Answers are grounded in this paper and cite their sources.
Natural language processing reveals modest differences in sex-related clinical features in memory clinic patients, suggesting variations in documentation practices.
• Natural language processing was applied to 1,036 routine memory clinic letters to examine sex differences in documented clinical features • We found showed substantial overlap in documented clinical features between women and men, with modest differences in a small subset of features • No sex differences remained statistically significant after correction for multiple comparisons • Documented clinical features reflect a combination of symptom occurrence, patient reporting, and clinician documentation practices • The study demonstrates the applicability of NLP for investigating documentation patterns in routine clinical data Sex differences in signs and symptoms of cognitive impairment are well described in research cohorts, but whether they are reflected in routine clinical documentation is unknown. We applied natural language processing to clinical letters from 915 memory clinic patients in the Utrecht Patient-Oriented Database. Sex differences in documented clinical features were assessed with logistic regression, with and without age adjustment. Benjamini–Hochberg correction was applied for multiple comparisons. We included 1,036 clinical letters from 915 patients (47% women; mean age 66 ± 14 vs. 67 ± 13 years, p = 0.6). The most common clinical features were memory problems (56.5% overall) and forgetting intended actions (49.6%). After correction for multiple comparisons, no significant sex differences remained. At the nominal level, women more often had documentation of stress (OR 0.39), headache (OR 0.63), and calendar use for memory support (OR 0.66); men more often had documentation of getting angry easily (OR 1.72) and short-term memory problems (OR 1.70). This exploratory study identified suggestive but non-significant sex differences in documented clinical features in memory clinic practice. Differences may reflect variation in occurrence, reporting, or documentation habits, and warrant replication in larger, diagnosis-stratified samples.
No takes yet. Share an insight, caveat, or question.
Kuipers et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: