Friday 25 November 2011

The weird world of US ethics regulation

There has been a lot of interest over the past week in the Burzynski Clinic, a US organisation that offers unorthodox treatment to those with cancer. To get up to speed on the backstory see this blogpost by Josephine Jones.
As someone who spends more of my time than I’d like grappling with research ethics committees, there was one aspect of this story that surprised me. According to this blogpost, the clinic is not allowed to offer medical treatment, but is allowed to recruit patients to take part in clinical trials. But this is expensive for participants. The Observer piece that started all the uproar this week described how a family needed to raise £200,000 so that their very sick little girl could undergo Burzynski’s treatment.
I had assumed that this trial hadn’t undergone ethical scrutiny, because I could not see how any committee could agree that it was ethical to charge someone enormous sums of money to take part in a research project in which there was no guarantee of benefit. I suspect that many people would pay up if they felt they’d exhausted all other options. But this doesn’t mean it’s right.
I was surprised, then, to discover that the Burzynski trial had undergone review by an Institutional Review Board (IRB - the US term for an ethics committee). A letter describing the FDA’s review of the relevant IRB is available on the web. It concludes that “the IRB did not adhere to the applicable statutory requirements and FDA regulations governing the protection of human subjects.”  There’s a detailed exposition of the failings of the Burzynski Institute IRB, but no mention of fees charged to patients. So I followed a few more links and came to a US government site that described regulatory guidelines for ethics committees, which had a specific section on Charging for Investigational Products. It seems the practice of passing on research costs to research participants is allowed in the US system.
There has been considerable debate in academic circles about the opposite situation, where participants are paid to take part in a study. I know of cases where such payments have been prohibited by an ethics committee on the grounds that they provide ‘inducement’, which is generally regarded as a Bad Thing, though there are convincing counterarguments. But I am having difficulty in tracking down any literature at all on the ethics of requiring participants to pay a fee to take part in research. Presumably this is a much rarer circumstance than cases where participants are paid, because in general people need persuading to take part in research. The only people who are likely to pay large sums to be a research participant are those who are in a vulnerable state, feeling they have nothing to lose. But these are the very people who need protection by ethics committees because it’s all too easy for unscrupulous operators to exploit their desperation. Anyone who doesn’t have approval to charge for a medical treatment could just redescribe their activities as a clinical trial and bypass regulatory controls. Surely this cannot be right.

