EXPERIENCE AND REASON |
An Evidence-Based Approach to Predicting Low IQ in Very Preterm Infants From the Neurological Examination: Outcome Data From the Indomethacin Intraventricular Hemorrhage Prevention Trial
* Department of Pediatrics, Division of Pediatric Neurology, Maine Medical Center, Portland, Maine
Department of Pediatrics, Brown University Medical School, Providence, Rhode Island
Departments of Pediatrics, Neurology, Neurosurgery, Epidemiology, and Public Health, Yale University School of Medicine, New Haven, Connecticut
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSSTATISTICAL METHODSRESULTSCOMMENTCONCLUSIONSEMBARRASSMENT OF RICHESREFERENCES |
|---|
We evaluated whether the degree of cerebral palsy (CP) at age 3 in very preterm children is predictive of full-scale intelligence quotient (FSIQ) <70 at age 8 by calculating likelihood ratios (LRs) for findings on the neurologic examination. Data from the follow-up phase of the Indomethacin Intraventricular Hemorrhage Prevention Trial, which includes periodic neurologic examination and neuropsychometric testing, were used. Information was available on 366 of 440 (83%) children with birth weight of 600 to 1250 g who survived. Neurologic examination at age 3 was grouped by presence and type of CP, and the Weschler Intelligence Scale for Children–Third Edition FSIQ at age 8 was grouped dichotomously (<70 or
70). CP was identified in 35 of 366 3-year-olds (9.5%). An FSIQ <70 was identified in 47 of 366 children at 8 years old (12.8%). FSIQ <70 occurred in 14 of 17 children with tri- or quadriplegia (82%), 8 of 18 children with di- or hemiplegia (44%), and 25 of 331 children without CP (7.5%). Useful LRs were calculated for tri- or quadriplegia (30), di- or hemiplegia (5.7), and children without CP (0.55). These LRs have greater impact on posttest odds for FSIQ <70 than those for birth weight <1000 g, history of bronchopulmonary dysplasia, and Stanford-Binet Intelligence Score <70 at age 3. We conclude that the neurologic examination at 3 years old predicts FSIQ <70 at age 8 with LRs that allow evidence-based parental counseling and intervention planning.
Key Words: evidenced-based medicine • premature infant • cerebral palsy • mental retardation
Abbreviations: LR, likelihood ratio • IQ, intelligence quotient • CP, cerebral palsy • BPD, bronchopulmonary dysplasia • SBIS, Stanford-Binet Intelligence Score • WISC-III, Weschler Intelligence Scale for Children–Third Edition • FSIQ, full-scale IQ • VIQ, verbal IQ • PIQ, performance IQ • PPV, positive predictive value • VLBW, very low birth weight
Increasingly, physicians are asked to provide evidence-based estimates of outcomes. We asked whether a useful estimate of the chance of mental retardation in very preterm survivors of neonatal intensive care could be obtained from the neurologic examination. Our objective was to calculate positive likelihood ratios (LRs) for intelligence quotient (IQ) <70 at age 8 based on the neurologic examination at 3 years of age. Follow-up data from the Indomethacin Intraventricular Hemorrhage Prevention Trial include a neurologic examination at ages 36, 54, 72, and 96 months as well as detailed psychometric testing.1,2 The antecedents of cerebral palsy (CP) in this cohort have been reported previously and include birth weight, bronchopulmonary dysplasia (BPD), and ventriculomegaly on cranial ultrasound scanning at 40 weeks postconceptional age.3 To calculate LRs, the neurologic examination results were stratified based on classical descriptions of CP. The categories of di- and hemiplegia and tri- and quadriplegia reflect a gradation of central nervous system involvement. LRs for other neonatal and postnatal factors associated with adverse outcomes were calculated to estimate the effectiveness of the neurologic examination in predicting low IQ compared with these other factors.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSSTATISTICAL METHODSRESULTSCOMMENTCONCLUSIONSEMBARRASSMENT OF RICHESREFERENCES |
|---|
From September 5, 1989 through August 31, 1992, preterm infants with birth weights of 600 to 1250 g were enrolled by 6 hours of age at each of the 3 participating institutions.1,2 The initial phase of the study examined whether low-dose indomethacin given intravenously would prevent the development of new intraventricular hemorrhage or extension of preexisting intraventricular hemorrhage. Serial cranial ultrasound examinations were performed, starting with an initial ultrasound between 5 and 11 hours of age. Subsequent cranial ultrasounds were performed on postnatal days 1, 2, 4, 5, 7, 14, and 21. A final cranial ultrasound was performed at 40 weeks postconceptional age. At each ultrasound examination, the presence and grade of intraventricular hemorrhage were noted, as well as the presence of cystic periventricular leukomalacia and ventriculomegaly. In addition, multiple demographic and clinical parameters were recorded for each patient during the neonatal period.
