Association between pre-diagnostic prevalence of 30 common diseases and the subsequent risk of dementia: a population-based retrospective cohort study in Taiwan

Background: Dementia associates with long-term progression, but the health status of individuals years before diagnosis remains poorly understood. We therefore conducted a large retrospective cohort study to examine the association between early-life morbidity and incident dementia. Methods: We designed a large-scale retrospective matched-cohort study using data from the National Health Insurance Research Database of Taiwan. Individuals newly diagnosed with dementia (2010-2013) were identified as the pre-dementia cohort and matched 1:3 with non-dementia controls. We assessed prevalence and outpatient visits for 30 common conditions (infectious to various of cancers.) between 2000 and 2013, with a backtracking time of 10 years for each individual. Conditional logistic regression was used to estimate odds ratios (ORs) for disease prevalence, while Negative Binomial regression was applied to calculate Incidence Rate Ratios (IRRs) for outpatient visit. Results: From 806,292 individuals, 19 of 30 diseases showed significant associations. Notably, the pre-dementia cohort exhibited a significantly lower prevalence of and fewer hospital visits for 13 conditions compared to controls. These included acute respiratory infections, peripheral inflammatory diseases, and certain digestive system malignancies. Conversely, a significantly higher prevalence was observed in only six conditions, predominantly those associated with direct or indirect brain injury, as well as cancers of the respiratory and urinary systems. Conclusion: Pre-dementia patients showed lower prevalence of most correlated common diseases. These unexpected findings are consistent with the hypothesis that altered peripheral immune function may be involved in the etiology and pathogenesis of dementia, which is beneficial for normal health but detrimental to the brain. Meanwhile, conditions that directly or indirectly contribute to brain damage are more likely to be positively associated with dementia. However, given the observational nature of our study and the absence of immunological biomarkers, these mechanistic interpretations remain speculative and require further validation.