In this rolling section, I have written short comments upon some reports of “new research findings” that have appeared in the health or science pages of the media. I am often critical of the way research reports are portrayed in the media as they tend to be over hyped. My commentaries are not researched or referenced but my personal assessment based upon my existing knowledge and experience. Where it is accessible for free then I have read the original paper. In some cases I have refrained from posting comments because I cannot readily access the paper. If I have commented without seeing the full paper then this is stated before the comments. I only regularly read one newspaper and the BBC news web-site so if readers find stories elsewhere that they think I might be able to comment upon then please send a link to firstname.lastname@example.org . I have started also posting new pieces from this section in the main body of the text so that I can give direct URL links to these posts.
Posted on 24/02/2018 A third of UK adults ‘underestimate calorie intake’
This headline appeared on the BBC news web-site (19/02/2018) and related headlines appeared in other media outlets (e.g. The Daily Telegraph ). These headlines were prompted by research from the Office of National Statistics ONS which claim that calorie intake measured objectively using a high-tech method (doubly labelled water, DLW) is much higher than the values estimated from those self-reported in 4-day food diaries used in the National Diet and Nutrition Survey (NDNS). The purpose of the study, which is not discussed here, was to see if the self-reported energy intakes in NDNS food diaries could be corrected for underreporting. The headline values are shown in the table below:
Table 1 Energy intakes derived from self-reporting (4-day food diaries) and their “real” intake using DLW in 197 subjects
|Intake from self-reporting (calories, kcal)||2126||1576|
|“Real intake” from DLW (calories, kcal)||3119||2393|
|Difference as % of self-reported intake||+47%||+52%|
So the calorie intake derived from the DLW in these 200 subjects is about 50% higher than that reported in their food diaries, which is almost 1000 calories extra per day in the men. Only 3 of the subjects recorded slightly higher energy intake in their diaries than was measured by DLW.
Many other groups have reported similar findings (see for examples). This source also reviews evidence that when someone other than the subject is responsible for recording dietary intake (e.g. parents of children) there is generally good agreement between intake as determined by “self” reporting and DLW. If we accept these ONS findings then they have major implications for nutrition-related research and for surveys that are aimed at surveillance of the diets and nutritional status of the population.
Potential problems with this research
The high-tech method used for comparison is known as the doubly labelled water method (DLW). It is an expensive technique and is only suitable for testing relatively small samples; in this case around 200 people had their energy intake measured and compared to their food diary data. Thousands of people produce food diaries every year as part of the rolling NDNS programme so it would be impractical and prohibitively expensive to use this technique on the whole sample. DLW is regarded as the gold standard method of assessing the “real” energy intake. What it actually measures though is the energy expenditure and it is assumed that the subject is in approximate energy balance i.e. not gaining or losing weight during the measurement period. This seems like a sound and reasonable assumption over the group as a whole. Another problem is that the DLW and food diaries were not done at the same time but once again provided that the sampling method used was reasonable then this would not be expected to produce a consistent difference over the sample, although it might affect individual results. Despite these flaws, which are acknowledged by the report’s authors, one should still be able to rely on the overall data summarised in table 1.
There are three possible explanations for the differences seen:
- Most of the sample were successfully dieting and losing weight when recording their diaries – subjects were asked at the outset of the study if they were dieting
- That subjects are not recording a substantial proportion of the foods that they ate during the diary period
- Subjects ate much less than normal during the diary period maybe because they were made more aware of their eating or because this made it easier to complete their diaries.
Whichever of these explanations is true then it means the NDNS food diary does not give a true reflection of what typical subjects would normally eat during a 4 day period. Maybe more than one of these reasons contributes to the big gap between real and self-reported energy intake.
Implications for nutritional surveillance of the population
These data reported by the ONS give no indication what foods are most affected by this underreporting; the method cannot be used for this purpose. If the underreporting is spread across all types of foods then this means that estimates of the number of people with inadequate intakes of vitamins and minerals derived from the NDNS (i.e. number below the LRNI) will be exaggerated and people will generally be consuming more than the food diary data suggest. For many key nutrients there is also biochemical assessment of the subject’s vitamin or mineral status so maybe higher weighting should be given to conclusions from these rather than from intake estimates derived from food diaries.
If, as intuitively seems probable, there is selective omission or temporary avoidance of foods perceived as “naughty” options e.g. alcohol and sugary or fatty snacks and drinks then the real diets of the UK population may be significantly worse in terms of targets for reducing fats, sugar and alcohol than NDNS diary data suggest.
Implications for the likely causes of current high levels of obesity in the UK
Historical data from the old UK National Food Survey and other methods looking at the rate of food usage in the UK (food balance sheets) suggested that between say 1950 and 1990 the average per capita calorie intake in this country declined by at least a quarter. More recent data from NDNS food diaries have also suggested a small but statistically significant decline in average calorie intake between 2008 and 2014 in men but not women. Since the 1950s rates of overweight, obesity and average body mass index have all increased very markedly in adults and also in children. If these historical data are to be trusted then calorie intakes have declined substantially at the same time as the population has got much fatter. The almost inevitable conclusion to be drawn from these data is that the main driver of the big increases in obesity and overweight is a decline in the activity level of the population. Even though people are “apparently” consuming fewer calories than they did in the past, this drop in energy intake has not been enough to compensate for the drop in energy expenditure and these extra calories have been stored as fat in the nation’s adipose tissue. There is clear evidence of a decline in population activity levels not least because of decreased need for physical work in jobs, transport and in completing household and garden chores. There is also evidence of a substantial decline in reported activity as one goes from BMI categories normal, overweight to obese but is this cause and effect or effect and cause? Does inactivity cause weight gain and obesity or does being overweight and obese discourage physical activity?
The historical data referred to earlier might be less susceptible to systematic under recording because it is not based on a subject reporting their own personal food intake.
