¹University of Cambridge, Department of Psychiatry, Cambridge, United Kingdom, ²Cambridgeshire and Peterborough NHS Foundation Trust, CPFT, Cambridge, United Kingdom

Background: TD is a movement disorder caused by exposure to dopamine receptor antagonists such as antipsychotics. TD is estimated to affect 20-30% of antipsychotic-treated patients, but is underdiagnosed. Unstructured EHR data, such as text notes recorded by clinicians, can allow for identification of patients lacking a formal diagnosis.

Aims & Objectives: In this study we estimated TD prevalence in a UK EHR database using unstructured free-text records, then investigated the demographic and clinical characteristics of patients with recorded TD and their clinical outcomes, compared to antipsychotic-treated patients without recorded TD.

Method: We conducted a retrospective study using de-identified EHR data from the Cambridgeshire and Peterborough NHS Foundation Trust (CPFT) via the CPFT Research Database (NHS Research Ethics 22/EE/0264). For patients with recorded TD, the index date was defined as the date of referral before the first TD description, and for patients without recorded TD, the index date was the date of referral before their first recorded antipsychotic prescription. Patients were included if they were >=18 years old with a minimum 6 months of baseline data before and 12 months of follow-up data after the index date. Patients without recorded TD were included if they had a recorded antipsychotic prescription. Patients with Huntington’s disease were excluded. Patients with recorded TD were identified by classification of text records relevant to TD. Demographic and baseline clinical variables were compared between patients with and without recorded TD using non-parametric tests. Poisson or negative binomial regression was performed to investigate the relationship between TD and clinical outcomes. In multivariable analyses, baseline demographic and clinical variables were used as covariates.

Results: ​​​​While there were no patients with a structured TD diagnosis, 515 patients were identified with either confirmed (n=313) or possible (n=202) TD from free-text records, and 19,801 patients were identified without recorded TD. Of these n=296 and n=6244 respectively met inclusion criteria, indicating TD prevalence of 4.7%. Patients with TD were significantly more likely to have amisulpride (X2=7.65, p=0.006), clozapine (X2=18.03, p<0.001), flupentixol (X2=91.13, p<0.001), or multiple antipsychotics (X2=29.20, p<0.001) as their first recorded antipsychotic, and less likely to have quetiapine (X2=86.58, p<0.001). In follow-up, patients with TD had an average of 16.1 inpatient days (SD:42.0) and 2.4 outpatient referrals (SD:6.6) compared to 6.3 days (SD:26.6) and 1.9 referrals (SD:4.99) for patients without recorded TD. Multivariable regression revealed that in follow-up, recorded TD was associated with a significantly higher number of unique antipsychotics (IRR:1.62, 95% CI:1.47-1.79, p<0.001) and significantly more outpatient visits (IRR:1.31, 95% CI:1.05-1.65, p=0.019). Univariate analysis indicated that patients with recorded TD had significantly more inpatient days (IRR:2.57, 95% CI:1.35–5.68, p=0.009), but this was no longer significant in multivariable analysis (IRR:1.9, 95% CI:0.86–5.28, p=0.122).

Discussion & Conclusions: ​​​​TD is significantly underdiagnosed in structured healthcare records. Analysis of free text can identify these patients, which is increasingly important considering the emergence of novel TD treatments. TD was associated with a significantly higher number of unique antipsychotics and outpatient referrals. Further analysis will investigate the timing of first documentation of specific antipsychotics in relation to first TD description.