Tag Archives: correlation causation

Does breastfeeding ‘cut leukaemia risk’? No good evidence

Occasionally my work and home life collide – I read some stories in the press about research claiming breastfeeding for six months or more could cut the risk of childhood acute leukaemia. The research was shaky, so I wrote this for the Leukaemia & Lymphoma Research blog (where I work).


Reports coming out in the media (such as Express, Mail, Mirror) are suggesting that breastfeeding for six months or more can lower the child’s risk of developing leukaemia. But where did these findings come from and how reliable are they?

In this post, we take a look at the research that led to these reports, and suggest that the results are not very robust, the conclusions are overstated and the claims likely to cause unnecessary alarm.

Greater than the sum of its parts?

The new research, published in the peer-reviewed journal JAMA Pediatrics, combines many previous studies that looked at whether children who had been breastfed (and for how long) went on to develop childhood leukaemia. The studies in the new analysis were a mixture of size and quality, so on their own couldn’t lead to any definitive conclusions. But when put together and re-analysed – known as a meta-analysis – it is intended to boost the power of the findings.

The new study includes 18 studies that met a certain quality threshold. The researchers, based at University of Haifa in Israel, report that children who had been breastfed for six months or more had up to a 19% lower risk of developing leukaemia than children who had never been breastfed or had been breastfed for under six months.

From this, they conclude this is sufficient evidence of a protective effect to further promote the health benefits of breastfeeding and encourage greater uptake amongst new mothers.

But we are not persuaded these claims truly stack up.

Correlation is not causation

The strength of a meta-analysis is that it tries to make sense of all the best research on the subject. But all of the studies within the new analysis looked at the association between breastfeeding and leukaemia, and therefore cannot tell you about true causes. Other background factors that affect both the likelihood of breastfeeding and leukaemia risk could have been missed or overlooked.

For example, parental affluence may affect the decision to breastfeed. But affluence will also affect a host of behaviours, like attendance at nursery, exposure to infections, decision to vaccinate, time of weaning, and many, many more. We don’t know for sure what factors do influence leukaemia risk – and parental affluence is just given as an example here merely to illustrate the complexity of background factors – but simply linking breastfeeding and leukaemia risk without consideration of other potential influencing factors is far too premature.

Experts who conduct these types of studies are well aware of these issues and always attempt to account for background factors, like socio-economics, lifestyle, gender, ethnicity, and so on, but it’s always hard to eliminate them altogether. Nevertheless, we were alarmed when we spoke to our statistical experts who noticed that the meta-analysis unusually relies on crude data that did not appropriately account for background factors.

They suggested that the authors’ relative inexperience with this type of analysis has led to a number of flaws.

Hidden biases

The experts we spoke to pointed out a gross error in the data – one dataset, that had been used in two different publications, is included twice. This could distort the statistical robustness or the size of any effect, and could be serious enough to consider a correction or withdrawal of the paper.

There are also a few ways these data could have been skewed to give wrong or exaggerated results. One is a bias in those participating in these studies. Many were based on phoning mothers at home or by a self-administered questionnaire. This introduces a potential participation bias, where the people surveyed and who agreed to take part were not representative of the population as a whole. This may mean certain groups, such as more educated or time-rich parents, were overrepresented in the comparison group (children who didn’t get leukaemia), suggesting a larger proportion of children who did not get leukaemia were also breastfed.

Almost all of the studies asked mothers or parents to remember the duration of breastfeeding, sometimes many years later. This introduces a possible recall bias, where parents may not have accurately remembered what they did or their responses were affected by knowing that their child had had leukaemia. And because childhood acute leukaemia is thankfully relatively rare – only three to four children in every 100,000 are affected each year in the UK – the small numbers could have inflated these biases.

It would be far better to recruit a large random sample of people, collect data in real-time, and then look at whether children went on to develop leukaemia. This is more costly and time-consuming, but it would help diminish some of these potential biases because the particpants and research questions would be defined up-front. 

The authors acknowledge these limitations in their discussion section of their paper, which is why we were surprised by the strength and certainty of their claims in the conclusions section.

What would this mean in the real world anyway?

There is some evidence that links proper immune system development to a reduced likelihood of developing leukaemia as a child and certain genetic faults present at birth can raise the risk substantially, but we are still far from truly understanding all the many different factors at play. It is an important area to be able to understand who is most at risk and what factors can alter their likelihood of developing this disease if we are to prevent some of these cases, but we are not convinced that this new analysis provides strong evidence for a significant role of breastfeeding.

We should also stress that because of the relative rarity of childhood leukaemia, even if the authors’ claims were true, it would still affect only a tiny number of children. And it would still only affect children who are already at risk because of key genetic faults that occur in the womb.

