Patterns of symptoms before a diagnosis of first episode psychosis: a latent class analysis of UK primary care electronic health records.

Chen, Ying; Farooq, Saeed; Edwards, John; Chew-Graham, Carolyn A.; Shiers, David; Frisher, Martin; Hayward, Richard; Sumathipala, Athula; Jordan, Kelvin P.

Abstract

Background: The nature of symptoms in the prodromal period of first episode psychosis (FEP) remains unclear. The objective was to determine patterns of symptoms recorded in primary care in the 5-years before FEP diagnosis.

Methods: The study was set within 568 practices contributing to a UK primary care health record database (Clinical Practice Research Datalink). Patients aged 16-45 years with a first coded record of FEP, and no antipsychotic prescription more than one year prior to FEP diagnosis (n=3,045) were age, gender and practice matched to controls without FEP (n=12,180). 55 symptoms recorded in primary care in the previous 5 years, categorised into eight groups (mood-related, ‘neurotic’, behavioural change, volition change, cognitive change, perceptual problem, substance misuse, physical symptoms), were compared between cases and controls. Common patterns of symptoms prior to FEP diagnosis were identified using latent class analysis.

Results: Median age at diagnosis was 30 years, 63% were male. Non-affective psychosis (67%) was the most common diagnosis. Mood-related, ‘neurotic’, and physical symptoms were frequently recorded (>30% of patients) before diagnosis, and behavioural change, volition change, and substance misuse were also common (>10%). Prevalence of all symptom groups was higher in FEP patients than controls (adjusted odds ratios 1.33-112). Median time from first recorded symptom to FEP diagnosis was 2-2.5 years except perceptual problem (70 days). The optimal latent class model applied to FEP patients determined three distinct patient clusters. ‘No or minimal symptom cluster’ (49%) had no or few symptoms recorded; ‘affective symptom cluster’ (40%) mainly had mood-related and ‘neurotic’ symptoms; and ‘multiple symptom cluster’ (11%) consulted for three or more symptom groups before diagnosis. The multiple symptom cluster were more likely to have drug-induced psychosis, be female, obese, and have a higher morbidity burden. Affective and multiple symptom clusters showed a good discriminative ability (C-statistic 0.766; sensitivity 51.2% and specificity 86.7%) for FEP, and many patients in these clusters had consulted for their symptoms several years before FEP diagnosis.

Conclusions: Distinctive patterns of prodromal symptoms may help alert general practitioners to those developing psychosis, facilitating earlier identification and referral to specialist care, thereby avoiding potentially detrimental treatment delay.

Citation

Chen, Y., Farooq, S., Edwards, J., Chew-Graham, C. A., Shiers, D., Frisher, M., …Jordan, K. P. (2019). Patterns of symptoms before a diagnosis of first episode psychosis: a latent class analysis of UK primary care electronic health records. BMC Medicine, 17, Article 227. https://doi.org/10.1186/s12916-019-1462-y

• Journal Article Type: Article
• Acceptance Date: Nov 5, 2019
• Publication Date: Dec 4, 2019
• Publicly Available Date: May 26, 2023
• Journal: BMC Medicine
• Publisher: Springer Verlag
• Peer Reviewed: Peer Reviewed
• Volume: 17
• Article Number: 227
• DOI: https://doi.org/10.1186/s12916-019-1462-y
• Keywords: first episode psychosis, symptom cluster, general practice, medical record research, latent class analysis, epidemiology
• Publisher URL: https://doi.org/10.1186/s12916-019-1462-y
• PMID: 31801530