ava i l ab l e a t www.sc i enced i rec t . com

www.e l sev i e r. com/ l oca te / ea r l humdev

Early Human Development (2008) 84, 311–317

Comparison between the perinatal risk inventory andthe nursery neurobiological risk score for predictingdevelopment in high-risk newborn infantsPatrizia Zaramella a,⁎, Federica Freato a, Anna Milan a, Davide Grisafi a,Andrea Vianello b, Lino Chiandetti a

a NICU—Department of Pediatrics, University of Padova, 35128 Padova, Italyb Epidemiology and Community Medicine Unit, University of Padova, 35128 Padova, Italy

Received 22 March 2007; received in revised form 25 July 2007; accepted 20 August 2007

Abbreviations: PERI, Perinatal RiskDevelopment Index; PDI, Psychomotor⁎ Corresponding author. Department o

Tel.: +39 049 8213573 05 06, Tel.: +39E-mail address: zaramella@pediatr

0378-3782/$ - see front matter © 200doi:10.1016/j.earlhumdev.2007.08.003

Abstract

Background: The availability of a score for predicting neonatal outcome prior to dischargemay helpus to define the risk of developmental disorders in very low birth weight infants.Aim: To compare Scheiner's Perinatal Risk Inventory (PERI) with Brazy's Neurobiological Risk Score(NBRS) when applied at discharge, in predicting developmental delay at 24 months of age.Study design: To evaluate the predictive power of the two tests, we measured their sensitivityand specificity in predicting outcome (Mental Development Index, MDI, Psychomotor Develop-ment Index, PDI, and Amiel–Tison Neurological Examination) in an observational study.Subjects: 102 very low birth weight infants (BW b1500 g) admitted to our NICU at the PediatricDepartment of Padova University.Results: In the cohort studied, 75.5% of the patients had a normal MDI, while 24.5% showed adelayed performance (8.8% mildly and 15.7% severely so); the PDI was normal in 74.5% patients,whilst 25.5% had a delayed performance (9.8% mildly and 15.7% severely so). According to theAmiel–Tison test, neurological performancewas normal in 66% patients, impairedwithout disabilityin 19% and impaired with disability in 15%. NBRS showed a sensitivity and specificity respectively of0.96 and 0.23 (MDI), 0.96 and 0.24 (PDI), 0.94 and 0.25 (Amiel–Tison test); for PERI were 0.88 and0.54 (MDI), 0.77 and 0.51 (PDI), 0.82 and 0.57 (Amiel–Tison test). The PERI and NBRS can predict theMDI with an AUC N0.8 and the PDI or Amiel–Tison findings with an AUC of 0.7–0.8. No significantdifferences were found between the areas under the ROC curves using the NBRS and the PERI.

KEYWORDSPerinatal risk inventory;Nursery neurobiologicalrisk score;Very low birth weightinfant;Outcome

Inventory; NBRS, Nursery Neurobiological Risk Score; VLBWI, Very Low Birth Weight Infant; MDI, MentalDevelopment Index; BSID-II, Bayley Infant Developmental Scale.f Pediatrics, Neonatal Intensive Care Unit, University of Padua, Via Giustiniani, 3, 35128 Padova— Italy.347 67 82 927(Mobile); fax: +39 049 8213502.ia.unipd.it (P. Zaramella).

7 Elsevier Ireland Ltd. All rights reserved.

Table 1 Neonatal data

Perinatal dataMale sex 46 (45%)Multiple birth 24 (23.5%)Cesarean section 70 (68.6%)Birth weight, g 1016.7

(±259.05)Birth weight b1000 g 50 (49%)Gestational age, wk 29.16 (±7.41)Gestational age ≤28 wk 55 (54%)Small for gestational age 21 (20.6%)

Measures of chronic illnessBPD 39 (38.2%)Oxygen dependence at 36 wks 14 (13.7%)Hospital stay, d 68.6 (±33.7)

Cranial ultrasound resultsNormal 24 (23.5%)Hyperechoic white matter 9 (8.8%)Intraventricular hemorrhage 39 (38.3%)

grade 1 14grade 2 18grade 3 4grade 4 3

White matter lesions (periventricularleukomalacia and porencephalic cysts)

24 (23.5%)

Hydrocephalus 6 (5.9%)

Data expressed as n (%) or mean±SD.

