. Association Between Age at Diabetes Onset and Subsequent Risk of Dementia. JAMA. 2021 Apr 27;325(16):1640-1649. PubMed.


Please login to recommend the paper.


  1. In this large and well-conducted cohort study, a higher risk of all-cause dementia was observed in participants with Type 2 diabetes (T2D) compared to those without T2D. This is consistent with lots of previous research—we know T2D is not good for our brains.

    What this study adds is to address the question of whether the duration of diabetes and the age at which it was first diagnosed change the way we observe that increased risk of dementia. Not too surprisingly, they found evidence participants with T2D diagnosed at younger ages had higher risk of dementia compared to participants without diabetes. They did not directly compare the participants with T2D to each other.

    What they found was that, for example (Table 2), if you survive to age 70 without being diagnosed with dementia, being diagnosed with T2D at or after 65 is not associated with higher risk of dementia compared to no diagnosis of T2D, but being diagnosed with T2D at ages 60 or younger is associated with about twice the risk of dementia compared to no diagnosis of T2D.

    Whether preventing T2D would be important for preventing dementia is an important question. While there is some disagreement in the field, I am comfortable saying that I think the primary way T2D affects brain health is through the damage it does to our blood vessels rather than through increasing risk of Alzheimer’s disease. We just don’t see a strong relationship between T2D and AD pathology in autopsy studies, while we do see strong relationships with brain infarctions.

    Preserving our vascular health (and preventing T2D) could have benefits with regard to preventing all-cause dementia in that we think that cerebrovascular damage in the brain means that it takes less AD pathology to cause someone to have symptoms of dementia compared to a brain with less cerebrovascular damage. And, because we think T2D shares common risk factors with lots of other chronic conditions and diseases (e.g., hypertension, coronary heart disease), successful efforts to prevent T2D would likely spill over and prevent some of those other conditions as well, further protecting brain health.

    So, even though only 79 of 542 dementia cases in this study were in participants diagnosed with T2D, perhaps suggesting that preventing T2D would not be terribly effective in reducing the overall burden of dementia,  population-level efforts to prevent T2D would likely also have helped the 463 of 542 participants with dementia but who did not have T2D. Heart health is brain health.

    View all comments by Erin Abner
  2. While there is a plethora of studies evaluating the association between the presence of diabetes and dementia, scarcely any are focused on assessing the effect of age at diabetes onset, diabetes duration, and diabetes severity both at onset and during the course of the disease. These diabetes-related factors are often not independent and may even moderate the effect of one another. Further understanding of how such factors are related to the risk of dementia may help enhance individualized dementia risk assessment in patients with diabetes and helps piece together clues about the etiology of dementia.

    This study sheds light on one of these diabetes-related factors, age at onset, and provides evidence of yet another negative impact of early onset Type 2 diabetes. Indeed, the authors tell a compelling story about dementia risk based on diabetes age of onset, duration, prediabetes, and concomitant vascular comorbidities. Importantly, this study uses prospective cohort data linked with longitudinal real-world data over a long period of follow-up (median=31.7 years) and accounts for a wide range of relevant confounders, such as comorbidities, genetics (ApoE allele), education, and other lifestyle factors. The findings indicate that not all diabetes patients have a similar risk of dementia; in fact, the risk of dementia is higher for those diagnosed with Type 2 diabetes at a younger age. For every five years younger age at diabetes onset, the risk of dementia is increased by approximately 24 percent. The strongest link between diabetes and dementia was found in those with a diabetes onset between 50 and 55 years of age, compared to those without diabetes at age 55. Risk of dementia doubled (Hazard ratio 2.14) for those with diabetes at age 55 compared to those without. Surprisingly, this risk of dementia dissipated as the age of diabetes onset increases. For those diagnosed between age 65 and 70, there was no difference in dementia risk compared to those without diabetes at age 70.

