 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
* Department of Pediatrics, University of Cincinnati College of Medicine and Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio
Department of Pediatrics, Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio
Research Triangle Institute, Research Triangle Park, North Carolina
|| University of Alabama–Birmingham, at Birmingham, Birmingham, Alabama
¶ Women and Infants Hospital, Providence, Rhode Island
# Child Policy Research Center, Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONCONCLUSIONAPPENDIXREFERENCES |
|---|
Methods. Data from a cohort of 1483 ELBW infant survivors who were born January 1993 through December 1994 and cared for at centers in the Neonatal Research Network of the National Institute of Child Health and Human Development were examined retrospectively. Children who were compliant with an 18- to 22-month follow-up visit, who visited but were not measured, or who made no visit were compared regarding 4 outcomes: 1) Bayley Scales of Infant Development, 2nd edition, Mental Developmental Index (MDI) <70 and 2) Psychomotor Developmental Index (PDI) <70, 3) presence or absence of cerebral palsy, and 4) weight <10th percentile for age. Logistic regression models were used to predict likelihood of these outcomes for children with no follow-up evaluation, and predicted probability distributions were compared across the groups.
Results. Compared with children who were lost to follow-up, those who were compliant with follow-up were more likely to have been 1 of a multiple birth, to have received postnatal glucocorticoids, and to have had chronic lung disease. These factors were significantly associated with MDI and PDI <70 in the compliant group. Chronic lung disease was associated with increased risk of cerebral palsy (CP). MDI and PDI scores <70 were found in 37% and 29% of children who were evaluated at follow-up, respectively. Prediction models revealed that 34% and 26% of infants in the no-visit group would have had MDI and PDI scores <70. Compliant children tended to have greater incidence of MDI <70 compared with those predicted in the no-visit group but not PDI <70. CP was identified in 17% of the compliant group and predicted for 18% of the no-visit group. Predicted probabilities of having CP were marginally higher among the no-visit infants compared with those who were compliant with follow-up. There were no statistically significant somatic growth differences among the compliant, visit but not measured, and no-visit groups.
Conclusion. ELBW infant survivors who weighed 401 to 1000 g at birth and who are compliant with follow-up evaluations may have worse Bayley Scales of Infant Development, 2nd edition, MDI scores than infants with no visit. Thus, follow-up studies based on infants who are compliant with follow-up care may lead to an overestimation of adverse outcomes in ELBW survivors.
Key Words: neurodevelopmental • outcomes • cerebral palsy • follow-up • extremely low birth weight infants
Abbreviations: ELBW, extremely low birth weight • NICHD, National Institute of Child Health and Human Development • NICU, neonatal intensive care unit • CP, cerebral palsy • BSID-II, Bayley Scales of Infant Development, 2nd edition • MDI, Mental Developmental Index • PDI, Psychomotor Developmental Index • IVH, intraventricular hemorrhage • NEC, necrotizing enterocolitis • CLD, chronic lung disease
Improvements in care of extremely low birth weight (ELBW; <1000 g birth weight) infants have led to increased survival among this very high-risk population. Well-designed long-term follow-up studies of ELBW populations are crucial to obtain reliable neurodevelopmental outcome information for counseling families and possibly to alter perinatal and neonatal care. As many as 10% to 26% of infants are lost during each year of follow-up.1–3 Inability to evaluate all members of a cohort may lead to a bias in population-based estimates of long-term outcomes.
Many factors may be responsible for loss to follow-up. Some investigators espouse that relatively healthy infants are more likely to participate in longitudinal studies because families with disabled infants may be reluctant to acknowledge publicly the disability.1,3 Others speculate that impaired infants are more likely to return for follow-up evaluation, whereas parents of healthier children may not believe that there is a need for sophisticated evaluation. Still others have proposed that lower socioeconomic status, parental education, and ethnicity may be important determinants of compliance with follow-up recommendations.1,2
Conclusions from previous European and Australian cohort studies4–6 may not be applicable to US populations. To evaluate the possibility of a biased estimation of US population-based outcomes related to suboptimal compliance with follow-up, we developed models to predict neurologic, developmental, and growth outcomes of infants who are lost to follow-up. We compared selected prenatal and postnatal characteristics of National Institute of Child Health and Human Development (NICHD) Neonatal Research Network survivors, 401 to 1000 g birth weight, who completed a follow-up evaluation at 18 to 22 months' postmenstrual age (compliant group) with 2 other groups: those who were lost to follow-up (no-visit group) and those who were evaluated incompletely (visit but not measured group).
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONAPPENDIXREFERENCES |
|---|
Outcomes
Evaluation at 18 to 22 months included administration of the Bayley Scales of Infant Development, 2nd edition (BSID-II),9 neurologic examination, and measurement of somatic growth. The BSID-II was administered by certified examiners.8 Scores of <50 on the Mental Developmental Index (MDI) and the Psychomotor Developmental Index (PDI) were coded as 49. Scores on MDI or PDI of <70 were >2 standard deviations below the mean for US infants.