Saturday 19 November 2011

Your Twitter Profile: The Importance of Not Being Earnest

I’m always fascinated by the profiles of people who follow me on Twitter. One of the things I love about Twitter is its ability to link me up with people who I’d never otherwise encounter. It’s great when I find someone from the other side of the world who’s interested in the same things as me. There are, of course, also those who just want to promote their product, and others, like Faringdon Motor Parts and Moaning Myrtle (@toiletmoans) whose interests in my tweets are, frankly, puzzling. But the ones that intrigue me most are the ones with profiles that create an immediate negative impression - or to put it more bluntly, make me just think "Pillock!" (If you need to look that up, you’re not from Essex).
Now language is one of my things - I work on language disorders, and over the years I’ve learned a bit about sociolinguistics - the influence of culture on language use. And that made me realise there were at least two hypotheses that could explain the occasional occurrence of offputting profiles. The first was that I am being followed by genuine pillocks. But the other was that there are cultural differences in what is regarded as an acceptable way of presenting yourself to the world. Maybe a turn of phrase that makes me think "pillock" would make someone else think "cool". And perhaps this is culturally determined.
So what, to my British ear, sets off the pillock detector? The major factor was self-aggrandisement. For instance, someone who describes themselves as "a top intellectual", "highly successful", "award-winning", or "inspirational".
But could this just be a US/UK difference? The British have a total horror of appearing boastful: the basic attitude is that if you are clever/witty/beautiful you should not need to tell people - it should be obvious. Someone who tells you how great they are is transgressing cultural norms. Either they really are great, in which case they are up themselves, as we say in Ilford, or they aren’t, in which case they are a dickhead. When I see a profile that says that someone is "interested in everything, knows nothing", "a lazy pedant", or "procrastinaor extraordinaire", I think of them as a decent sort, and I can be pretty sure they are a Brit. But can this go too far? Many Brits are so anxious to avoid being seen as immodest that they present themselves with a degree of self-deprecation that can be confused by outsiders with false modesty at best, or neurotic depression at worst.
A secondary factor that sets off my negative reactions is syrupy sentiment, as evidenced in phrases such as: "empowering others", "Living my dream", or "I want to share my love". This kind of thing is generally disliked by Brits. I suspect there are two reasons for this. First, in the UK, displays of emotion are usually muted, except in major life-threatening circumstances: so much so that when someone is unabashedly emotional they are treated with suspicion and thought to be insincere. And second, Polyannaish enthusiasm is just uncool. The appropriate take on life’s existential problems is an ironic one.
I was pleased to find my informal impressions backed by by social anthropologist Kate Fox, in her informative and witty book "Watching the English" (Hodder & Stoughton, 2004). Humour, she states, is our "default mode", and most English conversations will involve "banter, teasing, irony, understatement, humorous self-deprecation, mockery or just silliness." (p 61). She goes on to describe the Importance of Not Being Earnest rule: "Seriousness is acceptable, solemnity is prohibited. Sincerity is allowed, earnestness is strictly forbidden. Pomposity and self-importance are outlawed." (p. 62). Fox doesn’t explicitly analyse American discourse in the book, but it is revealing that she states: "the kind of hand-on-heart, gushing earnestness and pompous Bible-thumping solemnity favoured by almost all American politicians would never win a single vote in this country - we watch these speeches on our news programmes with a kind of smugly detached amusement." (p 62).
Anthropologists and linguists have analysed trends such as these in spoken discourse, but I wondered whether they could be revealed in the attenuated context of a Twitter profile. So in an idle moment (well, actually when I was supposed to be doing something else I didn’t want to do) I thought I’d try an informal analysis of my Twitter followers to see if these impressions would be borne out by the data. This is easier said than done, as I could find no simple way to download a list of followers, and so I had to be crafty about using "SaveAs" and "Search and Replace" to actually get a list I could paste into Excel, and when I did that, my triumph was short-lived: I found it’d not saved Location information. At this point, my enthusiasm for the project started to wane - and the task I was supposed to be doing was looking ever more attractive. But, having started, I decided to press on and manually enter location for the first 500 followers. (Fortunately I was able to listen to an episode of the News Quiz while doing this. I started to like all those eggs with no Location recorded). I then hid that column so it would not bias me, and coded the profiles for three features: (a) Gender (male/female/corporate/impossible to tell); (b) Self-promotion: my totally subjective rating of whether the profile triggered the pillock-detector; (c) Syrupy: another subjective judgement of whether the profile contained overly sentimental language. I had intended also to code mentions of cats - I was convinced that there was a British tendency to mention cats in one’s profile, but there were far too few to make analysis feasible. I was a victim of confirmation bias. So were my other intuitions correct? Well, yes and no.
For the analysis I just focused on followers from the US and UK. The first thing to emerge from the analysis was that pillocks were rare in both US and UK - rarer than I would have anticipated. I realised that, like mentions of cats, it’s something I had overestimated, probably because it provoked a reaction in me when it occurred. But, I was pleased to see that nonetheless my instincts were correct: there were 7/97 (7.2%) pillocks in the US sample but only 2/153 (1.3%) in the UK . The sample size is really not adequate, and if I were going to seriously devote myself to sociolinguistics I’d plough on to get a much bigger sample size. But nevertheless, for what it’s worth, this is a statistically significant difference (chi square = 5.97, p = .015 if you really want to know). Syrup followed a similar pattern: again it was rare in both samples, but it was coded for 3/153 of the UK sample compared with 7/97 of the US. I’d coded gender as I had thought this might be a confounding factor, but in fact there were no differences between males and females in either pillocks or syrup. Of course, all these conclusions apply only to my followers, who are bound to be an idiosyncratic subset of people.
My conclusion from all this: we need to be more sensitive to cultural differences in self-expression. Looking over some of the profiles that I categorised as "pillock" I realise that I’m being grossly unfair to their owners.  After all, on a Twitter profile, the only information that people have about you comes from the profile - and your tweets. So it really is preposterous for me to react negatively against someone telling me they are an "award-winning author": that should engender my interest and respect. And, because this is a profile, and not a conversation, if they didn’t tell me, I wouldn’t know. And we really ought to cherish rather than mock those who try to bring a bit of love and kindness into the world. But somehow….
I hope that Americans reading this will get some insight into the tortuous mindset of the Brits: if we come across as dysfunctionally insecure losers it’s not that we really are - it’s that we’d rather you thought that of us than that we were boastful.