    A similar observation was not observed in those with prediabetes, which is somewhat surprising given that some of the purported pathophysiologic mechanisms would be expected to be present, such as hyperinsulinemia and insulin resistance. Another important finding from this paper is the relevance of vascular complications in those with Type 2 diabetes, particularly stroke. Compared to those without a stroke, coronary heart disease or heart failure, the risk of dementia doubled in those with diabetes and stroke and was five times higher in those with diabetes and all three vascular complications.

    While our efforts to prevent diabetes should continue, the main take-home message from this important evidence is the need to focus on preventing, or at least delaying, the early onset of diabetes. Why? Because if we can delay diabetes by even five years, the number of cases of dementia will decrease. Current trends are concerning, however, as over the past decade, the average age of diabetes diagnosis has decreased and the prevalence of diabetes among younger age groups has increased dramatically (Koopman et al., 2005; Alberti et al., 2004). The prevalence of prediabetes is also increasing worldwide (Pal et al., 2021); however, the lack of association between prediabetes and incident dementia is reassuring. Lastly, the significantly negative impact of glucose-related macrovascular complications is worrisome. Nonetheless, both findings are critical in aiding clinicians and patients in building a better disease-management program.   

    What may be the population-level impact of preventing or delaying diabetes? The most recent study focused on dementia prevention has estimated the population-attributable fraction of incident all-cause dementia due to diabetes to be 3.2 percent, however, this estimate was reduced to 1.2 percent after adjusting for communality with the other eight modifiable risk factors, including obesity, hypertension, and physical inactivity (Livingston et al., 2017). This indicated that more than 1 percent of all-cause dementia cases can be attributed purely to diabetes. Although this might seem minimal, it is not negligible, given the increase in the prevalence of dementia and its public health burden and substantial social, psychological, and health impacts on patients and caregivers. It is noteworthy to mention that the impact of diabetes on the different types of dementia is potentially not equivalent, with most literature indicating a stronger association with vascular dementia compared to Alzheimer disease.

    Although this study provides insight on the age of onset of diabetes, while carefully controlling for disease duration, the association between other clinically relevant diabetes-related factors and dementia remains poorly understood. For example, this study does not provide insight on the differential effects of diabetes medications, which may impact dementia risk (McMillan et al., 2018). In short, this study moves the research direction beyond looking at diabetes as a binary, yes-or-no condition, and considers its heterogenous nature and how diabetes patients differ among each other both at onset and over the course of the disease.


    . Changes in age at diagnosis of type 2 diabetes mellitus in the United States, 1988 to 2000. Ann Fam Med. 2005 Jan-Feb;3(1):60-3. PubMed.

    . Type 2 diabetes in the young: the evolving epidemic: the international diabetes federation consensus workshop. Diabetes Care. 2004 Jul;27(7):1798-811. PubMed.

    . Time trends in the incidence of clinically diagnosed type 2 diabetes and pre-diabetes in the UK 2009-2018: a retrospective cohort study. BMJ Open Diabetes Res Care. 2021 Mar;9(1) PubMed.

    . Dementia prevention, intervention, and care. Lancet. 2017 Jul 19; PubMed.

    . Impact of pharmacological treatment of diabetes mellitus on dementia risk: systematic review and meta-analysis. BMJ Open Diabetes Res Care. 2018;6(1):e000563. Epub 2018 Nov 16 PubMed.

    View all comments by JM Gamble
  3. We know that cardiovascular risk factors, such as diabetes or hypertension, are also risk factors for dementia. Dementia takes at least 20 years to develop, so identifying risk factors is difficult because we have to go back in time to midlife.

    A number of changes happen just before or at the time of dementia diagnosis, such as involuntary weight loss, that make it hard to interpret risk factors just before dementia onset: For example, we know that people who lose weight in old age are more likely to be in the process of developing dementia, which makes it hard to establish an association with weight or BMI in those older cohorts. Weight may confound blood glucose.

    Additionally, a risk factor needs time to act. This is probably why the authors looked at five and 10 years before onset, which would be a bare minimum to detect a true risk factor.

    This is a large longitudinal cohort study with prospectively collected data, with a follow-up over 30 years. It's the best kind of study available for this kind of research question since it is impossible to randomize people to get diabetes in a clinical trial.

