This is a cross-sectional study conducted between October 2021 and July 2022 as part of a larger, cross-sectional study validating a new questionnaire evaluating emotional states in people with BD in a current mood episode. Study interviews were administered at the Mood Disorders Clinic in St. Joseph's Healthcare Hamilton (SJHH) in Hamilton, Canada, or using online video platforms, such as Zoom. Depending on the current COVID-19 restrictions at SJHH and the preference of interested participants, the screening process and data collection were conducted either in-person or remotely. The study was approved by the Hamilton Integrated Research Ethics Board (HiREB) under protocol #8193.

Potential participants that were 18 and older were invited for an interview to determine their eligibility for the study according to the inclusion and exclusion criteria, which differed for the BD and HC groups. Participants from the BD and HC groups were matched by age and sex at the time of recruitment. For the current study, the inclusion criteria for BD were (1) a diagnosis of BD according to DSM-5 criteria using the Mini International Neuropsychiatric Interview (MINI)22 and (2) currently experiencing a depressive, (hypo)manic, or (hypo)manic episode with mixed features episode according to the MINI. Exclusion criteria for the BD group were (1) severe psychotic symptoms, determined according to the presence of current psychotic symptoms on the MINI and the severity of these symptoms based on clinical judgment, (2) severe suicide risk, indicated by a score of ≥17 on the suicidality module of the MINI, and (3) severe cognitive disability that could impact on their understanding of the assessments based on clinical judgment. In contrast, inclusion criteria for healthy controls were (1) absence of any psychiatric diagnosis according to the MINI for DSM-5 and (2) did not demonstrate any severe cognitive disability that could impact their understanding of the assessments based on clinical judgment. The study was advertised using paper flyers in the community, as well as online advertisements posted on advertising websites and social media outlets (Facebook, Instagram, Reddit, and Kijiji). After receiving verbal and written information about the study, all participants provided informed consent prior to the screening interview. Assessments consisted of a diagnostic interview, clinical assessments of current mood state, and lifestyle assessments.

Demographic information was collected for all participants, which included questions regarding the individual's age, sex, years of education, highest educational degree attained, and current occupational status (working/studying). In this study, we only included sociodemographic variables that were assessed in the larger study based on previous validation studies.

The data were collected directly on REDCap, a secure web platform for collection and storage of data.28 After, the data were transferred to SPSS v.28.0 where the statistical analyses were performed.29 We tested the normality of continuous variables using the Shapiro–Wilk test for normality.30 To describe the sample characteristics between the two groups, HC vs BD, we performed chi-square and Mann–Whitney U tests, for categorical variables and nonparametric continuous data, respectively. We also performed a Mann–Whitney U test to analyze these two groups on the total SMILE scores. We used a non-parametric test for the bivariate analysis because the distribution of the outcome (SMILE scores) was asymmetric. We used Student's t-test for the comparisons of the domains with normal distributions of the scores, and Mann–Whitney U test for the comparisons with asymmetric distribution of the scores. We created two box plot graphs using the Prism GraphPad v.9 to demonstrate the findings of the comparison between HC and BD, and the comparison between HC, BD in a depressive episode, and BD in a manic (with and without mixed features) episode across the different SMILE domains (Figs. 1 and 2). We also calculated effect size of the differences between the groups using r values for nonparametric comparisons and Cohen's d for parametric comparisons. We calculated the post hoc statistical power of the main comparison using Epi Info™. Multivariate linear regression was performed to evaluate the effect of independent factors (predictors) on the SMILE score. The analyses included variables that presented an association of p<0.20 with the outcome (SMILE scores). We did not include the ‘highest degree of education’ variable in this analysis, as it measures the same construct as the years of education variable. The final model was reached using a manual stepwise removal of each non-statistically significant variable. Furthermore, we compared total scores on the SMILE scale, as well as scores for each domain of the SMILE across HCs, BD in a current depressive episode, and BD in a current manic (with and without mixed features) episode. Significance for all comparisons was set at p<0.05.

Fig. 1.

Comparison of total SMILE scores between healthy controls (HCs) and individuals with bipolar disorder (BD) in a current mood episode. Healthy individuals reported significantly higher median scores on the SMILE scale (median: 133.5 [IQR: 125.75–141.25]) than individuals with BD in a current mood episode (median: 104.0 [IQR: 95.0–115.0], p<0.001), suggesting unhealthier behaviors in individuals with BD.

BD: bipolar disorder; HC: healthy control; SMILE: Short Multidimensional Inventory Lifestyle Evaluation.

(0.06MB).

Fig. 2.

Comparison of SMILE scores across different lifestyle domains between healthy individuals, individuals with bipolar disorder in a depressive episode, and individuals with bipolar disorder in a manic or manic with mixed features episode.

