{"id":3569,"date":"2015-04-13T11:36:40","date_gmt":"2015-04-13T10:36:40","guid":{"rendered":"https:\/\/www.zoeharcombe.com\/?p=3569"},"modified":"2016-08-21T12:41:22","modified_gmt":"2016-08-21T11:41:22","slug":"being-overweight-reduces-dementia-risk","status":"publish","type":"post","link":"https:\/\/www.zoeharcombe.com\/2015\/04\/being-overweight-reduces-dementia-risk\/","title":{"rendered":"Being overweight reduces dementia risk?"},"content":{"rendered":"
The headline on Friday 10th April 2015 was \u201cBeing overweight \u2018reduces dementia risk<\/a>\u2019\u201d. The original article is here<\/a>. Sadly on pay per view only, so it cost me $31.50 for 6 pages.<\/p>\n I tweeted early on Friday morning \u201cBeing overweight may be associated with reduced incidence of dementia but \u2018reduces risk of\u2019? Really?\u201d<\/p>\n I couldn\u2019t see a plausible mechanism and nor could the researchers, who confessed to having been \u201csurprised\u201d. As the BBC article noted, \u201cAny explanation for the protective effect is distinctly lacking.\u201d<\/p>\n However, twitter being the marvel that it is, served up a plausible mechanism, from a Seattle physician, within minutes of my tweet. We\u2019ll get the study and the usual pitfalls out of the way first and then see what could really be interesting about this finding.<\/p>\n The study<\/strong><\/p>\n The study was the usual \u201clet\u2019s examine lots of data\u201d format, which is the most common method for grabbing headlines at the moment. Data from almost two million people (1,958,191) was reviewed. The average age at the start of study was 55. The average length of follow-up was 9 years. Hence this was a huge review with almost 20 million person years of data.<\/p>\n There were 45,507 cases of dementia recorded during the follow-up period. The paper presented this as 2.4 cases per 1000 person years (i.e. their calculation, not mine).<\/p>\n BMI (body mass index) was recorded at the start of the study for each participant. This measurement was taken some time between 1992 and 2007 \u2013 when the participants reached the age of 40 or older. We know nothing<\/em> about what happened to weight over the average nine years of follow-up. Those who were underweight at the start of the study could have been obese by the time they developed dementia and vice versa \u2013 we know none of this.<\/p>\n The study confirmed that dementia was strongly associated with age and so age needed to be adjusted for in the analysis of the data.<\/p>\n The numbers in the newspaper headlines came from Table 2 in the paper and from the top set of numbers (adjusted for age and sex only). However, even when the researchers adjusted for everything they could think of (smoking, diabetes, medications, alcohol etc) the numbers hardly changed, so we can see these as valid figures.<\/p>\n The Table 2 numbers took normal BMI as \u201chealthy weight\u201d and gave that a reference base of 1.0. They then presented the incident rate of dementia, relative to 1.0, as follows for the different weight categories:<\/p>\n – Underweight (BMI <20) = 1.34 (this was where the \u201c34% higher risk\u201d headlines came from).<\/p>\n – Overweight (25-29.9) = 0.82 (this was where the \u201c18% lower risk\u201d came from).<\/p>\n – Obese (30-34.9) = 0.76 (this was where the \u201c24% lower risk\u201d came from).<\/p>\n – Very obese (BMI >40) = 0.71 (this was where the \u201c29% lower risk\u201d came from).<\/p>\n The common pitfalls<\/strong><\/p>\n We should know these really well by now:<\/p>\n 1) This is association not causation.<\/p>\n We may have observed that people who wore red socks to the baseline health check went on to develop dementia. We should not jump to the conclusion that wearing red socks increases the risk of dementia.<\/p>\n 2) This is relative risk, not absolute.<\/p>\n The fact that there were 2.4 cases of dementia per 1,000 person years means that I\u2019m not<\/em> worried about any of this already. Applying this 2.4 number to the incident rate in each of the different weight groups, results in the following:<\/p>\n – The incident rate for underweight people becomes 3.2 cases per 1,000 person years.<\/p>\n – The incident rate for overweight people becomes 2.0 cases per 1,000 person years<\/p>\n – The incident rate for obese people becomes 1.8 cases per 1,000 person years<\/p>\n – The incident rate for very obese people becomes 1.7 cases per 1,000 person years<\/p>\n 2.0 per 1,000 person years vs. 2.4 per 1,000 person years suddenly doesn\u2019t seem quite so dramatic, don\u2019t you think? But those are the actual real, absolute, incident rates behind the headlines.<\/p>\n Possible things going on<\/strong><\/p>\n It could be the case that more obese people died and were therefore not available for follow-up (and not alive to develop dementia). An interesting study on weight and mortality here <\/a>was covered in a much more readable way by the brilliant Dr Malcolm Kendrick here<\/a>. The study showed that obesity was associated with more deaths, but that being overweight wasn\u2019t. Hence, this cannot serve as an explanation for this dementia weight finding, which showed that the incidence of dementia was lower with every increase in weight category they measured from underweight to very obese.<\/p>\n The researchers themselves offered the possible explanation that heavier people may have taken in more nutrients and that some of these may have had mind protection benefits. I\u2019m not sure that there is any evidence for this (and they didn\u2019t seem that convinced by it). When I heard the lead researcher on Radio 4 on Friday morning, he really did hold the view that he was surprised and couldn\u2019t explain the findings.<\/p>\n