17.1)
> 15 mg/L	47 (13.6)	36 (11.8)	11 (26.8)	0.00a
Trough levels drawn per recommended guidelines	124 (34.3)	111 (34.6)	13 (31.7)	0.57
Sample drawn before the fourth dose or a subsequent dose	56 (44.1)	50 (43.9)	6 (46.2)	0.88
Vancomycin trough concentration 1 (μg/mL)b	8.30 [4.89, 12.54]	8.15 [4.88, 12.22]	10.28 [4.74, 16.78]	0.11
Vancomycin trough concentration 2 (μg/mL)b	n = 40	n = 32	n = 8
Median [quartiles]	10.47 [4.64, 13.21]	10.47 [4.48, 12.30]	10.53 [6.00, 22.35]	0.28
Number of vancomycin courses administered				0.28
One course	246 (69.0)	220 (68.5)	26 (63.4)
Two courses	93 (25.7)	79 (24.6)	14 (34.1)
Three courses	23 (6.3)	22 (6.9)	1 (2.4)
Received multiple vancomycin coursesc	116 (32.0)	101 (31.5)	15 (36.6)	0.50
Dose holding based on vancomycin trough concentration	17 (4.7)	13 (4.0)	4 (9.8)	0.10
Dose adjustment based on vancomycin trough concentration	37 (10.2)	28 (8.7)	9 (22.0)	0.00*
Nephrotoxic medication use				0.94
Yes	258 (71.3)	229 (71.3)	29 (70.7)
No	104 (28.7%)	92 (28.7%)	12 (29.3%)
Number of nephrotoxic medications per patient
1	231 (63.8)	205 (63.9)	26 (63.4)	1.0
2	26 (7.2)	23 (7.2)	3 (7.3)
3	1 (0.3)	1 (0.3)	0 (0.0)
Nephrotoxic medications
Amikacin	180 (49.7)	165 (51.4)	15 (36.6)	0.07
Gentamycin	11 (3.0)	9 (2.8)	2 (4.9)	0.47
Piperacillin/tazobactam	95 (26.2)	80 (24.9)	15 (36.6)	0.11
Not classically nephrotoxic medications
Hydrocortisone	7 (1.9)	3 (0.9)	4 (9.8)	0.00*
Levetiracetam	5 (1.4)	2 (0.6)	3 (7.3)	0.01*
Paracetamol	45 (12.4)	41 (12.8)	4 (9.8)	0.63
Phenytoin	7 (1.9)	3 (0.9)	4 (9.8)	0.00*
Baclofen	1 (0.3)	0 (0.0)	1 (2.4)	0.11
Meropenem	103 (28.5)	86 (26.8)	17 (41.5)	0.05*
Furosemide	8 (2.2)	6 (1.9)	2 (4.9)	0.23
Cefotaxime	27 (7.5)	23 (7.2)	4 (9.8)	0.55
Use of caffeine citrate (nephroprotective)				0.88
Yes	111 (30.7)	98 (30.5)	13 (31.7)
No	251 (69.3)	223 (69.5)	28 (68.3)
Caffeine dose (mg/kg)	4.67 [2.87, 8.33]	4.30 [2.86, 8.28]	6.25 [4.90, 9.02]	0.10
Caffeine frequency	1.0 [1.0, 1.0]	1.0 [1.0, 1.0]	1.0 [1.0, 1.0]	0.33
Caffeine duration (days)	20.0 [8.5, 32.0]	18.0 [8.0, 30.8]	28.0 [17.5, 34.0]	0.03*
Cumulative caffeine dose per kilogram	77.7 [34.1, 167.7]	72.7 [28.6, 136.8]	181.3 [89.1, 204.1]	0.00*

Values are presented as n (%), mean ± SD, or median [quartiles]

*Statistically significant

^aHighest trough level (> 15 mg/L) against the other levels

^bVancomycin trough concentration 1 was taken initially; vancomycin trough concentration 2 was taken later during therapy for monitoring