Conclusions: In assessing the prognosis for individual babies, the physician can choose either thePERI or the NBRS to predict PDI, MDI or Amiel–Tison performance.© 2007 Elsevier Ireland Ltd. All rights reserved.

312 P. Zaramella et al.

1. Introduction

Thanks to advances in neonatal intensive care, very low birthweight infants (VLBWI) have better chances of survival [1].Care seems to have reached the limits of viability, althoughsome concerns exist about major neonatal morbidity [2–6]and the neurodevelopmental sphere for extremely low birthweight infants [7–11]. Severity of illness in NICU graduates isbroadlymeasured by BW, GA and Apgar scores, [12] which canpredict outcome, but the strong correlation of their outcomewith brain matter lesions has led neonatologists to search fora global definition of developmental risk. Finer's studies [13]led to the introduction of a scoring system that includes anoverall measure of handicap. In addition to the obstetricitems and infant emergency room scales, [14] the scoresavailable for neonatal mortality and acute morbidity are theCRIB (Clinical Risk Index for Babies) [15] and the SNAP (Scorefor Neonatal Acute Physiology) [16,17]. The NTISS (NeonatalTherapeutic Intervention Scoring System) [18,19] assessesillness severity on the basis of 62 therapeutic measures. Afterseparating therapeutic measures from severity of illness, theNICHD (National Institute of Child Health and HumanDevelopment) proposed a score [20] based on variablescollected before admission to the NICU.

To predict the risk of impaired outcome in pretermnewborns, there are two scales for use in assessments priorto discharge that consider perinatal medical problems, i.e.Scheiner's Perinatal Risk Inventory (PERI), [21] and Brazy'sNursery Neurobiological Risk Score (NBRS) [22]. These scalesare based on the correlation between certain perinatal eventsand outcome, and on the assumption that a perinatal injurywill produce a brain cell lesion and thus an abnormal outcome.The PERI comprises 18 items indicating an increasing severityor duration of an adverse event on an ordinal scale. TheNBRS isbased on 13 items that describe clinical and therapeutic eventsby means of a geometrical progression of scores.

The availability of an objective score for predictingoutcome prior to discharge and in the short term may help usto define the risk of developmental disorders. Neonatologistsmay then have a tool for focusing on prognosis and the child'sconsequent referral to the post-dischargeassessment services.

The main purpose of the study was to compare the PERIwith the NBRS at discharge, evaluating their sensitivity andspecificity in predicting neurodevelopmental variables. TheMDI, PDI and Amiel–Tison neurological examination wereassessed in our cohort of VLBWI followed up at 24 months ofage.

2. Patients and methods

Between January 1994 and December 1996, 291 VLBWI wereadmitted to our NICU. Sixty-five (22.3%) died (neonatalmortality rate 21.6%; postnatal mortality rate 0.68%). Thirty-two of the survivors (all outborn) were transferred to otherhospitals, 1 was excluded because he was diagnosed as

having Down's Syndrome, and 91 were not followed upbecause their parents refused.

A cohort of 102 VLBWI with no major congenital anomaliesfollowed up at the Pediatrics Department of the University ofPadova was thus enrolled for this study. Neonatal details aresummarized in Table 1.

2.1. Cerebral ultrasound

US was performed on days 2–4 of life, then weekly and nearthe time of discharge, and also on the basis of any clinicalneeds, using the HDI 3000 CV operating system (AdvancedTechnology Laboratories, Bothell, WA, USA). Cranial ultra-sound hemorrhagic findings were collected according toPapile et al. [23].

2.2. Scheiner's perinatal risk inventory and Brazy'sneurobiological risk score

Prior to discharge, the PERI and NBRS were applied twice toeach infant by two members of the research team (PZ and FF)with an inter-rater reliability of 0.97. The PERI items arescored from 0 to 3 using an ordinal scale, so the scores forindividual items are summarized. This scale is structured sothat a higher score coincides with a greater risk of subsequent

313Comparing PERI and NBRS to predict developmental outcome in newborn neonates

developmental abnormality. The NBRS has items scored as 0 ifabsent or graded from 1 to 4 according to a geometricalprogression to indicate the severity or duration of an event.The PERI items are: Apgar score, gestational age andappropriateness of the infant's weight, infections, seizures,head growth, EEG, brain scans, ventilation, polycythemia,hypoglycemia, hyperbilirubinemia, and long-term physicalillnesses such as BPD. The NBRS items cover the need forventilation, acidosis, seizures, intraventricular hemorrhage,(hydrocephalus), periventricular leukomalacia and cerebralatrophy, infections and hypoglycemia. Both scores cover thewhole hospital stay and can be established at discharge. Cut-offs set at≥6 (NBRS) and≥10 (PERI) identifymost infants withabnormal outcomes [21,22].

