BackgroundRapidly progressive dementia (RPD) is a syndrome originating from various diseases. Recent advances have allowed a better understanding of its categories and spectrum; however, it remains challenging to make an accurate differential diagnosis and prognosis prediction.MethodsThis study was a retrospective evaluation of all participants admitted to the neurology department of a single center in China from January 2015 to December 2019. The screened patients met the RPD criteria and their characteristics were collected to explore a diagnostic pattern of RPD. In addition, outcomes of RPD were evaluated with the Glasgow Outcome Scale (GOS), activities of daily living scale (ADL), and simplified Mini-Mental State Examination (MMSE), and different prognostic analysis methods were performed to determine the prognostic factors of RPD.ResultsA total of 149 RPD patients among 15,731 inpatients were identified with an average MMSE value of 13.0 ± 4.6 at baseline. Etiological epidemiology revealed infectious, neurodegenerative and toxic/metabolic diseases as the three largest groups, accounting for 26.2%, 20.8% and 16.8% of all cases, respectively. In particular, prevalence rates of Creutzfeldt–Jakob disease (13.4%), Alzheimer’s disease (11.4%), carbon monoxide poisoning (8.1%), neurosyphilis (5.4%) and dementia with Lewy bodies (5.4%) were highest in this series. A recommended diagnostic framework for RPD etiology was thus established. Follow-up evaluations showed a negative correlation between age and GOS scores (r=-0.421, P < 0.001), as well as age and simplified MMSE scores (rs =- 0.393, P < 0.001), and a positive correlation between age and ADL scores (rs =0.503, P < 0.001), and significantly different GOS, ADL and simplified MMSE scores across various etiologies (P = 0.003; F = 9.463, P < 0.001; F = 6.117, P < 0.001).ConclusionInfectious, neurodegenerative and toxic-metabolic entities were the most common RPD categories, and establishing a practical approach to RPD etiology would allow better disease management.