^cPatient received two or more courses of vancomycin therapy

AKI, acute kidney injury; SD, standard deviation

Epidemiology of treatment with vancomycin

The dose per kilogram did not differ between groups (p = 0.68) (Table 2). Vancomycin was most frequently administered every 12 h in 51.7% (187/362) of patients, with dosing frequency differing according to AKI status (p = 0.05). The duration of therapy and cumulative per kilogram dose were similar (p = 0.85 and p = 0.15, respectively). Trough levels > 15 mg/L were more frequent in the AKI group (26.8% [11/41] vs. 11.8% [36/321]; p = 0.01). Therapeutic drug monitoring (TDM) practices showed that approximately one-third (34.3%) of trough levels were obtained according to recommended guidelines (p = 0.57). Vancomycin trough levels showed a weak but significant positive correlation with maximum SCr during therapy (rho = 0.219; p < 0.001). A similar correlation was observed among patients whose trough measurements were obtained according to recommendations (rho = 0.222; p = 0.013) (Fig. 3). Dose adjustments based on vancomycin trough concentrations were more common in the AKI group (22.0% [9/4]) vs. 8.7% [28/321]; p = 0.01), whereas dose holding showed a non-significant trend (9.8% [4/41] vs. 4.0% [13/321]; p = 0.10) (Table 2).

Fig. 3.

Scatter diagram of the relation between vancomycin trough level and the maximum reported serum creatinine level during vancomycin therapy. (a) All patients with a reported trough level (n = 345, Spearman correlation: rho = 0.219; p < 0.001). (b) Patients with trough level drawn per-recommended guidelines (n = 124, Spearman correlation: rho = 0.222; p = 0.013)

Nephrotoxic medications were administered to approximately 71.3% of neonates, with no significant difference between those who subsequently developed AKI and those who did not (p = 0.94) (Table 2). None of the nephrotoxic medications differed significantly between groups, including piperacillin/tazobactam, amikacin, and gentamicin. Among medications not classically known to be nephrotoxic, hydrocortisone (9.8% [4/41] vs. 0.9% [3/321]; p = 0.00), levetiracetam (7.3% [3/41] vs. 0.6% [2/321]; p = 0.01), and phenytoin (9.8% [4/41] vs. 0.9% [3/321]; p = 0.00) were significantly more commonly used in patients who subsequently developed AKI. A trend toward higher meropenem use was observed in patients with AKI (41.5% [17/41] vs. 26.8% [86/321]; p = 0.05). Other medications, including cefotaxime and furosemide, did not differ significantly between groups. The use of the nephroprotective agent caffeine citrate was similar across groups (approximately 30.7%; p = 0.88). Neonates who developed AKI had a longer duration of caffeine therapy (28.0 vs. 18.0 days; p = 0.03) and higher cumulative caffeine exposure per-kilogram (181.3 vs. 72.7 mg/kg; p = 0.00).

Clinical outcomes

A comparison of the clinical outcomes by AKI status is presented in Table 3. There was no significant difference in the duration of NICU stay or in the total length of hospital stay between neonates with and without AKI (p > 0.09 and p = 0.45, respectively). Mortality during NICU admission was significantly higher in patients who developed AKI (53.7% [22/41] vs. 15.0% [48/321]; p < 0.001). Death was reported in 70 cases, of which septic shock was the predominant cause, accounting for 76.8% (n = 54) of all documented deaths. This was followed by intraventricular hemorrhage (17.4%, n = 12) and encephalopathy (2.9%, n = 2). Necrotizing enterocolitis and respiratory failure were the least frequently known causes. Ototoxicity rates did not differ significantly (p = 0.31), and baseline SCr was similar between groups (p = 1.00). Patients with AKI had higher peak SCr (1.25 [0.94, 1.43] vs. 0.59 [0.49, 0.73] mg/dL; p < 0.001) and greater median percentage post-therapy decrease in SCr concentration after vancomycin discontinuation, but the difference was not statistically significant (46.3 [26.1, 53.6] vs. 23.3 [9.5, 38.6] mg/dL; p = 0.09).