2.3. Neurodevelopmental outcome

A longitudinal follow-up was performed at a mean correctedage of 23.07±7.71 months using the Amiel–Tison neurologi-cal test and the Bayley Infant Developmental Scale (BSID-II).Three classes of neurodevelopmental condition: A = unim-paired, B = impaired without disability (mildly deviant), C =impaired with disability (severely deviant) were definedusing the Amiel–Tison neurological test [24]. The BSID-II wasused recording both the Mental Development Index (MDI) andthe Psychomotor Development Index (PDI) and consideringthem as continuous variables. The MDI measures environ-mental responsiveness and sensory and perceptual abilities,memory, learning, and early language and communicationabilities; the PDI measures both gross and fine motor skills.MDI and PDI scores of ≥85 are considered normal; valuesbetween 70 and 84 indicate a mildly delayed performanceand ≤69 a significantly delayed performance [25].

The neurological and developmental findings were usedto establish two groups, i.e. children were classified simplyas disabled or not. We defined children as disabled if theyhad severe cognitive delay (MDI≤69) and/or cerebral palsyon neurological examination (Amiel–Tison class C andPDI≤69) and/or if they were blind or deaf (requiringbilateral hearing aids).

2.4. Parental education level

The parents' level of education can be used as a marker ofsocioeconomic status. This variable was assessed on the basisof the self-reported highest level of education completed, interms of years of schooling, classified as follows: less thanhigh school (5–11 yr), high school or equivalent (13 yr), andmore than high school (N13 yr).

2.5. Statistical analysis

To evaluate the discriminatory power of the two tests, wemeasured their sensitivity and specificity in predictingoutcome, as well as the positive and negative predictivevalues, using a cut-off of 6 for the NBRS and 10 for the PERI.

The two scores were validated for our cohort of 102children using the receiver operating characteristic (ROC)curve, which indicates the ability of a test to discriminatebetween populations (developmentally delayed vs normal)and is particularly useful in describing the accuracy of a test.

The ROC curve is obtained by plotting the tradeoff betweenspecificity and false positive rate (or 1—specificity) over therange of cut-off points. A test is effective when it has a highsensitivity (true positive rate) and includes few falsepositives, so the AUC (area under the curve) is near 1.0. AnAUC of 1.0 represents a perfect test, while 0.5 represents aworthless test, and a test with an AUC≥0.8 is consideredsufficiently accurate [26]. We also compared the AUCs forthe two scoring systems, calculating the z values (cut-off≥ 1.96, for pb0.05) [27]. Descriptive statistics are givenas means±standard deviations or proportions.

Odds ratios (OR), sensitivity, specificity and predictivevalues were used, as appropriate.

Pearson's product-moment correlation was used to corre-late parametric data (MDI, PDI, parental education level) withNBRS or PERI scores. Spearman's rank correlation was used toassess the relationship between non-parametric data (Amiel–Tison scale) and NBRS or PERI scores. A p of b0.05 was con-sidered significant.

2.6. Ethical approval

The study was approved by the Hospital's Ethics Committeeand informed consent was obtained from parents beforetheir child's participation in the study.

3. Results

The incidence of cerebral palsy in our group was 17/102(16.7%), and consisted of 4 cases of severe tetraparesis, 9 ofspastic diplegia, and 4 of spastic hemiplegia. Motor delay wasalso seen in 4 patients. According to the Amiel–Tison test,neurological performance was normal in 66% patients,impaired without disability in 19% and impaired with disabilityin 15%.

Brainstem auditory evoked potentials were abnormal in 7infants (6.8%), and 4 had hearing loss on clinical evaluation.Visual evoked potentials were abnormal in 7 cases (6.9%),and 1 child was blind. At ophthalmologic examination,strabismus was recorded in 15 (14.7%) and refraction errorsin 24 (23.5%).

