Back

Among patients with pneumonia-related sepsis, do those with mental health problems have different radiographic findings than those without mental health problems?

Ng, H. A. H.; Puca, D.; Perla, J.; Richman, M.

2026-05-01 radiology and imaging
10.64898/2026.04.30.26352126 medRxiv
Show abstract

IntroductionPneumonia is a common source of sepsis and carries significant morbidity and mortality. Prior research suggests patients with severe mental illness may receive disparate care and experience worse outcomes from infectious and cardiovascular conditions. Whether mental illness influences the radiographic presentation of pneumonia-related sepsis, which can guide antimicrobial selection and clinical decision making, remains poorly-understood. MethodsThis retrospective study examined chest imaging findings in 202 adult septic patients with respiratory source infection presenting to a large tertiary care emergency department between December 2017 and December 2019. Patients were stratified by the presence (n = 51, 25.2%) or absence (n = 151, 74.8%) of a comorbid mental illness, defined broadly to include neurocognitive, mood, personality, psychotic, and substance use disorders. Radiographic findings, including parenchymal abnormalities, pleural effusions, and laterality, were systematically abstracted from the medical record and compared between groups using Chi-squared testing. ResultsPatients with mental illness were more likely to have at least one radiographic finding (84% vs. 74%), though this difference did not reach statistical significance (p = 0.14). The average number of findings per patient was nearly identical between groups (3.14 vs. 3.12, p = 0.38), and no individual radiographic feature differed significantly between cohorts. DiscussionThese findings suggest that, contrary to our hypothesis, mental illness may not be associated with the radiographic appearance of pneumonia in septic patients. Larger, diagnosis-specific studies are needed to evaluate whether specific psychiatric subgroups exhibit distinct imaging patterns.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.