# Smoking and obsessive–compulsive symptoms in patients with schizophrenia, schizoaffective disorder or bipolar disorder using electronic mental health records

## TL;DR

People with severe mental illness who have obsessive-compulsive symptoms or disorder are more likely to smoke, based on a large electronic health record analysis.

## Contribution

This study is the first to use natural language processing on electronic mental health records to explore the link between obsessive-compulsive symptoms and smoking in severe mental illness.

## Key findings

- Individuals with obsessive-compulsive symptoms or disorder were significantly more likely to have ever smoked.
- The adjusted odds ratio for ever smoking was higher in those with obsessive-compulsive symptoms or a diagnosis of obsessive-compulsive disorder.

## Abstract

Comorbid obsessive–compulsive disorder (OCD) or obsessive–compulsive symptoms (OCS) are common in people with severe mental illness (SMI; including schizophrenia, bipolar disorder and schizoaffective disorder), with little known about associations with smoking.

To estimate the association between OCD/OCS and smoking status among people with SMI in a huge electronic database.

Using the Clinical Records Interactive Search (CRIS) platform for data of service users in the South London and Maudsley (SLaM) NHS Foundation Trust, tobacco smoking status was retrospectively detected through an algorithm of natural language processing technique, categorising into ‘current smoker’, ‘ex-smoker’ and ‘non-smoker’ by the clinical notes of SMI individuals during 2007–2015. A hierarchical assignment rule was applied following the order of ‘smoker’, ‘ex-smoker’ and then ‘non-smoker’ in an individual. Logistic regression was used to examine the association between smoking and OCS in people with SMI for univariable and multivariable analyses.

We identified 15 479 SMI individuals (56% male; mean age 41 years old), with 90.4% ever smoked. Among them, 2320 (15%) had OCS (without OCD), while 2174 (14%) had a clinical diagnosis of comorbid OCD. After adjusting for demographics and functional status as confounders, both SMI individuals with OCS only and an OCD diagnosis were significantly more likely to have ever smoked (adj. odds ratio 1.47, 95% CI 1.23, 1.76 and adj. odds ratio 1.33, 95% CI 1.11, 1.60, respectively) compared with those without OCD/OCS.

In this large-scale analysis of people with SMI, we found that individuals with OCS or OCD were more likely to have ever smoked.