Why dementia is often not diagnosed by subtype: care pathways and imaging access in the United Arab Emirates.

Background: Whether a person with dementia receives a specific subtype diagnosis rather than a nonspecific label is a marker of diagnostic completeness. Subtype shapes safe management, including psychotropic prescribing, behavioural and psychological symptom management, and eligibility for Alzheimer's disease therapies, yet nonspecific labels remain common where specialist assessment is unevenly accessed. We examined which patient, clinical, and service-pathway factors were associated with specific subtype assignment in a dementia cohort in the United Arab Emirates (UAE), and whether an exploratory model could help identify patients who might remain without a subtype. The outcome reflects the specificity of the recorded diagnosis, not adjudicated accuracy. Methods: We conducted a retrospective cohort study of patients with dementia seen at the Behavioural Sciences Institute, Al Ain Hospital from 1 January 2010 to 31 December 2019. The primary outcome was specific subtype assignment versus unspecified dementia. Factors examined were age, sex, nationality, brain imaging utilisation, comorbidity burden, symptom cluster, location of first assessment, and year. We fitted multivariable logistic regression with internal validation, sensitivity analyses, multiple imputation, scanned-subgroup analysis, and decision-curve analysis. Results: Of 825 patients, 786 entered the complete-case analytic sample; 336 (42.7%) had an unspecified diagnosis and 450 (57.3%) received a specific subtype label. Brain imaging showed the strongest adjusted association with specific subtype assignment (adjusted odds ratio = 4.00; 95% confidence interval = 2.84-5.63). Lower odds of specific assignment were associated with three or more comorbidities (0.40, 0.29-0.57), behavioural presentation (0.60, 0.39-0.94), emergency department route (0.29, 0.16-0.51), other route (0.31, 0.20-0.48), and neurology inpatient route (0.44, 0.22-0.90). Among scanned patients, multimorbidity and nonclinic routes remained associated with unspecified labelling. The prescan model showed positive net benefit across thresholds of 0.20-0.50. Conclusions: Diagnostic completeness was associated less with demographics than with imaging access, clinical complexity, and route of entry to care. Maintaining access to imaging within dementia pathways, and to timely specialist psychiatric assessment for acute and nonspecialist routes, may be associated with more complete subtype recording in the UAE and comparable Gulf systems. An exploratory prescan model identified patients at risk of remaining without a recorded subtype, though external validation is required before clinical use.