There is a lot of pressure on new mothers, some mothers cannot breastfeed and many factors can affect how long new mothers can breastfeed for. Parents of children who develop leukaemia can also feel a lot of guilt, even though we know some children will unfortunately get the disease whether they’ve been breastfed or not.

Stories based on problematic research do not help anyone.

  • For some expert opinion, including our Research Director and Prof Eve Roman (an epidemiologist whose work we support), see the Science Media Centre
  • Reference: Amitay & Keinan-Boker (2015). Breastfeeding and Childhood Leukemia Incidence A Meta-analysis and Systematic Review. JAMA Pediatr. 2015;169(6):e151025. doi:10.1001/jamapediatrics.2015.1025

Should babies watch TV?

This question seems to trouble many parents, and can cause a lot of guilt too.

“Will the TV numb my baby’s brain?”

“Are they destined for a sedentary life?”


This is why an interview last week with psychologist Annette Karmiloff-Smith on the BBC’s The Life Scientific caught my ear (thanks to a pointer from mum-in-law, Jenny). It’s a fascinating insight into how babies learn to learn, and how their brains develop to understand the world around them. You can listen here: The Life Scientific.

But on TV watching, Prof Karmiloff-Smith, an expert in developmental disorders, argues that if the subject matter of the programme is carefully chosen and scientifically based, then the TV can be better for a child’s learning than even a book.

This was largely in response to advice reissued by the American Academy of Pediatrics (AAP) that babies under two shouldn’t watch any TV or DVDs. There are three main concerns: poorer language skills, a negative effect on sleep, and less time spent taking part in other types of unstructured play that are critical for the proper development of mental capabilities.

This is based on a growing body of scientific research. TV/DVD watching is common: in the US at least, by two years old over 90% of children regularly watch TV, spending an average of 1-1.5 hrs a day in front of the box. Very young babies (under 1.5 years old) cannot, however, really understand TV programmes, and are instead mainly attracted by obvious changes like applause or visual surprises.

Children learn new words or actions better when an adult is teaching it to them live, rather than via a television screen, and the worry is that parents talk to their kids less when the TV is on. And a growing number of studies suggest that children who spend longer watching TV/DVDs have delayed language development, at least in the short-term, and may also develop a worse attention span.

A child’s play may also be hindered by the distraction of a TV that’s on in the background, so the AAP advise to turn it off altogether. Many parents also use TV/DVDs as a sleep aid, but there is evidence that bedtime viewing may lead to more disturbed and shorter sleep.

Karmiloff-Smith, on the other hand, argues that we live in a media saturated world and it’s unrealistic to expect parents to shut down all media use. This view has support from some of the evidence cited in AAP report itself. Despite the original recommendation in 1999 that parents should be discouraged from letting their babies watch TV/DVDs, over 90% of them in the US currently do so by the time their child is two years old. What’s more, the average age that TV is introduced is 9 months, so the advice is clearly not striking a loud enough chord.

From my experience, I can certainly appreciate this. The AAP report says that many parents use the TV so that they can have a shower or cook dinner. Absolutely! Even these seemingly mundane activities can feel like an exercise in military-like efficiency when you’re looking after a child. A 10-minute respite when they’re quiet and content gazing at a TV or prodding an iPad can be just too tempting.

It’s also interesting to consider that throughout history many new technologies have been treated with caution. Dr Vaughan Bell, a psychologist based at King’s College London, has highlighted how the printing press, popularisation of the radio, and now the Internet have been damned for ruining kids’ brains.

Karmiloff-Smith goes on to say that, rather than banning TV for babies, TV programmes just need to be made better and based on science developments. For instance, the visual system is attracted by movement, but most kids’ TV programmes have their focus on the centre of screen. Instead, objects and features that come in from the sides, move across screen and encourage the child to interact promotes the active participation that’s good for mental development. For very young babies, moving image media may even have advantages over books, which are static and whose main attraction is the rustling of the pages.

The caveat in this is that Karmiloff-Smith reveals herself to be a scientific consultant to a DVD company that is designing such programmes. This could cause suspicion of a financial conflict of interest. But her honesty and gusto make me suspect that she became a consultant so that she could promote these ideas, rather than the other way around.

She finished the interview by emphasising that parents still need to interact with their children and the TV shouldn’t be used as a babysitter. But we should think more carefully about which types of media can stimulate the visual and auditory systems, so as to help train the attention and memory systems early.