Sunday 13 November 2011

Vitamins, genes and language

Thiamine chloride  (source: Wikipedia)
In November 2003, a six-month-old boy was admitted to the emergency department of  a children’s hospital in Tel Aviv. He had been vomiting daily for two months, was apathetic, and had not responsed to anti-emetic drugs. The examining doctor noticed something odd about the child’s eye movements and referred him on to the neuro-ophthalmology department. A brain scan failed to detect any tumour. The doctors remembered a case they had seen 18 months earlier, where a 16-year-old girl had presented with episodic vomiting and abnormal eye movements due to vitamin B1 deficiency.  They injected the child with thiamine and saw improvement after 36 hours. The vomiting stopped, and over the next six weeks the eye movements gradually normalised. When followed up 18 months later he was judged to be completely normal.
This was not, however, an isolated case. Other babies in Israel were turning up in emergency departments with similar symptoms. Where thiamine deficiency was promptly recognised and treated, outcomes were generally good, but two children died and others were left with seizures and neurological impairment. But why were they thiamine deficient? All were being fed the same kosher, non-dairy infant formula, but it contained thiamine. Or did it? Analysis of samples by the Israeli Ministry of Health revealed that levels of thiamine in this product were barely detectable, and there was an immediate product recall. The manufacturer confirmed that human error had led to thiamine being omitted when the formula had been altered.
The cases who had been hospitalised were just the tip of the iceberg. Up to 1000 infants had been fed the formula. Most of these children had shown no signs of neurological problems. But a recent study reported in Brain describes a remarkable link between this early thiamine deprivation and later language development. Fattal and colleagues studied 59 children who had been fed thiamine-deficient formula for at least one month before the age of  13 months, but who were regarded as neurologically asymptomatic. Children who had birth complications or hearing loss were excluded. The authors stress that the children were selected purely on the basis of their exposure to the deficient formula, and not according to their language abilities. All were attending regular schools.  A control group of 35 children was selected from the same health centres, matched on age.
Children were given a range of language tests when they were 5 to 7 years of age. These included measures of sentence comprehension, sentence production, sentence repetition and naming. There were dramatic differences between the two groups of children, with the thiamine-deficient group showing deficits in all these tasks. The authors argued that the profile of performance was identical to that seem in children with a diagnosis of specific language impairment (SLI), with specific problems with certain complex grammatical constructions, and normal performance on a test of conceptual understanding that did not involve any language.
Figure 1 An example of a picture pair used in the comprehension task. 
The child is asked to point to the picture that matches a sentence, 
such as ‘Tar’e li et ha-yalda she-ha-isha mecayeret’ 
(Show me the girl that the woman draws). From Fattal et al, 2011.