After the newborn phase of the trial, this cohort of patients was assessed periodically. Pertinent to the present study, patients received comprehensive physical and neurologic examinations at 3 years of age to evaluate for the presence of CP. Authors (W.C.A., L.R.M., and B.R.V.) at each institution made the diagnosis of CP based on the standard definitions of hypertonicity, hyperreflexia, and spastic or dystonic movement of the affected limbs. The patients with CP were placed in the standard categories of spastic diplegia, hemiplegia, triplegia, and quadriplegia based on the number of limbs affected. Also at 3 years of age, each patient had psychometric testing using the Stanford-Binet Intelligence Score (SBIS). At 8 years of age, the cohort was tested by using a group of neuropsychometric instruments including the Weschler Intelligence Scale for Children–Third Edition (WISC-III).4 The test was administered and scored in the usual manner. The WISC-III test yielded a full-scale IQ (FSIQ) score for each child that we report here as dichotomous results: either <70 or 70. The WISC-III component scores for verbal IQ (VIQ) and performance IQ (PIQ) are reported here in the same manner.
|
STATISTICAL METHODS |
|---|
|
TOPABSTRACTMETHODSSTATISTICAL METHODSRESULTSCOMMENTCONCLUSIONSEMBARRASSMENT OF RICHESREFERENCES |
|---|
The WISC-III FSIQ dichotomous results at age 8 were compared with the degree of CP diagnosed at age 3. Because the data were not distributed evenly, the Mantel-Haenszel
2 exact test for trend was used to evaluate statistical significance of the data. Similar analysis was performed for the WISC-III VIQ and PIQ compared with degree of CP at age 3. |
RESULTS |
|---|
|
TOPABSTRACTMETHODSSTATISTICAL METHODSRESULTSCOMMENTCONCLUSIONSEMBARRASSMENT OF RICHESREFERENCES |
|---|
The WISC-III FSIQ scores for the 366 of 440 (83%) of surviving patients tested at 8 years old have been published previously along with details of other neuropsychometric tests.5 Results were dichotomized into 2 categories (FSIQ <70 or
70) and are presented in Table 1. Forty-seven children tested had FSIQ scores that were <70 (12.8%). Of the patients with tri- or quadriplegia, 82% had FSIQ <70; of patients with di- or hemiplegia, 44% had FSIQ <70; and finally, of the 331 patients without CP, 7.5% had FSIQ <70. This trend was statistically significant (P < .0001). The WISC-III subscores for VIQ and PIQ were similarly dichotomized, and these data are shown in Tables 2 and 3. The presence of CP is more likely to produce PIQ <70 than VIQ <70 (94% compared with 65% with tri- or quadriplegia and 50% compared with 22% for di- or hemiplegia, respectively). As with FSIQ, VIQ and PIQ <70 were statistically associated with the degree of CP.