Table 3.

Clinical outcome in neonatal patients with and without AKI

Variable	Total (n = 362)	Without AKI (n = 321)	With AKI (n = 41)	p Value
Duration of NICU stay (days)a	23.0 [14.0, 44.8]	22.0 [13.5, 42.0]	32.0 [17.0, 59.0]	0.09
Length of hospital stay (days)	28.0 [16.0, 47.2]	27.0 [16.0, 46.0]	33.0 [15.5, 52.5]	0.45
Mortality	70 (19.3)	48 (15.0)	22 (53.7)	< 0.001*
Ototoxicity				0.31
No	171 (47.2)	155 (48.3)	16 (39.0)
Yes	5 (1.4)	4 (1.2)	1 (2.4)
Not reported	186 (51.4)	162 (50.5)	24 (58.5)
Baseline serum creatinine (mg/dL)	0.62 [0.49, 0.75]	0.62 [0.49, 0.75]	0.65 [0.46, 0.76]	1.00
Maximum creatinine concentration during vancomycin therapy (mg/dL)	0.62 [0.49, 0.78]	0.59 [0.49, 0.73]	1.25 [0.94, 1.43]	< 0.001*
Minimum serum creatinine after vancomycin discontinuation (mg/dL)	0.45 [0.37, 0.57]	0.44 [0.36, 0.53]	0.62 [0.44, 0.80]	< 0.001*
Decrease in creatinine level after vancomycin discontinuation (%)	24.9 [10.5, 42.2]	23.3 [9.5, 38.6]	46.3 [26.1, 53.6]	0.09

Values are presented as median [quartiles] or n (%)

*Statistically significant

^aLength of stay calculated among survivors only

AKI, acute kidney injury; NICU, neonatal intensive care unit

Predictors of AKI and mortality

In the univariable logistic regression of factors associated with AKI (Table 4), analysis of all neonates showed that higher weight was associated with reduced AKI risk (odds ratio [OR], 0.65; p = 0.03), whereas lower postmenstrual age (OR = 0.90; p = 0.007), low birth weight (OR = 1.59; p = 0.00), and extent of prematurity (OR = 1.49; p = 0.01) were associated with increased AKI risk. Most other variables were not significant; meropenem showed a borderline association (p = 0.05). Notably, caffeine citrate (OR = 1.06; p = 0.88) did not demonstrate a nephroprotective effect against AKI.

Table 4.

Univariable logistic regression analysis of potential risk factors for AKI in neonates who received vancomycin