Implications of these results for observational research
There are hundreds of studies published each year that measure dietary intakes of subjects and correlate this with risk markers (like blood cholesterol or blood pressure), disease incidence or mortality. Many of these studies use methods for assessing dietary intakes that are much less robust than those used in the NDNS sometimes they even crude assessment of past diets. These studies frequently report some association between consumption of a food, group of foods or a nutrient and the risk of some disease or perhaps the overall risk of dying. These associations are usually relatively small and often have relatively weak statistical significance. These associations are then corrected by complex mathematical models for a variety of confounding variables like smoking, alcohol use, body mass index, activity level, economic status, education level and other dietary variables to try to ascertain whether the association is likely to be cause and effect. As I have discussed in other posts, (such as 1 2 and 3 )this process of correction is extremely problematical. Often there is only limited information available about some or all of the likely confounding variables and one is never sure what all of the confounders are. What these data from the ONS now also seem to confirm is that the dietary assessment methods used in these studies may well be giving a poor indication of what subjects are actually habitually eating. How can one ever have enough confidence in the results of most such studies to alter one’s personal dietary choices let alone to give sound public health guidance to the population?
Posted on 16/02/2018 Ultra-Processed foods “linked to cancer”
The headline is taken the BBC new web-site (16/02/2018) and many newspapers (e.g. The Guardian) and other media also carried headlines about eating “ultra-processed foods” being linked to higher cancer risk. These headlines were in response to a paper published in the BMJ by a group of mainly French scientists. In my opinion this paper and the media coverage it has generated epitomizes what is wrong with much of current nutrition research. It uses weak associations from dubious and flawed data to reinforce existing prejudices (prejudices which I partly share).
This study found that people who eat more of what is termed “ultra-processed foods” have a higher overall risk of developing cancer. Whatever their opinion about the merits of “ultra-processed foods” most nutritionists would have predicted this outcome because the people who eat lots of these “ultra-processed foods” are likely to be very different from those who eat little of them. For example those who ate a lot of these foods were more likely to smoke, were less active, ate more calories and were more likely to be taking oral contraceptives. Of course, the authors did collect data about many other dietary, lifestyle and socioeconomic characteristics of the participants and used sophisticated mathematical models to make multiple “corrections” for these and quite a few other so-called confounding variables. There is, however, no statistical magic wand that precisely corrects for all the potential confounders. In several cases, the information on confounding variables is a relatively crude categorisation, for example:
- Categorisation of physical activity levels into low, medium and high
- Categorization of education level into three categories – less than high school degree, less or more than two years after high school degree
- Smokers were just divided into current smokers or never/former smokers.
The less precise the information about confounders is, then the less reliable the correction process is and, of course, one can never be sure that all of the potential confounders have been corrected for. The dietary assessment process is also relatively crude (two or more online 24h recall questionnaires) as dietary assessments of free-living people always are. The sample was a self-selecting one and was not representative of the French population e.g. it was predominantly female (78%) and tended to be better educated, more health conscious, to eat less “ultra-processed foods” and they had substantially lower rates of cancer than the French population as a whole.
The size of the cohort (c105000) makes this sound like an impressively large study but the statistical analysis of course compares rates of cancer in different categories. The overall rate of cancer in the study group was 79 cases per 10,000 people per year c.f. 97 cases per 10,000 per year in the general French population (comparison standardised for age and sex).
Epidemiological studies have been important in the discovery of important links between lifestyle and behavioural factors and disease risk e.g. smoking and lung cancer, sleeping position and cot death, asbestos exposure and the normally rare, lung condition mesothelioma. However, the differences seen between exposed and unexposed in such examples is large; certainly more than doubled and often more than 10 times. When one is dealing with such large differences in risk then it is difficult to believe that such associations are not cause and effect and are an artefact caused by some unexpected or improperly corrected confounding variable. In this paper, the French authors suggest that increased cancer risk is 12% (or between 6 and 18%) for a 10% increase in use of “ultra-processed foods”. Given the cancer rate in this cohort, this amounts to just a few extra cases per 10,000 people per year; it now becomes much more credible, likely even, to suggest that this difference could be explained by the necessarily crude methodology and any flaws in the correction process. They may be measuring the bias in their study.
I would also suggest that even if we accept the findings at face value they are of very limited practical value for public health guidance. The category of “ultra-processed foods” is a very broad and diverse one it includes:
- Mass produced packaged breads and buns
- Sweet or savoury snacks
- Industrialised confectionery and desserts including many dairy products
- Sodas and sweetened drinks including calorie-free versions
- Meat products which have been preserved with anything other than salt e.g. bacon, ham, sausages etc.
- Instant noodles and soups
- “Ultra-processed fruits and vegetables”
- Ready meals including chilled and frozen.
They list a large range of processes and additions which would make any food “ultra-processed”.
The authors suggest many theoretical mechanisms by which “ultra-processed foods” might increase cancer risk. These range from the traditional low in vitamins and fibre but high fat, saturated fat, salt and added sugar, through deliberate food additives to contaminants from packaging or produced during processing. Only some of these “potential hazards” will apply to each of the categories above.
The foods categorized as “ultra-processed” make up a substantial proportion of the food eaten by many (most?) people living in an industrialised country like France and the UK. How are public health advisers, let alone ordinary consumers, supposed to interpret this weak, blanket-association with one type of disease into practical dietary change? Should consumers be advised to reduce or avoid a huge number of foods because some of them may be weakly associated with an increased cancer risk? A return to a bygone age where most people bought largely fresh ingredients to make home-cooked food or bought bread and other freshly processed foods from a local artisans seems like a pretty unlikely prospect, certainly for the typical French or British person working relatively long hours to generate a limited income.
The authors call for other large-scale observational studies with different populations and settings. I cannot see how this would substantially improve the usefulness of these findings. A weak association with a cohort of half a million is still a weak association and unless it is extremely well-resourced then it seems more likely to amplify the problems mentioned rather than reduce them. Large, very well resourced studies designed specifically to look for dietary links to cancer such as the major EPIC programme (European Prospective Investigation into Nutrition and Cancer) are able to use more robust methods of assessing diet and potential confounding variables.