The mean (±SD) MDI in the study cohort was 100.87(±19.84), 77 patients (75.5%) had a normal MDI, and theindex revealed a delay in 25 cases (24.5%), which was mild in9 (8.8%) and severe in 16 (15.7%) patients; 24 of these 25babies had a NBRS≥6 and 22 had a PERI≥10.

The mean PDI was 97.64 (±15.62) and was normal in 76patients (74.5%), mildly delayed in 10 (9.8%) and significantlydelayed in 16 (15.7%); of these 26 babies with a mildly orsignificantly delayed performance, 25 had a NBRS≥6 and 20had a PERI≥10.

The level of the parents' education (2 fathers data aremissing) was as follows: 55 (53.9%) mothers and 59 (59%)fathers had studied for less than 11 years (to less than highschool level), 35 (34.3%) mothers and 28 (28%) fathers hadstudied for 13 years (high school equivalent), and 12 (11.8%)mothers and 13 (13%) fathers had been educated for morethan 13 years (beyond high school). Parents had received amean (±SD) 11.01 (±3.32) and 10.68 (±3.68) years ofschooling for mothers and fathers, respectively (t-test forindependent samples p=0.45, NS). There was a positive

314 P. Zaramella et al.

correlation between MDI values and the fathers' educationallevel (r=0.22, p=0.026), but no such correlation emergedwith the mothers' schooling (r=0.11, p=0.248).

The correlations between the NBRS and the outcomeindexes (MDI, PDI and Amiel–Tison scale) were respectively—0.41 (p=0.00001),— 0.25 (p=0.01) and 0.39 (p=0.0003). Thesame indexes showed a correlation of — 0.46 (p=0.000001),—0.26 (p=0.007) and 0.41 (p=0.00001) with the PERI values.

The sensitivity and specificity, and the positive andnegative predictive values of the NBRS cut-off ≥6 and PERIcut-off ≥10 are shown in Table 2.

When the NBRS was used in MDI, PDI and Amiel–Tisonprediction, the AUC values were 0.839, 0.7506 and 0.7798.Using the PERI, the AUC values were 0.8512, 0.7293, 0.7932,respectively.

To better compare the accuracy of the NBRS and PERIscores, we measured the difference in the areas under theROC curves drawn by calculating the sensitivity and 1—specificity in predicting MDI (z=0.074, NS), PDI (z=0.108, NS)and Amiel–Tison score (z=0.062, NS) (Fig. 1). No significantdifferences emerged between the areas under the ROCcurves obtained using the NBRS and the PERI.

4. Discussion

Outcome predictions based on birth weight and gestationalage may be less accurate in very premature babies due totheir poor general conditions and related clinical course.Mortality is still the main outcome considered in models usedat neonatal units to predict the prognosis in VLBWI [1]. Littlework has been done, on the other hand, to compare thepredictive efficacy of different perinatal scoring systems(NBRS and PERI). The latter approach is based on theassumption that a biomedical event would predict outcome,i.e. the greater the number and severity of the perinatal

Table 2 Relative risk (RR), sensitivity, specificity, positivepredictive value (PPV+) and negative predictive value (NPV−)in relation to cut-offs for NBRS≥6 and PERI≥10

NBRS (≥6) PERI (≥10)

MDIRR 5.5 5.8Sensitivity 0.96 0.88Specificity 0.23 0.54PPV+ 0.29 0.38NPV− 0.95 0.93

PDIRR 5.7 2.6Sensitivity 0.96 0.77Specificity 0.24 0.51PPV+ 0.30 0.35NPV− 0.95 0.86

Amiel–TisonRR 3.7 3.7Sensitivity 0.94 0.82Specificity 0.25 0.57PPV+ 0.38 0.49NPV− 0.89 0.86

events, the greater the likelihood of the infant revealingdevelopmental sequelae.

Among the perinatal events, neonatal cerebral ultrasoundabnormalities have been shown to predict neurodevelop-mental outcome, [28] but a sizeable number of impairedpreterm children had normal cerebral ultrasound [29]findings or low risk [30,31] IVH, grades I–II [32].

