Validation of the STARZ neonatal acute kidney injury risk stratification score in an independent prospective cohort

Abstract

OBJECTIVE:

A rapid AKI risk assessment score would allow for improving management and outcomes. STARZ (Sethi, Tibrewal, Agrawal, Raina, waZir) score was developed for acute kidney injury (AKI) risk stratification of critically ill neonates. This is the first independent validation for the novel score outside the original enrolled centres.

STUDY DESIGN:

750 neonates were included in the study. The STARZ score was calculated after 12 hours of admission. Neonates admitted in NICU and receiving IV fluids for at least 48 hours were included.

RESULTS:

A total of 8.8% neonates had AKI in the first 7 days post admission. The duration of hospital stay was significantly higher among neonates with AKI [10.5 (7–19) vs. 7 (5–10) days; p < 0.001]. Mortality risk was 6.4 times higher among those with AKI [8 (12.1%) vs. 13 (1.9%); p < 0.001; RR (95% CI): 6.38 (2.74–14.83)]. In this study, the STARZ neonatal scoring model showed a sensitivity of 89.4% in detecting AKI with a 90.9% specificity and a high negative predictive value of 98.9%. The area under ROC was 0.958 (0.934–0981) - a high discriminative power.

CONCLUSIONS:

The STARZ score allows for AKI risk stratification, providing opportunity for therapeutic interventions which may improve outcomes in critically ill neonates.

Keywords

AKI risk stratification score Neonatal AKI validation

Abbreviations

ACOG

American College of Obstetricians and Gynecologists

AKI

Acute kidney injury

Confidence interval

EMR

Electronic medical record

IQR

Inter-quartile range

ISN

International Society of Nephrology

Intravenous

KDIGO

Kidney disease improving global outcomes

LMIC

Low and middle income countries

NICU

Neonatal intensive care unit

NSAIDs

Non-steroidal anti-inflammatory drugs

PDA

Patent ductus arteriosus

PPHN

Persistent pulmonary hypertension of the newborn

PPV

Positive pressure ventilation

RAI

Renal angina index

ROC

Receiver operating characteristic

Relative risk

STARZ

Sethi, Tibrewal, Agrawal, Raina, waZir

1 Background

Acute kidney injury (AKI) in neonates was previously under-reported and understudied. Published in 2017, the aptly named AWAKEN study, was the first international, multicentre analysis of neonatal AKI, its risk factors and consequences in short term [1]. AKI in neonates has now been established as an important contributor to morbidity and mortality [1 –3]. Mortality has been found to be increased in neonates with even milder forms of AKI [3]. Owing to injury occurring in maturing kidneys in neonates, development of AKI even predisposes to chronic kidney disease in the long term [2, 4]. Thus, AKI should be actively sought and managed at the earliest. Risk stratification of neonates for AKI can help in better management and outcomes.

The International Society of Nephrology (ISN) ‘0 by 25’ AKI minimisation initiative, emphasises the need for early recognition of ‘at risk’ individuals and the utility of risk scores in identifying them, especially in resource poor areas, where access to tertiary care and interventions such as kidney replacement therapy are often suboptimal [5]. A separate neonatal modification of Kidney Disease–Improving global outcomes (KDIGO) AKI definition which recognises the unique features of this subset population facilitates better management [6]. But a scoring system, especially those linked to electronic medical record (EMR) alerts, would assist in identifying an ‘at risk’ neonate, wherein the application of preventive measures, even secondary if not primary, could allow for avoidance of further insult in a vulnerable patient and timely interventions when required.

Common illness severity scores in neonates are Neonatal Therapeutic Intervention Scoring System (NTISS), the Transport Risk Index of Physiologic Stability (TRIPS), the Clinical Risk Index for Babies (CRIB), Score for Neonatal Acute Physiology (SNAP), the simplified age–weight–sex (SAWS) score, and NMR-2000 score. In older children, illness severity scores used are Pediatric RISk of Mortality (PRISM) and Pediatric Index of Mortality (PIM 1, 2, 3). Except SAWS and NMR-2000, neonatal scores are not validated in the developing world [7, 8].

