Approximately 463 million people live with diabetes mellitus (DM) worldwide (Lancet, 2017; Federation I.D., 2019; Li et al., 2020). It is considered as “the third brain disease” and presents considerable threat by exacerbating glomerular microcirculation, arteriosclerosis, retinopathy, metabolic abnormalities and neuropathy and has, thus, become a public health issue of global concern (Zheng et al., 2018). In China, the prevalence of diabetes has reached 30.2% in people aged 60 or older and is increasing yearly as the process of ageing society of China accelerates (Li et al., 2020). Dementia is a global health challenge owing to its destructive effects on the brain (Wortmann, 2012), and its prevalence has reached 6.0% in adults aged 60 or older in China (Jia et al., 2020). DM and dementia have caused huge economic losses and medical burdens in China (Xu et al., 2017; Liu et al., 2019). Given that DM is a booster for dementia, DM treatment and its relationship with dementia is worthy of special attention in China (Biessels et al., 2006; Frankish and Horton, 2017; Callisaya et al., 2019; Marseglia et al., 2019).

High glycaemia can cause vascular injury and affect blood flow (Lyu et al., 2020). Insulin resistance, a frequent symptom of DM, can increase β-amyloid levels and induce the hyperphosphorylation of the tau protein (Biessels and Despa, 2018; Lyu et al., 2020). Nevertheless, patients with DM undergoing active treatment have lower abnormally aggregated protein levels and lighter vascular injuries than those without treatment (McIntosh and Nation, 2019; van Sloten et al., 2020; Yaribeygi et al., 2020). In addition, metformin, an insulin sensitiser, can significantly reduce the rate of cognitive decline and the risk of dementia in patients with DM (Samaras et al., 2020). Using one or more diabetes medications, such as biguanides and sulfonylureas, changes the link between tau pathology and DM and slows down the progression of cognitive decline (McIntosh and Nation, 2019). Therefore, the active treatment of diabetes has been recommended for the prevention of dementia.

As a progressive disease, a long period of cognitive decline occurs before dementia (Ma et al., 2015; Pal et al., 2018). Strengthening the early prevention of cognitive impairment is the most effective measure to control the onset of dementia. Not only elderly people but also middle-aged people with DM show poorer cognitive performance and faster cognitive decline (Tuligenga et al., 2014). A retrospective study suggested that the association between DM and Alzheimer's disease (AD) is stronger in middle-aged people than in elderly people (Smolina et al., 2015), implying that age plays an important role in the relationship between DM and AD. Regardless of age, there are always subjects who develop diabetes but are not actively treated. However, whether or not diabetes treatment exerts the same cognitive benefits in middle-aged and older adults is unclear.

In this study, we attempted to evaluate the relationship between diabetes treatment behaviour and cognition in middle-aged and elderly adults and further explore the impact of age-related risk factors on the relationship between diabetes treatment behaviour and cognitive decline to evoke aggressive treatment behaviours in diabetes patients.

In this study, we selected the diabetes-free group as the reference group. We found that patients with diabetes had greater cognitive gains from active treatment than those without diabetes, but significant gains were observed only in elderly adults. The results were more obvious in patients without depression and those who never smoked and drunk and had normal weights (BMI 18.5–23.9 kg/m²).

Antidiabetic agents have anti-inflammatory, antioxidant and anti-autophagy effects (Yaribeygi et al., 2020), which are the potential mechanisms against the neurodegenerative complications of diabetes and other neurological disorders. The inflammatory hypothesis has been proposed in the pathology of diabetes complications. Antidiabetic agents can suppress the expression of inflammatory mediators and thereby alleviate inflammatory responses (Yaribeygi et al., 2019a). Diabetes disrupts the physiologic redox state, therefore promoting an increase in free radicals and the development of oxidative stress, which contribute to vascular complications. Antidiabetic medications may protect microcirculation and macrocirculation mainly through their hypoglycemic effect, reducing glucose-induced free radical generation (Yaribeygi et al., 2019b). Antidiabetic agents can also induce autophagy pathways and modulate this process, improving cognitive function (Ashrafizadeh et al., 2019).

Diabetes is a risk factor for dementia both in midlife and late-life people (Tuligenga et al., 2014; Marseglia et al., 2019). Some studies demonstrated that the effect of diabetes on dementia risk decreases with age, suggesting that the age of exposure to diabetes is related to the development of dementia (Xu et al., 2009; Cheng et al., 2012; Smolina et al., 2015; Barbiellini et al., 2021). As previously reported, diabetes onset in midlife was associated with a higher risk of dementia compared with late-life diabetes (Barbiellini et al., 2021). This phenomenon was also found in other cardiovascular risk factors. Recently, some studies indicated that the associations of several cardiovascular risk factors and vascular disorders with the risk of dementia were age-dependent, and these factors were associated with a high risk of dementia in midlife but not necessarily in late-life (Anstey et al., 2014; Legdeur et al., 2019; Olaya et al., 2019). Therefore, elderly people with diabetes are supposed to have better cognitive outcomes than middle-aged people. Our results showed that treatments for diabetes had protective effects on the cognitive function of elderly participants. However, the same phenomenon was not observed in middle-aged people. These findings suggested that the effect of diabetes treatment on the improvement of cognitive function varies with age. The subgroup analysis in this study showed that elderly participants with stroke in the treated-diabetes group were related to better cognitive function, indicating that these patients received greater cognitive benefits from active treatment of diabetes. It is possible that while treating diabetes, cardiovascular disease has also been improved. Moreover, previous studies have shown that some drugs for the treatment of cardiovascular disease also have a protective effect on cognition (Goldstein et al., 2009), so the protective effect was more obvious in these people. However, the mechanisms underlying the age-dependent effect of dementia risk are still unclear, although our study provided a novel insight for further basic research and physiological and pathological research.

