How Your Health Habits Cluster May Shape Your Risk of Anxiety and Bipolar Disorder

What if the single number that summarizes your cardiovascular health is hiding the very information that matters most for your brain? A new analysis of nearly a quarter-million adults in the UK Biobank suggests exactly that. Researchers applied a statistical technique called latent profile analysis to the American Heart Association’s Life’s Essential 8 framework—the eight modifiable factors that include diet, physical activity, nicotine exposure, sleep, body mass index, blood glucose, blood lipids, and blood pressure—and found that these components do not vary independently. Instead, they cluster into five distinct health patterns, and the pattern a person belongs to predicts their risk of developing anxiety disorders or bipolar disorder over nearly fourteen years of follow-up.

The study, published in BMC Psychiatry, followed 242,947 UK Biobank participants aged 40 to 70 who were free of schizophrenia, anxiety disorders, and bipolar disorder at baseline. Rather than collapsing the eight LE8 components into a single composite score, the research team used latent profile analysis, a person-centred method that identifies subgroups of individuals who share similar configurations across multiple measured variables. The approach is grounded in finite mixture modeling: participants are assigned probabilistically to latent classes, and model fit is evaluated with tools such as the Akaike and Bayesian information criteria, entropy values that quantify classification accuracy, and a bootstrap likelihood ratio test that compares successive class solutions. In this case, the five-profile solution achieved an entropy of 0.95—well above the 0.80 threshold for excellent classification—while the six-profile solution converged to a suboptimal fit, supporting the choice of five.

The five profiles that emerged were strikingly heterogeneous. The first, labeled the Low LE8 score group, averaged a composite score of 55.96 and was dominated by near-universal current smoking, reflected in a nicotine exposure score of just 0.06 out of 100. The second profile, characterized by elevated adiposity, blood pressure, and glucose, contained the oldest participants, with a mean age of 59.31 years, and showed markedly poor metabolic markers despite relatively preserved lifestyle behaviors. The third and largest group, comprising roughly 31 percent of the cohort, combined low physical activity with poor sleep health while maintaining near-average metabolic indicators. The fourth profile showed elevated blood pressure and dyslipidemia as its primary deficits, with otherwise favorable behaviors. The fifth, the High LE8 score group, averaged 80.59 and exhibited consistently positive standardized scores across nearly all dimensions, along with the youngest mean age of 52.45 years and the most favorable socioeconomic indicators.

Over a median follow-up of 13.8 years, the researchers identified 11,419 incident cases of anxiety disorders, 341 cases of bipolar disorder, and 40 cases of schizophrenia, using ICD-10 codes drawn from hospital inpatient records, primary care data, death registries, and self-reported medical histories. Cox proportional hazards regression, adjusted for age, sex, ethnicity, education, household income, the Townsend Deprivation Index, employment status, long-standing illness, alcohol intake, and family history of severe depression, revealed that all four comparison profiles carried significantly lower anxiety risk than the Low LE8 score group. The High LE8 score group showed the strongest protection, with a hazard ratio of 0.65, translating to a 35 percent risk reduction. The elevated blood pressure and dyslipidemia group followed at 0.68, while the adiposity-glucose and inactivity-sleep groups showed hazard ratios of 0.77 and 0.78, respectively.

The bipolar disorder findings were even more dramatic in relative terms. Despite the smaller number of cases, all four comparison profiles showed substantial risk reductions ranging from 49 to 63 percent. The elevated blood pressure and dyslipidemia group had the lowest hazard ratio at 0.37, followed by the inactivity-sleep group and the High LE8 score group, both at 0.41, and the adiposity-glucose group at 0.51. Notably, the pattern was not simply a monotonic function of total LE8 score. The elevated blood pressure and dyslipidemia group, despite a mean composite score of 70.93 that was lower than the High LE8 group’s 80.59, showed the strongest bipolar protection. This dissociation between composite score and profile-specific risk is precisely what the latent profile approach is designed to detect and what conventional scoring methods would obscure.

The biological plausibility of these associations rests on well-characterized pathways linking cardiovascular health to brain function. Chronic nicotine exposure pharmacologically activates the hypothalamic-pituitary-adrenal axis and increases sympathetic arousal, which may paradoxically heighten long-term vulnerability to anxiety and mood instability despite the subjective anxiolytic effects that drive smoking behavior. Physical inactivity is associated with reduced brain-derived neurotrophic factor-mediated hippocampal neuroplasticity, while poor dietary quality has been linked to elevated systemic inflammatory markers and oxidative stress. In the sleep-inactivity profile, chronic sleep insufficiency has been shown to impair prefrontal cortical regulation of amygdalar reactivity, and physical inactivity may limit the exercise-mediated downregulation of stress axes and anti-inflammatory signaling. These converging mechanisms offer a coherent framework for why the Low LE8 score group, with its convergence of adverse exposures, carried the highest neuropsychiatric burden.