The STARZ (Sethi, Tibrewal, Agrawal, Raina, waZir) score derived from our previous multicentre study of 763 neonates in neonatal intensive care units (NICUs) across India provides a comprehensive AKI risk analysis, it included 10 variables related to AKI risk [7]. This scoring model detected AKI with a sensitivity of 92.8%, specificity of 87.4%, positive predictive value of 80.5%, negative predictive value of 95.6% and accuracy of 89.4% [7]. A recent multicentre validation study for STARZ AKI score, was conducted in 11 NICUs in India, involving 744 neonates [8]. It also reiterated a high AKI predictive ability of STARZ score in neonates under intensive care [8].

The present study is the first independent validation of the STARZ score outside the primary enrolled centres, in a single NICU in Northern India.

2 Materials and methods

2.1 Study design

This study was conducted at a single centre in northern India, Ludhiana, in a level III NICU. The primary aim was to utilise STARZ score in assessing AKI risk in neonates admitted to NICU. The present centre was not a part of the derivation cohort for the STARZ score. It was an observational, prospective study between September 2019 and August 2020 (data was collected prospectively to validate our original derivation cohort). Informed consent was obtained from one of the parents. The study was approved by the Institutional Ethics Committee.

2.2 Inclusion criteria (similar to the original study)

Neonates (≤28 days) who were admitted in NICU; and received intravenous (IV) fluids via peripheral vein/ central vein for at least 48 hours for hydration and/or nutrition were included in the study [7].

2.3 Exclusion criteria (similar to the original study) [7]

Death within 48 hours of admission

Presence of any lethal chromosomal anomaly, such as Trisomy 13, 18, anencephaly etc.

Requirement of congenital heart surgery within the first 7 days of life, as the condition may possibly be accompanied by congenital kidney abnormalities

Neonates receiving routine care in nursery without receiving IV fluids for < 48 hours

2.4 Data collection

An exhaustive clinical history was documented on entry into NICU by attending doctor. Baseline demographics regarding date of birth, gender, date of NICU admission, gestational age, birth weight, length, head circumference, need for positive pressure ventilation in delivery room, Apgar scores, need for resuscitation, mode and site of delivery, reason for admission were recorded. Maternal variables were also documented including age, gravida, parity, any pregnancy associated complications or chronic illnesses, medications, peripartum infections; maternal risk factors were identified as per standard ACOG guidelines (American College of Obstetricians and Gynecologists).

2.5 Daily assessment

Daily weight of the neonate was taken, along with fluid intake and output, blood pressure, medication use including nephrotoxic drugs, laboratory investigations such as serum creatinine, urea, sodium, total leucocyte count, markers of sepsis (blood cultures, urine cultures, and other appropriate cultures) were recorded. Haematological and kidney function parameters were recorded on a daily basis for the first week and then weekly thereafter till discharge of patient. Urine was measured via urinary catheter or weighing of diapers. In babies with acute kidney injury, renal function tests were investigated daily till resolution of AKI.

Cardiac disease was considered as significant if the newborn was detected to have persistent pulmonary hypertension of the newborn (PPHN), hemodynamically significant patent ductus arteriosus (PDA), cardiogenic shock and other congenital cardiac disease.

AKI staging was based on the modified neonatal Kidney Disease: Improving Global Outcomes (KDIGO) definitions. According to the modified neonatal KDIGO definition, AKI was classified into three stages based on increase in SCr from a previous value, decrease in urine output or both.

Neonates who developed AKI were managed as per hospital protocol, utilising diuretics, kidney replacement therapy wherever required. AKI was defined as an increase in serum creatinine of 0.3 mg/dl or more or 50% or more from the previous lowest value, or a urinary output of < 1 ml/kg/h on postnatal days 2–7, according to the KDIGO workgroup AKI definition modified for neonates [6].

2.6 Statistical analysis

The STARZ score, calculated any time after 12 hours of admission in NICU, delineates AKI risk in the first 7 days post admission. As some included variables are available at admission and certain others can be obtained within 12 hours post admission, the score needs calculation after 12 hours of admission. The score was derived from our previous multicentre study [7]. The study utilised a multivariable logistic regression using step-wise backward elimination method to identify 10 variables with significant AKI association (7 categorical, 3 continuous) [7]. Each variable was assigned a score according to previously detailed methodology; the total score can range from 0 to 100 [7].

