Sepsis on dementia risk: A population-based cohort study with dose-dependent analysis

Purpose

Emerging evidence links sepsis-related inflammation to dementia risk, but the dose-dependent effects of recurrent sepsis episodes remain unclear. This study assessed whether sepsis increases dementia risk and explored a potential dose-response relationship between sepsis frequency and dementia.

Methods

We conducted a retrospective cohort study using Taiwan's National Health Insurance Research Database (2005–2022), enrolling patients aged ≥18 years hospitalized with sepsis. An index period (2008–2013) was used to identify patients without prior dementia. A fixed 12-month landmark period (calendar year 2014) was applied to assess the number of sepsis episodes. Patients who died during the landmark period were excluded. Propensity score matching was performed to generate well-balanced sepsis and non-sepsis groups. Follow-up for dementia and mortality began after the landmark period and continued for up to 64 months. Cox and Fine-Gray models were used to account for competing risks of death.

Results

Sepsis was associated with a significantly increased risk of all-cause dementia (HR 1.59; 95 % CI, 1.47–1.72; P < 0.0001). A dose-response relationship was observed: patients with multiple sepsis episodes had the highest dementia risk (sHR 1.63; 95 % CI, 1.39–1.91). Vascular dementia showed the strongest association, with a higher incidence in the sepsis group (1.2 % vs. 0.6 %, P = 0.0003).

Conclusions

Our findings provide robust evidence of a dose-dependent association between sepsis and increased dementia risk, particularly vascular dementia, even after adjusting for competing mortality risks.

Summary

This study addresses the limitations of previous research by not only employing propensity score matching (PSM) to balance confounding factors between the sepsis and non-sepsis groups but also using an index period and landmark period design to better explore potential causal relationships. These periods ensure that dementia onset occurred after sepsis and allow for the examination of dose-response relationship between sepsis episode frequency and dementia. Furthermore, this is the largest study to date involving sepsis patients, providing more robust evidence than prior studies, which were often smaller and lacked adjustments for competing risks of death.