Date of Award
2025
Degree Type
Dissertation
Degree Name
Doctor of Philosophy in Health Sciences
Department
Nutrition
First Advisor
Maya Vadiveloo
Abstract
Cardiovascular Disease (CVD) remains the leading cause of morbidity and mortality worldwide. Lifestyle modifications, particularly improvements in dietary patterns, are effective approaches for CVD prevention and management. Epidemiological evidence has consistently linked greater adherence to heart-healthy diet patterns such as Dietary Approaches to Stop Hypertension (DASH) diet and reduced risk of CVD. Nevertheless, diet is not yet integrated in routine care and is often not measured comprehensively using full dietary assessment tools in clinical trials when diet is not the primary focus, due to the burdens of high costs, time constraints, and competing demands. These infeasibility challenges have led to a call in the field for validated rapid diet screeners that are able to adequately capture diet patterns linked to CVD risk reduction, and the American Heart Association (AHA) established theoretical and clinical criteria for validation of rapid screening tools.
Like many clinical trials, the Stenting vs. Aggressive Medical Management for Preventing Recurrent Stroke in Intracranial Stenosis (SAMMPRIS) study, a multi-center randomized controlled trial conducted from 2008 to 2013, used a set of non-validated questions to capture intake of key DASH components in lieu of a full dietary assessment. Evaluating whether these questions accurately captured DASH adherence is important to establish their utility in both routine clinical care and future clinical research. Therefore, three studies were conducted in this dissertation to develop and validate a DASH diet screener score using the questions from in the SAMMPRIS trial based on established methods that follow AHA guideline on rapid screening tool development and validation.
The first study developed a DASH screener score following the weighting strategies of an established DASH score (Fung DASH score) and examined its construct validity in three 2-year cycles of National Health and Nutrition Examination Survey (NHANES, 2009-2014). The DASH screener used 11 non-validated SAMMPRIS questions to create 8 components. Each component was multiplied by its corresponding weight and summed to obtain an overall score (0-100), with higher values indicating better adherence. Construct validity was examined by analyzing whether the screener score had variable distribution, correlated with the validated score, differentiated groups with known diet quality differences, and was concordant with the validated score. Results showed that construct validity was demonstrated with strong correlations between the total and most component scores, the ability to distinguish known-group differences, and strong concordance between the DASH screener score and Fung DASH score.
The second study examined the predictive validity of the DASH screener score on CVD biomarkers in a generally healthy adult population in NHANES (2009-2014). Survey-weighted multivariable linear and logistic regression models were used to examine the associations between the DASH screener score and CVD biomarkers including systolic blood pressure (SBP), diastolic blood pressure (DBP), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), total cholesterol (TC), and glycohemoglobin (HbA1c). Higher DASH screener scores were consistently associated with lower SBP and HDL-C, but not with DBP, HbA1c, TC, and LDL-C, providing some evidence of predictive validity of the DASH screener.
The third study prospectively examined the predictive validity of the DASH screener score in a high-risk clinical population in the SAMMPRIS trial. Multivariable-adjusted linear regressions were used to examine (1) the associations between baseline DASH screener score and CVD biomarkers at baseline, 4-mo, 1-yr and 2-yr follow-up, and (2) the associations between up to 1-yr Δ in DASH screener score and CVD biomarkers at 1-yr and 2-yr follow-up. Multivariable-adjusted logistic regressions were used to examine the associations between up to 1-yr Δ in DASH screener score and achieving CVD biomarker targets at 1-yr and 2-yr follow-up. Linear mixed-effects models were used to examine the association between cumulative mean of the DASH screener score and repeated measures of continuous CVD biomarkers over 2 years. Baseline DASH screener score was associated with lower DBP and LCL-C at 4-mo follow up, and lower DBP at 1-yr follow up in one of the treatment arms; A positive increase in up to 1-yr Δ in DASH screener score was associated with higher likelihood of achieving the SBP target at 1-yr follow up, and greater cumulative mean of DASH screener score was associated with lower DBP over 2 years of follow up, providing some evidence suggestive of predictive validity.
These studies provided strong evidence of construct validity for assessing DASH diet adherence and suggested some degree of predictive validity for evaluating diet-related CVD risk through biomarkers in general and clinical populations. Future research is needed to assess predictive validity with hard CVD outcomes to better determine the potential utility of the screener in routine care and future clinical research.
Recommended Citation
Yao, Qisi, "DIET SCREENER FOR ASSESSING THE DIETARY APPROACHES TO STOP HYPERTENSION DIET ADHERENCE: DEVELOPMENT AND VALIDATION" (2025). Open Access Dissertations. Paper 4534.
https://digitalcommons.uri.edu/oa_diss/4534