Claudia Daniela Lupu, Vlad-Dan Cotuțiu, Victoria Birlutiu
Hospitalized COVID-19 frequently involves several organ systems at once, yet neuropsychiatric manifestations are often examined in isolation. In a retrospective single-center cohort study, a five-axis phenotype atlas was constructed (respiratory, cardiovascular, neuropsychiatric, hepato-gastrointestinal, and renal-urinary) from ICD-linked hospital records of 1361 adults hospitalized with confirmed SARS-CoV-2 infection at a Romanian county emergency hospital between March 2020 and January 2025. Candidate neuropsychiatric associations were evaluated under prespecified cross-axis adjustment, treatment- and indication-bias gates, penalized outcome models, and pandemic-era comparison. Multi-organ involvement was almost universal: 91.0% of admissions carried at least two axes. Neuropsychiatric positivity reached 22.9% but was dominant in 0.8%, while 98.1% of these patients carried another chronic organ axis. Of the 42 initial laboratory candidates, three correlated electrolyte associations remained after adjudication, suggesting that these associations were more consistent with shared renal-metabolic burden than with neuropsychiatric-specific biology. Renal-urinary involvement was independently associated with intensive care admission (OR 1.97) and in-hospital death (OR 2.58); neuropsychiatric positivity was not. The previously reported Late Omicron enrichment of severe neurological presentations remained evident after age and sex adjustment (adjusted OR 3.35) and was the only one of 35 prespecified claims retained as neuropsychiatric-specific. Renal-urinary rather than neuropsychiatric involvement carries the adjusted prognostic weight in hospitalized COVID-19, and, given the dependence of retrospective record-derived phenotypes on clinical documentation, multi-organ adjudication should precede organ-specific inference from hospital records.