|
|
|
The utility of the neurologic examination at age 3 as a test to predict FSIQ <70 at age 8 is shown in Table 4. The data from Table 1 are used to determine the detection rate (sensitivity), false-positive rate (1-specificity), and positive LR (detection rate divided by the false-positive rate) for FSIQ <70. For comparison with the neurologic examination results, Table 4 also includes the detection rate, false-positive rate, and LRs for other common neonatal and postnatal factors associated with adverse outcomes. The neonatal factors are birth weight <1000 g, presence of BPD, and ventriculomegaly on cranial ultrasound scan at 40 weeks postconceptional age as defined in previous publications.1,2 Of postnatal factors assessed at 3 years of age in the Indomethacin Intraventricular Hemorrhage Prevention Trial, the SBIS is shown, because it provides an IQ estimate most comparable with the WISC-III FSIQ. The first column in Table 4 shows the percentage of children with the neonatal or postnatal factor who have FSIQ <70 at 8 years old (the positive predictive value [PPV] for the factor) followed by the detection rate, false-positive rate, and LR. As noted in Table 4, the detection rate for FSIQ <70 is 30% for tri- or quadriplegia and 17% for di- or hemiplegia. Children without CP on the neurologic examination at age 3 accounted for 53% of all children with FSIQ <70. The false-positive rate for tri- or quadriplegia as a test for FSIQ <70 is 1% (that is, only 1% of all children with FSIQ
70 had tri- or quadriplegia). The false-positive rate for di- or hemiplegia is 3%. Of the children without CP, FSIQ 70 was found in 96%. The positive LRs calculated from the data for the neurologic examination at age 3 ranged from 30 for tri- or quadriplegia to 0.55 for no CP. Of the neonatal factors, only ventriculomegaly had an LR comparable to the neurologic examination and had a PPV of 56%. An SBIS IQ <70 at 3 years old had a PPV of 39.5% for FSIQ <70 at 8 years and had an LR less than that for di- or hemiplegia (4.8 vs 5.7).
|
|
COMMENT |
|---|
|
TOPABSTRACTMETHODSSTATISTICAL METHODSRESULTSCOMMENTCONCLUSIONSEMBARRASSMENT OF RICHESREFERENCES |
|---|
LRs are a statistical tool used to assess the predictive power of a test. The LR is applied to the pretest odds for an outcome to demonstrate the effect of a positive test result in predicting the outcome of interest. The pretest odds are the odds of finding the condition of interest before any test being performed. LRs >1 increase the pretest odds of finding the outcome of interest, and LRs <1 reduce the pretest odds. An LR of 1 indicates that the test has no effect on the pretest odds.6 LRs are calculated from data found in 2 x 2 tables. Table 1 contains the data necessary to give the values in Table 4. Only the FSIQ is found in Table 4, because it is a global estimate of IQ. However, as shown in Tables 2 and 3, VIQ and PIQ are also statistically associated with the degree of CP on the neurologic examination at age 3. PIQ is more likely to be affected in children with CP, because it depends in part on fine motor skills.7,8 Table 4 shows the LRs for multiple factors thought to be predictive of low IQ available from the Indomethacin Intraventricular Hemorrhage Prevention Trial. Of the factors found in the neonatal history of very preterm infants, only the presence of ventriculomegaly on the 40-week-postconceptional-age ultrasound scan has an LR that would be helpful in predicting FSIQ <70. Birth weight <1000 g and BPD are statistically associated with low IQ. However, they have LRs close to 1 and are not helpful in predicting FSIQ <70 when the clinician is dealing with a very preterm infant. Of the factors evaluated at 3 years old, it is interesting that a neurologic examination finding of CP outperforms an SBIS IQ <70.
The diagnosis of CP can be made reliably by age 3 (and often by age 1) on neurologic examination. However, cognitive deficits may not be identified for many years, because intelligence testing in young children is difficult to obtain, requires testing by experienced psychometricians, and (as demonstrated in Table 4) is not as predictive of FSIQ at age 8 as would be expected. Early detection of patients at increased risk for a major cognitive deficit is important for 2 reasons: instituting early developmental interventions and preparing the parents for the likely cognitive abilities of the child. At age 3, children with CP are likely to be receiving physical therapy but may not be receiving developmental therapy. Having an evidence-based rationale for treatment may be helpful in dealing with third-party payers.
Roth et al9 reported that the neurologic examination of preterm infants at age 1 predicted future cognitive and physical impairment at 14 years old. Nine percent of the patients studied had a "disabling neurologic impairment" at 1 year of age. This finding had a 98% PPV for a disabling neurologic impairment at 14 years old, which suggests that the neurologic examination at age 1 was a reliable predictor of late neurologic impairment. Furthermore, the study showed that, of the patients with an identifiable neurologic deficit at age 1, 63% (19 of 30) required assistance in school at age 14. As shown in Table 1, CP at 3 years old in the Indomethacin Intraventricular Hemorrhage Prevention Trial had a 63% (22 of 35) chance of having FSIQ <70 at 8 years old. This degree of cognitive impairment would require assistance in school.