They also call for further studies to try to discriminate between the various potential causes of any association between “ultra-processed foods” and cancer e.g. between nutritional factors, contaminants or additives. Investigation of many of these factors has already been ongoing for decades so exactly what further studies do the authors think are justified by their findings?
What these authors have found is a weak association between the total consumption of a huge category of foods and a slightly increased cancer risk. I can see no practically useful conclusions that can be drawn from this work even if one accepts the findings at face value. I can foresee it generating a steady stream of related studies that will soak up research resources and increase the mountain of published papers that do not really add to our scientific understanding or our ability to give clear and consistent advice about dietary improvement. On the positive side, these future papers will add extra lines to the curricula vitae and enhance the careers of their authors and help fill the mass of available space in some of the tens of thousands of scientific journals.
Ioannidis and Trepanowski (2018) acknowledge the importance of diet to a good health but say that the influence of individuals foods and nutrients is ambiguously tiny or non-existent. They go on to say that:
“Substantial reliance on observational data for which causal inference is notoriously difficult also limits the clarifying ability of nutrition science”
Posted on 1/09/2017 “Life-saving fruit and vegetable diet need only be three portions – study”
This headline appeared in The Guardian newspaper on 29/08/2017 and was triggered by an article published by a large Canadian-led team in the Lancet also on 29/08/2017. Many other media outlets also covered this story and some of the headlines suggested that the current five-a-day recommendation is too high e.g. “Forget the five-a-day mantra” in the Metro and “Forget five-a-day!” in the Daily Mail.
In a previous post, I discussed other recent scientific papers and media coverage which were interpreted as suggesting that, rather than being reduced, the current 5-a-day recommendation should actually be increased to 7 or even 10-a-day. It was back in 1990 that the World Health Organisation (WHO) first made the recommendation that we should all eat at least 5 x 80g portions of fruit and vegetables each day i.e. a total of 400g/day. This led to 5-a-day health promotion campaigns in other countries including the UK. In this previous post, I summarised some of the evidence that underpinned this recommendation. The association between high fruit and vegetable consumption and reduced cardiovascular and total mortality has been consistently reported in many different studies over several decades although there is disagreement about whether it is also associated with reduced cancer risk. I concluded in my previous post, that increases in fruit and vegetable consumption had the most benefit as intakes rose up to 5 portions per day and any extra benefits of even higher intakes were certainly much smaller and maybe non-existent. As 75% of the UK population fail to reach 5-a-day and almost half fail to reach three a day (240g) then arguments about possible small additional benefits of increases beyond 5 or 7-a-day seem rather pointless; in almost 15 years, the UK 5-a-day campaign has managed to increase average UK intakes from about 3.4 to 3.5 portions per day.
The results in this new Lancet paper are derived from a very large, elaborate and expensive international cohort study, the Prospective Urban Rural Epidemiology (PURE) study. This PURE study started recruiting subjects in 2003 and now has more than 135,000 subjects from 613 communities in 18 countries with differing average income levels. Subject data is collected at the time of recruitment and at subsequent 3 year intervals. Information collected includes aspects of the subject’s medical history, lifestyle, diet, body size and biochemical and genetic determinations from collected blood samples. Information about fruit and vegetable intake is obtained using a food frequency questionnaire (e.g. how often do you eat food X?). Despite the size and complexity of the study, the method used to evaluate diet is, of necessity, relatively crude. The paper includes data from low-income countries whereas most previous data is dominated by findings from Europe, the USA, China and Japan. These low income countries include some where fruit and vegetable consumption is low, but intake of legumes is high.
The authors’ own conclusions were that higher consumption of fruit, vegetables and legumes (i.e. things that grow in a pod like peas, beans and peanuts) was associated with lower cardiovascular, non-cardiovascular and total mortality i.e. those with high fruit and vegetable consumption were less likely to die during the average seven and a half years of follow-up. They suggested that maximum benefits were achieved with 3-4 servings of fruit and vegetables per day. They used 125g as their measure of portion size so this translates to 375-500g/day which seems to be me to be totally consistent with the very longstanding WHO recommendation of 5x80g portions or 400g/day. Several of the media sources do accept this point within their articles despite the misleading headlines. The Daily Telegraph actually implies this point in its headline “3 a day” is enough to slash death risk – as long as you have a decent serving”. At one point when discussing the implications of their work, the authors do seem to me to be implying that the new target of 375g is a materially different target to the WHO’s 400g/day.
The paper may well contain new useful findings about health and diet in some lower-income countries but does it really suggest, as some headlines imply, that current guidelines in affluent countries like the UK should be changed? To suggest that 375g/day target is materially different to 400g/day seems to me to be a nuance that is beyond the sensitivity of the relatively crude dietary assessment methods used. Whether it is more practical to take our approximately 400g/day in 5x80g or 3x125g portions seems like an inconsequential detail when around half of the UK population fail to reach 250g/day and probably around 1 in 5 people fail to reach 100g/day.
My main criticism is of the media headlines that have accompanied this paper. In April 2014, February 2017 and now in August 2017 we have had deluges of headlines about what is the optimal intake of fruit and vegetables. These headlines seem to imply that there is a great discrepancy between the “experts” with estimates ranging from 3 to 10-a day. When I read the three papers that led to these headlines, I am struck by the consistency of the findings rather than the disagreements. All show that cardiovascular and total mortality is lower in those who eat good amounts of fruits and vegetables. All suggest that least most of the benefits are obtained when daily intake approaches 400g/day i.e. about 5x80g portions or just over 3x125g portions. Of course, what none of these studies can prove is that increased fruit and vegetable consumption is a direct cause of this decreased mortality. It is a fundamental limitation of this type of epidemiological study that it can only demonstrate association not cause an effect; this is discussed at length in another recent post. Absolute proof that high fruit and vegetable intake causes reduced mortality can only be obtained by a large controlled experiment which is clearly impractical to perform.