The data was entered in an online database (similar to the original cohort). The data specific to the index centre was extracted and exported to Microsoft excel. Statistical software (SPSS version 20) was used for the statistical analyses. All the variables were tested for normality using Kolmogorov-Smirnov test. Categorical variables were summarized as frequencies and percentages, while continuous variables as medians and inter-quartile range (IQR; 25th to 75th percentiles).Univariate analysis (using the chi-square or Fischer exact test for categorical variables and the Wilcoxon’s rank-sum test for continuous variables) was carried out to assess the unadjusted relationship between the variables and AKI incidence post-admission in NICU. The risk of AKI has been reported as relative risk (RR) along with its 95% confidence interval (95% CI).

In the STARZ scoring model, a score was assigned to each of the included 10 variables, with a total score ranging from 0 to 100, and cut-off score≥31.5; where higher score indicates greater probability of AKI incidence [7]. In the present study, the predictive validity of the STARZ scoring system was validated based on the standard statistical measures [sensitivity, specificity, positive predictive value, negative predictive value, accuracy and area under the receiver operating characteristic (ROC) curve]. A two-sided p value < 0.05 was considered to be statistically significant.

Table 1
Univariate association of different categorical variables with AKI incidence within 7 days post-admission in NICU

Variables AKI p value Unadjusted RR (95% CI)

Yes [n (%)] No [n (%)]

Infections¹ 26 (39.4%) 17 (2.5%) <0.001 10.69 (7.27–15.72)

Diabetes¹ 27 (40.9%) 8 (1.2%) <0.001 14.14 (9.92–20.16)

Chronic hypertension¹ 10 (15.2%) 3 (0.4%) <0.001 10.12 (6.86–14.95)

Pre-eclampsia¹ 19 (28.8%) 21 (3.1%) <0.001 7.18 (4.68–11.00)

Intrauterine growth retardation¹ 14 (21.2%) 0 (0%) <0.001 NA

Oligohydramnios¹ 8 (12.1%) 20 (2.9%) 0.002 3.56 (1.88–6.72)

Polyhydramnios¹ 5 (7.6%) 31 (4.5%) 0.2 1.63 (0.70–3.80)

Mode of delivery [Vaginal Delivery] 35 (53%) 178 (26%) <0.001 2.85 (1.80–4.49)

Site of delivery [Outborn] 35 (53%) 216 (31.6%) <0.001 2.25 (1.42–3.55)

Gestational age < 28 weeks 6 (9.1%) 27 (3.9%) 0.04 2.17 (1.01–4.66)

Birth weight < 1,000 gm 10 (15.2%) 36 (5.3%) 0.004 2.73 (1.49–5.00)

Appropriate for age 51 (77.3%) 542 (79.2%) 0.7 0.90 (0.52–1.56)

Male 42 (63.6%) 434 (63.5%) 0.9 1.01 (0.62–1.63)

Apgar score at 5 minutes < 9 18 (58.1%) 76 (16.3%) <0.001 5.92 (3.01–11.66)

PPV in the delivery room 43 (65.2%) 93 (13.6%) <0.001 8.44 (5.27–13.52)

Age at NICU admission < 25.5 hours 38 (57.6%) 280 (40.9%) 0.003 1.41 (1.12–1.76)

Need for surfactant 27 (40.9%) 94 (13.7%) <0.001 3.60 (2.29–5.65)

Perinatal asphyxia 25 (37.9%) 74 (10.8%) <0.001 4.01 (2.56–6.29)

Feeding Intolerance 18 (27.3%) 69 (10.1%) <0.001 2.86 (1.75–4.68)

Necrotising enterocolitis 9 (13.6%) 39 (5.7%) 0.02 2.31 (1.22–4.38)

Sepsis 11 (16.7%) 43 (6.3%) 0.005 2.58 (1.44–4.63)

Significant cardiac disease² 20 (30.3%) 13 (1.9%) <0.001 9.93 (6.37–13.89)

Seizures 7 (10.6%) 7 (1%) <0.001 6.24 (3.50–11.12)

Use of nephrotoxic drugs³ 58 (87.9%) 299 (43.7%) <0.001 7.98 (3.87–16.48)

Use of furosemide 3 (4.5%) 1 (0.1%) 0.002 8.88 (4.81–16.40)

Need for inotropes⁴ 42 (63.6%) 59 (8.6%) <0.001 11.25 (7.13–17.74)

Serum creatinine≥0.98 mg/dl⁵ 46 (69.7%) 17 (2.5%) <0.001 25.08 (15.88 –39.62)

Urine output < 1.32 ml/kg/hr⁵ 52 (78.8%) 38 (5.6%) <0.001 27.24 (15.75 –47.09)