Previous studies have shown that certain anatomic or structural abnormalities found on imaging studies in preterm infants correlate with future cognitive deficits. Ventriculomegaly detected by cranial ultrasonography is one particularly well-recognized predictor of future cognitive deficits in preterm very low birth weight (VLBW) infants. Whitaker et al10 reported that the presence of ventriculomegaly and grade IV intraventricular hemorrhage in preterm VLBW infants was predictive of poor cognitive performance at age 6. Ment et al11 reported that 56% of preterm VLBW infants with evidence of ventriculomegaly at 40 weeks postconceptional age had FSIQ scores <70 at 4.5 years of age. Similar studies of correlation between imaging abnormalities and outcome have been reported for magnetic resonance imaging scanning.12–14
How would a pediatrician use the LRs in Table 4 to advise parents? Table 5 demonstrates their use in 2 different scenarios based on the pretest odds of FSIQ <70 found on the WISC-III at 8 years old in our study of 1:7 (12.8%). Pretest odds are defined as the odds of finding a particular condition (in this case, FSIQ <70) before any test being performed. Posttest odds are the odds of finding the condition of interest if there is a positive test result (in this case, the odds of finding FSIQ <70 if there is a positive test result; ie, a degree of CP on physical examination at 3 years old). In the first scenario, the odds of FSIQ <70 are calculated based only on the neurologic examination. In the second scenario, the pretest odds have been adjusted for the presence of ventriculomegaly. The LR associated with tri- or quadriplegia of 30 has a large impact on outcome prediction no matter the pretest clinical situation in these infants. Although this information may be sobering and disheartening to many parents, it provides a realistic estimation of the child’s outcome. Di- and hemiplegia affect the pretest odds of low FSIQ less dramatically. However, even in children without knowledge of the cranial ultrasound scan at 40 weeks corrected age (scenario 1), the posttest odds of 1:1 are sufficient to warrant developmental intervention as well as physical therapy. The absence of CP at age 3 can be reassuring to parents, because it reduces the odds to 1:13, a risk of only 7%. Although this is an improvement, it should be contrasted with the general population risk of IQ <70 of 
1:24 (4%) and suggests the need for ongoing surveillance for delays. Finally, although it is somewhat reassuring, a normal examination at age 3 in the second scenario in Table 5 does not reduce the odds enough to suggest that developmental intervention should be abandoned, because ventriculomegaly on the 40-week cranial ultrasound scan is a strong predictor of low IQ alone.11
|
|
CONCLUSIONS |
|---|
|
TOPABSTRACTMETHODSSTATISTICAL METHODSRESULTSCOMMENTCONCLUSIONSEMBARRASSMENT OF RICHESREFERENCES |
|---|
The information obtained from this study reveals that the diagnosis of CP at age 3 not surprisingly correlates with an increased likelihood of having FSIQ <70 at 8 years of age. The abnormal neurologic examination at age 3 is more strongly predictive of FSIQ <70 than intelligence testing at age 3. The LRs calculated from these data can be used to estimate the chance of adverse cognitive outcome in very preterm infants in an evidence-based, semiquantitative way, which should allow the physician to counsel parents and arrange developmental therapy for those patients at increased risk.
|
EMBARRASSMENT OF RICHES |
|---|
|
TOPABSTRACTMETHODSSTATISTICAL METHODSRESULTSCOMMENTCONCLUSIONSEMBARRASSMENT OF RICHESREFERENCES |
|---|
"Walt Kelly’s comic-strip character Pogo observed, ‘We is faced with insurmountable opportunities.’ And the poet Ogden Nash commented, ‘Progress is great, but it has gone on far too long.’"
Carroll J. Change: resistance is futile. Globe and Mail. September 3, 2003
Submitted by Student
|
ACKNOWLEDGMENTS |
|---|
This work was supported by National Institute of Neurologic Disorders and Stroke grants NS 27116 and NS 35476 and National Center of Research Resources grant RR 06022.
|
FOOTNOTES |
|---|
Received for publication May 30, 2003; Accepted Oct 29, 2003.
Address correspondence to Walter C. Allan, MD, Foundation for Blood Research, 69 US Rt One, Scarborough, ME 04074. E-mail: allan{at}fbr.org
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSSTATISTICAL METHODSRESULTSCOMMENTCONCLUSIONSEMBARRASSMENT OF RICHESREFERENCES |
|---|
- Ment LR, Oh W, Ehrenkranz RA, et al. Low-dose indomethacin and prevention of intraventricular hemorrhage: a multicenter randomized trial.