The association between high fruit and vegetable intake and lower mortality has now probably been proven “beyond a reasonable doubt”. It seems highly likely to be a cause and effect relationship and certainly more than meets the “on the balance of probabilities” requirements of a civil court. Further epidemiological studies, no matter how large, expensive or sophisticated will probably not materially alter that judgement. The consensus belief among nutritional scientists is that effective intervention to increase fruit and vegetable intake in those with currently low intakes will reduce cardiovascular disease and total mortality. There is less certainty about whether it will also reduce cancer. Not much has really changed since 1990!
Posted on 11/6/2017 Measles “tragedy” kills 35 across Europe
This headline was taken from the BBC news web-site (11/06/17) and discusses findings by the WHO that 35 people across Europe (31 in Romania) have died as a result of measles in the past year. Italy has had 3300 cases and the death of a six-year-old boy is the most recent of these measles deaths. Measles deaths have also been reported in Germany and Portugal in the last year. The WHO regional director for Europe is quoted as saying that “every death or disability caused by this vaccine-preventable disease is an unacceptable tragedy”. If 35 people have died then many more will have experienced a serious illness and some of these are likely to have suffered permanent impairment as a result of their measles.
This BBC report is one of more than a dozen BBC reports over the past year dealing with local outbreaks, attempts to control them and attempts by governments to improve vaccination rates. Several of these reports highlight the role of Andrew Wakefield’s discredited claim that the MMR jab is linked to autism, as a major reason why vaccination rates have fallen below the 95% rate necessary to achieve herd immunity. This Wakefield case, the anti-vaccination movement and its impact has been the subject of one of my earlier full blog articles and I have recently added a full case-study of Andrew Wakefield’s activities.
Some of the other measles-related articles from the BBC news web-site in the past year:
22/06/17 600 children get MMR jab after measles outbreak in Newport
26/06/17 Germany vaccination: Fines plan as measles cases rise This article reports that the German government is planning fines of up to ϵ2500 for parent who fail to seek medical advice on vaccinating their children. Although it will not be an offence to refuse vaccinations.
19/05/17 Italy makes 12 vaccinations compulsory for children This is a report that the Italian government has ruled that children must be vaccinated against 12 common illnesses (including measles, mumps and rubella, MMR) before they can enrol at state schools. Rates of vaccination against measles have fallen to below 80% in Italy.
9/06/17 Parents eye Austrian asylum in Italy vaccination dispute This describes threats by some anti-vaccine parents in the German-speaking Tyrol region of northern Italy to seek asylum in Austria rather than allow their children to be vaccinated.
28/06/17 Italian father in passionate vaccines plea to Veneto governor This describes a letter written by an Italian father, whose daughter whose has low immunity after chemotherapy, pleading with the regional governor of Veneto to abandon his legal challenge to this Italian government ruling.
28/03/17 Measles outbreak across Europe
10/10/16 Vaccination offer after Edinburgh measles outbreak
23/09/16 Carmarthenshire measles outbreak linked to festival
8/08/16 Measles spreading at music festivals
The effects of Wakefield’s discredited 1989 paper involving at least partially fabricated results from just 12 subjects are still being felt across Europe and the rest of the world.
Posted on 24/4/2017 Dairy-free diets warning over risk to bone health
This headline appeared on the BBC new web-site on 12/04/2017 and was prompted by a warning from the National Osteoporosis Society issued in response to a survey that they had conducted involving 2000 UK adults. Dairy products (milk, cheese and yoghurt) have for many years been the biggest source of calcium in the British diet so they suggest that their finding from this survey that many young people are restricting their intakes of dairy produce threatens their calcium adequacy and long-term bone health i.e. increased risk of osteoporosis in later life. Osteoporosis is a serious loss of bone mineral that affects almost half of older woman and a fifth of older men and increases the risk of wrist, vertebral and hip fractures. As bone mineral is a calcium compound then the seemingly reasonable rationale is that lowered calcium intake reduces deposition of bone mineral in childhood and early adulthood so that when bone starts to thin in older people the bones will become weaker quicker.
The BBC article also quotes from a Food Standards Agency survey that indicates that almost half of 16-24 year olds claim to be intolerant to cow’s milk although most of them had no medical diagnosis to support this belief. This self-reported level of intolerance is many times higher than reported by older adults. As the article implies, this is almost certainly a massive overestimate of the true level of cow’s milk intolerance in young people. Of course, a proportion of people will be intolerant to cow’s milk products and as these are consumed by most Briton’s this is well documented and often reported as a reason to avoid cow’s milk. It seems likely that if goat’s milk, soya milk or some other alternative were consumed on the same scale then the real level of intolerance to these might be at least as high as that for cow’s milk.
The article notes that the recommended adult intake for calcium is 700 mg/day but rises to 1000 mg/day for those aged 11-18 years. These are the estimated needs of those in the group with the highest requirement and are set as intakes that should guarantee adequacy for everyone in the group. The high values for adolescents reflect the rapid skeletal growth; during the five years of adolescence boys accumulate an average of 200 mg/day of extra calcium in their skeleton. The article quotes from UK surveys which suggest that a quarter of teenagers consume less that 400 mg/day of calcium which is the value set by an expert panel as the threshold below which intake is assumed to be inadequate. As a nutritionist, I would regard intake of any essential nutrient that is below this minimum threshold value as a significant problem even if it does not lead to immediate, obvious symptoms of deficiency. I would also regard avoidance of a whole category of foods like dairy products as undesirable. However once someone has made a considered decision to avoid dairy then the role of the nutritionist or dietitian is to make suggestions and recommendations that allow this personal choice to be healthily implemented. The wide commercial availability of many fortified dairy substitutes has made this a relatively easy task.
I would take issue with this BBC article on three grounds:
- The role of most dairy substitutes as good sources of calcium and vitamin D is underplayed
- The role of vitamin D in ensuring efficient utilisation of dietary calcium and as a requirement for bone formation and maintenance is underplayed
- The link between dietary calcium intake and bone health is not as clear-cut as this piece implies.