    I'd be interested to know more if this group checked kidney function. In a study from our group, kidney function decline was a strong risk factor for dementia. Low kidney function is a consequence of diabetes, so it would be interesting to see if part of the effect of diabetes was mediated via a decline in kidney function (Xu et al., 2020).

    I would also like to know if adequate treatment of diabetes or specific treatments reduce the risk for dementia. There are a number of things we can do to prevent diabetes, and those will probably have a greater impact at population level. Still, it is important to know which treatments to choose in diabetic patients. In a recent Ph.D. thesis from our group, certain newer antidiabetic medications were associated with slower cognitive decline and better survival in patients with dementia (see KI News January 2021). 


    . Population attributable fraction of dementia risk in individuals with reduced kidney function. Alzheimer's & Dementia, 07 December 2020

    View all comments by Sara Garcia-Ptacek
  4. This is a very well-conducted study, with notable strengths including using large U.K. national datasets with more than 10,000 persons aged 35 years and older, with a median follow-up of 32 years. Results are novel and convincing, showing that a younger age of onset of diabetes is associated with increased dementia risk.

    The study adds to the current data on the importance of preventing diabetes or managing existing diabetes, to decrease brain complications such as cognitive decline and dementia.  Given the global increase in numbers of people with diabetes and with dementia, this study underscores the importance of further study on the topic, including mechanistic studies elucidating the link between diabetes and brain function, to further advance scientific knowledge, shift clinical care practices, and ultimately improve public health.

    View all comments by Zoe Arvanitakis
  5. I congratulate the authors on a very well conducted study providing another piece of the puzzle to the endocrine-neurological conundrum of diabetes and dementia. The first mentions of a putative increased risk of dementia in diabetes are approximately 20-30 years old; however, the field is far from exhausted, and we still know too little to provide actual targeted interventions on a global scale. The work by Amidei and colleagues certainly provides a roadmap as far as diabetes onset is involved. Although the findings may have been somewhat expected—the duration of the exposure to a noncommunicable factor certainly plays a role in the development of many outcomes—there were to this date few large-scale studies with sound methodology to quantify the effect of when diabetes affects risk of dementia.

    The graded association with five-year decrements of time certainly provides consistency. It challenges the rigorous work of Livingston and colleagues (Livingston et al., 2020) on two fronts—(a) diabetes may be a risk factor for dementia from mid-life all through the late life and (b) diabetes is likely responsible for more than the estimated 1.1 percent of dementia cases (Livingston et al., 2020). 

    Certainly, this research moves the needle of intervention timing further back and highlights the life-course perspective we should be adapting as a gold standard when evaluating risk factors for dementia. Clinically, it is evident we should press toward primary prevention of diabetes Type 2 in mid-life and even sooner.

    While the design was elegant, I cannot agree with all the interpretations—exempli gratia the lack of effect of prediabetes may simply lie in the inability to measure the glucose variability precisely. Second, an inclusion of marginal structural modeling to deal with time-dependent confounding in the time-updated models would have been a welcome improvement. A further interesting avenue would have been the evaluation of diabetes in patients with already manifest dementia, as approximately 20 percent of dementia patients live with diabetes (Bunn et al., 2014). 

    However, the main message of the article resonates: The pandemic of diabetes contributes significantly to the epidemic of dementia, and with earlier diabetes onset we can expect more damage to the brain vasculature and neuronal dyshomeostasis. This is an exciting time for researchers in the field, and I hope we will be able to offer specific pharmacological treatments for patients suffering from diabetes to prevent the onset of dementia.


    . Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. Lancet. 2020 Aug 8;396(10248):413-446. Epub 2020 Jul 30 PubMed.

    . Comorbidity and dementia: a scoping review of the literature. BMC Med. 2014 Oct 31;12(1):192. PubMed.

    View all comments by Juraj Sečník

Make a Comment

To make a comment you must login or register.

This paper appears in the following:


  1. Diabetes in Mid-Life Drives Up Dementia Risk