	All neonates (n = 345)			Subgroup with trough levels measured before next dose (n = 124)
Variable	OR	95% CI	p Value	OR	95% CI	p Value
Postnatal age	0.96	[0.90, 1.03]	0.28	0.93	[0.83, 1.05]	0.26
Postmenstrual age	0.90	[0.84, 0.97]	0.00*	0.90	[0.79, 1.03]	0.13
Sex: male	1.12	[0.58, 2.16]	0.74	2.11	[0.55, 8.1]	0.28
Weight	0.65	[0.44, 0.95]	0.03*	0.82	[0.42, 1.56]	0.54
Low birth weight	1.59	[1.17, 2.17]	0.00*	1.41	[0.80, 2.45]	0.23
Extent of prematuritya	1.49	[1.1, 2.03]	0.01*	1.34	[0.78, 2.31]	0.29
Delivery mode (C/S vs. SVD)	1.01	[0.53, 1.94]	0.98	2.75	[0.8, 9.45]	0.11
Respiratory distress syndrome (diagnosis)	2.24	[0.91, 5.50]	0.08	2.43	[0.51, 11.55]	0.27
Sepsis (diagnosis)	1.63	[0.84, 3.14]	0.14	2.73	[0.84, 8.91]	0.10
Neonatal seizure (diagnosis)	1.26	[0.53, 3.02]	0.60	1.65	[0.32, 8.44]	0.55
Neonatal jaundice (diagnosis)	0.32	[0.08, 1.39]	0.13	0.00b	[0.000]	1.00
Duration of vancomycin therapy	1.07	[0.99, 1.15]	0.11	1.01	[0.85, 1.2]	0.90
Vancomycin trough concentration	1.01	[0.99, 1.04]	0.34	1.03	[0.10, 1.1]	0.18
Trough concentration categoryc	1.31	[0.95, 1.8]	0.10	2.13	[1.19, 3.9]	0.01*
Multiple courses of vancomycin therapy administered	1.26	[0.64, 2.48]	0.51	1.6	[0.50, 5.31]	0.43
Use of nephrotoxic medications	0.97	[0.48, 1.99]	0.94	4.05	[0.50, 32.55]	0.19
Number of nephrotoxins medications	0.97	[0.55, 1.71]	0.91	1.57	[0.56, 4.37]	0.39
Aminoglycosides	0.64	[0.33, 1.23]	0.18	0.56	[0.18, 1.79]	0.33
Piperacillin/tazobactam	1.74	[0.88, 3.44]	0.11	4.01	[1.24, 13.03]	0.02*
Meropenem	1.94	[0.99, 3.78]	0.05	0.51	[0.11, 2.5]	0.41
Caffeine citrate	1.06	[0.53, 2.13]	0.88	0.45	[0.09, 2.14]	0.32

*Statistically significant

^aExtent of prematurity was coded from 1 to 4 for full term, late preterm, very preterm and extreme preterm, respectively

^bFor neonatal jaundice, the OR was 0.000 due to extreme values, and the 95% CI could not be calculated

^cTrough level categories were coded from 1 to 4 for < 5, 5–10, 10–15, and > 15 mg/L, respectively

AKI, acute kidney injury; OR, odds ratio; CI, confidence interval; C/S, cesarean section; SVD, spontaneous vaginal delivery

In the subgroup with vancomycin trough levels measured before the next dose (Table 4), higher trough category (OR = 2.14; p = 0.01) and piperacillin/tazobactam use (OR = 4.01; p = 0.02) were significantly associated with increased AKI risk. Other variables, including vancomycin trough concentration and co-administration of nephrotoxic medications (e.g. amikacin, gentamicin, piperacillin/tazobactam), were not significant; sepsis (p = 0.10) and delivery mode (p = 0.11) showed marginal associations but did not reach statistical significance.

Using backward elimination, the final logistic regression model for AKI (Table 5) identified only the postmenstrual age as a predictor of AKI (OR = 0.90; p = 0.00). After excluding several other clinical variables, including weight, clinical reasons, and medication exposures (Online Resource 3), each one-week increase in postmenstrual age was associated with a 10% reduction in the odds of the outcome.

Table 5.

Multivariable logistic regression model for AKI and mortality predictors after backward elimination

Variable	OR	95% CI	p Value
AKI predictors (final model)
Postmenstrual age	0.90	0.83–0.96	0.00*
Mortality predictors (final model)
Weight (kg)	0.26	0.16–0.41	< 0.001*
Caffeine citrate	0.21	0.10–0.45	< 0.001*
Aminoglycosides	0.54	0.29–1.02	0.058
Trough concentration category	1.31	0.97–1.76	0.08
AKI incidence	5.88	2.63–13.12	< 0.001*

*Statistically significant

AKI, acute kidney injury; OR, odds ratio; CI, confidence interval

Backward elimination identified key predictors of mortality (Table 5). Higher weight was identified as a protective factor, with each unit increase being associated with a 74% reduction in the odds (OR = 0.26). Caffeine citrate use was associated with a 79% reduction in odds (OR = 0.21). AKI incidence was associated with approximately sixfold higher odds of mortality (OR = 5.88). Other variables in the initial model were excluded for non-significance (Online Resource 4).