Variables	AKI	p value	Unadjusted RR (95% CI)
Infections¹	26 (39.4%)	17 (2.5%)	<0.001	10.69 (7.27–15.72)
Diabetes¹	27 (40.9%)	8 (1.2%)	<0.001	14.14 (9.92–20.16)
Chronic hypertension¹	10 (15.2%)	3 (0.4%)	<0.001	10.12 (6.86–14.95)
Pre-eclampsia¹	19 (28.8%)	21 (3.1%)	<0.001	7.18 (4.68–11.00)
Intrauterine growth retardation¹	14 (21.2%)	0 (0%)	<0.001	NA
Oligohydramnios¹	8 (12.1%)	20 (2.9%)	0.002	3.56 (1.88–6.72)
Polyhydramnios¹	5 (7.6%)	31 (4.5%)	0.2	1.63 (0.70–3.80)
Mode of delivery [Vaginal Delivery]	35 (53%)	178 (26%)	<0.001	2.85 (1.80–4.49)
Site of delivery [Outborn]	35 (53%)	216 (31.6%)	<0.001	2.25 (1.42–3.55)
Gestational age < 28 weeks	6 (9.1%)	27 (3.9%)	0.04	2.17 (1.01–4.66)
Birth weight < 1,000 gm	10 (15.2%)	36 (5.3%)	0.004	2.73 (1.49–5.00)
Appropriate for age	51 (77.3%)	542 (79.2%)	0.7	0.90 (0.52–1.56)
Male	42 (63.6%)	434 (63.5%)	0.9	1.01 (0.62–1.63)
Apgar score at 5 minutes < 9	18 (58.1%)	76 (16.3%)	<0.001	5.92 (3.01–11.66)
PPV in the delivery room	43 (65.2%)	93 (13.6%)	<0.001	8.44 (5.27–13.52)
Age at NICU admission < 25.5 hours	38 (57.6%)	280 (40.9%)	0.003	1.41 (1.12–1.76)
Need for surfactant	27 (40.9%)	94 (13.7%)	<0.001	3.60 (2.29–5.65)
Perinatal asphyxia	25 (37.9%)	74 (10.8%)	<0.001	4.01 (2.56–6.29)
Feeding Intolerance	18 (27.3%)	69 (10.1%)	<0.001	2.86 (1.75–4.68)
Necrotising enterocolitis	9 (13.6%)	39 (5.7%)	0.02	2.31 (1.22–4.38)
Sepsis	11 (16.7%)	43 (6.3%)	0.005	2.58 (1.44–4.63)
Significant cardiac disease²	20 (30.3%)	13 (1.9%)	<0.001	9.93 (6.37–13.89)
Seizures	7 (10.6%)	7 (1%)	<0.001	6.24 (3.50–11.12)
Use of nephrotoxic drugs³	58 (87.9%)	299 (43.7%)	<0.001	7.98 (3.87–16.48)
Use of furosemide	3 (4.5%)	1 (0.1%)	0.002	8.88 (4.81–16.40)
Need for inotropes⁴	42 (63.6%)	59 (8.6%)	<0.001	11.25 (7.13–17.74)
Serum creatinine≥0.98 mg/dl⁵	46 (69.7%)	17 (2.5%)	<0.001	25.08 (15.88 –39.62)
Urine output < 1.32 ml/kg/hr⁵	52 (78.8%)	38 (5.6%)	<0.001	27.24 (15.75 –47.09)

¹Maternal variables. ² Significant cardiac disease included persistent pulmonary hypertension of the newborn (PPHN), hemodynamically significant patent ductus arteriosus (PDA), cardiogenic shock and other congenital cardiac disease. ³Nephrotoxic drugs included vancomycin or colistin or amphotericin B. ⁴Inotrope drugs included dopamine or dobutamine or epinephrine or norepinephrine. ⁵During the first 12 hours post-admission in NICU. AKI: Acute Kidney Injury; RR: Relative Risk; NICU: Neonatal Intensive care unit; CI: Confidence Interval; NA: Not applicable; gm: gram; PPV: Positive Pressure Ventilation; mg: milligram; dl: decilitre; ml: millilitre; kg: kilogram.