I’ve written before about the various kinds of programmes and the various contexts in which kids can watch TV, which may have different effects on child development. And some of the evidence cited in the AAP report highlights these complexities. The effects on children’s attention, for instance, seem to depend on the programme content and style, with problems seen not when the content is deemed educational but only when it’s geared towards entertainment. And when a parent watches a programme with an infant and talks them through it, the child tends to become more attentive and responsive. The AAP report also points to evidence that watching Sesame Street can have a negative effect on expressive language in children under two. But the same study showed that watching other programmes, such as the North American-based shows Dora the Explorer, Blue’s Clues, Arthur, Clifford, or Dragon Tales, was associated with greater vocabularies and higher expressive language scores. So it appears that not all ‘screen time’ is equal.

The AAP report seems to fall into the trap of treating all TV and DVD viewing as the same:

For the purposes of this policy statement, the term “media” refers to television programs, prerecorded videos, Web-based programming, and DVDs viewed on either traditional or new screen technologies.

Another major limitation of the AAP report is that all of the cited studies are, by necessity, observational. These investigations are good at highlighting whether two factors are associated with each other, but they cannot tell you whether one causes the other. As the report itself asks, are children with poor language skills simply placed in front of the TV more? Are children with shorter attention spans more attracted to screens? Are parents who are less attentive on the whole, more prone to resort to screen time? If so, then turning the TV off would not necessarily lead to more parent-child interactions.

And some results are just contradictory. One study in the US showed that when the mother’s educational status and household income were taken out of the equation, the association between TV viewing and poor language development disappeared. This appears to have been glossed over by the AAP.

So how do I answer my original question?

The AAP are right to caution against a lot of TV for under twos (over four hours a day, say), as this is when the damaging effects are really apparent. But Karmiloff-Smith is also right to say it’s unrealistic to expect no TV at all, and that the right programme in the right environment is fine and potentially beneficial.

And I’ll leave you with this quote in Time from Dr Dimitri Christakis, a paediatrician at Seattle Children’s Hospital:

Ask yourself why you’re having your baby watch TV. If you absolutely need a break to take a shower or make dinner, then the risks are quite low. But if you are doing it because you think it’s actually good for your child’s brain, then you need to rethink that, because there is no evidence of benefit and certainly a risk of harm at high viewing levels.

“It must be his teeth”: teething symptom myths

I wrote some time ago about how our little one was teething and the measures we’d taken to alleviate his discomfort. Well, this little chapter is still in full swing, with a mere two teeth erupted so far. But this isn’t a place to write long diary entries detailing my child’s progress; instead, I want to bring some science to light.

I’ve heard a number of physical signs regularly attributed to teething, such as rosy cheeks, diarrhoea, green faeces, fever, gnawing, irritability and drooling to name a few. It seems these beliefs are fairly common, as surveys of parents and health professionals have shown.

It gets interesting, however, when you look at some of the research done in this area. There seems to be some disagreement over which symptoms are actually caused by teeth eruption, and whilst it’s clear that individual babies will display different teething signs, it’s also apparent that some beliefs are not borne out by the evidence.

One of the difficulties is that the onset of teething (6-12 months) often occurs around the same time that babies become particularly susceptible to a variety of infections and upsets. This is mainly due to the decline in immunity imparted by maternal antibodies, as well as changes in behaviours that see infants actively interact with their environment.

There certainly seems a fair amount of evidence that many signs assumed to be caused by teething are actually caused by something else, such as meningitis, bacterial infection and herpes simplex virus infection. As the NHS Clinical Knowledge Summary bluntly puts it: “Teething does not cause children to become systemically unwell”. 

Another problem is our familiar foe: limited experimental design. Studies in this area are rather limited in number, often rely on self-reporting rather than objective measurement, deal with correlations not causations, and many look back at clinical data rather than tracking babies as they develop. These complications may be exacerbated by the fact that many health professionals hold erroneous beliefs too, which influences the data collected.

So what do we know?

Symptoms often misattributed teething:

  • Diarrhoea: This is one of the most common symptoms attributed to teething, but no solid data exist to suggest this is due to teething in the majority of cases. It has been tentatively suggested that slightly looser stools may occur during teething and this could lead to mild nappy rash.
  • Fever: Teething may cause a small rise in body temperature, but a feverish temperature above 38°C is unlikely to be due to teething.
  • Runny nose: The jury’s out on this one, but the reported associations are weak and this symptom is more likely to be a due to a wider problem.
  • Wakefulness: While teething may cause some disruption of sleep, this is probably over-exaggerated by parents, and may be partly down to changing sleep patterns and the formation of attention-seeking habits. I was also told by a midwife that teeth move more during the night, causing greater wakefulness – I initially thought not, as it seems more reasonable that the distress is apparent when there isn’t anything to distract the infant. Any data on this latter point would be gratefully received!
  • Green faeces: This is one I’ve heard a number of times, sometimes with an explanation of a change in the stomach acid balance. I can’t find anything in the literature (but please come forth and proffer!) but, on the face of it, it doesn’t seem to chime with idea that teeth eruption does not cause systemic upset.

