OBJECTIVE To assess the value of health and behavioural problems at 8 weeks as predictors of behavioural problems at 8 months in a whole year birth cohort.
STUDY DESIGN Prospective birth cohort study.
SETTING The socially and ethnically diverse city of Coventry.
MAIN OUTCOME Parent reported behavioural problems at 8 months.
METHOD Parent reported infant health and behaviour data were collected, using a validated questionnaire administered by the family health visitor at 8 weeks and 8 months, on 1541 infants participating in the Coventry cohort study. Sociodemographic data were collected at the health visitor’s initial visit. Unadjusted relative risks (with 95% confidence intervals (CI)) of behaviour problems at 8 months by sociodemographic variables and health and behavioural problems at 8 weeks were estimated. Adjustment for confounding was made by logistic regression.
RESULTS Infants reported to have behavioural problems at 8 weeks had a significant risk of parent reported behavioural problems at 8 months (adjusted relative risk, 3.44; 95% CI, 1.95 to 6.09) after adjustment for other health outcomes and sociodemographic factors. Of infants with behavioural problems by 8 weeks of age, 19.1% were reported to have behavioural problems at 8 months.
CONCLUSIONS Infants whose parents report behaviour problems by 8 weeks of age are at higher risk of behavioural problems at 8 months. However, despite the higher risk, the proportions of infants identified by behaviour at 8 weeks were too small for the early outcomes to be useful as predictors of behaviour at 8 months in the whole infant population.
- behavioural problems
- birth cohort study
- predictive value
Statistics from Altmetric.com
Stability of behavioural problems over time has been noted in previous studies but most of these have first identified behavioural difficulties around 5 years of age.1-3 Infant sleeping, crying, and feeding behaviour has been reported to show little stability over the first 9 months of life but “high fussers” at 6 weeks tended to remain fussers at 9 months.4 Temperamental difficulties noted by mothers at 4 months of age have been shown to be associated with an increased risk of maternally reported behavioural problems at 6 years of age,5 and infants with problems of feeding and crying behaviour at 4 months have a higher risk of behavioural problems in their 4th year.6 An association between onset of asthma before 3 years of age and behavioural problems at 4 years has been reported,7 as has an association between length of hospital admission in the preschool years and behavioural problems at 6 years of age.8
However, there are limited published data on the association between other adverse early infancy health outcomes and behavioural problems. Our study reports on the association between parent reported health outcomes collected at 8 weeks of age and parent reported behavioural problems at 8 months in a longitudinal whole year birth cohort. The value of these early outcomes as predictors of behavioural problems in later infancy is assessed.
Two thousand five hundred and eighty infants born within the Coventry city boundary in 1996 were enrolled into the Coventry cohort study by the family health visitor at the birth visit. Sociodemographic data were collected by the health visitor on to a standard proforma. Data on parent reported infant health and illness were collected as part of routine child health surveillance at 8 weeks and 8 months using a validated measure, the “Warwick child health and morbidity profile”,9 incorporated into the “parent held record” and administered by the health visitor. Full details of the development and validation of the Warwick child health and morbidity profile have been published.9 The profile was validated in field testing with 228 parents of infants in three health service settings: a child development unit, child health clinics, and paediatric outpatient departments. Test–retest and interobserver reliability achieved satisfactory levels using the weighted κ statistic. Internal and external validity were satisfactory.
Data were available at 8 weeks and 8 months for 10 domains: general health status; acute minor illness status; behavioural status; accident status; acute illness requiring medical attention; hospital admission (overnight stay); immunisation status; chronic illness status; functional health status; and health related quality of life. Some domains, such as general health status and functional health status, have a certain amount of overlap but measure different aspects of parental perception of the effects of illness on the infant’s well being. For example, children can be perceived by the parents as “healthy” despite limitations in daily living or the presence of a chronic condition such as asthma.
Responses to each domain, which each have four categories, were collapsed down to binary variables as follows:
- General health status: very healthy and healthy versus not healthy and poor health.
- Acute minor illness status: none or fewer than expected versus more than expected.
- Behavioural status: no behavioural problems versus any behavioural problems.
- Accident status: no accidents versus any accidents.
- Acute illness requiring medical attention: none versus any.
- Hospital admission: none versus any.
- Immunisation status: up to date versus not up to date.
- Chronic illness status: none versus any.
- Functional health status: no impairment versus impairment.
- Health related quality of life: not affected versus affected.
Sociodemographic variables were dichotomised as follows:
- Housing tenure: rented versus owner occupied.
- Marital status: lone parent versus married/cohabiting.
- Mother’s age: < 20 versus > 20 years.
- Ethnicity: Indian subcontinent versus rest.
