Cardiovascular risk assessment methods yield unequal risk predictions: a large cross-sectional study in psychiatric secondary care outpatients

Background: Patients with a mental illness are more likely to develop, and die from, cardiovascular diseases (CVD), necessitating optimal CVD-risk (CVR)-assessment to enable early detection and treatment. Whereas psychiatrists use the metabolic syndrome (MetS)-concept to estimate CVR, GPs use absolute risk-models. Additionally, two PRIMROSE-models have been specifically designed for patients with severe mental illness. We aimed to assess the agreement in risk-outcomes between these CVR-methods.

Methods: To compare risk-outcomes across the various CVR-methods, we used somatic information of psychiatric outpatients from the PHAMOUS-, and MOPHAR-database, aged 40–70 years, free of past or current CVD and diabetes. We investigated: (1) the degree-of-agreement between categorical assessments (i.e. MetS-status vs. binary risk-categories); (2) non-parametric correlations between the number of MetS-criteria and absolute risks; and (3) strength-of-agreement between absolute risks.

Results: Seven thousand twenty-nine measurements of 3509 PHAMOUS-patients, and 748 measurements of 748 MOPHAR-patients, were included. There was systematic disagreement between the categorical CVR-assessments (all p < 0.036). Only MetS-status versus binary Framingham-assessment had a fair strength-of-agreement (κ = 0.23–0.28). The number of MetS-criteria and Framingham-scores, as well as MetS-criteria and PRIMROSE lipid-scores, showed a moderate-strong correlation (τ = 0.25–0.34). Finally, only the continuous PRIMROSE desk and lipid-outcomes showed moderate strength-of-agreement (ρ = 0.91).

Conclusions: The varying methods for CVR-assessment yield unequal risk predictions, and, consequently, carry the risk of significant disparities regarding treatment initiation in psychiatric patients. Considering the significantly increased health-risks in psychiatric patients, CVR-models should be recalibrated to the psychiatric population from adolescence onwards, and uniformly implemented by health care providers.

Trial registration: The MOPHAR research has been prospectively registered with the Netherlands Trial Register on 19th of November 2014 (NL4779).