I have some methodological quibbles with the paper. The authors excluded three control children who did poorly on the syntactic tests because they were outliers - this seems wrong-headed if the aim is to see whether syntactic problems are more common in children with thiamine-deficiency than in those without. The non-language conceptual tests were too easy, with both groups scoring above 95% correct. To convince me that the children had normal abilities they would need to demonstrate no difference between groups on a sensitive test of nonverbal IQ. My own experience of testing children’s grammatical abilities in English is that ability to do tests such as that shown in Figure 1 can be influenced by attention and memory as well as syntactic ability, and so I think we need to rule out other explanations before accepting the linguistic account offered by the authors. I’d also have liked a bit more information about how the control children were recruited, to be certain they were not a ‘supernormal’ group - often a problem with volunteer samples, and something that could have been addressed if a standarized IQ test had been used. But overall, the effects demonstrated by these authors are important, given that there are so few environmental factors known to selectively affect language skills. These results raise a number of questions about children’s language impairments.
The first question that struck me was whether thiamine deficiency might be implicated in other cases outside this rare instance. I have no expertise in this area, but this paper prompted me to seek out other reports. I learned that thiamine deficiency, also known as infantile beriberi, is extremely rare in the developed world, and when it does occur it is usually because an infant is breastfeeding from a mother who is thiamine deficient. It is therefore important to stress that thiamine deficiency is highly unlikely to be implicated in cases of specific language impairment in Western societies. However, a recent paper reported that it is relatively common in Vientiane, Laos, where there are traditional taboos against eating certain foods in the period after giving birth. The researchers suggested that obvious cases with neurological impairments may be the extreme manifestation of a phenomenon that is widespread in milder form. If so, then the Israeli paper suggests that the problem may be even more serious than originally suggested, because there could be longer-term adverse effects on language development in those who are symptom-free in infancy.
The second question concerns the variation in outcomes of thiamine-deficient infants. Why, when several hundred children had been fed the deficient formula, were only some of them severely affected? An obvious possibility is the extent to which infants were fed foods other than the deficient formula. But there may also be genetic differences between children in how efficiently they process thiamine.
This brings us to the third question: could this observed link between thiamine deficiency and language impairment have relevance for genetic studies of language difficulties? Twin and family studies have indicated that specific language impairment is strongly influenced by genes. However, one seldom finds genes that have a major all-or-none effect. Rather, there are genetic risk variants that have a fairly modest and probabilistic impact on language ability.
Robinson Crusoe Island
A recent study by Villanueva et al illustrates this point. They analysed genetic variation in an isolated population on Robinson Crusoe Island, the only inhabited island in the Juan Fernandez Archipelago, 677 km to the west of Chile. At the time of the study there were 633 inhabitants, most of whom were descended from a small number of founder indviduals. This population is of particular interest to geneticists as there is an unusually high rate of specific language impairment.  A genome-wide analysis failed to identify any single major gene that distinguished affected from unaffected individuals. However, there was a small region of chromosome 7 where there genetic structure was statistically different between affected and unaffected cases, and which contained genetic variants that had previously been found linked to language impairments in other samples. One of these, TPK1 is involved in the catalysis of the conversion of thiamine to thiamine pyrophosphate. It must be stressed that the genetic association between a thiamine-related genetic variant and  language impairment is probabilistic and weak, and far more research will be needed to establish whether it is generalises beyond the rare population studied by Villanueva and colleagues. But this observation points the way to a potential mechanism by which a genetic variant could influence language development.
To sum up: the importance of the study by Fattal and colleagues is two-fold. First, it emphasises the extent to which there can be adverse longer-term consequences of thiamine deficiency in children who may not have obvious symptoms, an observation which may assume importance in cultures where there is inadequate nutrition in breast-feeding mothers. Second, it highlights a role of thiamine in early neurodevelopment, which may prove an important clue to neuroscientists and geneticists investigating risks for language impairment.

Fattal I, Friedmann N, & Fattal-Valevski A (2011). The crucial role of thiamine in the development of syntax and lexical retrieval: a study of infantile thiamine deficiency. Brain : a journal of neurology, 134 (Pt 6), 1720-39 PMID: 21558277  

Villanueva P, Newbury DF, Jara L, De Barbieri Z, Mirza G, Palomino HM, Fernández MA, Cazier JB, Monaco AP, & Palomino H (2011). Genome-wide analysis of genetic susceptibility to language impairment in an isolated Chilean population. European journal of human genetics : EJHG, 19 (6), 687-95 PMID: 21248734