 Symptoms more likely to be caused by teething:

  • Drooling: Excessive saliva can form and this may be seen by an infant dribbling more than usual. This isn’t conclusive, though, as salivary glands become active around 2–3 months of age and constant drooling can be expected then.
  • Gnawing: Teething infants may gnaw on cold, hard objects or on their fingers to temporarily help with teething discomfort.
  • Mild irritability: The pain associated with teething, which is mostly associated with an inflammatory response within the gums, might cause grizzliness, disturbed sleep, ear rubbing and a decreased appetite.
  • Rosy cheeks: This is some support in the literature that flushed, red cheeks are associated with teething, although this is not clear-cut.

As I said earlier, different babies will experience different symptoms and it’s worth bearing in mind that no single symptom can definitively ‘diagnose’ teething.

Many of these beliefs appear to have some root in history as far back as Hippocrates in the 4th century, when teething was thought to be a deadly disease (“dentition difficilis”, Latin for ‘difficult teeth’). Teeth eruption, it was believed, caused a disturbance in the infant’s nervous system, leading to severe systemic upset. This was, again, most probably due to coincidental timing of onset of teeth eruption and an increased likelihood of serious infection, which in those days often led to infant death. Worse still, many of the treatments used for teething right up until the 19th century were actually toxic, such as opiates, lead, mercury salts, bromide and salt.

So while some studies have suggested that teeth eruption is associated with certain physical symptoms, none has really been able to establish causal relationships, and others have found no associations at all. It may be that a diagnosis of teething can relieve parents’ anxieties about an upset infant. Understanding true causal factors, however, is important to prevent misdiagnosis – attributing symptoms to teething could miss more serious conditions that require immediate medical attention.

P.S. I was hampered in my access to a lot of the research in this area, and so if I’ve missed anything then please let me know. A good argument to support Open Access publishing!

Is breastfeeding all it’s cracked up to be?

Zoe Williams wrote an intriguing article in The Guardian that challenged some widely held assumptions and beliefs about the benefits of breastfeeding (‘The backlash against breastfeeding’). It did, as you might imagine, generate strong reactions from those who vehemently espouse that “breast is best”. It was written on the back of a recent front cover of Time magazine that caused an even greater furore (‘Are you mom enough’), which championed a form of ‘attachment parenting’ that encouraged breastfeeding until the child is well over one year old.

I did feel that Zoe Williams’ article raised some interesting points, some of which I wanted to address here.

But I also had the strong suspicion that the claims made for [breastfeeding’s] benefits – the higher IQ, the protection against obesity, the superior bonding, the warding off of disease both now and for ever, both for baby and for mother – were mostly bogus. A lot of the reasoning seemed syllogistic (babies born into low-income families end up fatter; low-income mothers breastfeed less than high-income mothers; therefore breastfeeding prevents obesity) or frankly lame.

This relates to the concept I’ve written about before, that correlation does not imply causation. As she quotes Joan B Wolf later in the article: “Breastfeeding cannot be distinguished from the decision to breastfeed, which could represent a more comprehensive commitment to healthy living.”  

While this is true, that is not to say that the researchers who conduct these types of observational studies are not aware of this limitation. It is hard to do interventionist, controlled studies, as you would do in a lab or in randomised controlled trials to test a new medicine. Imagine randomly assigning new mothers to one of two groups – breastfeeding or formula-feeding – and getting them to stick to this regimen so that you can measure the outcomes of the children. Not likely, which is why such interventionist studies have been limited (1 & 2).

So researchers work with the tools available to them. They try to include appropriate comparator groups and make statistical adjustments to account for potential confounding variables. Not perfect by any stretch, but the best available methods.

I knew a lot of mothers who formula fed; they didn’t seem to love their babies less.

Ignoring the submission to anecdote, which is a mortal sin to evidence-based thinkers, a more worthwhile point to address is an argument I’ve often heard that goes approximately: “Well, they say breastfeeding is best, but I formula-fed and my child turned out alright”.

There are two key flaws in this reasoning. One is that by advocating breastfeeding, it is implying that formula-feeding is bad. No. It’s just there’s less chance that any outcomes associated with breastfeeding will also be seen with formula-feeding. The second misconception is that there’s a complete separation, or dichotomy, between the two groups. But here we’re comparing averages of large study groups – there will be huge amounts of variability around the averages and a lot of overlap, and only the right statistical analyses can reveal any true differences.

Put simply, there will be mothers of formula-fed babies who form strong attachments (to carry on the example in the article) and mothers of breastfed babies who do not.