Discussion

This study evaluated the incidence and risk factors of vancomycin-associated AKI in neonates admitted to a tertiary NICU in Saudi Arabia. Among 362 neonates treated with vancomycin for > 48 h, 41 (11.3%) developed AKI. Of these, 51% were classified as stage 1 according to the mKDIGO definition. Neonates who developed AKI were more premature and had lower birth weight.

Vancomycin therapy was prescribed in approximately 43% of admitted neonates. Although global prevalence of NICU-specific vancomycin data are limited, reported antimicrobial use varies widely (< 10% to > 50%) [26], and our findings fall within expected ranges. This supports the generalizability of our results despite differences in NICU admission patterns, reflecting global variability in clinical practice.

The observed AKI incidence (11.3%) aligns with previously reported rates (2.7–20%) [13]. Differences across studies likely reflect variations in patient populations and AKI definitions. Among the studies included in the systematic review by Gao et al. [13], only one cohort study from the USA used the mKDIGO; the others used KDIGO [27], Acute Kidney Injury Network (AKIN) [28], or Risk, Injury, Failure, Loss of function, End-stage kidney disease (RIFLE) classifications [29]. Although these criteria appear similar, they can yield different AKI incidence estimates and affect risk factor identification.

Extremely low birth weight (< 1 kg) and very or extremely preterm neonates were at high risk of AKI, consistent with kidney immaturity and reduced drug clearance [30]. Shock and neonatal seizures were also frequent among neonates with AKI. These conditions may exacerbate kidney hypoperfusion and tubular injury, supporting the multifactorial nature of kidney injury [31]. In contrast, sex, nationality, feeding pattern, and delivery mode were not associated with AKI.

In the current study, vancomycin dose per-kilogram did not differ between groups. However, trough concentrations were significantly higher in neonates who developed AKI, particularly at levels > 15 mg/L. Although a weak but statistically significant positive correlation was observed between trough levels and peak SCr, this finding should be interpreted as descriptive rather than confirmatory due to inconsistent timing of vancomycin trough measurements and lack of correspondence with peak creatinine. Overall, higher trough levels in the AKI group likely reflected altered pharmacokinetics, requiring closer monitoring.

Subtherapeutic vancomycin trough concentrations were observed in approximately 65% of patients receiving standard dose (< 10 mg/L); however, this finding should be interpreted cautiously because TDM timing was suboptimal, with only one-third of samples obtained within the recommended 30 min prior to the next dose. This may reflect several challenges such as nursing workload, variable dosing intervals, and limited staff awareness of optimal sampling timing. Inconsistent adherence to appropriate timing of trough measurements may have led to inaccurate estimation of vancomycin exposure and variability in assessing its relationship with AKI, underscoring the need for standardized TDM protocols. Similar inconsistencies in vancomycin monitoring, particularly inaccurately timed blood sampling, have been reported in previous studies, with implications for dosing accuracy and early detection of nephrotoxicity [32, 33].

Although recent guidelines from the Infectious Diseases Society of America and the Pediatric Infectious Diseases Society recommend individualized, AUC-guided vancomycin dosing based on pharmacokinetic principles, its practical implementation in neonatal settings remains challenging. Limited feasibility of repeated blood sampling, inconsistent sampling times, lack of well-validated neonatal-specific AUC targets, and rapidly changing neonatal pharmacokinetics often necessitate continued reliance on trough-based monitoring in most NICUs [34, 35].

In the current study, only a minority of patients received recommended TDM, highlighting the requirement for quality improvement. Structured nephrotoxic medication stewardship initiatives, such as the Nephrotoxic Injury Negated by Just-in-Time Action (NINJA) and neonatal adaptations (NAKI), have promoted standardized kidney monitoring, enhanced staff education, and automated sampling alerts, which can improve monitoring compliance and prevent AKI incidence through early identification [36, 37].

