The Journal of Pediatrics
Volume 148, Issue 4 , Pages 427-429, April 2006

Understanding neurodevelopment in premature infants: Applied chaos theory

  • William MacKendrick, MD

      Affiliations

    • Corresponding Author InformationReprint requests: William MacKendrick, MD, Dept of Pediatrics, Evanston Hospital, Walgreen Building #1505, 2650 Ridge Ave, Evanston, IL 60201.

Department of Pediatrics, Evanston Hospital, Evanston, Illinois

Article Outline

Abbreviations: MR, Magnetic resonance

 

Understanding dynamic systems processes in health and disease is one of the core tasks of modern medicine. The prevalent paradigm of process analysis in medicine is rooted in linear dynamics—that is, human biologic systems are generally assumed to follow a predictable pattern of cause and effect by which a given event can be shown to directly and predictably influence a given outcome. In neonatology, perhaps the most important system processes to understand are those of brain development and subsequent neurodevelopmental outcome in the prematurely born infant. Can brain development and neurodevelopmental outcome in the premature infant be adequately modeled using linear dynamic principles?

See related article, p 438

In this issue of The Journal, Ambalavanan et al seek to provide outcome models that use antenatal and early postnatal clinical data to predict death or neurodevelopmental impairment at 18 months corrected age in a population of extremely low birth weight infants.1 To do so, they applied a relatively new and potentially powerful technique of classification tree analysis to a dataset.2 This technique is intended to provide an optimal binary decision tree that the clinician can “walk down” to determine the likelihood of a primary outcome occurring for a particular infant with the available clinical data. Three models were developed: 1 using antenatal data only; 1 using antenatal data plus data from the first 3 days of life; and 1 using antenatal data plus data from the first 8 days of life. The investigators found that adding postnatal data to antenatal data did not improve the predictive power of the model; in addition, the predictive accuracy of all 3 models was only 61% to 62%. The authors correctly conclude that such predictive accuracy is “…not adequate for clinical decision making.”1

Why were Ambalavanan et al not able to achieve better predictive accuracy with their models? As they point out, the available dataset was limited to those variables collected as part of the trial; other, more novel variables that might improve model accuracy may exist. However, the most important consideration is likely to be the implicit assumption that linear dynamics are capable of adequately modeling the combined outcome of death or neurodevelopmental impairment at 18 months. It has been fairly well shown that most neonatal mortality can be successfully predicted by using illness severity scoring systems3 and clinical intuition.4 That observation leaves the question posed earlier: can brain development and neurodevelopmental outcome in the premature infant be successfully modeled using linear dynamic principles?

The traditional view of brain development in the premature infant has followed the principles of linear dynamics, by which infants who do not sustain intraventricular hemorrhage or show evidence of white matter injury with cranial ultrasound scanning during their neonatal course are assumed to subsequently have normal brain development. However, studies of premature brain development with newer magnetic resonance (MR) imaging techniques have cast considerable doubt on that assumption. In a study with advanced 3-dimensional volumetric MR imaging techniques, Inder et al5 found significantly lower cortical gray matter volumes in former premature infants when studied at term as compared with full-term control subjects. Further analysis of the data presented shows that this difference remained significant even when infants with evidence of white matter injury were excluded. In a similar study with volumetric MR techniques, Kesler et al6 compared regional brain volumes in mid-childhood between a group of former premature infants and full-term control subjects. They found significant differences in regional brain volumes and an overall reduction in cerebral tissue volume among former premature infants. None of the former premature infants in Kesler’s study had sustained severe intraventricular hemorrhage or periventricular leukomalacia. Peterson et al7 used functional MR imaging to evaluate brain activation pathways during language processing tasks at 8 years of age in a group of 23 former premature infants, none of whom had sustained severe neonatal brain injury. When compared with full-term control subjects, the former premature infants had profound differences in language processing pathways, generally using the same pathways to perform high-level semantic processing that were used to perform low-level decoding tasks in the full-term control subjects. Taken together, these studies suggest that brain development may be significantly altered in many, if not all, premature infants, not just in those with gross evidence of brain injury in the neonatal period.

If premature infants without obvious brain injury nonetheless have abnormalities of brain development, one would expect that they might have evidence of neurodevelopmental impairment as a result. There is evidence to suggest that this is true in the short term. In a study from the NICHD Neonatal Network, Laptook et al8 reported on 18- to 22-month neurodevelopmental outcomes in a cohort of infants with birthweights <1000 g who had normal results on early and late cranial ultrasound scans during their neonatal intensive care unit stays. Despite the absence of apparent brain injury, almost 10% of these infants had evidence of cerebral palsy, and 25% had mental development index scores in the significantly handicapped range (<70) on the Bayley scales.