Pediatrics.1994; 93
:543
–550
[Abstract/Free Full Text] - Ment LR, Oh W, Ehrenkranz RA, et al. Low-dose indomethacin therapy and extension of intraventricular hemorrhage: a multicenter randomized trial. J Pediatr.1994; 124 :951 –955[CrossRef][Web of Science][Medline]
- Allan WC, Vohr B, Makuch RW, Katz KH, Ment LR. Antecedents of cerebral palsy in a multicenter trial of indomethacin for intraventricular hemorrhage.
Arch Pediatr Adolesc Med.1997; 151
:580
–585
[Abstract/Free Full Text] - Weschler D. Manual for the Weschler Intelligence Scale for Children. 3rd ed. San Antonio, TX: Psychological Corporation; 1991
- Vohr BR, Allan WC, Westerveld M, et al. School-age outcomes of very low birth weight infants in the indomethacin intraventricular hemorrhage prevention trial. Pediatrics.2003; 111(4) . Available at: www.pediatrics.org/cgi/content/full/111/4/e340
- Sackett DL, Haynes RB, Guyatt GH, Tugwell P. Clinical Epidemiology–A Basic Science for Clinical Medicine. 2nd ed. Boston, Massachusetts: Little, Brown and Company; 1991:69–152
- Goodman R, Yude C. IQ and its predictors in childhood hemiplegia. Dev Med Child Neurol.1996; 38 :881 –890[Web of Science][Medline]
- Stiers P, Vanderkelen R, Vanneste G, Coene S, De Rammelaere M, Vandenbussche E. Visual-perceptual impairment in a random sample of children with cerebral palsy. Dev Med Child Neurol.2002; 44 :370 –382[CrossRef][Web of Science][Medline]
- Roth S, Wyatt J, Baudin J, et al. Neurodevelopmental status at 1 year predicts neuropsychiatric outcome at 14–15 years of age in very preterm infants. Early Hum Dev.2001; 65 :81 –89[CrossRef][Web of Science][Medline]
- Whitaker AH, Feldman JF, Van Rossem R, et al. Neonatal cranial ultrasound abnormalities in low birth weight infants: relation to cognitive outcomes at six years of age.
Pediatrics.1996; 98
:719
–729
[Abstract/Free Full Text] - Ment LR, Vohr B, Allan W, et al. The etiology and outcome of cerebral ventriculomegaly at term in very low birth weight preterm infants.
Pediatrics.1999; 104
:243
–248
[Abstract/Free Full Text] - Aida N, Nishimura G, Hachiya Y, Matsui K, Takeuchi M, Itani Y. MR imaging of perinatal brain damage: comparison of clinical outcome with initial and follow-up MR findings. AJNR Am J Neuroradiol.1998; 19 :1909 –1921[Abstract]
- Asao C, Korogi Y, Kondo Y, Yasunaga T, Takahashi M. Neonatal periventricular-intraventricular hemorrhage. Subacute and chronic MR findings. Acta Radiol.2001; 42 :370 –375[Web of Science][Medline]
- Valkama AM, Paakko EL, Vainionpaa LK, Lanning FP, Ilkko EA, Koivisto ME. Magnetic resonance imaging at term and neuromotor outcome in preterm infants. Acta Paediatr.2000; 89 :348 –355[CrossRef][Web of Science][Medline]
PEDIATRICS (ISSN 1098-4275). ©2004 by the American Academy of Pediatrics
CiteULike 
Connotea 
Del.icio.us 
Digg 
Facebook 
Reddit 
Technorati 
Twitter What's this?
This article has been cited by other articles:
|
|
 |
|
  K. Mikkola, N. Ritari, V. Tommiska, T. Salokorpi, L. Lehtonen, O. Tammela, L. Paakkonen, P. Olsen, M. Korkman, V. Fellman, et al. Neurodevelopmental Outcome at 5 Years of Age of a National Cohort of Extremely Low Birth Weight Infants Who Were Born in 1996-1997 Pediatrics, December 1, 2005; 116(6): 1391 - 1400. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