By analogy, there will be some people that do all the ‘right’ things – not smoke, eat healthily, exercise, apply sun cream – who will get cancer and there will those who smoke 40-a-day and drink to excess who do not. It doesn’t disprove the link, it’s all about stacking the odds in your favour.

(I should clarify that the evidence I’ve seen for superior bonding associated with breastfeeding is somewhat weak, but I referred to it here to address a more general point about the common use anecdotal evidence).

This struck me as a bizarre place to have arrived at; where even to talk about the evidence behind the benefits of one type of infant feeding over another is heretical.

Absolutely correct. We should constantly gather and scrutinise data, and if strong enough evidence emerges that contradicts our prior assertions, then we should be prepared to change our minds. I applaud the fact that this article was written, even if there are bits I wouldn’t necessarily or wholeheartedly go along with, as we should always be ‘talking about the evidence’.

The questioning of the orthodoxy is taken as a direct attack on babies.

[Charlotte Faircloth, sociologist]: “… Everything has got very heated, and very moralised. How you feed your kids is no longer a personal decision. There’s this idea that you can breastfeed your way out of poverty, or if you don’t breastfeed your kid’s going to be fat or have a low IQ…” She pauses. “It’s all got a bit out of hand.”

Williams is right that some breastfeeding advocates are evangelical and overly ideological about breast milk, to the point that some mothers feel over-bearing pressure to breastfeed. As Charlotte Fairclough is quoted, it’s a personal decision. Fairclough also touches on my earlier point that there’s no binary distinction in the outcomes of children who are breastfed and those who are not – breastfeeding doesn’t determine the outcome but may make it more probable (and, if it does, it will certainly still only be one factor amongst many).

Williams then quotes French feminist Elisabeth Badinter:

“when it’s recommended that you breastfeed your child for one year – six months exclusively, with nothing else, day and night, on demand – there are obviously consequences for a couple.”

“There are women for whom breastfeeding is a true pleasure. It’s very good for them and it’s very good for the baby. But to breastfeed a baby if the mother herself doesn’t like it? It’s a catastrophe. The decision to breastfeed is an intimate and private decision. No one should be able to interfere.”

In my mind, this touches on something I have written about before – that to lay down a hard-and-fast prescription for parents to follow is not always straightforward or productive. A more nuanced approach is often required that takes into account the familial circumstances and considers the parents’ well-being as well. It would be no good hectoring parents to follow a particular path, if it leads to a significant worsening of their own physical or mental health, especially if the knock-on effect is a deterioration in family relationships.

It seems highly unlikely that this would allow people to be good parents nor provide a decent environment for the child (a brief look at the scientific literature suggests this assumption isn’t entirely unreasonable, e.g. thisthis and this). It also risks attaching blame to a parent for any perceived character flaw in their child, and guilt-tripping parents in this way seems a counter-productive approach.

The research about ear infections, respiratory disease and diabetes is very mixed. Neither fussiness around new foods nor constipation are classic or very salient markers of good health. The obesity studies are debatable. The weight loss of breastfeeding mothers is taken from a WHO report, based on two interventionist studies in Honduras. It’s nothing like that straightforward: it does use 500 calories a day, but if your appetite increases at the same time, it is amazing how quickly and easily you can pop 500 calories into your mouth. Meanwhile, Wolf points out, no study on maternal cancer has “distinguished the effects of breastfeeding from the behaviour of women who breastfeed”

Again, I agree, it’s not straightforward! But I would, perhaps, strike a more positive tone.

The studies aren’t perfect (Williams also quotes the American Academy of Paediatrics: ‘There are a lot of methodological problems’), but that’s no reason to just give up on evidence altogether. The best available evidence, as far as I can see, is saying that breastfeeding is likely to bring a range of benefits to the child, and possibly some to the mother. A Cochrane Review from 2009 includes many citations that support various benefits of breastfeeding, as well as a discussion of some of the uncertainties.

I also agree that the weight loss data must be taken in context, such as whether you are dealing with developed or developing countries, and some effects, such as iron status, can be achieved through supplementation. So, in some instances, there’s no need to be over-zealous about breastfeeding, as long as adequate healthcare advice is in place.

“The notion of risk [has been] transformed from a dichotomy to a continuum.” It’s no longer a case of “safe” versus “dangerous”; rather, everything carries some risk and you announce your fitness as a parent to the world by interpreting hazards in the most credulous, fervent way. The onus isn’t on the researcher to prove the point any more – the onus is on the parent, or parent-to-be, to prove that they’ll believe the researcher.

I’m not sure this is true and seems an over-generalisation. There’s a huge variety in parental attitudes and some are incredibly questioning, and whether and how this has changed in recent years is uncertain without any data. I think it misrepresents the role of a researcher, too – that the onus is on them to ‘prove a point’. Scientists don’t (or shouldn’t) undertake work to prove already held beliefs, rather they formulate hypotheses and design experiments to support or reject those theories.

