"Diet drinks triple your risk of stroke and dementia," the Daily Mail reports, as US research found a link between daily intake and increased risk. However, the chain of evidence is not as strong as reported.
The researchers analysed data from an ongoing US cohort study to see if consumption of sugar or artificially sweetened drinks was linked with risk of stroke or dementia 10 years later. Several thousand people were included in the study, and during follow-up 3% had a stroke and 5% developed dementia.
Overall, when taking account of all health and lifestyle factors that could have an influence (confounders), the researchers actually found no link between artificially sweetened drinks and risk of dementia.
The figures reported in the media came from a model that wasn't adjusted for all confounders, such as diabetes, that could explain part of the link.
For stroke the links with artificially sweetened drinks were inconsistent. There were no overall links when looking at longer term patterns.
The study does not give definitive "cause and effect" proof that drinking artificially sweetened drinks will lead to stroke or dementia. Still, the lead author's reported statement that it is healthier (not to mention cheaper) to just drink water is sound advice.
The study was carried out by researchers from Boston University School of Medicine and Tufts University, Boston. The long-term cohort study is funded by the National Heart, Lung, and Blood Institute, the National Institute on Aging and the National Institute of Neurological Disorders and Stroke.
Individual researchers of this particular analysis received funding from additional sources including the National Health and Medical Research Council. The researchers declare no conflict of interest.
The Guardian gave a good overview of the research while making clear that no cause and effect had been proven.
The Daily Mail's headline – "Diet drinks TRIPLE your risk of stroke and dementia" – is somewhat misleading as it is based on unadjusted data. Though the authors themselves included this information in the abstract of the study.
Several independent experts in the field have also advised caution in taking the results of this research as being conclusive, until further research is conducted.
This was a prospective cohort study, using data collected from the ongoing Framingham Heart Study Offspring, to see whether drinking sugar or artificially sweetened drinks was linked with risk of stroke or dementia.
The researchers say how previous research has linked both types of soft drink with cardiovascular disease, like stroke, but dementia had yet to be examined.
This type of large cohort can find links, but it's very difficult to prove that any individual factor, such as drinks, is directly responsible for a health outcome. Food questionnaires can be subject to inaccurate recall and it's hard to account for all other health and lifestyle factors that can have an influence.
The Framingham Heart Study Offspring cohort started in 1971, enrolling 5,124 people living in the community of Framingham, Massachusetts. They had assessments every four years up to 2014.
The assessments at waves five (1991–95), six (1995–98), and seven (1998–2001) included food frequency questionnaires that assessed dietary intake over the previous 12 months. This included questions on sugar-sweetened and artificially-sweetened drinks, among many other food and drink items. Responses for intake ranged from "never or less than once per month" to "six or more per day".
The researchers pooled the most common response categories for each drink to come up with ranges that weren't directly comparable:
The 10 year risk of new-onset stroke or dementia started from the last food and drink assessment wave (1998–2001) onwards.
Stroke cases were identified by monitoring hospital admissions, medical records, and asking about stroke at each assessment cycle. Diagnoses of stroke were confirmed by doctors using valid criteria.
Dementia was detected by routine Mini-Mental State Examination at each study assessment. Those with cognitive impairment were flagged for a full review by doctors, and diagnoses of dementia were again made using valid diagnostic criteria.
Links between drinks and stroke or dementia were adjusted for the following confounders:
The researchers analysed 2,888 people aged above 45 years (average 62) for the stroke assessment, and 1,484 adults aged above 60 years (average 69) for the dementia assessment.
During follow-up there were 97 cases of stroke (3% of the cohort), 82 of which were caused by a clot (ischaemic). There were 81 new diagnoses of dementia (5%), 63 of which were consistent with Alzheimer's.
When adjusted for all confounders, there was no link between any consumption of either total sugary drinks or sugar-sweetened drinks and risk of stroke.
They did find significant links for artificially sweetened drinks when looking at recent history. Recent consumption of artificially-sweetened drinks (at the 1998–2001 assessment) was linked with risk of stroke: compared with none:
The risk was slightly higher when restricting the analysis to ischaemic stroke cases only.
For overall intake across all assessment periods, however, there was no significant link between artificially-sweetened drinks and strokes – this was only found again when restricting to the smaller number with ischaemic stroke.
The reported figure of a tripled increase with artificially-sweetened drinks comes from an HR of 2.67 (95% 1.26 to 6.97) for one or more drinks per day – but this was in the model that was not fully adjusted for all health factors – and for ischaemic stroke only.
In the fully adjusted model there were no significant links for risk of dementia (or Alzheimer's specifically) from total sugary drinks, sugar-sweetened drinks or artificially-sweetened drinks.
Again, the reported figure of a tripled risk increase with artificially-sweetened drinks comes from an HR of 2.89 (95% 1.18 to 7.07) for one or more drinks per day – but this was again in the model that was not fully adjusted – and for Alzheimer's only.
The researchers conclude: "Artificially sweetened soft drink consumption was associated with a higher risk of stroke and dementia."
The researchers used data from a large ongoing cohort study to look for links between consumption of sugary and artificially sweetened drinks and risk of stroke or dementia.
This cohort study benefits from the large overall sample size, long period of data collection, careful and valid diagnostic assessments, and adjustments for a number of confounders. However, care must be taken when interpreting these results – particularly if latching on to the maximal tripled risk figures reported in the media.
There are several points to consider:
The new number of strokes and dementia in this study was small, just 3% and 5% of the cohort, respectively. The most common category for consumption of artificially sweetened drinks in the full cohort was actually zero.
The paper doesn't report how many of the 97 people with stroke or 81 with dementia were in the highest consumption categories, but it's likely to be few. The numbers will get even smaller when restricting to the 82 with ischaemic stroke and 63 with Alzheimer's.
Analyses with smaller numbers can be less accurate, as indicated by the rather wide confidence intervals on the tripled associations.
Variable consumption measures
As said above, the researchers grouped consumption categories according to the most common response. The categories for the three different drinks aren't consistent, which makes it quite difficult to compare them with one another.
Overall this makes it very difficult to conclude with any certainty that artificially sweetened drinks carry more risk than sugary drinks.
In the fully adjusted model, links between artificially sweetened drinks and stroke were only found for the last food and drink assessment taken at wave seven.
There was no statistically significant link for stroke when looking at cumulative intake over all assessments – only when restricting to ischaemic strokes only.
Overall this makes it difficult to give any conclusive answer about the strength of the links with artificially-sweetened drinks. Arguably the cumulative, overall pattern, should give the most reliable indication – and this found no link.
Recall in food frequency questionnaires
Food frequency questionnaires are a validated measure for assessing food and drink consumption. However, people may not be able to accurately recall how much and how often they have consumed a particular drink over the past year.
Potential influence of confounders
As highlighted, the triple figures came from models that hadn't adjusted for health related factors. Full adjustment for all health and lifestyle factors gave more tentative links. Even then this analysis may not have been able to adjust for all factors that could be having an effect.
If there is a link it may not have been directly caused by artificially sweetened drinks. For example people with diabetes or obesity may be more likely to consume artificially sweetened drinks and are also more likely to develop stroke and some forms of dementia.
Lastly, this is a sample from one region of the US only. Lifestyle habits – including drink consumption – may differ and the findings may not have been the same if studying other samples.
Overall the various limitations mean that this study does not give definitive proof that drinking artificially sweetened drinks will increase risk of stroke or dementia.
When it comes to the healthiest drink to have on a day to day basis you can't go wrong with plain old tap water.