Neonatal severity of illness measures have severalimportant uses. They enable a comparison of outcome riskacross NICUs. Perinatal severity scores may be biased asconcerns mortality because most of them require thecollection of a broad range of physiological measurementsthat may not be routinely collected at an NICU. Wepreviously found GA a better measure of neurodevelopmen-tal outcome than the CRIB in a population of extremely lowbirth weight infants [33] and surmised that this was due tothe CRIB items attributing little weight to GA.

Our main purpose here was therefore to compareScheiner's Perinatal Risk Inventory with Brazy's Neurobiolo-gical Risk Score when applied at discharge. Both scores seemuseful in revealing the illness severity factors associated withoutcome and disability. The main differences between thetwo lie in the latter including apnea with bradycardia, andthus also blood pH and hypotension, and patent ductusarteriosus items, whereas only the PERI considers dysmorphicfeatures, head growth and polycythemia. Many studies haveshown that long-term outcome, especially in terms ofcognitive and language measures, can be related to socialfactors [34,35] and it has been demonstrated that themother's level of education has a relevant effect on languageand cognitive development in preterm babies at 3.5 years old[36]. Low socioeconomic status and low level of schooling inthe mother reportedly correlate with the mental outcome inpreterm and term babies [37,38]. Given the correlationalnature of our study, it may be that other variables (such as amother encouraging the father to become more or lessinvolved, or the father's perception of self-efficacy) mighthelp to explain the discrepancy we found, with regards to thefather and no mother relation to the child MDI, vis-à-vis thedata in the literature [39].

The two scores reveal no differences in terms of thevarious items for predicting neurodevelopmental outcome.This is presumably due to the balanced weighting of thespecific items in a global assessment method. Many variablesshould be more predictive of Amiel–Tison neurobehavioralperformance, and consequently of cognitive development orpsychomotor development.

One of the weaknesses of this study includes the lowenrollment level of our group (less than 50% of the sample),casting doubts on our ability to draw general conclusionsbased on this group. Our previous data obtained in a group ofextremely preterm babies admitted during the same periodand enrolled in a larger proportion (89%) nonethelessdemonstrated a severe disability rate of 19% as opposed to15.7% in the present cohort, in higher GA and BW infants. Sowe can assume that our group suffered from little or no bias,as though it was no representative of the extremely low birthweight population.

In conclusion, we compared two methods for scoringdevelopmental risk at discharge to seek the best way topredict developmental outcome in high-risk neonates. ThePERI and NBRS are both useful in predicting outcome and,

Figure 1 ROC curves for NBRS and PERI scores in predicting MDI (A), PDI (B), and Amiel–Tison values (C).

315Comparing PERI and NBRS to predict developmental outcome in newborn neonates

judging from the AUCs, they measure cognitive develop-ment more accurately (AUCN0.8) than psychomotor devel-opment or Amiel–Tison (AUC 0.7–0.8) outcome. Thesescoring systems could be used to provide useful prognostic

insight and facilitate the planning of measures for contain-ing major sequelae after discharge. The function of suchscores is to select patients that do not need to be followed(with a sensitivity of 0.96%), targeting follow-up programs

316 P. Zaramella et al.

to the children at greatest risk. Using this screeningmethods would enable us to reduce the number of infantsfollowed up unnecessarily, instead physicians should usethem as tools for an early identification of children suitablefor monitoring, by excluding false negative patients. At thesame time, they can be used to calculate and measureglobal NICU outcome in high-risk infants and may thusprovide a measure of the quality of neonatal intensive careservices. Furthermore, as VLBW children may exhibit mildto moderate disabilities in later life and at school, [40]these disabilities are more difficult to detect early. Sofurther studies are warranted to provide targeted follow-upprograms, as well as to design appropriate developmentalcare and services.

References

[1] Gould JB, Benitz WE, Liu H. Mortality and time to death in verylow birth weight infants: California, 1987 and 1993. Pediatrics2000;105:37.

[2] Himmelmann K, Hagberg G, Beckung E, Hagberg B, Uvebrant P.The changing panorama of cerebral palsy in Sweden. IX.Prevalence and origin in the birth-year period 1995–1998.Acta Paediatr 2005;94:287–94.

[3] Mikkola K, Ritari N, Tommiska V, Salokorpi T, Lehtonen L,Tammela O, et al. Neurodevelopmental outcome at 5 years ofage of a national cohort of extremely low birth weight infantswho were born in 1996–1997. Pediatrics 2005;116:1391–400.

