Abstract Body: Introduction
Stroke and dementia are among the leading causes of mortality globally. This can be mitigated through targeting modifiable risk factors. Identification of those at-risk through screening tools could be facilitated by inclusion of self-reported risk factors rather than reliance on clinical data. We aimed to determine whether self-reported risk factors are reliable in identifying at-risk individuals and whether self-reported risk factors can predict stroke and dementia-related mortality.
Methods
In this study of cross-sectional data from the National Health and Nutrition Examination Survey (NHANES) from 1999-2018 linked to death certificate records up to December 2019, we included participants with available data on both self-reported and measured hypertension, hypercholesterolemia, diabetes, kidney disease, hearing impairment and overweight. We assessed the ability of self-reported risk factors to dichotomize risk on a population level, compared confusion matrices and F1-scores between different questions, and compared the prospective validity of self-reported and measured risk factors through Cox proportional hazards models.
Results
Included participants ranged from 16,966 participants for hearing impairment to 59,319 for BMI. Reliability of self-reported risk factors was high in overweight (F1 0.81) and diabetes (F1 0.71), moderate in hearing impairment (F1 0.59) and hypertension (F1 0.56) and low in hypercholesterolemia (F1 0.49) and kidney disease (F1 0.25). Compared to measured risk factors, self-reported risk factors were associated with similar hazard ratios for stroke and dementia mortality.
Conclusions
The reliability and validity of self-reported risk factors for stroke and dementia differs between risk factors and depends on the phrasing of the questions, but overall predicts mortality similar to measured risk factors. Screening tools that incorporate self-reported risk factors could improve identification of high risk individuals in communities with lower health literacy or access to care.