3 Results

The study flow chart has been depicted in Fig. 1. Of the included 750 neonates, 8.8% (n = 66) had AKI post-admission in the NICU with the majority having stage 1 AKI [66.7% (n = 44)], followed by stage 2 AKI [18.2% (n = 12)] and stage 3 AKI [15.1% (n = 10)]. The probable causes of AKI were sepsis [60.6% (n = 40)], septic shock [57.6% (n = 38)], dehydration [18.2% (n = 12)], and nephrotoxic drugs [28.8% (n = 19)] (considering overlapping causes in few cases).

Fig. 1

Study flow chart.

For all the included subjects, the proportion of neonates with age at admission in NICU < 25.5 hours was 42.4%; with gestational age < 28 weeks was 4.4%; requiring positive pressure ventilation (PPV) in the delivery room was 18.1%; with sepsis was 7.2%; with significant cardiac disease was 4.4%; with use of nephrotoxic drugs was 47.6%; with use of inotropes was 13.5%; with serum creatinine during first 12 hours post-admission≥0.98 mg/dl was 8.4%; and with urine output < 1.32 ml/kg/hr during first 12 hours post-admission in NICU was 12%. The data for other variables are provided in Supplementary Table 1. The median (IQR) hospital length of stay was 7 (5–11) days. A total of 2.8% (n = 21) neonates died during the hospital stay.

Table 2

STARZ scoring model

Variables		Assigned Score
Age at entry in NICU (hours)	<25.5	6
	≥25.5	0
PPV in the delivery room	Yes	7
	No	0
Gestational age (weeks)	<28	7
	≥28	0
Sepsis	Yes	6
	No	0
Significant cardiac disease¹	Yes	10
	No	0
Urine output (ml/kg/hr)²	<1.32	7
	≥1.32	0
Serum creatinine (mg/dl)²	≥0.98	20
	<0.98	0
Use of nephrotoxic drugs³	Yes	11
	No	0
Use of furosemide	Yes	9
	No	0
Use of inotropes⁴	Yes	17
	No	0

¹Significant cardiac disease included persistent pulmonary hypertension of the newborn (PPHN), hemodynamically significant patent ductus arteriosus (PDA), cardiogenic shock and other congenital cardiac disease. ²First 12 hours post admission in NICU. ³Nephrotoxic drugs included vancomycin or colistin or amphotericin B. ⁴Inotrope drugs included dopamine or dobutamine or epinephrine or norepinephrine. NICU: Neonatal Intensive Care Unit; PPV: Positive pressure ventilation; hr; hour; ml: millilitre; kg: kilogram; mg: milligram; dl: decilitre.

3.1 Univariate analysis

The variables that were associated with significantly higher risk of AKI were maternal infections during pregnancy [unadjusted RR (95% CI): 10.69 (7.27–15.72)]; maternal diabetes during pregnancy [14.14 (9.92–20.16)]; maternal hypertension [10.12 (6.86 –14.95)]; pre-eclampsia [7.18 (4.68 –11.00)]; oligohydramnios [3.56 (1.88 –6.72)]; outborn site of delivery [2.25 (1.42–3.55)]; gestational age < 28 weeks [2.17 (1.01 –4.66)]; birth weight < 1,000 gm [2.73 (1.49 –5.00)]; Apgar score < 9 at 5 minutes [5.92 (3.01 –11.66)]; requirement of positive pressure ventilation (PPV) in the delivery room [8.44 (5.27 –13.52]; age at admission in NICU [< 25.5 hours] [1.41 (1.12–1.76)]; necrotising enterocolitis [2.31 (1.22 –4.38)]; sepsis in

Table 3
The predictive validity of the STARZ scoring model based on the standard statistical measures

Predicted based on STARZ scoring model	Observed based on the data
	AKI	Non - AKI	Total
AKI (≥31.5 score)	59 (89.4%)	62 (9.1%)	121 (16.1%)
Non-AKI (<31.5 score)	7 (10.6%)	622 (90.9%)	629 (83.9%)
Total	66 (100%)	684 (100%)	750 (100%)

AKI: Acute Kidney Injury.

NICU [2.58 (1.44 –4.63)]; significant heart disease [9.93 (6.37 –13.89)]; seizures [6.24 (3.50 –11.12)]; use of nephrotoxic drugs [7.98 (3.87 –16.48)]; use of furosemide [8.88 (4.81 –16.40)]; need for inotropes [11.25 (7.13 –17.74)]; serum creatinine during first 12 hours post-admission in NICU≥0.98 mg/dl [25.08 (15.88–39.62)]; or urine output < 1.32 ml/kg/hr during first 12 hours post-admission in NICU [27.24 (15.75–47.09)] [Table 1].