Abnormal brain development in premature infants may be linked to neurodevelopmental impairment in the short term, but a corollary relationship to long-term neurodevelopmental outcome is much less clear. There is evidence that cognitive function may significantly improve during childhood in former premature infants,9 and infants with mild motor deficits at 1 year of age may be free of such deficits later in childhood.10 Hack et al have recently reported that the mental development index score at 20 months on the Bayley scales is poorly predictive of cognitive function at school age.11 Outcomes studies of premature infants in adolescence and young adulthood suggest that significant adaptive processes occur in the neurodevelopmental and psychosocial realms. In adolescence, former premature infants generally rate their quality of life and self-esteem as good, as do their full-term peers.12, 13 There is also evidence that former premature infants tend to engage in less risk-taking and illicit behavior as young adults.14 Although former premature infants clearly have significantly higher rates of handicap than their full-term peers in young adulthood,14 the absolute rates of major handicap in former premature infants are relatively low.

It may be argued, then, that linear dynamics fail to lead to an adequate understanding of brain development and neurodevelopmental outcome in premature infants. Linear dynamics would suggest that avoidance of brain injury should result in normal brain development, but brain development appears to be abnormal in many, if not all, extremely premature infants despite the absence of gross brain injury. Linear dynamics would suggest a direct relationship between brain development and short- and long-term neurodevelopmental outcome, but that relationship appears to be decidedly complex and indirect. On the face of things, we are left with a rather nihilistic and seemingly illogical set of conclusions: despite our best efforts, we may be unable to achieve normal brain development in premature infants; however, such abnormal brain development may mean relatively little for the ultimate outcome of these infants. However, the aformentioned conclusions can be logically understood, and even cast in a positive light, when they are viewed as the results of processes governed by the principles of nonlinear dynamics, more popularly known as chaos theory.

Chaotic systems have certain characteristic properties: they are exquisitely sensitive to the initial conditions imposed on the system; cause and effect are not proportional; and they appear to exhibit random, disorganized behavior that is not truly random but is governed by complex, nonlinear relationships.15 Predicting exact future outcomes in a chaotic system is not possible, but for many chaotic systems, a set of possible outcomes can be identified. One of the best-known examples of a chaotic system in medicine is the human heart rate in health and disease. In healthy individuals, the heart rate in any given interval exhibits chaotic behavior that can be described by using fractal analysis.16 In disease states such as severe congestive heart failure, the heart rate loses its fractal characteristics and becomes monotonous or follows a sinusoidal pattern.

The processes that drive human brain development are not well understood, so the assignment of any particular model to describe those processes is entirely speculative. However, if a relatively straightforward process such as the heart rate can be described as a chaotic system, then the immensely more complex process of brain development must surely also follow the principles of chaos theory.

When this premise is accepted, the finding of abnormal brain development in premature infants without actual brain injury is easily understood. Chaotic systems are extremely sensitive to the smallest of perturbations, and the very act of being born prematurely obviously brings profound changes to the neurosensory inputs that will help to shape brain development. Significant departures from normal brain development must then be inevitable in premature infants. Furthermore, because even the smallest changes in initial conditions may have profound effects in the future, it can be argued that despite any future advances in neonatal care, the brain development that occurs in an infant born at term can never be precisely replicated in a premature infant.

Although premature birth may unavoidably alter brain development, that development will presumably remain a chaotic process in the absence of gross injury that severely limits potential for development. Chaos implies nonlinearity, and so one should expect a range of outcomes, many of them favorable, in premature infants despite initial abnormalities in brain development. As development continues postnatally, small changes in the inputs that drive brain development will continue to have potentially large impacts, and certain of those changes will presumably tend to drive the system toward the more optimal of the possible outcomes of the system. A potential example of this principle is the significant influence of maternal socioeconomic status and education on neurodevelopment in former premature infants.6, 17

Modern medicine strives toward total understanding of human biologic processes, but many of those processes may not be adequately understood using linear dynamic principles. Chaos theory can provide a framework for understanding such processes, but as James Goodwin18 has noted, it does not promise a total understanding of the system in question. Ambalavanan et al have made an admirable attempt to provide us with a tool that will accurately predict adverse outcome in extremely low birth weight infants, but their success was likely limited by the nonlinear, chaotic nature of human neurodevelopment. Chaos theory provides a framework for understanding why brain development is likely to always be inherently different in premature infants as compared with their full-term counterparts. It also underscores that there is always hope for improvement with time, except in the child who is severely brain-injured, while simultaneously acknowledging that we cannot predict the future.