The schizophrenia results demand careful interpretation. With only 40 incident cases, the analysis was severely underpowered, and the authors explicitly frame these findings as exploratory. Beyond the statistical limitation, the UK Biobank cohort largely excludes the typical age of onset for schizophrenia, which typically emerges in late adolescence and early adulthood. Ninety percent of the 40 cases occurred after age 50, and 55 percent after age 60, meaning the captured cases almost exclusively represent late-onset presentations, which are thought to involve neurodegenerative processes rather than the neurodevelopmental abnormalities that predominate in early-onset disease. A further complication, which the authors call a therapeutic paradox, is that atypical antipsychotics induce metabolic side effects including weight gain, sedation, and glycemic elevation, potentially introducing reverse causality into observational data. The absence of significant associations should therefore not be read as evidence that lifestyle and metabolic health are irrelevant to schizophrenia risk.

Several sensitivity analyses bolstered confidence in the primary findings. To minimize reverse causation, participants diagnosed within the first six months, one, two, three, and five years of follow-up were sequentially excluded, with consistent results across all thresholds. Multiple imputation by chained equations was used to include participants with missing LE8 or covariate data, yielding a substantially larger analytic sample with concordant findings. When conventional LE8 total score categories—low, moderate, and high—were used as an alternative exposure, higher scores were associated with lower anxiety and bipolar risk in a monotonic gradient, but this approach could not distinguish the profile-specific patterns that the latent analysis revealed. Subgroup analyses stratified by sex, age, ethnicity, socioeconomic status, education, employment, illness status, alcohol intake, and family history of depression produced broadly consistent results, with the anxiety findings particularly stable across virtually all strata.

The clinical implications are tentative but intriguing. If distinct health patterns carry differential neuropsychiatric risk, then prevention strategies might be tailored to profile rather than to composite score. For individuals in the Low LE8 score group, integrated interventions combining smoking cessation with comprehensive lifestyle modification could be prioritized. For the largest subgroup—the inactivity-sleep cluster—structured exercise programs combined with sleep hygiene interventions represent a plausible target. The authors emphasize that these observational findings generate hypotheses for future interventional studies rather than prescribing immediate clinical action, and that the profile-specific risk patterns require prospective validation before implementation in practice.

Limitations temper the enthusiasm. The cohort is 96.21 percent white and generally affluent, limiting generalizability to more diverse populations. LE8 components were assessed only at baseline, so changes in health behaviors over the follow-up period could not be captured. Unmeasured confounders, including genetic predisposition, social support networks, and environmental stressors, cannot be entirely excluded despite comprehensive covariate adjustment. Neuropsychiatric disease identification relied on clinical diagnoses recorded in healthcare databases, potentially under-ascertaining milder cases that never receive clinical attention. And the bipolar disorder findings, while consistent in direction across all four comparison profiles, rest on only 341 incident cases, limiting the precision of individual estimates. Nevertheless, the study demonstrates that the architecture of health—how behaviors and biomarkers co-occur within individuals—may carry information about neuropsychiatric vulnerability that a single summary number cannot convey, and it positions person-centred, pattern-based approaches as a promising frontier for understanding how modifiable lifestyle factors shape the risk of mental illness.

Subject of Research: Associations between latent Life's Essential 8 cardiovascular health profiles and incident neuropsychiatric diseases in the UK Biobank

Article Title: Life’s essential 8 health profiles and risk of incident neuropsychiatric diseases: a nationwide prospective cohort study

Article References: Huangfu, H., Jia, H., Chen, C., Xu, Y., Zhou, X., Tang, Y., & Gao, X. (2026). Life’s essential 8 health profiles and risk of incident neuropsychiatric diseases: a nationwide prospective cohort study. BMC Psychiatry, 26(1), Article 714. https://doi.org/10.1186/s12888-026-08341-1

Image Credits: AI Generated

DOI: 10.1186/s12888-026-08341-1

Keywords: Life's Essential 8, UK Biobank, latent profile analysis, anxiety disorders, bipolar disorder, schizophrenia, cardiovascular health, neuropsychiatric diseases, lifestyle factors, Cox proportional hazards, sleep health, physical activity