Sex differences in prodromal features preceding bipolar disorder: evidence from electronic health records integrating natural language processing

Introduction

Bipolar disorder (BD) typically emerges during adolescence and early adulthood, yet its prodromal phase is characterised by heterogeneous presentations, contributing to frequent misdiagnosis and substantial delays in diagnosis(1-3). Early clinical features often overlap with depression, anxiety and behavioural disturbances, limiting the utility of single symptom predictors (4). While previous work has identified potential early markers, sex differences in prodromal trajectories remain insufficiently characterised. Electronic health records (EHRs), combined with natural language processing (NLP), provide an opportunity to capture symptom evolution in routine clinical settings and may help refine early detection strategies.

Objectives

To examine sex differences in diagnoses, clinical features and treatment patterns preceding first recorded bipolar-spectrum disorder, and to explore whether these reflect distinct prodromal characteristics.

Methods

We conducted a retrospective cohort study using de-identified EHR data from the South London and Maudsley NHS Foundation Trust, accessed via the Clinical Record Interactive Search platform. Individuals were identified through two complementary ascertainment pathways: (a) a structured diagnosis group, including individuals with at least one recorded ICD-10 diagnosis of bipolar-spectrum disorder (F30.x/F31.x), and (b) a NLP-identified group, including individuals with evidence of BD derived from validated NLP algorithms applied to unstructured clinical text. Variables included mood, anxiety, neurodevelopmental and substance use disorders, as well as symptom-level features such as irritability and aggression. Medication use prior to diagnosis was also examined. Sex-stratified comparisons were conducted using χ² tests.

Results

A total of 10,799 individuals with bipolar-spectrum disorder were identified across both ascertainment pathways. Of these, 4,434 were identified through structured diagnoses and 6,365 through natural language processing. Marked sex differences were observed across previous diagnosis, symptoms and medications in the structured diagnosis group (Table 1). Males showed higher rates of prior psychosis diagnosis, irritability and aggression symptoms, as well as cannabis use. Neurodevelopmental conditions, including ADHD and ASD, were also more frequent in males. In contrast, females more frequently presented with anxiety and depressive disorders, as well as eating disorders. Treatment patterns differed by sex, with higher antipsychotic use in males and higher antidepressant use in females. These differences were also observed in the NLP-identified group and were more pronounced for psychosis, aggression and substance use.

	Female (n=2749)	Male (n=1671)	p-test
Prior psychosis diagnosis	8,7%	12,8%	<0.001
Prior anxiety diagnosis	7.5%	4.8%	<0.001
Prior depression diagnosis	11.2%	6.6%	<0.001
Prior Attention-deficit/hyperactivity disorder (ADHD)	2.5%	4.1%	0.023
Prior Autism Espectrum disorder (ASD)	0.9%	3.2%	<0.001
Prior eating disorder	1.9%	0.2%	<0.001
Prior cannabis use	32.5%	43.6%	<0.001
Prior irritability	38.7%	42.5%	0.041
Prior agression	38.3%	46.9%	<0.001
Prior antipsychotic medication	44.5%	50.1%	0.001
Prior antidepressant medication	33.1%	23.1%	<0.001

Conclusions

These findings indicate distinct prodromal pathways to BD by sex, with predominantly externalising and substance-related presentations in males and internalising, affective profiles in females. They highlight the need to adapt clinical assessment and early identification strategies to these differing presentations, as current approaches may not adequately capture this heterogeneity. Future research should account for these differences when defining and validating prodromal markers.

References

1. Howes OD, Lim S, Theologos G, Yung AR, Goodwin GM, McGuire P. A comprehensive review and model of putative prodromal features of bipolar affective disorder. Psychol Med. 2011;41(8):1567-77.
2. Skjelstad DV, Malt UF, Holte A. Symptoms and signs of the initial prodrome of bipolar disorder: a systematic review. J Affect Disord. 2010;126(1-2):1-13.
3. Van Meter AR, Burke C, Youngstrom EA, Faedda GL, Correll CU. The Bipolar Prodrome: Meta-Analysis of Symptom Prevalence Prior to Initial or Recurrent Mood Episodes. J Am Acad Child Adolesc Psychiatry. 2016;55(7):543-55.
4. Salazar de Pablo G, Guinart D, Cornblatt BA, Auther AM, Carrion RE, Carbon M, et al. Demographic and Clinical Characteristics, Including Subsyndromal Symptoms Across Bipolar-Spectrum Disorders in Adolescents. J Child Adolesc Psychopharmacol. 2020;30(4):222-34.