Amiel-Tison neurologic assessments10 were performed by certified, masked developmentalists who had been trained to reliability in a 2-day, hands-on workshop on neurologic assessment. The neurologic assessment included an evaluation of active tone, strength, reflexes, passive tone, and posture. Infants were scored as normal when no abnormalities were observed in the neurologic examination. The workshop also established interexaminer agreement on the definition of CP. CP was defined as a nonprogressive central nervous system disorder characterized by abnormal muscle tone in at least 1 extremity and abnormal control of movement and posture. The neurologic assessment results of all infants who were classified as having CP were reviewed and approved by a central review committee.8 New examiners, added after the workshops, were trained to criteria by each site's certified examiner.
Gender- and age-specific weight percentiles from National Center for Health Statistics11 growth curves were used to classify study infants as above or below the 10th percentile for expected body weight on the basis of their body weight measured during the follow-up visit. A body weight below the 10th percentile was defined as an outcome.
The definitions of intraventricular hemorrhage (IVH), necrotizing enterocolitis (NEC), chronic lung disease, growth retardation, and small for gestational age have been described previously.8
Statistical Analysis
Stepwise logistic regression modeling procedures with an entry and stay significance of .1 were used to develop prediction models for the likelihood of 4 outcomes of interest using data from the compliant cohort of children who returned for follow-up evaluation: MDI <70, PDI <70, CP, and body weight <10th percentile for age. The predictor variables selected were available for all 3 study subgroups: compliant, visit but not measured, and the no-visit group. Initially, independent variables were selected from predictors identified in the retrospective network follow-up study,8 from among known associates of poor outcome12–16 and from prediction models for neurologic outcome developed at the University of Alabama-Birmingham.17,18 These factors were maternal age, infant gender, infant race, plurality, gestational age, birth weight, weight <10th percentile, grade 3 or 4 IVH, Bell stage II or greater NEC, use of antenatal glucocorticoids, chronic lung disease (CLD), and use of postnatal glucocorticoids. Because of observed, unexplained differences among network centers in 18- to 22-month outcomes,8 final models that included study center among the independent variables were developed. All potential predictor variables were categorical except for birth weight and gestational age, which were included as continuous variables. To use the results from these models to predict the outcomes among those without follow-up evaluations, we restricted potential predictor variables to those that were available for all children regardless of follow-up study compliance. Data regarding maternal education, whether the child was living with the biological mother, and family income were unavailable for children who did not participate in the follow-up study and therefore could not be included. Prediction models were initially constructed using all observations with nonmissing values for each outcome so that the number of observations used varied depending on outcome. However, to simplify presentation of odds ratios and confidence intervals for significant predictors, we also constructed the models using data from children with all 4 outcomes nonmissing for each outcome (same number of observations for each outcome). Results from the 2 analyses were nearly identical.
Distributions of each variable that was found to be a significant predictor for an outcome among compliant children were compared among the compliant group, the no-visit group, and the visit but not measured group. Statistical significance for differences across these groups was determined using 
2 or Fisher exact tests for categorical variables and Wilcoxon tests for continuous variables. P < .05 was considered statistically significant.
The prediction equations derived from the logistic regression models were used to compute predicted probabilities of each outcome for all children, including those for whom the outcomes were missing. Predicted probabilities from the compliant, visit but not measured, and no-visit groups were combined, and decile cutpoints were determined for this distribution. Infants were assigned to categories on the basis of these decile cutpoints. The distribution of infants in the probability categories was compared between the compliant group and the no-visit and visit but not measured groups using a row mean score 2 test. All analyses were completed using SAS software.19
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONAPPENDIXREFERENCES |
|---|
|
|
|
|
Using the model prediction equation, 26% of the children in the no-visit group and 29% of the children in the visit but not measured group would have had PDI scores <70. No significant differences were detected in the distribution of predicted probabilities for PDI <70 between compliant and no visit (P = .3) or between the compliant and the visit but not measured groups (P = .6; Table 4).
CP
Among the 1151 children who attended the 18- to 22-month visit, 1107 (96%) were evaluated for CP and 191 (17%) were identified with CP. IVH grade 3 or 4, CLD, and NEC stage 2 or higher were associated with an increased risk of CP (Table 2). Antenatal glucocorticoid use was associated with a decreased likelihood of CP. Among these variables, CLD occurred more frequently among compliant children than among those in the no-visit group (Table 3).