3.2 Scoring model validation

The STARZ scoring model [Table 2] included 10 variables; with total score ranging from 0 to 100, and cut-off score of 31.5 indicating higher risk of AKI, as based on derivation cohort [7]. Each of the 750 neonates was assigned the score based on the data of these variables. The mean, median and range score among all the 750 neonates were 17.5, 13 and 0–84, respectively. The probability of AKI was < 20% up to score 36, 20 -<40% for score 37–50, 40% –<60% for score 51–60, 60 –<80% for score 61–67, and≥80% for score≥68. This scoring model was observed to have a sensitivity of 89.4% [59/66], specificity of 90.9% [622/684], positive predictive value of 48.8% [59/121], negative predictive value of 98.9% [622/629] and accuracy of 90.8% [681/750] [Table 3]. Based on the STARZ cut-off score≥31.5, an area under the ROC curve was observed to be 0.958 (0.934–0.981), p < 0.001; indicating high discriminative power [Fig. 2].

Fig. 2

Receiver operating characteristic (ROC) curve for the validation cohort.

3.3 Mortality and Length of stay

The median duration of hospital length of stay was significantly higher among neonates with AKI as compared to those without AKI [10.5 (7 –10) days; p < 0.001]. Mortality risk was 6.4 times higher among those with AKI as compared to those without AKI [8 (12.1%) vs. 13 (1.9%); p < 0.001; RR (95% CI): 6.38 (2.74 –14.83)].

4 Discussion

Development of AKI can worsen a child’s prognosis when receiving intensive care. The neonatal population is at heightened risk of AKI and its adverse consequences, owing to peculiarities arising from gestational issues, hypoxic events at birth, congenital abnormalities and their long term effects on nephrogenesis and maturation [1 –3]. AKI has been established as an independent predictor of mortality among neonates across all gestational ages and clinical spectrum [1, 2]. AKI in neonates also increases length of hospitalisation [1, 9]. As a significant proportion of neonatal deaths occur early after birth, establishing illness severity should be done expeditiously to allow for therapeutic interventions [10]. Risk stratification for AKI can help identify vulnerable neonates and allow for prognostication and overall management to improve outcomes.

Various biomarkers which may help in predicting neonatal AKI have been studied, with promising results [6]. However, biomarkers are still not in routine clinical practise due to non-standardisation, unfamiliarity and limited availability especially in resource limited settings, making their use unfeasible in low and middle income countries (LMICs).

Basu and colleagues derived and validated the renal angina index (RAI) in a multi cohort of children (>28 days old) [11]. The index included assessment of creatinine, fluid balance, illness severity in form of need of intensive care, ventilatory and inotropic support. An admitting score of > 8 helped predict AKI by day 3, outperforming the KDIGO AKI criteria. A score less than 8 was found to have high negative predictive values (92–99%) [11]. Thus, RAI has become an extremely useful tool for AKI prediction, and it has been applied to various population subsets including adults. However, it has not been exclusively validated in the neonatal population. Though a few studies did include neonates among study subjects, data pertaining to RAI and its accuracy in neonates, has not been studied [12]. A major drawback in RAI which could limit its use in neonates is the exclusion of events at birth and gestational age which are significant AKI risk factors in this population [1].

The STARZ score was derived from our previous multicentre, prospective cohort study conducted in level 2–4 NICUs across India. Accounting for the distinctive events of parturition, heart disease, clinical presentation as well as therapeutic measures taken during intensive care of the neonate, the STARZ score was developed, which provides for a collective AKI risk assessment in neonates [7]. In a similar validation study for the STARZ sore done in multiple NICUs across India, involving 744 neonates, the score showed great utility in AKI risk prediction with sensitivity of 82.1%, specificity 91.7%, positive predictive value 81.2%, negative predictive value 92.2% and accuracy 88.8% [11].

The variables incorporated into the score include gestational age at entry into NICU. The AWAKEN study found AKI to be highest in neonates with lowest gestational age (22 weeks to < 29 weeks) [1]. Similarly, the trends from the US national database for acute kidney injury in premature infants showed majority of premature babies with AKI were < 1000 gms (74.3%) and≤28 weeks gestation (63.9%). AKI was significantly associated with higher mortality and significant increase in length of stay. AKI was associated with comorbidities such as necrotizing enterocolitis, patent ductus arteriosus, bronchopulmonary dysplasia, intraventricular hemorrhage, and septicaemia [13].

