Failure to thrive (FTT) is a term used to describe children whose growth is relatively poor in infancy,¹ and is associated with deficits in development and social interaction. As the most objective clinical finding associated with the condition of FTT is poor weight gain,² the term “growth faltering” is preferable for these infants as it avoids the pejorative use of the word “failure”.

In the complex literature on FTT, many factors have been reported to be associated with the condition,³ including biological factors such as parental size,⁴ social factors including deprivation, maternal educational level and family size,⁵ and a wide range of physical conditions.⁶ In most cases, the final pathway is nutritional intake inadequate for the metabolic and growth needs of the child. However, although there is a large literature on the factors associated with FTT,³ many of the studies are limited as a result of small sample size, being hospital or clinic based, or using retrospective data—and the research field is plagued by diagnostic inconsistency between studies.¹ The Avon Longitudinal Study of Parents and Children (ALSPAC; http//www.alspac.bristol.ac.uk) provides a unique opportunity to investigate FTT in a whole population of infants, using prospectively collected information. We have already described the familial, social and prenatal factors associated with FTT in this cohort,⁴ and now report on the postnatal factors associated with poor infant growth.

METHODS

All births to women resident in the former Avon Health Authority area, with an expected date of delivery between 1 April 1991 and 31 December 1992, were eligible for enrolment in ALSPAC; >80% of the known births from the geographically defined catchment area were included, resulting in a total cohort of 14 062 live births. This study population has social and demographic characteristics in common with those recorded in the UK national census surveys.⁷

Of the 14 062 live births in the study, a small proportion (0.7%) was lost to follow-up mainly because the family moved out of the study region. Infants with major congenital abnormality likely to affect feeding (eg, Down’s syndrome, cleft palate; 89/13 970, 0.6%) and infants born before 37 or after 41 completed weeks’ gestation (893/13 970, 6.4%) were excluded. We also excluded 1292 infants who had incomplete weight data, resulting in a final sample of 11 900 infants (83% of the original cohort). As typically found in the literature, there were more exclusions among social classes III, IV and V than from social classes I and II (26.6% v 19.1%), but the final sample contained a broad spectrum of social background including many families with limited economic resources.

Weight data were extracted from the Avon Child Health Computer system, using measurements made as part of the local pre-school child health surveillance programme. Measurements were taken at birth, at 8 weeks (range 1–3 months) and at 9 months (range 6–12 months). All weights were standardised to z scores adjusting for differences in sex and age (gestational age in weeks for weight at birth and infant age in weeks for subsequent weights). Growth was assessed by calculating the difference in z scores between two time points and adjusting for regression towards the mean using correlates provided by the British 1990 Growth Reference.^8,9 This increasingly used technique reflects infant growth more accurately, as it accounts for the smaller infants who tend to grow faster and the larger infants who tend to grow slower. Cases of growth faltering were defined as those infants below the 5th centile for weight gain, corresponding to a conditional growth score of −1.645. Controls were all the other infants in the cohort above the 5th centile. Data on feeding were extracted from questionnaires completed by the infant’s parents at the following ages (response rate): antenatal (88%), 4 weeks (87%), 6 months (81%) and 15 months (84%).

Ethical permission for the ALSPAC was granted by the ethics committees of the United Bristol Healthcare Trust, Frenchay and Southmead NHS Trusts and the study was also monitored by the ALSPAC Ethics and Law Advisory Committee.

Statistical methods

Correlation was calculated as Pearson’s product moment coefficient for normally distributed data. Odds ratios (ORs), 95% confidence intervals (CIs) and p values (all quoted as two sided) were calculated for both the univariate and multivariate analyses. In the univariate analysis, the significance of differences was determined using the χ² test with Yates’s continuity correction (or Fisher’s exact test when an expected cell count was <5). Variables significant at the 5% level in the univariate analysis were entered into subsequent multivariate models constructed in SAS¹⁰ using the stepwise method for selection of variables, the dependent variable indicating whether the infant was in the slowest growing 5% of the cohort.

RESULTS

The mean (standard deviation (SD)) birth weight of the term infants was 3459 (410) g, with the mean birth weight SD scores for boys and girls close to 0, indicating that the sample was very similar to the UK standard population at birth. Only 2.5% of this term cohort was born <2500 g and classified as low birth weight. We found 528 cases of growth faltering from birth to 8 weeks and 495 cases from 8 weeks to 9 months (table 1).

Birth to 8 weeks

The univariate analysis of maternal and infant factors associated with slow growth in the first 8 weeks showed that infants from older mothers (>32 years, the highest quintile of the age distribution of mothers in the ALSPAC) and shorter mothers (<160 cm, the lowest quintile of mothers’ heights) were more likely to be cases of growth faltering. Poor growth was associated with reported feeding problems, and with infant illness or admission to hospital. No significant associations were found with social class or parental education.