Using the model prediction equation, 18% of children in the no-visit group and 15% of children in the visit but not measured group would have received a diagnosis of CP. The distribution of predicted probabilities of having CP was marginally higher among children in the no-visit group compared with the compliant group (P = .05). No statistically significant difference was found in the probability distributions between the compliant and visit but not measured groups (P = .08; Table 4). Risk distributions derived from the initial model did not include study center as a covariate and showed that the predicted likelihoods of MDI <70 and PDI <70 were less among infants in the no-visit group (P = .03 and P < .01, respectively).20
Growth
Body weight was measured in 1096 (95%) of 1151 children who attended the 18- to 22-month visit. Weights ranged from 6.3 to 21.7 kg (mean: 10.1 kg) with 482 (44%) children below the 10th percentile. Male gender, small for gestational age, more advanced gestational age, and IVH grade 3 or 4 were associated with higher likelihood of follow-up weight <10th percentile (Table 2). None of these characteristics differed significantly between the compliant and no-visit groups.
Using the model prediction equation, 42% of children in the no-visit and 41% of children in the visit but not measured groups were predicted to have body weight <10th percentile at 18 to 22 months' postmenstrual age. No significant differences were detected in the distributions of predicted probabilities of weight <10th percentile between the compliant and no-visit groups (P = .1) or between the compliant and visit but not measured groups (P = .1; Table 4).
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONAPPENDIXREFERENCES |
|---|
Consistent with network observations,8 other follow-up studies have found that ELBW infants have severe neurologic abnormalities, developmental delay, and functional deficits.12–14,21,22 Determining the presence or absence of impairment from a population-based perspective requires that a representative sample of infants be evaluated reliably at an age at which results can be used to predict long-term outcomes.14 For the sample to be truly representative, infants who are evaluated must be similar to those who are not evaluated with respect to attributes that influence the outcomes of interest.
Conclusions based on previous longitudinal studies of ELBW infants are limited by subject loss to follow-up. Wolke et al5 evaluated subject loss in a German population by performing a prospective study of NICU graduates who were born at <32 weeks' gestational age. Physician reporting was used to obtain information regarding infants who were lost to follow-up. By using the general developmental quotients from the Griffiths Mental and Developmental Scales for Neurodevelopmental Assessment, these investigators found that infants who were lost to follow-up had more motor and neurosensory disability and that mothers were of a lower educational level in comparison with those who were compliant with follow-up. The authors postulated that parents of severely disabled infants have not come to terms with their disabilities and avoid situations in which the disabilities may be discussed. The authors concluded that exclusion of this population in follow-up studies might lead to overly optimistic reports of neurodevelopmental outcomes of prematurely born infants. Tin et al6 found similar results in a prospective study of survivors of English NICUs who were born at <32 weeks' gestational age. The authors stated that conclusions drawn from neurodevelopmental outcome studies limited to subjects who are compliant with follow-up visits may be biased by overestimating the proportion of survivors with better outcomes.
Callanan et al4 in Australia evaluated very low birth weight survivors who were followed with difficulty—defined as those who missed their first appointment for no valid reason, missed their appointment and their families were reluctant to attend, and did not miss their appointment but were reluctant to attend. Infants who were followed with difficulty tended to have both increased disability and lower neurodevelopmental test scores at 2 and 5 years of age. Mothers of the difficult-to-follow group were more likely to have single marital status and have <12 years of education.
Predicted outcomes in the current study contrasted with European and Australian studies.4–6 BSID-II MDI scores were more often <70 in the compliant group than predicted for the no-visit group. The current study evaluated ELBW infants who averaged 26 weeks' gestation and were of 401 to 1000 g birth weight. Wolke et al5 and Tin et al6 included infants who were of <32 weeks' gestation and were primarily >1000 g. In the current study, the compliant group had significantly higher percentages of infants from multiple gestations, greater use of postnatal glucocorticoids, and greater likelihood of CLD compared with the no-visit group.23–29 Differences in the patient populations evaluated may explain differences between the current study and the European and Australian reports. The sample sizes differed among these longitudinal studies: current study, 1483; Tin et al,6 796; Wolke et al,5 396; and Callanan et al,4 204. The follow-up rates also differed: current study, 78%; Tin et al, 89%; Wolke et al, 78%; and Callanan et al, 94%. Single-center studies of smaller cohorts may have allowed for easier tracking and more intensive encouragement.
The multiple risk factors associated with CP, including poor prenatal care, antenatal steroid use, inadequate nutrition, and lower socioeconomic status, likely have a cumulative impact on neurodevelopment.11,16,18,30–41 For example, in the current study, the compliant group tended to receive antenatal steroids more often (38% vs 36%), which has been shown to be protective against the development of CP.31,38,42–44 The group that was lost to follow-up tended to have caregivers with single marital status and less prenatal care than the compliant group. When the 3 study groups were compared for the proportion with follow-up weight <10th percentile, no differences were observed, suggesting that infant growth or weight gain are not important factors in the decision to return for follow-up.