Relative risks with 95% confidence intervals (CI) were calculated for behavioural problems at 8 months by health outcomes at 8 weeks and sociodemographic characteristics. To adjust for potential confounding, health outcomes and sociodemographic factors associated with a significant unadjusted risk of behavioural problems at 8 months were entered into a logistic regression model in SPSS for Windows10 as dichotomised predictor variables, with behaviour at 8 months as the dependent variable. Variables that failed to reach significance on univariate analysis were not entered into the logistic regression model. Predictor variables with significance of < 0.05 in the logistic regression model were taken as independently associated with the dependent variable. For predictor variables independently associated with behaviour at 8 months, the value of the variable as a predictor of behavioural problems at 8 months in the whole cohort population was estimated.
Data were available for 1541 infants at 8 weeks and 8 months. The attrition rate between birth and 8 months was 37%. There was a higher attrition rate among lower social groups (table 1).
Unadjusted and adjusted relative risks for behaviour problems at 8 months by health outcomes at 8 weeks and family sociodemographic characteristics are shown in table 2. Only behavioural problems at 8 weeks retained a significant risk after adjustment.
Of the infants reported to have behavioural problems at 8 weeks, 19.1% were reported to still have problems at 8 months. These infants constituted 28.5% of the infants reported to have behavioural problems at 8 months.
Behavioural problems in childhood and adolescence have been the focus of increased interest, both in terms of their impact on child health and education services and their social policy implications.11 Behavioural problems now constitute the largest single category of disabilities in childhood.12Interest has grown in the early identification of behavioural problems to enable more effective interventions to be tested.13
The results of our study indicate that parent reported behavioural problems in early infancy are independently associated with an increased risk of parent reported behavioural problems in later infancy. However, despite a significantly increased risk adjusted for confounding, behavioural problems in early infancy are only weakly predictive of problems in later infancy, constituting only 28.5% of the infants reported to have behavioural problems at 8 months.
Most studies of the stability of behavioural problems over time report on the relation between behaviour around 5 years of age and behaviour in later childhood and adolescence. Most studies of temperamental difficulties in infancy also report on the risk of behavioural problems in later childhood, but not in later infancy.5 6 14 15Our study focuses on the predictive value of outcomes in early infancy for behavioural problems in late infancy. A study measuring the amounts of time spent asleep, awake content, feeding, fussing, and crying in a cohort of infants at 2, 6, 12, and 40 weeks of age4reported moderate stability of sleep between 6 weeks and 9 months, although not of crying. Infants classified as “high fussers” in early infancy were likely to retain this characteristic at 9 months. Comparison with the results of this study is difficult because different outcome measures were used. Behavioural problems in the 1st year of life are less well defined and less fully studied, but represent a considerable burden on parents and child health professionals. Subsequent data as part of the follow up of the cohort should be relevant to prediction of stability of early infancy behavioural problems into later childhood.
Account should be taken of a number of issues when interpreting these results. The Warwick child health and morbidity profile is designed as a simple measure of parent reported infant health and illness experience for use in routine child health surveillance. The measure of behavioural problems is based on parent reported problems with infant behaviour. It is not an objective score, derived from a standardised questionnaire, but relates directly to the parents’ own experience of their infant and the problems they present. Feeding, sleeping, and crying formed the bulk of these problems at 8 weeks, and sleeping at 8 months.
The attrition rate in our study is high and there is a differential rate of attrition by social group. It is possible that the experience of infant behavioural problems may be different in cohort drop outs. However, it seems unlikely that this would be sufficient to affect significantly the results related to stability of behavioural problems between 8 weeks and 8 months.
In our study infants of lone mothers and mothers aged < 20 years were at increased risk of behavioural problems at 8 months, which is consistent with the reported social gradient in childhood behavioural problems.16 On adjustment, the risks were no longer significant. Because a higher proportion of infants with these family characteristics dropped out of the study, it is possible that the regression analysis underestimates the influence of marital status and maternal age in the model.
Despite these limitations, it is reasonable to conclude that infants reported by their parents to have behavioural problems at 8 weeks are at increased risk of parent reported behavioural problems at 8 months, but this early infancy outcome has insufficient predictive value to constitute a useful predictor of late infancy behavioural problems. However, the use of a simple measure of parent reported infant health and illness experience, such as the Warwick child health and morbidity profile, may prove useful in routine child health surveillance in identifying infants whose parents perceive them as having behavioural problems so that early intervention can be instituted.
We thank the Coventry parents and health visitors for their invaluable contribution to the success of the cohort study. The Coventry cohort study has been supported by grants from local charities, Coventry Health Authority, and “Babes in Arms”.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.