The article does list several non-dairy sources of calcium like nuts, seeds, fish and fortified white flour. It does say also that almond and soya milk may contain calcium if fortified. On a recent visit to the supermarket, I looked at labels on a number of dairy substitutes and all of those I looked at, including those of the best know UK brand, had been fortified with calcium and also importantly with vitamin D. The manufacturers’ web-sites also indicated that most of their products were fortified with calcium and vitamin D. This means that people who, for whatever reason, choose to replace dairy foods with the most common commercial alternatives will probably not reduce their overall calcium intake and may well increase their vitamin D intake because ordinary UK milk is not a good source of vitamin D. If dairy avoidance is just part of a whole range of avoidances or one aspect of a bizarre diet then this may threaten calcium adequacy but probably also threatens the adequacy for other nutrients. Severe and poorly planned calorie-restriction to achieve an unhealthy level of leanness, especially in female adolescents, is one such scenario.
Vitamin D is probably a more critical determinant of bone health than calcium intake. Vitamin D is necessary for the efficient absorption of dietary calcium and in severe vitamin D deficiency (rickets), insufficient calcium is absorbed to allow proper bone mineral even if dietary calcium levels are not low. The average UK diet probably only provides about a quarter of vitamin D needs. We rely upon production in our skin when it is exposed to summer sunlight to produce most of our vitamin D needs. Poor vitamin D status is common in many sectors of the UK population because of a combination of low dietary intake and inadequate sunlight exposure. People who simply swap dairy foods for non-dairy substitutes may actually increase their vitamin D intake and in other posts on this blog (for example) I have advocated the fortification of a staple UK food or foods with vitamin D. Weight-bearing exercise is known to increase bone mass in children and young adults – so increased outdoor activity has two benefits for bones as it also increases exposure to sunlight. (Note that outdoor grown or wild mushrooms are a good vegetarian source of vitamin D. Bulk-produced UK mushrooms do not contain vitamin D if grown in the dark, although even these will produce good amounts of vitamin D if left in direct sunlight for an hour or so).
The link between dietary calcium intake and bone health is complicated by factors like the need for good vitamin D status to absorb calcium efficiently and the increase in bone mass caused by weight-bearing exercise. In some populations where consumption of milk by adults and older children has not been the cultural norm, there has been little or no problem with osteoporosis despite relatively low calcium intakes. In some other western countries, osteoporosis is a major public health problem despite calcium intakes that are relatively high. Relatively high calcium intake is no guarantee of good long-term bone health and good bone health can be associated with relatively low calcium intakes if other dietary and lifestyle factors are favourable.
Posted on 17/4/2017 The problem with scientific publishing and how to fix it.
This headline appeared in the “Economist explains” section of The Economist magazine on 30th March 2017. It was brought to my attention by my daughter Kate Webb. Scientific journals have been used to disseminate scientific information dating back at least as far as 1665 when Philosophical Transactions (of the Royal Society, London) was first published. Peer review is now seen as the main method of quality control used for selecting which papers are worthy of publication but it has only been the norm for most journals since the middle of the 20th century. Journal editors send submitted papers to one or more anonymous experts in the field to evaluate the paper and make recommendations about whether the paper should be published or not and any modifications needed before publication.
The author of the Economist article points out that journal publications now serve another major purpose that is often more of a priority than simple dissemination. The number of peer-reviewed papers, especially those published in the top journals, is used as the key indicator of a scientist’s research prowess and thus a major determinant of their career progression. S/he argues that this unintended role of journal articles means that scientists have an incentive to withhold their results until they can publish a more complete and powerful piece. S/he argues that such delays in release of important data might be harmful and cites the recent Zika crisis where s/he claims that research sponsors had to persuade publishers to declare that scientists would not be penalised for early release of their data.
The article author goes on to highlight the large rise in the number of flawed and retracted papers in recent years and notes that the rates of retraction are higher in the top international journals like Nature and Science. S/he says that this shows that they are no longer the guardians of quality that they claim to be. One probable reason for retraction rates being higher in the top journals is that published articles will be subject to much greater critical scrutiny than those published in low-level journals that may never be read, sometimes not even read properly by the peer reviewers. Any flaws in major papers published in top journals are thus more likely to be exposed and reported to the editor and these editors may be much more willing to retract papers shown to contain irretrievably flawed or fabricated data. Higher retraction rates could thus be seen as a sign of better quality control in these top journals. Seriously flawed or fabricated data published in low level journals are likely to go unnoticed, unreported and unretracted
In a rather throwaway sentence, the article author also suggests that papers published in elite journals are no more statistically robust than those in lesser journals. In a previous post, I have discussed some of these statistical issues and I am sure that they are not confined to papers in lesser journals, for example:
- Small underpowered studies, which makes falsely statistically significant results more likely
- Small effect sizes which may be just a measure of the inherent bias in the study
- Multiple testing, e.g. correlating multiple variables until one gets a statistically significant association (by chance!)
- Multiple modelling i.e. sequential correcting for confounding variables until results become or remain statistically significant
- Slight manipulation of results so that they reach the magic “<5% probability of occurring by chance” so that they can be claimed as “statistically significant”.
The author of this Economist piece suggests three reforms which s/he believes would improve and accelerate science and allow health agencies to respond more quickly to epidemics and speed up the development of new treatments:
- Scientists should put their academic papers into publicly accessible repositories before they are published
- Peer reviewers should no longer be anonymous and the reviews should also be published
- Alternatives to journal publications as the path to career advancement for scientists should be sought. S/he mentions number of views of preprints and whether clinical data has been used in guidelines for doctors as possible examples.
I have some issues with these specific suggestions such as those below.
- If peer reviews are to be published then without strict safeguards they may simply become an opportunity for reviewers to promulgate their own views and theories in lengthy essays without the need for supporting data or peer review.
- Science, like most human activities, is competitive and competition may be a major motivating factor fuelling scientists’ creative energies. To persuade scientists to prematurely release their “new early exciting data” for others to exploit is going to require a major cultural shift.
