Scheduled or on-demand feeding Effects on children s educational

Scheduled or on-demand feeding? Effects on children’s educational outcomes, psycho-social development and sleeping patterns Maria Iacovou University of Cambridge UNICEF UK Baby Friendly Initiative Conference Glasgow, 27 th-28 th November 2013

What we know already what you feed a baby when you feed a baby

What we know already what you feed a baby when you feed a baby

Questions Do feeding schedules affect babies’ and children’s development? Growth and health? Cognitive development? Psycho-social adjustment? Later sleeping patterns? And mothers’ wellbeing?

Data: ALSPAC Ø Avon Longitudinal Study of Parents and Children Ø About 10, 500 babies born in 1990 and 1991 Ø Based in and around Bristol Ø Information from: Ø Both parents, 3 x during pregnancy Ø Both parents, repeatedly over childhood Ø Weights, measures and various tests Ø Teacher assessments Ø Data from national tests

Variable of interest: feeding schedules Asked at 4 weeks: “Is your baby fed (either by breast or bottle) on a regular schedule (e. g. every 4 hours)? ” “Yes, always” (7. 2%) “Try to” (23. 4%) “No, fed on demand” (69. 4%).

1. Maternal wellbeing Ø “Getting enough sleep? ” Ø “Feeling: weepy or tearful? irritable? exhausted? ” Ø Maternal confidence and enjoyment scales Ø “Motherhood is something a woman learns naturally” Ø “. . . Pretty sure I’m doing the right thing for my child”. Ø Depression measures: Edinburgh & Crown-Crisp scales. Continuous variables are standardised; discrete variables are collapsed into two categories.

2. Cognitive attainment SATs scores in Reception class (age 5), KS 1 (age 7), KS 2 (age 11) and KS 3 (age 14) Ages 5 and 7: Reading, writing, maths Ages 11 and 14: English, maths, science Separate IQ test at age 8 All standardised (0, 1)

3. Psycho-social outcomes Strengths & Difficulties questionnaire Teacher assessments in Years 3 & 6 (ages 8 & 11) Five domains Hyperactivity Emotional problems Peer problems Conduct problems Prosocial score Aggregate composed of the 4 “problem” scores

4. Sleeping 6 months, 18 months, 9 years: does child sleep regularly? 9 years: Refusing to go to bed, difficulty falling asleep, night waking, early morning waking, nightmares. 9 and 11 years: sleep durations

The ideal: a randomised controlled trial Mothers and babies randomly assigned to scheduling or demand groups Unfortunately, no such trial exists Feeding mode is a choice mothers make, based on various factors. . . . which may also affect their babies’ development Use statistical techniques which account for this selection

Who feeds to a schedule? Demand-feeding mothers in the sample are from more socially privileged groups. For example: 6% of schedule-feeding mothers have a degree 8% of “tried a schedule” mothers 18% of demand-feeding mothers “Tried-a-schedule” mothers are much more similar to schedule-feeding mothers than to demandfeeding mothers.

Analytical methods Ordinary Least Squares and logistic regression Control for factors including social class, education, marital situation, mother’s work, health behaviours, housing, breastfeeding, etc Propensity Score Matching Compare “like with like” Compare schedule-feeding mother-baby pairs with demand-feeding pairs who look just like them

Results: maternal wellbeing On almost all measures, schedule-feeding mothers report significantly higher levels of wellbeing than demand-feeding mothers. These differences persist for at least 3 years However, on the formal measures of depression, there is NO difference between the two groups Mothers who tried to schedule have the same outcomes as demand-feeding mothers.

Results: cognitive development After controlling for observable differences between the samples, demand-fed babies do significantly better than schedule-fed babies on all SATs tests and IQ tests. The outcomes of “tried-a-schedule” babies are no different to those of demand-fed babies Iacovou, M. and Sevilla-Sanz, A (2013). "Infant feeding: the effects of scheduled vs. on-demand feeding on mothers' wellbeing and children's cognitive development" European Journal of Public Health (2013) 23(1) pp 13 -19

Questions: How big is the effect? About 17% of a standard deviation Conservative estimate: about 3 “positions” in a class of 30 children What might be the mechanisms? Discuss this later

Cognitive attainment

Cognitive attainment Full

Cognitive attainment

A question on breastfeeding Could the relationship between demand feeding and better cognitive outcomes arise because of longer breastfeeding durations for demand-feeding mothers? Answer: yes, but only partially (about 2 points of the 17 -point difference comes via breastfeeding durations)

Perhaps the most important question of all We’ve controlled for observable differences between mothers. But what if unobservable differences are driving the relationship between feeding mode and children’s outcomes? Two reasons why this probably isn’t a major issue. First, the outcomes of “tried-to-schedule” babies. Second, look at this graph. . .

Is unobserved heterogeneity driving the effect?

Results: psycho-social adjustment

Year 6 SDQ scores, by sex

Year 6 SDQ scores, by sex

Results: psycho-social adjustment Results differ by gender For boys: no difference in outcomes between schedule- and demand-fed children. For girls: significant differences in outcomes, with demand-fed babies having a lower incidence of later problems. Effects are concentrated in the hyperactivity and emotional components of SDQ

Results: sleeping habits Question: do schedule-fed babies have better/different sleep habits than demand-fed babies? Answer: not really Schedule-fed babies are more likely to sleep regularly at 6 months But not at older ages At age 9, schedule-fed babies more likely to refuse to go to bed and to wake at night But differences are really very small

Results: sleep durations n n Schedule-fed babies: 20 mins longer sleep durations Fairly large difference: 0. 5 SD

School-day sleep durations, age 9 by feeding mode schedule demand

Implications Robust relationship Possible issues with parental reporting Next step: get hold of data on ALSPAC subsample who were “metered” Assess whether 20 mins extra sleep is a good thing

What are the mechanisms? Heterogeneity between mothers Possible, but unlikely to be a major factor Heterogeneity between babies Possible: investigate with observational studies

Possible causal mechanisms Breastfeeding duration Biological – frequency of feeding affecting concentration of lipids in milk? Demand-feeding encourages interactive “asking-andgetting”? Maternal responsiveness

Next steps Collaborate with psychologists to measure effects of feeding mode on (eg) stress hormones Collect more detailed data on the nature of feeding schedules (and demand feeding) More work on sleep!