There are lots of other points in and around the ones I’ve highlighted, a lot of which veer from my intention to address evidence-based decisions, but I do encourage you to read the whole article for some interesting discussion of feminism and social policy.

I’ll summarise by saying that I would still state that breastfeeding can potentially bring a range of benefits over formula-feeding, because that is where the balance of scientific evidence is, with little or no contrary evidence to dissuade me yet. And this should be reflected in the healthcare advice that is provided. I absolutely agree, though, that foisting this decision on mothers is, at best, unnecessary and, at worst, damaging. Healthcare advice must take context into account.

On a more general point, I welcome any sensible, critical look at the evidence behind claims, as Zoe Williams has done. It’s certainly made me re-evaluate – and in some instances, such as in the link to lowered risk of maternal breast cancer, made me temper – my own standpoint on the benefits of breastfeeding.

A final point, though. It was disappointing that an article that relied so heavily on rejecting or challenging some of the evidence that is used to promote the “breast is best” standpoint, did not once link to any primary research. A few quotes here and there from, no doubt, serious academics, and a link to a book, but nothing that could point the reader to the original research studies.

It is perennial complaint that evidence-based advocates level at popular journalists, so I don’t want to single out this author for special criticism. But it is something that could do with a cultural shift and, perhaps, mainstream commentators will start to follow the examples of their science colleagues.


Hat-tip to Mum-in-law, Jenny, for The Guardian article.

Demanding babies and fraught mothers

There was quite a bit of coverage at the end last week about a report published by Essex and Oxford-based researchers that looked at the effects of on-demand feeding versus scheduled feeding. The researchers analysed just over 10,000 thousand babies born in the 1990s and checked whether babies fed in a particular way at 8 weeks and 33 months achieved better academic results later in their lives and whether the well-being of the mothers was affected.

The authors of the study reported an association between being fed on-demand as a baby and higher IQ scores at 8 years and better SATs exam scores at 5, 7, 11 and 14 years. This held true for breastfed and bottle-fed babies. Another effect they uncovered was that mothers who fed on-demand reported lower confidence, higher fretfulness and worse sleep patterns.

Photo by Anton Nossik*

Now I’ve mentioned quite a few times the mantra that “correlation does not imply causation”, and so my immediate reaction to seeing this in The Guardian was ‘here we go again’. Journalists often overlook the limitations of correlative studies in favour of neat and tidy story – after all, “this behaviour causes this effect” has a bit more punch than “this behaviour is associated with this outcome, but it may be acting through a third or multiple independent variable(s)”.

For an entertaining and revealing example of this issue, I would recommend reading stand-up mathematician Matt Parker’s ‘mobile phone masts cause increase in birth rates’ hoax story.

My first thought when my wife told me about this story was that, rather than on-demand feeding directly causing the higher IQ scores, it could equally be that a third factor was at play. As a speculative example, it could be that an attentive personality primes a mother to feed on-demand and also, independently, to spend more time with their child on educational development. This is an important distinction because it would mean that mothers wouldn’t be able to improve their child’s academic chances by feeding on-demand, rather they would have to pay more attention to educational development.

The authors were well aware of possible confounders and tried to take into account other differences between the groups to narrow down the possibilities as much as possible. This involved adjusting for family income, the education level of the parents, different parenting styles, the age and sex of children, and general maternal health. After all this, there was still a significant difference – children who were fed on-demand as babies had, on average, IQ scores 4 points better than children who were fed to a schedule.

This study has an advantage over many other correlative studies in that a third group presented itself – mothers who wanted to feed to a schedule but were unable to and so fed on-demand, i.e. they wanted to be in one group (scheduled feeding) but ended up as if they were in the other group (on-demand feeding). Crucially, the children in this third group achieved the same elevated academic achievement as seen in the other on-demand group.

Photo by Tom Carmony**

This meant that the authors were able to suggest that the improved academic outcomes weren’t due to having the type of mother who wishes to feed to a schedule, but that they were due to the actual act of being fed on-demand. This relies on an assumption that the mothers in the scheduled feeding group and the attempted scheduled feeding group had similar characteristics, and would therefore have similar approaches in other ways that may affect educational development. This may not be entirely true as other, more subtle differences could exist between these groups, but it’s a tantalising result that merits further investigation. I guess measuring the amount of time spent with the child on learning and development would be something at which to look.

There will still be other factors at play and it is still essentially a correlation, but it starts to narrow down the possibilites. The authors themselves admit the limitations in their study, with the lead author Dr Maria Iacovou telling Science Daily:

“At this stage, we must be very cautious about claiming a causal link between feeding patterns and IQ. We cannot definitively say why these differences occur, although we do have a range of hypotheses. This is the first study to explore this area and more research is needed to understand the processes involved.”

