This study evaluates the characteristics of clinical studies on dementia to identify the reasons behind early discontinuation and non-publication, aiming to prevent these issues in the future.

We searched ClinicalTrials.gov for all clinical studies related to dementia up to September 2025. Data on enrollment, study location, Funder type, and intervention type were extracted and analyzed using binomial logistic regression.

A total of 3,787 eligible registered clinical studies were included in the analysis. Of these, 3,266 (86.2%) were completed, and 521 (13.8%) were discontinued. Unadjusted logistic regression analysis identified several significant predictors:

- Studies with small sample sizes (fewer than 100 participants) were significantly more likely to be discontinued (OR = 0.56, 95% CI [0.44–0.72], P < 0.001).

- Studies conducted at a single location were more likely to be completed than those conducted at multiple locations (OR = 1.45, 95% CI [1.14–1.85], P = 0.003).

- Studies with industrial funding were more likely to be discontinued compared to those with non-industrial funding (OR = 0.76, 95% CI [0.58–0.996], P = 0.046).

- studies Included only males were more likely to be completed than studies included both gender (OR = 0.46, 95% CI [0.22–0.99], P = 0.047).

There is evidence of non-dissemination bias in clinical studies of dementia. This bias raises ethical concerns about exposing volunteer participants to potential risks without advancing medical knowledge.