- Using published papers as the measure of a scientist’s worth is undoubtedly a distorting influence on scientific publishing. As in any other field, once a metric of “success” is accepted it ceases to become an incidental objective measure and starts to define the way that activity is conducted and managed. Acquiring peer-reviewed publications becomes a major objective in its own right rather than their being primarily a vehicle for disseminating high quality research data. To find better measures of the quality of a scientist’s research output is a desirable objective but to find something that is both objective and suitable for mass application is not easy. This is why paper counting has been the standard tool for so long.
My main criticism of this piece, however, is not in the suggested solutions but in the writer’s identification of the problem with scientific publishing. Delay in the publication of important data may be an occasional problem in specific circumstances. However, the major problem with scientific publishing is the deluge of poor quality data often published after only nominal peer review. Rather than holding back important data, many scientists are guilty of prematurely publishing inadequately substantiated data or splitting up data sets into small parcels that yield multiple publications rather than one substantial paper. A large proportion of what is published either cannot be reproduced by other scientists or is not of sufficient interest to persuade others to even try to reproduce it. A high proportion of the 2 million papers published each year will never be cited by any other scientist and many will have few, if any, readers. Release of unsubstantiated and potentially false data may open up research blind alleys that soak up research effort and resources possibly for decades afterwards (see previous post on yogurt as a proposed cause of ovarian cancer). I would be looking for ways of persuading the majority of scientists to wait until they are confident of the reproducibility and value of their data before rushing into print. Fewer but better papers would be my desired outcome.
Posted on 02/04/2017 “Fake research” comes under scrutiny
This headline appeared on the BBC news web-site on 27/03/2017 . The main point of the piece was that official data on the incidence of research fraud allegations underestimates by at least ten fold the number of cases revealed in a BBC survey of 23 of the UK’s research intensive universities. The article may well have been prompted by the start of an inquiry into research integrity by a House of Commons committee. The author notes suggestions that the UK should have an official regulatory body to oversee publicly funded research, modeled on the Office of Research Integrity (ORI) in the USA. Dr Ivan Oransky who helped found the web-site RETRACTION WATCH is quoted as confirming that in general only a tiny fraction of cases of research fraud identified by universities and funding agencies get reported. He says that in one frequently cited survey about 2% of researchers admitted to having committed some act of research misconduct. This is much higher than the amount of research misconduct that is actually reported and recorded.
The term research misconduct covers a spectrum of misbehaviour including data fabrication or falsification and plagiarism. Fabrication is the invention of data and falsification involves manipulating real data so that what is reported no longer reflects what was actually found.
2% sounds like quite a low figure but it is almost certainly an underestimate. Even in a survey said to be anonymous, how many people will admit to doing something that is regarded as the worst professional crime that a scientist can commit? Evidence from those found guilty of research fraud, is that they are usually multiple offenders and some have built long and successful careers based almost solely upon fabricated data (see for example Ranjit Chandra and Diederik Stapel on this blog). If we accept 2% as a minimum estimate then this means that at least 40,000 of the 2 million scientific papers published annually involve some form of misconduct. This is not confined to papers in low quality journals because the best journals have the highest rates of retracted papers (more intense scrutiny means that exposure of fake data is more likely).
Some scientists take a fairly relaxed or complacent view of bad and fraudulent research because they believe that it will be exposed or discredited when other scientists are unable to reproduce it. However, very few fraudulent scientists are exposed by this route and some churn out publications based upon fabricated data for decades before being exposed.
Why does “fake research” matter so much?
- It undermines the credibility of science and public trust in research. The realisation that “pop star” psychologist, Diederik Stapel’s reputation was founded upon a mass of fabricated data threatened for a time to undermine confidence in his field of social psychology research about the psychological processes underlying individual choice and economic decision-making.
- It is a misuse of scarce resources of the host institution and research funders. In 2005, world-renowned obesity researcher Eric Poehlman of the University of Vermont acknowledged that he had used fabricated data in 17 funding applications and pleaded guilty in a court in Vermont to defrauding the US National Institute of Health of $547,000. Prosecutors believe that he actually defrauded the NIH of $2.9million. He was sentenced to a year and a day in prison, one of the very few research fraudsters to receive a jail sentence or indeed any legal sanction.
- It damages the careers and lives of collaborators and research students. Ten of Diederik Stapel’s graduate students had their theses tainted by the public acknowledgement that they contained fabricated data and one young woman, who had just submitted her thesis, felt obliged to withdraw it. One of Haruko Obokata’s distinguished research supervisors, Yoshiki Sasai, hanged himself in his laboratory in August 2014 just a few months after the retraction from Nature of the two papers that he co-authored with her.
- Fraudulent data hinders the progress of research. It persuades other researchers to try to reproduce the false findings or leads them into unproductive areas. An editorial in the American Journal of Physiology in 2007 bemoans the amount of time, money and energy that had been expended by other groups around the world in trying to replicate the falsified data of Jatinder Ahluwalia about the way in which white blood cells kill ingested bacteria. The eminent geologist Sir Arthur Smith Woodward FRS is said to have spent the last 21 years of his life trying unsuccessfully to unearth further fossils and artefacts from the area where Charles Dawson claimed to have found the Piltdown man skull.
- Policy decisions, choice of medical treatment or public behaviour can be based upon false data and false conclusions. Scott Reuben’s fabricated data on the benefits of the COX-2 type of painkillers probably increased sales of these compounds by several billion dollars worldwide and influenced the pain management protocols of many physicians and surgeons around the world. The false data of Werner Bezwoda helped encourage the use of high dose chemotherapy (HDC) in cases of advanced breast cancer. It became difficult to recruit subjects onto clinical trials comparing HDC and conventional chemotherapy (in case they were allocated to the conventional treatment group) and so delayed the completion of these clinical trials by several years. Many thousands of patients underwent this expensive, gruelling and potentially life-ending treatment partly as a result of his fake positive data.
