Behind the Headlines

Thursday November 26 2009

Apple-shaped women face a doubled risk of dementia, according to the Daily Mail.

This news is based on Swedish research that measured the physical dimensions of 1,500 women and followed them for an average of 32 years. Those women with an “apple shape” (broader in the middle than the hips) faced an increased risk of dementia if they lived beyond 70 years.

This study was well designed and collected data over several years. However, the researchers were surprised to find no link between body mass index (BMI) and stroke risk, an association found in other studies. Other results also suggest the findings may have occurred by chance. Overall, this large study has not convincingly shown the link between apple-shaped women and dementia, and raises more questions than it answers. Further research into this area is needed.


Where did the story come from?

This research was carried out by Dr Deborah Gustafson and colleagues from the Institute for Neuroscience and Physiology in Sweden. The study was supported by the US National Institutes of Health and the Swedish Research Council. The study was published in the peer-reviewed medical journal Neurology.

Other newspapers reported this research in a similar way to the Daily Mail. The Daily Telegraph also highlighted that the researchers found no link between dementia and high BMI.


What kind of research was this?

This was a report of data from a cohort study conducted in Sweden between 1968 and 2000. This research report focuses on the links between physiological measurements and the risk of dementia.

The researchers explain that the relationship between weight and dementia is complex. Higher body fat in middle age and later life is thought to increase dementia risk, but there is also a trend for the elderly to lose weight within a few years of being diagnosed with dementia. The researchers feel that these relationships needed further study.

Given dementia sufferers’ shifts in body fat over time, it was important that this study included analysis of these changes over time. These types of factor can only reliably be analysed using an inception cohort study, which recruits people at the same stage in the development of a disease. In this case, the researchers performed an inception cohort study that recruited only women who had no diagnosis of dementia and followed them over time to see who might develop the condition.


What did the research involve?

In this study, called the Prospective Population Study of Women (PPSW), a representative sample of 1,462 women without dementia was recruited in 1968. They ranged in age from 38 to 60 years. At entry into the study, they underwent a range of clinical and psychiatric tests, and had various body measurements taken (weight, height, waist circumference and waist-to-hip ratio). These same tests were performed again in 1974, 1980, 1992 and 2000.

Dementia was diagnosed according to DSM-III-R criteria, a recognised diagnosis method of the American Psychiatric Association. The researchers also included a category for “probable and possible Alzheimer’s dementia”, and diagnosed the sub-types of dementia using recognised criteria and CT scanning where necessary. They particularly looked at a condition called Alzheimer’s dementia with cerebrovascular disease (ADCVD), in which Alzheimer’s was diagnosed in the presence of a history of stroke. Stroke is known to be linked to both dementia and BMI.

The researchers noted whether there was a history of previous heart attack, stroke or diabetes based on the participants’ self-reports. They also conducted blood tests, ECG assessments and assessments of lifestyle factors, including physical activity, alcohol consumption and cigarette smoking. They also asked about level of education and socioeconomic status, which they then defined as working class, middle class or upper class (based on 1968–69 survey responses).

The researchers used all these factors to adjust their analysis in order to minimise the influence of factors linked to dementia. This would increase their ability to detect any link, if one existed, with the weight and fat distribution measures.

The measurements in this study appear to have been carefully conducted, and the study’s findings were well reported.


What were the basic results?

Over 32 years, dementia occurred in 161 participants. The average age at diagnosis was 75.6 years (range 48 to 91 years). There were 75 cases of Alzheimer’s, 108 cases of Alzheimer’s in people with previous stroke (ADCVD) and 37 cases of pure vascular dementia.

There was an association between having a waist-to-hip ratio greater than 0.80 at the start of the study (e.g. 80cm waist with 100cm around the hips) and a 2.2-fold higher risk of dementia among surviving participants aged 70 or older. The result reported by the researchers was an odds ratio of 2.22 (95% confidence interval 1.00 to 4.94), meaning that it was only just statistically significant. The other measure of statistical significance, the p-value, was 0.04. This confirmed the borderline significance of the result, as a p-value of less than 0.05 is generally considered significant.

This finding was one of 40 statistical results relating to the relationship between physical factors measured in midlife and late life and dementia. The only other statistically significant result showed that a low BMI or waist circumference in examinations later in life (between the ages of 62 to 92) was associated with an increased risk of dementia. This was consistent with the researchers’ earlier suspicions that the closer a person gets to a diagnosis of dementia, the less likely they are to be overweight.


How did the researchers interpret the results?

The researchers say that, in the Swedish women, they observed an approximately twofold increase in the odds of later dementia in those with a high midlife measure of central body fat. They note that this finding is only true for those who survived for 32 years (to at least the age of 70) and who participated in the neuropsychiatric examination.



This well-designed study was reported accurately by researchers and news sources. It ran over a long period of time with careful follow-up and measurements. However, only one of the “adiposity” (relating to fat) measurements that the researchers looked at was significant, which suggests that the findings should be treated cautiously. In addition, as the single positive finding was only just statistically significant, it may have occurred by chance alone.

Other cautions are raised by the researchers:

  • It is often difficult to diagnose different types of dementia, and the researchers used strict criteria combined with reviews by psychiatrists. However, this review of notes, particularly among those who were lost to follow-up, might have reduced the accuracy of diagnosis.
  • There were small numbers of people with the sub-types of dementia, which limits the ability of the researchers to analyse these specific sub-groups.
  • The researchers note that, as the study was carried out in a group of Swedish women with similar characteristics (average weight 64.5kg and BMI 24.1), its results may not apply to other more varied groups. They say that, ideally, the study should be repeated in other populations.

Overall, this large study conducted over a long time has not convincingly shown the link between apple-shaped women and dementia, and it raises more questions than it answers. These questions can only be answered through further study.

Analysis by Bazian

Edited by NHS Choices

Links to the headlines

Apple-shaped women at double risk of dementia. The Daily Mail, November 25 2009

Apple-shaped women 'twice as likely to suffer dementia'. The Daily Telegraph, November 25 2009

Links to the science

Gustafson DR, Bäckman K, Waern M et alAdiposity indicators and dementia over 32 years in Sweden. Neurology 2009; 73: 1559-1566


How helpful is this page?

Average rating

Based on 0 ratings

All ratings

0  ratings
0  ratings
0  ratings
0  ratings
0  ratings

Add your rating