The 11 factors significant in the univariate analysis at the 5% level were entered for the multivariate analysis (table 2). Prenatal factors that retained significance were short maternal height, older mothers and lack of private transport. Major postnatal factors identified were difficulty in feeding and infant illness. The feeding symptoms significantly associated with weight faltering were weak sucking and difficulties in feeding, but not vomiting, dribbling or refusing feeds. Weak sucking was equally important in breastfed and bottle-fed infants: one in six infants in the cohort was reported by their parent to have weak sucking, and growth faltering was nearly twice as likely in this group.

8 weeks to 9 months

Of the 528 infants with growth faltering from birth to 8 weeks, only 30 (5.7%) also had poor growth for the second period. No data were available as to what interventions were made for those infants whose growth rate recovered. A similar range of maternal and infant factors was used in univariate analysis for the second period of infancy. Although only 0.6% of the cohort was of Asian origin (Indian subcontinent and Far East combined), growth faltering in the second period was 3–4 times as common in this group, which had a similar mean weight z score (–0.66) at birth, 8 weeks and 9 months. Other associated maternal factors were height and parity, but not age or educational attainment.

The median duration of breast feeding was 4 (interquartile range 1.5–8.5) months: 28% of infants were breast fed for >6 months, and 15% were breast fed for the whole 9-month period under study (fig 1). The mothers who breast fed for ⩾9 months tended to belong to a higher social class and were less likely to be smokers. Infants who were breast fed for ⩽9 months were born to larger families, were reported to have more difficulty in accepting solids and were more likely to refuse other milks (table 3).

Most infants in the study were weaned between 4 and 6 months, and those infants who received ⩽2 solid meals/day at 6 months were more likely to be cases of growth faltering than controls.

In contrast with the earlier growth period, infant illness was not associated with growth faltering between 8 weeks and 9 months. No associations were found with measures of social class or parental education.

The 12 factors identified in the univariate analysis to be significantly associated at the 5% level with growth faltering during this period were entered for multivariate analysis (table 4).

Distinctive maternal factors in the final model were ethnicity (Asian), height and parity. Distinctive infant factors were breast feeding after 6 months, feeding slowly, and taking small quantity of solids at 6 months.

DISCUSSION

We investigated the factors associated with weight gain in infancy by using a large representative sample from the UK population: the results suggest that the feeding characteristics of the infant are the dominant factors, rather than the socioeconomic status or the educational attainment of the mother. The strengths of the study are the use of a large well-characterised representative sample and prospectively collected feeding data, avoiding many of the methodological weaknesses of hospital and clinic-based samples and of retrospective data collection. The study is limited by the feeding data being derived from maternal report at the time of an ALSPAC questionnaire (4 weeks, 6 months and 15 months), whereas the weight data were collected at routine health service contacts (birth, 8 weeks and 9 months).

In the first 8 weeks, maternal age and height had an influence on early growth, but infant feeding difficulties were important, particularly weak sucking. There may be two underlying explanations for this association: oral–motor dysfunction^11–13 or differences in the infant’s appetite.^14,15 After age 8 weeks, weight gain was also related to maternal height, and the over-representation of Asian infants in the slowest growing 5% is also probably a consequence of shorter mothers (but could be a reflection of different infant feeding practices). The most important postnatal influences on growth were the efficiency in feeding, ability to successfully take solids and duration of breast feeding. Although these symptoms were reported by parents, who also influenced the timing of weaning, our results are consistent with behavioural studies¹⁵ reporting that children with FTT are fed as much and as frequently as controls but tend to refuse or reject offered food more often. An alternative explanation is that mothers may sense that the infant is not ready to wean (eg, not demanding solids, or showing immature oral–motor skills when offered tastes of solid food), and they continue to breast feed.

Observational studies have generally shown an association between prolonged breast feeding and slower weight gain.¹⁶ Evidence from the only relevant randomised controlled trial on breast feeding¹⁷ suggests that infants with lower appetites grow more slowly and hence are satisfied with breast milk for longer. However, it is debatable if such slow-growing breast-fed infants are at an overall disadvantage over the life course,¹⁸ as slow growth during infancy may actually have a long-term beneficial effect on reducing the risk of obesity and cardiovascular disease in adulthood.¹⁹

These results have several implications for clinical practice: firstly, a reminder that the early onset and persistence of slow or difficult feeding may be a warning of inadequate nutritional intake and possible growth faltering; and secondly, the importance of supporting parents in weaning their infants at an appropriate time developmentally. The World Health Organisation (http://www.who.int/child-adolescent-health) now recommends that mothers should be encouraged to exclusively breast feed and postpone the introduction of solids until age 6 months. However, the developmental “window” to wean on to solids is short: previous work from ALSPAC^20,21 has shown that infants unable to take lumpy solids at 9 months were more likely to have continuing feeding difficulties and poor weight gain in the second year of life. Infants who are still predominately breast feeding at age 9 months are likely to have smaller appetites and may have difficulties taking solids; these infants require careful assessment and their mothers need sensitive support to persist in offering weaning foods of appropriate consistency and variety.

Further research using this longitudinal study will clarify whether early feeding problems leading to growth faltering is a marker for persisting neurological abnormalities and whether it is associated with developmental difficulties later in childhood.