[4] Hack M, Fanaroff AA. Outcomes of children of extremely lowbirthweight and gestational age in the 1990′s. Early Hum Dev1999;53:193–218.

[5] Fanaroff AA, Stoll BJ, Wright LL, Carlo WA, Ehrenkranz RA,Stark AR, et al. NICHD Neonatal Research Network. Trends inneonatal morbidity and mortality for very low birthweightinfants. Am J Obstet Gynecol 2007;196:147.e1–8.

[6] Lemons JA, Bauer CR, Oh W, Korones SB, Papile L-A, Stoll BJ,et al. Very low birth weight outcomes of the National Instituteof Child Health and Human Development Neonatal ResearchNetwork, January 1995 through December 1996. Pediatrics2001;107:1–8.

[7] Neubauer AP, Voss W, Kattner E. Outcome of extremely lowbirth weight survivors at school age: the influence of perinatalparameters on neurodevelopment. Eur J Pediatr 2007 [Electro-nic publication ahead of print].

[8] Hintz SR, Kendrick DE, Vohr BR, Poole WK, Higgins RD. Changesin neurodevelopmental outcomes at 18 to 22 months' correctedage among infants of less than 25 weeks' gestational age born in1993–1999. Pediatrics 2005;115:1645–51.

[9] Hack M, Wilson-Costello D, Friedman H, Taylor GH, SchluchterM, Fanaroff AA. Neurodevelopment and predictors of outcomesof children with birth weights of less than 1000 g: 1992–1995.Arch Pediatr Adolesc Med 2000;154:725–31.

[10] Vohr BR,Wright LL, DusickAM,Mele L, Verter J, Steichen JJ, et al.Neurodevelopmental and functional outcomes of extremely lowbirth weight infants in the National Institute of Child Health andHuman Development Neonatal Research Network, 1993–1994.Pediatrics 2000;105:1216–26.

[11] Vohr BR, Wright LL, Poole WK, McDonald SA. Neurodevelop-mental outcomes of extremely low birth weight infantsb32 weeks' gestation between 1993 and 1998. Pediatrics2005;116:635–43.

[12] Pollack MM, Koch MA, Bartel DA, Rapoport I, Dhanireddy R, El-Mohandes AAE, et al. District of Columbia Neonatal Network. Acomparison of neonatal mortality risk prediction models in verylow birth weight infants. Pediatrics 2000;105:1051–7.

[13] Finer NN, Robertson CM, Richards RT, Pinnell LE, Peters KL.Hypoxic–ischemic encephalopathy in term neonates: perinatalfactors and outcome. J Pediatr 1981;98:112–7.

[14] Richardson DK, Tarnow-Mordi WO. Measuring illness severity innewborn intensive care. J Intensive Care Med 1994;9:20–33.

[15] Cockburn F, Cooke RWI, Gamsu HR, Greenough A, Hopkins A,McIntosh N, et al. The CRIB (clinical risk index for babies) score:a tool for assessing initial neonatal risk and comparing per-formance of neonatal intensive care units. Lancet 1993;342:193–8.

[16] Richardson DK, Gray JE, McCormick MC, Workman K, GoldmannDA. Score for Neonatal Acute Physiology: a physiologic severityindex for neonatal intensive care. Pediatrics 1993;91:617–23.

[17] Zupancic JA, Richardson DK, Horbar JD, Carpenter JH, Lee SK,Escobar GJ. Vermont Oxford Network SNAP Pilot ProjectParticipants. Revalidation of the Score for Neonatal AcutePhysiology in the Vermont Oxford Network. Pediatrics2007;119:156–63.

[18] Gray JE, Richardson DK, McCormick MC, Workman-Daniels K,Goldman DA. Neonatal therapeutic intervention scoring system(NTISS): a therapy-based severity of illness assessment tool.Pediatrics 1992;90:561–7.

[19] Lee SK, McMillan DD, Ohlsson A, Pendray M, Synnes A, Whyte R,et al. Variations in practice and outcomes in the Canadian NICUnetwork: 1996–1997. Pediatrics 2000;106:1070–9.