BD – D: bipolar disorder in a depressive episode; BD – M/M: bipolar disorder in a manic or manic with mixed features episode; HC: healthy control; SMILE: Short Multidimensional Inventory Lifestyle Evaluation; *p<0.05; **p<0.001.

(0.17MB).

The total sample of this study consisted of 96 participants (46 BD and 50 HC). The sociodemographic characteristics of the sample are described in Table 1. There were no significant differences between groups regarding sex (p=0.957) or age (p=0.445). Consistent with the well-known educational and occupational burden of BD, the HC group had completed significantly more years of education (p<0.001), attained a higher educational degree (p<0.001), and were more likely to be working/studying (88.0%) at the time of assessment than the BD group (60.9%, p=0.002). The groups also differed in scores on the clinical assessments, as expected. The HC group had significantly lower median scores on the MADRS (2.0, IQR: 0–3.0) and on the YMRS (1.0, IQR: 0–2.0) than the BD group (MADRS: 24.0, IQR: 20.0–30.5, p<0.001; YMRS: 8.5, IQR: 2.75–17.5, p<0.001).

Our analysis revealed significant differences with a large effect size on the total SMILE score between the HC (median: 133.5 [IQR: 125.75–141.25]) and BD (median: 103.5 [IQR: 95.0–114.25], r=0.75, p<0.001) groups (Fig. 1). The post hoc statistical power for the comparison between BD and HC was 100%. These differences indicate a significantly worse lifestyle among individuals with BD than healthy controls. Furthermore, our analysis revealed significantly lower scores in BD in comparison to HC with a large effect size across all domains on the SMILE scale (Table 2).

In the final adjusted linear regression model, we included years of education, current working/studying status, and diagnosis (HC or BD) as predictors of SMILE scores. After the removal of the least significant factor in the model (currently working/studying), we performed another linear regression with years of education and SMILE scores. Years of education was also non-significant in the model, and after the removal of this factor, diagnosis (HC/BD) remained as the only significant predictor of total SMILE scores, where a diagnosis of BD was predictive of worse scores on the SMILE. The multivariate linear regression models are presented in Supplementary Material 1.

For the analysis of scores on the SMILE scale across different mood states, we included 26 participants in a depressive state, 20 participants in a manic state (manic without mixed features=9, manic with mixed features=11), and 50 healthy controls. In addition to the total SMILE scores, we analyzed the scores of the 7 subdomains of the SMILE scale across the 3 groups (Table 3 and Fig. 2). Once again, participants in both BD groups had significantly lower median scores on the total SMILE scale, as well as all 7 subdomains of the scale than HC (p<0.05). In contrast, BD in a depressive episode and BD in a manic (with and without mixed features) episode did not show any differences in total SMILE scores or in any of the subdomains of the scale.

Findings from our study should be interpreted considering some limitations. Considering that this cross-sectional study was part of a larger study designed to validate an emotional reactivity scale in people with BD in a current mood episode, we did not include any data of euthymic individuals with BD, which would be helpful for understanding whether unhealthy lifestyles are associated to the disorder even when individuals are asymptomatic. Also, the SMILE scale has not been validated for use in a BD population yet, however, the validation study is currently ongoing. Furthermore, 37 out of the 46 individuals in the BD group had a positive screening for a depressive episode, which prevented us from including a manic only group in our analysis and led to the grouping of individuals in manic and manic with mixed features states. Also, our sample predominantly consisted of females, although this could be explained by the higher prevalence of females than males with BD.6 Lastly, we did not collect information regarding the socio-economic status, current medications use of participants, comorbid medical conditions and other psychiatric disorders which may be related to the observed behaviors on some of the SMILE domains. Despite these limitations, this is the first cross-sectional study to compare lifestyle across healthy individuals and people with BD in different mood episodes using a multidimensional approach based on contemporary notions of lifestyle.

Findings from this study confirm the association between multidimensional unhealthy lifestyle behaviors and symptomatic people with BD. We found that symptomatic individuals with BD reported significantly unhealthier lifestyle behaviors than healthy controls. This association was present across multiple areas of lifestyle, an important consideration for the onset and progression of physical and psychiatric comorbidities in BD. However, there were no differences among individuals with BD across different mood episodes, indicating unhealthy lifestyle patterns in BD regardless of the polarity of the mood episode. There is a strong need for future research on implementing a multidimensional approach to lifestyle assessments in BD across all mood states of the disorder, including euthymia, and evaluating the clustering of unhealthy lifestyle behaviors across different areas of everyday life that could potentially lead to worse clinical and functional outcomes for the affected individuals.