Many other studies have reported similar findings [14, 15]. Premature babies may have a smaller complement of nephrons, reducing renal reserve and making them at risk of AKI when exposed to stressors [16]. Kidney injury caused by hypoxic, ischemic insults as evidenced by need for positive pressure ventilation in delivery room also increase AKI risk [2]. Use of nephrotoxic drugs are important AKI risk factors. In this population, Non-steroidal anti-inflammatory drugs (NSAIDs) are often prescribed for patent ductus arteriosus. Antibiotics such as aminoglycosides, Amphotericin B and others are often used in neonatal sepsis [2]. Similarly, the development of sepsis in the newborn also increases AKI risk [16, 17]. A significant cardiac disease in a newborn can increase AKI risk by altering hemodynamics, hypoxia and hampering fluid balance [18]. The need for inotropes or furosemide in first 12 hours of NICU stay or need for intensive care within first 26 hours of birth, betray the severity of illness and the higher risk of AKI. The score also includes two parameters of renal injury, serum creatinine and urine output; though serum creatinine in the neonate early after birth may reflect maternal values and urine output maybe altered by use of diuretics [2]. But a serum creatinine > 0.98 mg/dl at 12 hours after NICU admission does predict a high risk for AKI as seen in our derivation cohort [7]. Neonates often develop non oliguric kidney failure [2] and thus a urine output < 1.32 ml/kg/hour was identified in our derivation [8] and validation cohorts with a higher risk of AKI and poor outcomes as opposed to the <1 ml/kg/hour criteria as per the neonatal KDIGO AKI modification [8]. Bezerra and colleagues reached similar conclusions when studying urine output and outcomes in critically ill neonates [19]. Thus, the score utilises easily obtainable clinical and laboratory information and is simple to use.

In the present independent validation study, all the above factors were also found to be associated with a higher risk of AKI on univariate analysis. The neonates who developed AKI also had longer lengths of hospital stay and a 6.4-fold higher risk of mortality. The use of STARZ scoring model showed a sensitivity of 89.4% in detecting AKI with a 90.9% specificity and a high negative predictive value of 98.9%. The area under ROC was 0.958 (0.934–0.981); p < 0.001 showing a high discriminative power. Thus the score helps predict AKI at 7 days and provides an opportunity to intervene and modify harmful drugs, optimise hemodynamics, apply care bundles and also manage AKI and its complications. Managing nutrition, acid base imbalances, fluid balances and providing kidney replacement therapy when required also improve outcomes; as evidenced by the AWAKEN study, the odds of mortality following AKI were higher in centres where monitoring was less frequent [1]. Thus, risk stratification using STARZ score can also allow for greater surveillance to help mitigate the complications of AKI.

Limitations of this study include it being a single centre cohort. Further studies in varied ethnicities and resource settings would allow for a greater understanding of the score and its utility. There is a need to apply the STARZ score in NICU cohort which cater to sicker babies for validation of the score. Our cohort had 33.5% outborn babies and 4.4% babies with gestation < 28 weeks, and 6% babies < 1 kg. Urinary or serum biomarkers were not done in the study. Comparing the score and incorporation with serum or urinary biomarkers would also provide valuable information of its AKI prediction ability. A major strength of the study, is its large cohort and meticulous data collection. In conclusion the STARZ score helps in anticipating AKI in critically ill neonates using simple laboratory and clinical parameters, allowing for interventions to improve outcomes.

Author Contributions

All authors made substantial contributions to conception and design, acquisition of data, analysis and interpretation of data; drafting the article or revising it critically for important intellectual content. All authors gave final approval of the version to be published.

Conflict of interest

The authors declare no conflict of interest.

Statement of financial support

No financial assistance received

Disclosure

None

Footnotes

The supplementary material is available in the electronic version of this article: .

References

Jetton

, Boohaker

, Sethi

, Wazir

, Rohatgi

, Soranno

, et al. Incidence and outcomes of neonatal acute kidney injury (AWAKEN): A multicentre, multinational, observational cohort study. Lancet Child Adolesc Health. 2017;1:184–94.

Andreoli

. Acute renal failure in the newborn. SeminPerinatol. 2004;28:112–23.