We consider that the possible increased likelihood of developmental morbidity in the compliant group may result in a subsequent perceived need for more frequent evaluation. Previous European studies reported that no-visit infants had greater NICU illness severity.5,6 Including these infants in long-term follow-up could result in worse neurodevelopmental outcome estimates for ELBW infants. On the basis of the current study's predictive models using a US population, we propose that ELBW infants who are lost to follow-up may actually be healthier and that perhaps the primary caregiver does not believe that there is a need for more sophisticated evaluation and care. If infants who were lost to follow-up in the current study had been included in the outcome evaluations, then overall neurodevelopmental outcomes may have been better. Alternative study designs are required to confirm this speculation.
The findings of this study are limited by the use of predictive models. Other factors not measured may have affected our model and thus had an impact on our results. For example, data regarding maternal educational status, marital status, and quality of the home environment were not measured and may have an impact on infant development.16,35,36,38,39,41
In an earlier analysis, study site was not used as a covariate to derive the risk distributions. There were significant differences in the BSID-II MDI and PDI scores comparing the compliant group and the group that was lost to follow-up.20 The compliant group tended to have greater neurodevelopmental impairment shown by low MDI and PDI scores. Variations in follow-up rates across centers and the type of children followed may have affected these results. The follow-up rates among survivors in this cohort varied from 56% to 97% across the 12 network centers.45 Patient outcomes are affected by treatment therapies that may have differed across the study centers. It has been reported that various facilities differ in interventional therapies, and therapeutic outcomes of care can be variable.8,46–49 How each individual center ensures follow-up after NICU discharge can also vary. Perhaps centers with poor follow-up compliance may be focusing their efforts on patients who are believed to have more significant diseases. Furthermore, a single center with a high follow-up rate and a high proportion of children with neurodevelopmental impairment among those followed may have made the compliant group misrepresentative of the study patient population. Thus, how each individual center ensures follow-up after NICU discharge may need to be examined and perhaps changed.
The most precise evaluation of long-term neurodevelopmental outcome of ELBW infants would be achieved with 100% follow-up, thus avoiding bias in estimating the likelihood of adverse neurodevelopmental outcomes of ELBW infants. The population who is not reluctant to be seen may be easier to retain as a result of the need for continual contact with health care services. Perhaps infants whose mother has the high-risk characteristics defined by Callanan et al4 can be identified in the perinatal period and more effort can be used to ensure follow-up. However, even with the most vigilant tracking practices and incentive programs, 100% follow-up is difficult, if not impossible, to achieve.
|
CONCLUSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONAPPENDIXREFERENCES |
|---|
As the probability of survival of infants previously believed to be on the edge of viability increases, it is important that studies provide unbiased estimates of longer term outcomes. Such estimates will allow better evaluations of the quality of current treatment strategies.
|
APPENDIX |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONAPPENDIXREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
Presented in part at the Pediatric Academic Societies' Annual Meeting, May 2000.
|
FOOTNOTES |
|---|
Reprint requests to (L.C.) Cincinnati Children's Hospital Medical Center, 3333 Burnet Ave, Cincinnati, OH 45229-3039. E-mail: lisa.castro{at}cchmc.org
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONAPPENDIXREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  S. M. Schulzke, G. C. Deshpande, and S. K. Patole Neurodevelopmental Outcomes of Very Low-Birth-Weight Infants With Necrotizing Enterocolitis: A Systematic Review of Observational Studies Arch Pediatr Adolesc Med, June 1, 2007; 161(6): 583 - 590. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. M Rees, A. Pierro, and S. Eaton Neurodevelopmental outcomes of neonates with medically and surgically treated necrotizing enterocolitis Arch. Dis. Child. Fetal Neonatal Ed., May 1, 2007; 92(3): F193 - F198. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. E. K. Stein, M. J. Siegel, and L. J. Bauman Are Children of Moderately Low Birth Weight at Increased Risk for Poor Health? A New Look at an Old Question Pediatrics, July 1, 2006; 118(1): 217 - 223. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
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]
|
|
|
|
|
|
E.T.M. Hille, L. Elbertse, J. B. Gravenhorst, R. Brand, S.P. Verloove-Vanhorick, and on behalf of the Dutch POPS-19 Collaborative Study Nonresponse Bias in a Follow-up Study of 19-Year-Old Adolescents Born as Preterm Infants Pediatrics, November 1, 2005; 116(5): e662 - e666. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Ho and S. Saigal Current Survival and Early Outcomes of Infants of Borderline Viability NeoReviews, March 1, 2005; 6(3): e123 - e132. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2004 American Academy of Pediatrics. All rights reserved. |
||
|
||