[20] Horbar JD, Onstad L, Wright E. Predicting mortality risk forinfants weighing 501 to 1500 grams at birth: a NationalInstitutes of Health Neonatal Research Network report. CritCare Med 1993;21:12–8.

[21] Scheiner AP, Sexton ME. Prediction of developmental outcomeusing a Perinatal Risk Inventory. Pediatrics 1991;88:1135–43.

[22] Brazy JE, Eckerman CO, Oehler JM, Goldstein RF, O'Rand AM.Nursery Neurobiologic Risk Score: important factors in predict-ing outcome in very low birth weight infants. J Pediatr1991;118:783–92.

[23] Papile LA, Burstein J, Burstein R, Koffler H. Incidence andevolution of subependymal and intraventricular hemorrhage: astudy of infants with birth weights less than 1,500 gm. J Pediatr1978;92:529–34.

[24] Amiel-Tison C, Stewart A. Follow up studies during the first fiveyears of life: a pervasive assessment of neurological function.Arch Dis Child 1989;64:496–502.

[25] Bayley N. Manual for the Bayley Scales of Infant Development.2nd ed. San Antonio, Tex: Psychological Corporation; 1993.

[26] Swets JA. Measuring the accuracy of diagnostic systems.Science 1988;240:1285–93.

[27] Hanley JA, McNeil BJ. A method of comparing the areas underreceiver operating characteristic curves derived from the samecases. Radiology 1983;48:839–43.

[28] Pinto-Martin JA, Riolo S, Cnaan A, Holzman C, Susser MW,Paneth N. Cranial ultrasound prediction of disabling andnondisabling cerebral palsy at age two in a low birth weightpopulation. Pediatrics 1995;95:249–54.

[29] Laptook AR, O'Shea TM, Shankaran S, Bhaskar B. NICHDNeonatal Network. Adverse neurodevelopmental outcomesamong extremely low birth weight infants with a normal headultrasound: prevalence and antecedents. Pediatrics 2005;115:673–80.

[30] Rennie JM. Neonatal cerebral ultrasound. Cambridge: Cam-bridge University Press; 1997.

[31] Volpe JJ. Neurology of the newborn. 3rd ed. Philadelphia: W.B.Saunders; 1995.

[32] Patra K, Wilson-Costello D, Taylor HG, Mercuri-Minich N, HackM. Grades I–II intraventricular hemorrhage in extremely lowbirth weight infants: effects on neurodevelopment. J Pediatr2006;149:169–73.

[33] Lago P, Freato F, Bettiol T, Ciandetti L, Vianello A, Zaramella P.Is the CRIB score (Clinical Risk Index for Babies) a valid tool in

317Comparing PERI and NBRS to predict developmental outcome in newborn neonates

predicting neurodevelopmental outcome in extremely lowbirth weight infants? Biol Neonate 1999;76:220–7.

[34] Molfese VJ, Holcomb L, Helwig S. Biomedical and social–environmental influences on cognitive and verbal abilitiesin children 1 to 3 years of age. Int J Behav Dev 1994;17:271–87.

[35] Resnick MB, Gueorguieva RV, Carter RL, Ariet M, Sun Y, Roth J,et al. The impact of low birth weight, perinatal conditions, andsociodemographic factors on educational outcome in kinder-garten. Pediatrics 1999;104:74.

[36] Sansavini A, Guarini A, Alessandroni R, Faldella G, Giovanelli G,Salvioli G. Are early grammatical and phonological workingmemory abilities affected by preterm birth? J Commun Disord2007;40:239–56.

[37] Duncan GJ, Brooks-Gunn J, Klebanov PK. Economic deprivationand early childhood development. Child Dev 1994;65:296–318.

[38] Jefferis BJ, Power C, Hertzman C. Birth weight, childhoodsocioeconomic environment, and cognitive development in the1958 British birth cohort study. BMJ 2002;325:305.

[39] Wood NS, Costeloe K, Gibson AT, Hennessy EM, Marlow N,Wilkinson AR, et al. The EPICure study: associations andantecedents of neurological and developmental disability at30 months of age following extremely preterm birth. Arch DisChild Fetal Neonatal Ed 2005;90:134–40.

[40] Lindeke LL, Mills MM, Georgieff MK, Tanner ME, Wrbsky PM.Health, growth, and use of community services in NICUgraduates at early school age. J Pediatr Health Care1998;12:299–304.