Although the number of nephrotoxic agents in the current study did not differ significantly between the groups, some medications had an impact on vancomycin nephrotoxicity. The univariable analysis revealed the association of piperacillin/tazobactam with nephrotoxicity in the subgroup of neonates with appropriately timed vancomycin trough measurements, which, however, did not withstand statistical adjustment and lost significance in the multivariable analysis. The synergistic nephrotoxic effect of this combination was reported in studies that included extremely low birth weight and very low birth weight infants, as well as in large systematic reviews and meta-analyses [38–40], which support our findings. This single-center cohort study may be subject to false-positive or false-negative results owing to factors related to the sample size compared with the number of variables entered in the regression model. The reported AKI associated with co-administration of vancomycin and piperacillin/tazobactam may reflect pseudo-nephrotoxicity, as piperacillin/tazobactam can reduce creatinine secretion by blocking tubular secretion through inhibition of organic anion transporters (OAT1 and OAT3) [41]. The co-administration of meropenem, which is not intrinsically nephrotoxic, was more prevalent in patients with AKI (41.5% vs. 26.8%, p = 0.05); however, this association was lost in the univariable and multivariable regression analyses of our neonatal cohort. The initial association likely reflected confounding by indication, as more severely ill neonates requiring broad-spectrum antibiotics inherently have a higher baseline risk of kidney injury, and this relationship was not sustained after adjustment for other clinical variables in the model. Consistent with our findings, previous pediatric cohort studies and a meta-analysis have reported no independent association between meropenem use and vancomycin-associated AKI despite higher observed trough concentrations [42, 43].

Although previous studies have identified aminoglycosides, among other nephrotoxic medications, as contributors to vancomycin-induced nephrotoxicity [44, 45], such an association was not observed in the current study in either the univariable or multivariable analysis. The absence of significant association between aminoglycoside use and AKI may reflect the relatively small number of affected neonates as well as cautious dosing practices and avoidance of prolonged therapy in our NICU. Although a borderline association with mortality was observed in multivariable analysis, this finding was not statistically significant and likely represents confounding by indication and underlying illness severity rather than a direct drug effect.

In the current study, patients with AKI experienced marked elevations in peak SCr during therapy (median increase to 1.25 [0.94, 1.43] vs. 0.59 [0.49, 0.73] mg/dL; p < 0.001). Following vancomycin discontinuation, the SCr levels decreased in both groups; however, the AKI group showed a more gradual reduction, suggesting ongoing kidney recovery or residual impairment. Although some patients with AKI may recover, concerns remain regarding the potential long-term kidney consequences of neonatal vancomycin exposure, including persistent impairment of kidney function. In the current cohort, none of the patients who recovered from vancomycin-associated AKI experienced long-term kidney outcomes, including clinical events pertaining to kidney function, dialysis requirements, and nephrology referral within 2 years of the initial vancomycin administration. The mortality rate was significantly higher in neonates with AKI (53.7%) than in those without AKI (15.0%; p < 0.001). Several studies have reported that neonatal AKI is associated with increased morbidity, higher mortality, and long-term kidney sequelae [3, 46, 47].

In the univariable regression analysis, markers of developmental immaturity, including lower postmenstrual age, low birth weight, and greater extent of prematurity, were significantly associated with AKI in our cohort, thus reflecting underlying delayed kidney maturation in preterm neonates and emphasizing the necessity for improved kidney surveillance in high-risk neonates [47]. Although some variables, such as sepsis, respiratory distress syndrome, and the use of nephrotoxic drugs, did not reach statistical significance in the univariable model, their ORs tended to be consistent with clinical expectations. Despite being a single-center study, the current study succeeded in showing the importance of these factors as determinants of vancomycin nephrotoxicity in neonates, as these were the same determinants confirmed in the meta-analysis by Gao et al. [13].