Dr Iacovou was also fairly even-handed on Channel 4 news, rightly highlighting that we still don’t know the mechanism for the observed effects and this is really the first step in establishing a causal relationship.

The final thing to note about the study, is that it was gratifying that the researchers looked at both sides of the issue – the outcomes for the baby (IQ, academic achievement) and the mother (well-being indicators). I’ve written before about the importance of looking at all knock-on effects, as it can be counter-productive to focus solely on the baby if it means a serious detrimental effect on the mother’s mental health.

If on-demand feeding really does drive academic development, then implementing a healthcare strategy to encourage this must take into account the negative impact this may have on the mother and include an appropriate support mechanism. Otherwise, the effectiveness of the intervention will be limited, as mothers will be more likely to give up in favour of a reassuring routine.

As Dr Iacovou said, “mothers are people too”.


Hat-tip to mum-in-law Jenny who sent me this story.

*CC licence: http://creativecommons.org/licenses/by/3.0/deed.en

**CC licence: http://creativecommons.org/licenses/by-nc-nd/2.0/deed.en_GB

My son, the enigma machine

Little Roo has been babbling for some time now. Until now, it has been the fairly standard oohs, aahs, yelps and giggles. But the other day, we heard something that sounded distinctly like “Dada”. The joy! The amazement! He can talk and he knows who I am!


Or does he?

I couldn’t help think of the entertaining talk by science communicator, author and libel crusader, Simon Singh, that I’ve seen a few times at various events. The story, in its essence, goes something like this: within the Hebrew text of the Bible (the ‘Torah’) lie many coded messages that foretell future events. All one needs to do is remove all punctuation and spacing to leave a single string of thousands of letters, and the predictions pop out like a wordsearch puzzle. Using this technique, the Bible successfully anticipated the assassination of JFK by Lee Harvey Oswald (“dallas” “president kennedy” “oswald”), Hitler’s evil intent (“hitler” “evil man” “nazi”) and “Newton”‘s theory of “gravity”. This theory was popularised by Michael Drosnin and his ‘Bible Code’ book – plus TWO follow-ups – which have supposedly sold millions of copies worldwide.

Then those pesky scientists, who like to destroy a nice story with ‘evidence’, came along to put the cat amongst the pigeons. One of them, Brendan McKay, a mathematician and computer geek, thought that if there was an array of thousands of letters, the chances of randomly coming up with some relevant words was well within reason, especially if you allow reading forwards, backwards, diagonally, skipping rows, etc. (as well as variations in Hebrew spelling). So he was challenged to find the same sort of predictions in a non-sacred text. He devised an algorithm – a series of calculation steps  – to mine the mystical and secret-containing book, Moby Dick.

It turns out that, using the same technique, Moby Dick predicts Leon Trotsky’s murder (“trotsky” “ice” “hammer”), Princess Diana’s death (“princess” “wales” “lady diana” “henri” “paul” “dodi”) and the 9/11 terrorist attacks (“world” “trade” and “center” “twin” crossed with “building” “Osama”). I also came across Dave Thomas’ work, which does a pretty good job at finding secret messages in a U.S. Supreme Court document (“hitler” “nazi” and so on), in Drosnin’s Bible Code itself (“vain” “hoax” “megalo-mania”), as well as predicting anti-Judeo-Christian messages in the Bible. These findings debunk the sanctity of the Bible and tell us that coincidences, no matter how spooky, are just that – coincidences.

There’s a video by Singh to explain it more graphically (the first 4 minutes are about the Bible Code):

Vodpod videos no longer available.

from 5×15

To take it back to Roo’s babble, most of time it sounds like this:


And this supposed random exploration of sounds and syllables is a part of how we develop the ability of speech and language. Babies move from gurgling vowel sounds at 2-3 months of age to add in consonants at 6-7 months, much like Roo is doing in the clip above. At this stage he’s as likely to say “gaga” or “baba” as he is to say “dada”, and we wouldn’t assume he’d developed an astoundingly early fascination with American songstress, Lady Gaga, or rapper of science – yes, a science rapper – Baba Brinkman.

From 6-12 months, Roo will continue to experiment with this vocal play, and it won’t be until he’s over a year old that sounds will start to carry meaning. For instance, he might make the sound a car makes or utter monosyllabic words in place of a more complicated phrase, and he will also start to understand the meaning of sounds he cannot yet say. At 18-24 months, he should then be at the stage to start using multisyllabic words in short sentences. There’s a correlation between early babbling and linguistic ability later on, as a review by Carol Stoel-Gammon explains:

…infants who produced more in the prelinguistic period (i.e. more vocalizations at age [3 months]; more CV syllables at age [1 year]) had superior performance on subsequent speech and language measures during childhood.