Charlton

, Boohaker

, Askenazi

, Brophy

, D’Angio

, Fuloria

, et al. Incidence and risk factors of early onset neonatal AKI. Clin J Am Soc Nephrol. 2019;14:184–95.

Askenazi

, Ambalavanan

, Goldstein

. Acute kidney injury in critically ill newborns: What do we know? What do we need to learn? Pediatr Nephrol. (2009;24:265–74.

Mehta

, Cerdá

, Burdmann

, Tonelli

, García-García

, Jha

, et al. International Society of Nephrology’s 0 by 25 initiative for acute kidney injury (zero preventable deaths by 2025): A human rights case for nephrology. Lancet. 2015;385:2616–43.

Jetton

, Askenazi

. Update on acute kidney injury in the neonate. Curr Opin Pediatr. 2012;24:19–6.

Wazir

, Sethi

, Agarwal

, Tibrewal

, Dhir

, Bajaj

, et al. Neonatal acute kidney injury risk stratification score: STARZ study. Pediatr Res. 2021. doi: 10.1038/s41390-021-01573-9. Epub ahead of print.

Sethi

, Raina

, Rana

, Agrawal

, Tibrewal

, Baja

, et

. Validation of the STARZ neonatal acute kidney injury risk stratification score. Pediatr Nephrol. 2022.doi: 10.1007/s00467-021-05369-1. Epub ahead of print.

Gadepalli

, Selewski

, Drongowski

, Mychaliska

. Acute kidney injury in congenital diaphragmatic hernia requiring extracorporeal life support: An insidious problem. J Pediatr Surg. 2011;46:630–35.

10.

Collaborators MDS. Causes of neonatal and child mortality in India: A nationally representative mortality survey. The Lancet. 2010;376:1853–60.

11.

Basu

, Zappitelli

, Brunner

, Wang

, Wong

, Chawla

, et al. Derivation and validation of the renal angina index to improve the prediction of acute kidney injury in critically ill children. Kidney Int. 2014;85:659–67.

12.

Sethi

, Raghunathan

, Shah

, Dhaliwal

, Jha

, Kumar

, et al. Fluid overload and renal angina index at admission are associated with worse outcomes in critically ill children. Front Pediatr. 2018;6:118.

13.

Elgendy

, Othman

, Younis

, Puthuraya

, Matar

, Aly

. Trends and racial disparities for acute kidney injury in premature infants: The US national database. Pediatr Nephrol. 2021;36:2789–95.

14.

Carmody

, Swanson

, Rhone

, Charlton

. . Recognition and reporting of AKI in very low birth weight infants. Clin J Am Soc Nephrol. 2014;9:2036–43.

15.

Koralkar

, Ambalavanan

, Levitan

, McGwin

, Goldstein

, Askenazi

. Acute kidney injury reduces survival in very low birth weight infants. Pediatr Res. 2011;69:354–8.

16.

Perico

, Askenazi

, Cortinovis

, Remuzzi

. Maternal and environmental risk factors for neonatal AKI and its long-term consequences. Nat Rev Nephrol. 2018;14:688–703.

17.

Momtaz

, Sabzehei

, Rasuli

, Torabian

. The main etiologies of acute kidney injury in the newborns hospitalized in the neonatal intensive care unit. J ClinNeonatol. 2014;3:99–102.

18.

Zheng

, Yao

, Han

, Xiao

. Renal function and injury in infants and young children with congenital heart disease. Pediatr Nephrol. 2013;28:99–104.

19.

Bezerra

, Vaz Cunha

, Libório

. Defining reduced urine output in neonatal ICU: Importance for mortality and acute kidney injury classification. Nephrol Dial Transplant. 2013;28:901–9.

Validation of the STARZ neonatal acute kidney injury risk stratification score in an independent prospective cohort

Abstract

OBJECTIVE:

STUDY DESIGN:

RESULTS:

CONCLUSIONS:

Keywords

Abbreviations

1 Background

2 Materials and methods

2.1 Study design

2.2 Inclusion criteria (similar to the original study)

2.3 Exclusion criteria (similar to the original study) [7]

2.4 Data collection

2.5 Daily assessment

2.6 Statistical analysis

Table 3 The predictive validity of the STARZ scoring model based on the standard statistical measures

4 Discussion

Author Contributions

Conflict of interest

Statement of financial support

Disclosure

Footnotes

References

Table 3
The predictive validity of the STARZ scoring model based on the standard statistical measures