- In extreme cases false data can kill people. The fabricated data of German anaesthesiologist, Joachim Boldt, helped to increase and prolong the use of certain hydroxyethyl starch preparations (HES) as fluid replacers in seriously sick or injured people. They are now listed as contraindicated for many procedures and are known to be associated with increased risk of adverse events and with higher risk of death. The falsified data of Boldt thus probably contributed to extra patient deaths through their unjustified use. As a result of the highly positive false data relating to HDC of Werner Bezwoda published in 1995, a small trial that sought to replicate Bezwoda’s treatment protocol was started in Seattle, Washington. Four of the six patients recruited, before the trial was abandoned, suffered serious and permanent cardiac damage (heart failure) because of the treatment and two died as a result of this.
Is it now time to consider whether research fraud should be made a specific criminal offence?
Posted on 24/03/2017 “Fish oil does not benefit baby intelligence, study finds”
This headline appeared in the “Australia” section of the BBC news web-site on 22/03/2017.
The headline was prompted by a research letter published in the Journal of the American Medical Association (JAMA). I have only read an extended summary of this letter which is a 7-year follow-up of a study published in 2010. I have been able to access and read this original report . These articles report the results of a controlled clinical trial of the effects of fish oil supplements given in the second half of pregnancy. They show that the supplements do not reduce postnatal depression nor improve the cognitive development and intelligence of the babies.
Maria Makrides and her Australian colleagues (2010) recruited 2400 women who were less than 21 weeks into their pregnancy. Half were given a fish oil capsule containing 800mg of DHA and half were given a placebo (vegetable oil). Almost all of these mothers were assessed for postnatal depression and 726 babies were assessed for cognitive and language development at 18 months old. A proportion of the babies were also assessed at 4 years old and 7 years (about 540 babies) – it was the letter reporting the results from this last follow-up that prompted the title headline. In the 2010 paper the authors found that the fish oil supplement made no difference to rates of postnatal depression in the mothers. The fish oil also made no difference in composite scores for cognitive (intelligence) or language development in their babies. The 7 year follow-up essentially confirms the findings at 18 months old. When they broke down the composite scores then there were a couple apparent small positive effects in the fish oil group but this was offset by small negative effects on behaviour in the supplemented group. Small differences in odd individual measures where multiple measures have been made should anyway be treated very sceptically. There was some evidence that gestation was slightly prolonged in the fish oil women and although firm conclusions were difficult to reach from this particular study, it is in line with other reports of the effects of fish oils.
Fish oil and cod liver oil supplements represent around 40% of the £400 to 500 million UK supplement market. People take these oils because they are a rich source of the long chain omega-3 fatty acids known as EPA and DHA. Cod liver oil also contains vitamins A and D and this is the historical rationale for its widespread use, especially in children.
Other than oily fish (like salmon, mackerel, herrings and sardines) and maybe meat from animals fed fish-meal, sources of EPA and DHA in the diet are scarce. Some vegetable oils like rapeseed oil and particularly flaxseed oil do contain substantial amounts of other shorter chain omega-3 fatty acids. In theory our bodies are able to make EPA and DHA from these short chain omega-3s. Our capacity to make EPA and DHA may, however, be limited, especially if our diets contain copious amounts of omega-6 fatty acids which are abundant in many vegetable oils and spreading fats. Large intakes of omega-6s may inhibit the conversion of short chain omega-3s to EPA and DHA.
There is reasonably convincing evidence that fish oil from food or supplements may slightly reduce the risk of having a heart attack. The best of this evidence is from studies with heart attack survivors where fish oil seems to reduce the risk of a second fatal heart attack. Some of this evidence comes from before the mass use of statins and any small, beneficial, population-wide effect of fish oil may be swamped by the impact of statins. Statins are taken by 6 million British adults and there are suggestions that this number should double.
Many people take cod liver oil or fish oil to “lubricate their joints” and alleviate or prevent the symptoms of osteoarthritis. There is a plausible mechanism for how fish oily might do this, but there is little or no substantial evidence of efficacy. There is some evidence that huge doses of fish oils may have some limited benefits for patients in relatively uncommon rheumatoid arthritis.
Fish oil supplements are also taken in the hope that they may alleviate the symptoms of depression. One of the original aims of the Makrides paper was to see if they reduced the risk or severity of postnatal depression. There is no convincing evidence for any beneficial effect upon depression or postnatal depression. There has been much SPECULATION about whether fish oil supplements might improve behaviour, attention span and academic performance in schoolchildren but there is no real evidence to support these claims. The work of Makrides and her colleagues indicates that antenatal fish oil does not improve a baby’s intellectual development.
Posted on 17/3/2017 B vitamins may have “protective effect” against air pollution
Note: I have only read the author’s summary of their work as it requires a payment to read the full paper.
The title appeared as a headline on the BBC new web-site 14/03/2017 and similar headlines were used by media outlets around the world . These 16 researchers exposed 10 subjects to either clean air or air polluted with small particulates known as PM2.5 for periods of 2 hours. They performed these exposures after pre-treatment for a month with either a B vitamin mixture or a placebo. They assessed changes in DNA in blood cells involved in human immunity after each of these four exposure situations. Exposure to the particulate pollutants produced a measurable change in DNA (increased methylation) which was apparently prevented by the vitamin supplement.
The doses of vitamins were extremely high, over 12 times EU Recommended Daily Allowance (RDA) for folic acid (B9), 30 times the RDA for vitamin B6 and 400 times the RDA for vitamin B12. The folic acid and B6 doses are over the Tolerable Upper Limit recommended by the European Food Safety Authority (no upper limit is set for B12). Any changes seen with these doses may be unrelated to the normal vitamin activity of the substance. Unless similar effects are seen at more normal vitamin doses this would undermine any practical application for these findings. Massive doses of vitamins are not harmless and severe toxicity associated with overdose of some is well established.
They used very short-term changes (2h) in a marker for DNA change. The implication is that this DNA change is a reliable predictor of likely long-term health damage (e.g. cancer) that might result from prolonged exposure to these particulates.
This was only a pilot study with just ten volunteers. The measured effect of the pollutant was only just statistically significant but did seem to be completely blocked by the supplements. Small scale studies with results that are anywhere near the borderlines of statistical significance should always be treated with extreme caution.