According to previous reports, depression increases the risks of dementia and cognitive impairment (Rubin, 2018; Canton-Habas et al., 2020; Yang et al., 2020). In this study, the protective effect of antidiabetic treatment on cognitive function was only observed in elderly individuals without depression. Moreover, depression attenuated the cognitive protective effect of diabetes treatment. Previous studies found that the level of risk of dementia associated with comorbid diabetes and depression was higher than that associated with diabetes alone (Johnson et al., 2015; Black et al., 2018). Depression-related hypothalamic-pituitary axis dysregulation would lead to the chronic elevation of cortisol level and may damage brain regions involved in cognition. Our findings were consistent with those of prior studies and demonstrated that depression plays an adverse role in well-managed diabetes patients. The main reason is that depression is correlated with poor adherence to recommendations for prescription drugs, physical activity and healthy diet in patients with diabetes (Katon et al., 2015; Downer et al., 2016; Lunghi et al., 2017; Choi and Smaldone, 2018). Therefore, preventive measures for depression are essential to the management of diabetes patients.

Overweight, obesity, smoking and drinking are regarded as risk factors for a series of adverse health outcomes, including dementia (Durazzo et al., 2014; Sabia et al., 2014; Ma et al., 2020). Overweight and obesity can increase vascular disorders and thereby cause dementia. The most common measurement of overweight and obesity is BMI, which is a predictor of dementia and cognitive decline (Gustafson et al., 2012). Other researchers suggested that people with dementia have low BMI, and decline in BMI and underweight years before the diagnosis of dementia are related to dementia (Cova et al., 2016; Kang et al., 2021). In this study, the protective effect of diabetes treatment on cognition was observed only in elderly people with normal weights (BMI: 18.5–23.9 kg/m²), whereas people who were underweight or overweight had worse cognitive functions, suggesting that low and high BMI are predictors for cognitive decline. Smoking is associated with vascular disorders, which are potential risk factors for dementia and cognitive decline (Durazzo et al., 2014). Moreover, smoking is related to adverse effects on brain neurobiology and function. Such effects cause dementia and mild cognitive impairment. According to subgroup analysis, people who never smoked had better cognition than former smokers. Our results supported the fact that smoking is harmful to cognitive function. Alcohol drinking may affect cognitive function. The subgroup analysis in our study showed that people who never drink can benefit from diabetes treatment, supporting the finding that drinking has adverse effects on cognitive function.

Possible reasons that resulted in the differential effect of diabetes treatment on cognitive function among middle-aged people and elderly people may be the discrepancy of their brain conditions. The brains of middle-aged people may still be resilient enough, so treatment has no effect yet. But the brains of the elderly may be more vulnerable and more sensitive to the treatment of diabetes. Another reason may be that the cognitive decrement of middle-aged patients has not declined in this study, and there is no significant difference among middle-aged people with different diabetes statuses, so the results did not show the effect of treatment.

The principal strength of this study is that it provides a nationally representative sample of community-dwelling middle-aged and older populations in China. The study examined the relationship between diabetes status and cognitive decline in midlife and late-life people and found that diabetes treatment effects vary with age. However, some limitations need to be acknowledged. First, some diabetic patients were unaware of their illnesses in the self-reported questionnaire and were diagnosed using glucose or HbA1c levels. Hence, they were included in the untreated-diabetes group. This situation may have caused misclassification bias. Second, the measurement of cognitive function was based only on the self-reported scales in the questionnaire, and guidance from professional physicians was lacking. Neuropsychological assessment in such a large-scale study is difficult to complete. On account of the limitation of the cognition assessment method in this study, it is difficult to generalise the results to other cognitive domains. Further studies can improve this study by conducting a more integrated and formal neuropsychological assessment. Third, information on medication adherence was insufficient, and some people in the treated-diabetes group may have not taken medications as prescribed. The exact type of antidiabetic medications was unclear. Therefore, which types of medication exerted effects were not identified in this study. Fourth, the covariables HbA1c and FBG were not considered because of the large number of missing data. Individuals without HbA1c and FBG data were categorised only with self-reported diagnosis information. Last, the reason that these people had not been treated for diabetes is unclear. The factors that affected diabetes treatment may have also affected cognition.

In summary, we revealed that diabetes treatment is associated with enhanced cognitive performance in late-life diabetes. However, the same phenomenon was not observed in midlife diabetes, thus warranting further studies. Despite that antidiabetic treatment had a protective effect on participants aged ≥60 years, depression, smoking, drinking and an abnormal BMI may have mitigated this effect.

Publicly available datasets were analysed in this study. This data can be found here: http://charls.pku.edu.cn/.

The studies involving human participants were reviewed and approved by the Biomedical Ethics Review Committee of Peking University (IRB00001052-11015). The patients/participants provided their written informed consent to participate in this study.

KW wrote the article. KW and HL performed the data analysis. HL and JZ drafted and critically revised the manuscript. LZ and SG provided clinical guidance. RZ, ZY, and ZH organised database. XW contributed to the study concept and design and reviewed the article. All authors contributed to the article and approved the submitted version.

This work was supported by the National Natural Science Foundation of China (82173607), the Basic and Applied Basic Research Foundation of Guangdong Province (2021A1515011684), the Open Project of the Guangdong Provincial Key Laboratory of Tropical Disease Research (2020B1212060042), and Guangzhou Municipal Science and Technology Project (202102080597).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.