Back to Article Outline

References 

  1. Ambalavanan N, Baibergenova A, Carlo WA, Saigal S, Schmidt B, Thorpe KE. Early prediction of poor outcome in extremely low birth weight infants by classification tree analysis. J Pediatr. 2006;148:438–444
  2. Schmidt B, Davis P, Moddemann D, Ohlsson A, Roberts RS, Saigal S, et al. Long-term effects of indomethacin prophylaxis in extremely-low-birth-weight infants. N Engl J Med. 2001;344:1966–1972
  3. Richardson DK, Corcoran JD, Escobar GJ, Lee SK. SNAP-II and SNAPPE-II (simplified newborn illness severity and mortality risk scores). J Pediatr. 2001;138:92–100
  4. Meadow W, Frain L, Ren Y, Lee G, Soneji S, Lantos J. Serial assessment of mortality in the neonatal intensive care unit by algorithm and intuition (certainty, uncertainty, and informed consent). Pediatrics. 2002;109:878–886
  5. Inder TE, Warfield SK, Wang H, Huppi PS, Volpe J. Abnormal cerebral structure is present at term in premature infants. Pediatrics. 2005;115:286–294
  6. Kesler SR, Ment LR, Vohr B, Pajot SK, Schnieder KC, Katz KH, et al. Volumetric analysis of regional cerebral development in preterm children. Pediatr Neurol. 2004;31:318–325
  7. Peterson BS, Vohr B, Kane MJ, Whalen DH, Schneider KC, Katz KH, et al. A functional magnetic resonance imaging study of language processing and its cognitive correlates in prematurely born children. Pediatrics. 2002;110:1153–1162
  8. Laptook AR, O’Shea TM, Shankaran S, Bhaskar B NICHD Neonatal Network. Adverse neurodevelopmental outcomes among extremely low birth weight infants with a normal head ultrasound (prevalence and antecedents). Pediatrics. 2005;115:673–680
  9. Ment LR, Vohr B, Allan W, Katz KH, Schneider KC, Westerveld M, et al. Change in cognitive function over time in very low-birth-weight infants. JAMA. 2003;289:705–711
  10. Nelson KB, Ellenberg JH. Children who “outgrew” cerebral palsy. Pediatrics. 1982;69:529–536
  11. Hack M, Taylor G, Drotar D, Schluchter M, Cartar L, Wilson-Costello D, et al. Poor predictive validity of the Bayley scales of infant development for cognitive function of extremely low birth weight children at school age. Pediatrics. 2005;116:333–341
  12. Saigal S, Feeny D, Rosenbaum P, Furlong W, Burrows E, Stoskopf B. Self-perceived health status and health-related quality of life of extremely low-birth-weight infants at adolescence. JAMA. 1996;276:453–459
  13. Saigal S, Lambert M, Russ C, Hoult L. Self-esteem of adolescents who were born prematurely. Pediatrics. 2002;109:429–433
  14. Hack M, Flannery DJ, Schluchter M, Cartar L, Borawski E, Klein N. Outcomes in young adulthood for very-low-birth-weight infants. N Engl J Med. 2002;346:149–157
  15. Lorenz EN. The essence of chaos. Seattle: University of Washington Press; 1993;
  16. Goldberger AL, Amaral LAN, Hausdorff JM, Ivanov PC, Peng C-K, Stanley HE. Fractal dynamics in physiology (alterations with disease and aging). Proc Natl Acad Sci U S A. 2002;99:2466–2472
  17. Vohr BR, Allan WC, Westerveld M, Schneider KC, Katz KH, Makuch RW, et al. School-age outcomes of very low birth weight infants in the indomethacin intraventricular hemorrhage prevention trial. Pediatrics. 2003;111:e340–e346
  18. Goodwin JS. Chaos and the limits of modern medicine. JAMA. 1997;278:1399–1400

PII: S0022-3476(05)01235-7

doi:10.1016/j.jpeds.2005.12.049

Refers to article:

  • Early prediction of poor outcome in extremely low birth weight infants by classification tree analysis

    N. Ambalavanan, A. Baibergenova, W.A. Carlo, S. Saigal, B. Schmidt, K.E. Thorpe, Trial of Indomethacin Prophylaxis in Preterms (TIPP) Investigators
    The Journal of Pediatrics April 2006 (Vol. 148, Issue 4, Pages 438-444.e1)

The Journal of Pediatrics
Volume 148, Issue 4 , Pages 427-429, April 2006

Article Tools