These authors have actually shown that massive doses of these B vitamins seemed to block short-term changes in an assumed marker for long-term harm (DNA methylation in white blood cells) in a very small pilot study.
This study is widely reported as if it were a major breakthrough rather than some very preliminary indications of a possible future research area. The authors themselves did caution about the limitations of their study.
As an analogy, high doses of antioxidant vitamins (e.g. A, C, and E), minerals (e.g. selenium) and antioxidant-rich foods (e.g. watercress) reduce short-term measures of oxidant damage to cells, including DNA damage. This has been widely interpreted for decades as an indication that increased antioxidants might reduce the risk of cancer and/or heart disease. Yet high quality clinical trials of antioxidant supplements lasting many years have found no indication that they reduce cancer, heart disease or total mortality risk. Several studies have reported that some of these supplements, particularly those that contain beta-carotene (vitamin A), cause net harm.
Posted on 20/3/2017 Obesity crisis: Is this the food that is making us all fat?
This headline and a picture of vegetable oil appeared in the business section of the BBC News web-site on 15/03/2017 ;an unexpected source for one of these posts.
My first reaction
I wrongly assumed after reading this headline that the rise in consumption of vegetable oil in the UK since the 1960s and 70s was being suggested as the reason why UK obesity rates had quadrupled over this period. There has undoubtedly been a major shift in our sources of fat:
- In 1975, 68 % of spreading fats sales were butter but two decades later this was just 25%
- In 1975 just 22% of cooking fat sales were vegetable oil but by 1996 this was 77%
- In 1975, 78% of cooking fats were animal fats (mainly lard) and this was down to 23% by 1996.
Since 1996, vegetable oil has increased to over 80% of cooking fat sales and reduced fat spreads (not readily available in 1975) now account for 60% of spreading fat sales. These changes in our sources of fat resulted in a substantial reduction in our consumption of saturated fat and an increase in polyunsaturated fat consumption. All fats, whether saturated or unsaturated are concentrated sources of calories. They all provide about 9 Kcal/g compared to around 4 for sugar, starch and protein. What about total fat intakes? The contribution of fat to our calorie intake rose during the years after WW2 as post-war shortages and rationing eased. From 1975 to the early 1990s, the fat contribution to total energy intake remained constant at around 40% despite the switch from saturated to unsaturated fat. Although the absolute amount of fat (grams) dropped during this period, this was in line with the overall reduction in calorie intake recorded over this period. Since the early 1990s the contribution of fat to the energy in our diets has actually fallen to around 35%. Therefore vegetable oil and spreading fats made from oil, initially replaced other fats and since the 1990s overall fat consumption has been falling in real terms. It does not seem logical, therefore, to blame rising sales of cheap vegetable oil for our current obesity crisis.
What triggered the headline?
The BBC article was prompted by a report produced by a Professor Benton at Leeds University about sources of UK food and the impact of globalization. This independent report sponsored by the supermarket chain Morrison’s is not primarily about nutrition but about where our food comes from. Some of the chapter headings illustrate the general nature of its content:
- The food we eat: where is it sourced?
- Is more food trade a good idea?
- The benefits of local production
- The UK capacity for greater self-sufficiency.
In his conclusions, Professor Benton argues that globalization drives diets towards major commodity crops including vegetable oils leading to a loss of diversity and a drive towards calorie rich but low nutrient diets. He argues that the global food trade means that rich countries are able to export some of the environmental costs of food production to other countries. He argues that whilst the UK cannot hope to be self-sufficient in food we could become much less reliant on imported food. Reading further into the BBC article it was not explicitly argued, except in the headline, that the UK obesity crisis was caused by eating more vegetable oil.
After a brief skim through Benton’s 36 page report, I find it difficult to see how it generated the BBC headline. Of course, Professor Benton was probably interviewed for the BBC article and this may have led to the strong suggestion of a link between vegetable oil and obesity. One could interpret the first part of this BBC report as suggesting that globalization and the massive production and trading in high calorie vegetable oils has allowed increased consumption of high calorie, low nutrient diets that have helped to create a climate, in many countries, where obesity increases.
Is wide availability of calorie rich foods really the cause of the global obesity epidemic?
If the BBC news story had been solely related to Professor Benton’s report then I might have abandoned my intention to post this commentary. However the reporter (Katie Hope) then goes on to talk about a study produced by a group at the London School of Economics entitled “Globesity”? The effect of globalization on obesity and caloric intake. Joan Costa-Font and Nuria Mas reported that globalization as determined by a OECD index is associated with a rise in both obesity and calorie intake. However, they concluded that “social globalization” i.e. changes in the way we work and live is a much stronger driver of obesity than increases in the availability of cheap calorie rich food. To put it simply obesity is increasing because people are expending less energy in working, shopping, socializing, travelling and doing household chores rather than because they are eating more.
In the UK, data from government sponsored surveillance reports like the old National Food Survey, The Family Expenditure Survey and their subsequent replacements have suggested that we eat, or at least buy, many fewer calories than we did in the late 1950s or early 1960s; perhaps a quarter to a third less. Yet over this same period, obesity rates have at least quadrupled. Almost a quarter of the UK adult population are now obese and around two-thirds are currently classified as either overweight or obese. We are getting fatter and yet apparently eating less which suggests that reduced activity and reduced energy expenditure are the major drivers of these obesity increases. There have also been big rises in the number of children who are classified as obese or overweight.
Cross sectional studies have found that when one classifies people as low, medium or high activity then obesity rates rise with decreasing activity. Several studies from around the world have found that number of hours of TV watching in children is not only linked to their current obesity but also predicts the future likelihood of becoming obese. Future studies will obviously have to widen the definition of “viewing” to include all activities that involves sitting and watching the screen of an electronic device.
Whilst the availability of a plentiful supply of cheap calorie rich foods is an essential prerequisite for rising obesity levels, it is decreased activity and reduced energy expenditure that is the main driver of rising obesity and especially the rising level of childhood obesity.