So, my initial sound clip, of Roo saying “dada”, is more likely to be a version of the Bible Code – a chance coming together of noises, among a cacophony of vocal explorations, to form a sound that is recognisable to us. Our temptation to believe that he’s forming meaningful speech seems to me to be a form of confirmation bias – the way in which we filter out the negatives and remember only the positives, thereby confirming our already-held beliefs. This is especially true if we retrofit the associations – there are many phrases that we could associate with JFK’s assassination or words that could be formed from two or three simple syllables, but we need only a couple to convince ourselves that ‘there’s something in it’.

We are tempted to add significance to coincidences all the time. When something relevant happens on a special day or when a psychic stumbles upon a personal fact (often in a room of hundreds of people), it’s often said that it’s “too spooky to be just coincidence, don’t you think?”. Practitioners of pseudoscience, like psychics and astrologers use this – intentionally or unintentionally – to convince readers that they have hit the nail on the head, when in actual fact they threw many nails in the vague direction of the person in the hope that one will strike its target.

Natural’s not in it: the problem with homeopathy for babies

I never thought I’d end up writing about homeopathy for babies, but some things just take you by surprise.

For those that are unaware, a homeopathic preparation starts by taking a substance, usually one that would cause an ailment, and dilute it down to such a degree that none of the original substance remains. The belief is that this dilution can then be used to treat the ailment and that the more diluted the preparation, the more effective the ‘remedy’ is. To give an idea of the level of dilution of most standard homeopathic preparations, the Merseyside Skeptics made ‘homeopathic vodka’ and tested it on a few willing volunteers.

The arguments against homeopathy have been made effectively elsewhere, so I won’t re-tread those well-articulated paths too heavily, but will sum it up briefly. To support homeopathy, proponents usually either ‘cherry-pick’ flimsy, uncorroborated evidence to try and prove efficacy, suggest that a placebo effect is still a positive effect (and so what’s the problem?), or simply argue that everyone has a choice to decide what treatments they use. The primary problem for me (for it’s a problem among many) is that patients replace or delay conventional treatment in favour of alternative treatment, often at a serious detriment to their health. This is exacerbated by the decision to make homeopathic treatments available on the NHS – justified by the Government with the patient choice argument – that lends validity to the practice in many people’s minds. The Science and Technology Committee, however, conclude unequivocally that it’s not valid.

So why do I bring this up here?

Well, we’re fairly sure that my little boy’s teething at the moment. This can often be a pretty painful process, as anyone who can remember those first adult teeth poking through or who have been unfortunate enough to gain some wisdom teeth. We would, of course, like to reduce Reuben’s discomfort as much as possible, and so it is with this in mind that my wife bought some teething granules, on the recommendation of some her friends who swear by this particular brand.

Now imagine my surprise when I whipped out the box, in anticipation of riding to my son’s rescue and alleviate his pain, only to discover that these were homeopathic teething granules. First, the surprise that these even exist; and second, the puzzlement that my wife, knowing my somewhat sceptical nature, had actually bought them for our son. On the second point, she assured me that she didn’t realise they were homeopathic (this fact is revealed only on the back of the packet) and was going only on the testimonials of her friends (common ‘evidence’ homeopaths produce). So I’ll give her the benefit of the doubt on that score.

But the first point, that homeopathic pain relief exists for babies, has been troubling me since. The preparation is a 6C dilution of Chamomilla (or camomile). This means that it has been diluted 10-12, or 0.000000000001 of the original substance. In the ‘homeopathic vodka’ preparation, this would have been reached by the 6th cup of water. So a pretty extreme dilution. Leaving aside the pro-homeopathic argument that camomile doesn’t cause teething (‘like with like’ theory), there’s no evidence of this substance’s pain relief qualities (as advertised) nor, as far as I could tell, of its often assumed calming properties (also this). So at a 10-12 dilution, it seems extraordinary that there would be enough active molecules to have an effect (unless one subscribes to the ‘weaker makes it stronger’ argument).

As soon I saw that this was a homeopathic treatment, I convinced my wife that we needed to buy a proper teething pain relief. We bought some teething gel, with some well-tested analgesic and antiseptic compounds in it. And this, to me, demonstrates the crux of the issue: if we’d persisted with the homeopathic treatment, then we would have delayed using the more reliable teething gel and could have caused our son a few nights of needlessly heightened pain.

I guess it may have also subsided independently of the homeopathic treatment and we would now be telling our friends about this wonderful, magical treatment for teething. And the bandwagon would roll on.