In the multivariable model, postmenstrual age was the only independent predictor of AKI, with every extra week of postmenstrual age associated with a 10% lower risk. This finding is consistent with a systematic review of critically ill neonates showing that reduced glomerular filtration capacity and kidney immaturity increase susceptibility to kidney damage in preterm neonates [13]. Although low birth weight was associated with AKI in univariable analysis, it was not retained after backward elimination, likely due to collinearity with postmenstrual age. Postmenstrual age more accurately reflects kidney maturity and physiological vulnerability, which may explain why it remained the sole independent predictor in the multivariable model. Moreover, AKI was a significant predictor of mortality after adjustment for multiple factors in the multivariable regression model. It was expected that factors that covaried with AKI would not be retained in the model. However, neonatal weight remained a significant and independent predictor of mortality.

In our cohort, caffeine use was similar between groups, although neonates with AKI had a longer duration of therapy and cumulative exposure. Caffeine use was not retained as a protective factor in the regression model evaluating AKI risk. However, caffeine was independently associated with substantially lower mortality (OR = 0.21; p < 0.00). Although previous studies suggest reduced AKI with early caffeine use [48, 49], this association should be interpreted as observational rather than causal, as higher cumulative exposure in our cohort likely reflects underlying illness severity and prematurity rather than a protective drug effect.

In our cohort, baseline SCr levels were within gestational age-appropriate reference ranges in all neonates. SCr levels fluctuated during the first two weeks of neonatal life due to physiological maturation of kidney function. As a result, inaccurate classification of normal baseline kidney function based on early SCr readings may occur, especially in preterm neonates. These limitations of serum creatinine in neonates have been well documented, as SCr rises several days after kidney injury and reflects glomerular filtration rather than structural damage, requiring cautious interpretation when defining AKI [30].

This study has several limitations. First, owing to its retrospective design, data collection was limited to information available in medical records. Inconsistent vancomycin monitoring may have contributed to variability in nephrotoxicity outcomes. AKI was defined using SCr alone without urine output criteria, and SCr follow-up measurements were not consistently obtained. This may have limited our ability to accurately assess the association between vancomycin exposure and nephrotoxicity. Furthermore, neonatal length was not consistently recorded throughout the study period, hindering the calculation of eGFR using the Schwartz equation; thus, maximum SCr levels were used as the sole indicator of kidney function. Because this was a single-center study with limited insights into the regional epidemiology, larger multicenter studies are needed to better characterize patterns of vancomycin-associated nephrotoxicity. The potential influence of unmeasured confounders cannot be excluded. Despite these limitations, this study provides valuable insight into the multifactorial nature of vancomycin-associated AKI in neonates and highlights the importance of optimizing TDM practices and improving early identification of kidney injury risk in this vulnerable population.

In conclusion, this study highlights the significant risk of vancomycin-associated AKI in neonates, particularly those with lower postmenstrual age. Lower neonatal weight and the presence of AKI were associated with increased mortality. These findings underscore the importance of accounting for neonatal maturity when assessing AKI risk, implementing optimized dosing strategies, ensuring consistent therapeutic monitoring, and early identification of high-risk patients to reduce nephrotoxicity and improve outcomes. Vancomycin pharmacokinetics are markedly altered in neonates, necessitating careful dose individualization.

Supplementary Information

Below is the link to the electronic supplementary material.

Author’s contribution

All authors contributed to the study conception and design. Material preparation, data collection, and analysis were performed by Hanouf Bafhaid, Ibraheem Bagbag, Abdullah Dearar, Rashed Monshi, Hanadi Alrammaal, Lujain Khan, Walaa Felemban, Kholod Alhazmi, and Roaa Alharbi. The first draft of the manuscript was written by Hanouf Bafhaid, and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

No funding was received for conducting this study.

Data availability

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

Declarations

Ethical approval and informed consent

The study was approved by the Research Ethics Committee at the Directorate of Health Affairs in Makkah Region (H-02-K-076–1222-866), and the requirement to obtain informed consent was waived.

Consent to publish

Not applicable.

Competing interests

The authors have no relevant financial or non-financial interests to disclose.