The studies presented below reveal some continuities in development, and highlight certain tests at preschool stage, which could serve as “predictors” of later performance. The hypothesis of the continuity of development and the predictive value of a early risk indexes is explored. Only studies which relate to occupational therapy practices are presented in this section.
Fallang et al (2005) suggested that even screening of early motor behaviours such as reaching and postural behaviour at the very early age of 4 and 6 months, can predict
47
neuromotor behaviour at 6 years. Despite the main inclusion criterion of the children (N=74) being LBW (<1750g), findings are presented here since the average GA was 29 weeks (SD: 2.9). This follow up study showed that in the cases of preterm children that do not present with cerebral palsy, non optimal reaching at 4 months was associated to minor neurological dysfunction (MND) at the age of six (p=.008), as assessed with the use of Touwen’s profile (Touwen, 1979) of MND and the Movement ABC Test (Henderson & Sugden, 1992). In a similar fashion, the quality of reaching at 6 months was related to the development of MND (p=.009), as measured by the instruments mentioned before. Caution is required though in the interpretation of the findings, since the sample size was relatively small (N=74) and the attrition rate reached 30%.
The study of Majnemer, Rosenblatt and Riley (1994) did not solely focus on prematurity as defined by duration of gestation but BW, SGA and perinatal complications. It is however presented here as many of these children were also born prematurely (average GA was 36.4 weeks). Majnemer, Rosenblatt and Riley (1994) suggested that occupational therapists can be “instrumental” (p.731) in identifying children, who have been in Neonatal Intensive Care Units (NICU), who might profit from early intervention services, and that early neurobehavioral assessments could guide practice. In their prospective study, 51 “high risk” infants were assessed for their neurobehavioral status at term, and then re-assessed at 1 and 3 years of age for detection of any developmental delays. The aim of the study was to evaluate the prognostic value of the Einstein Neonatal Neurobehavioral Assessment Scale (ENNAS) (Daum et al, 1977 cited by Majnemer, Rosenblatt and Riley, 1994) designed to assess features such as muscle tone, reflexes, spontaneous movements of the infant, in relation to the developmental progress of the children measured with the Griffith’s Scale. ENNAS appeared to have a predictive value that increased from one to three years. However the assertion of “predictability” has to be viewed with caution. Majnemer, Rosenblatt and Riley (1994) suggest that such assessments should always be used in combination with other information or the use of other evaluation tools before making any clinical decisions. Moreover, the main qualifying criterion for participation in the study was “being high risk” with prematurity ,as
48
defined by gestational age, being only one out of three “high risk” diagnostic criteria (the other two were: very low birth weight and perinatal asphyxia). This means that the sample was heterogeneous, and that the presented findings have to be viewed with caution when making any inferences about predictability of development of children born preterm.
In a similar fashion, in the OT-focused study of Lane, Soares Attanasio and Farmer Huselid (1994), the sensorimotor performance of a cohort of 30 preschool children who were born prematurely was associated with their neurological scores at the end of 18 months in order to determine the predictive strength of the Miller Assessment for Preschoolers (MAP) (Miller, 1982 cited by Lane, Soares Attanasio and Farmer Huselid, 1994). The children were drawn from a previous study following up 78 premature children from birth to 18 months. All children were born in the University Hospital in Edmonton, Canada. Based on the assessment at 18 months, children were classified to “neurologically normal” and “neurologically suspicious”. A third,
“neurologically abnormal” category, consisting of participants who were also born preterm, was excluded from this study. When assessed by the MAP between the age of four and five years (mean: 4.5), the “suspicious” group scored significantly lower than the normal group in some of the subtests and the total score, and were in the
“need for referral” and follow-up for areas such as coordination, and complex tasks.
Moreover, there was a significant predictive value (p<.05) of the examination at 18 months as to the sensorimotor performance at the age of four (as assessed by the Miller Assessment for Preschoolers/MAP), indicating that “soft” neurological signs could be indicative of later sensorimotor and possibly school difficulties. Caution is however recommended when interpreting these findings, as the sensitivity of the particular assessment tool (MAP) is, according to the same authors, not well established. Moreover, the sample size was relatively small (possibility for Type II error), and the participants were all drawn from one particular hospital. Moreover, no information is provided by Lane, Soares Attanasio and Farmer Huselid (1994) as to whether the selection of this setting was based on random sampling, or convenience.
The latter would be an exclusion criterion for the selection of this study.
49
More caution is recommended in the use of infant tests when predictions are made on the basis of a “risk index”. The interpretation of the findings of an assessment measure with a predictive value should be attempted with caution as it is often specific subtests and not total scores of measures, that are differentially predictive of different functions, and this, at different developmental stages/ ages. This means that a function could be a predictor of possible later delay at a developmental stage when it is supposed to reach its ascendancy (Siegel, 1982a; Siegel, 1982b). Biological factors of a “risk index” might be more significant determinants at young ages and early scores, while environment is more influential for functions that mature at later stages (Siegel, 1985a). Moreover the statistic measure of correlation, in a key number in the studies about prediction, is sensitive to the nature of distribution and the heterogeneity of the sample, and it is not useful in individual cases. Once more the issue of statistical against clinical significance for individual cases appears to be pertinent. Correlation values are not about predicting exact scores but knowing how likely it might be for a child to obtain scores in the below average range. Siegel (1989) suggests that an analysis useful for individual cases is the determination of true positive and true negatives of a test i.e. the children who presented with an
“abnormal score” and “normal score” (respectively) both during the early and late test.
Moreover she claims that assessment at two different developmental ages does not suggest that the skills assessed resemble each other. It rather means that a
“developmental lag” at an early stage could be indicative of later difficulties given the continuity of development. Finally assessing a child at an early stage and predicting a possible future problem might have positive clinical implications but could raise ethical issues relating to “labelling” the child and “sensitising” the parents (Siegel, 1989).
Aylward (2004) has more recently raised concerns about the idea of “prediction” of outcome for professionals who work with children, such as preterm children running a certain risk. The author claims that prematurity, LBW or perinatal histories can only be viewed as potentially indicative of later problems, and that they do no more than raise concern as to what the developmental outcomes might be. A host of factors such a neural organisation, timing, type and duration of possible perinatal insults and, last
50
but not least, the environmental effects can alter the course of development, and cannot be predicted accurately. Biological and environmental factors are interwoven in complex ways acting either synergistically or antithetically, leading to negative or positive outcomes. Both Aylward (2004) and Lane, Soares Attanasio and Farmer Huselid (1994) stress that the assessment of the child and its environment at certain time intervals could enhance prediction and that any decisions on appropriate interventions should be made in the light of these re-examinations.
The idea of “predicting” outcome becomes broader and possibly more functional when skills from various developmental domains are used in order to forecast school achievement (Fallang et al, 2005; Deutscher and Fewell, 2005). However none of this research is tied into OT practice, and it is therefore not presented here.