Development and Validation of a Large Synthetic Cohort for the Study of Cardiovascular Health Across the Life Span

Abstract

We developed and validated a synthetic cohort approach to examine lifespan cardiovascular risk factors (CRFs) and events, including cardiovascular disease (CVD) and death from ages 20-90 years. This study included 40,875 participants from 7 large, population-based longitudinal epidemiology studies (1948-2016). We multiply imputed the participant’s lifespan CRFs and events using the available records based on a joint multi-level imputation model. To validate the imputed values, we partially removed the observed data, then compared the imputed and observed values. The complete lifespan synthetic dataset reflected the original observed data trends well. Among our validation sample, the distributions of imputed CRFs and events were close to the observed but with less variability. Bland-Altman plots indicated that there was a slightly negative trend in general and the agreement bias was relatively small for the continuous CRFs. The hypothetical linear regression model suggested that the relationships between the CRFs and events were preserved in the imputed dataset. This approach generated valid estimates of CRFs and events across the lifespan for African American and White participants. The synthetic cohort may be accurate enough to be useful in assessing the origins and timing of accumulating cardiovascular risk that can inform efforts to avoid cardiovascular risk development.

Five‐Year Cumulative Cardiovascular Health and Clinical Events in Patients With Chronic Kidney Disease: The CRIC Study

“… 22 This method is recommended to impute multilevel data and is appropriate for handling the clustering of repeated longitudinal measures within individuals. 23 For variables with missing data, jomo applies a multilevel linear regression model that leverages other variables as fixed effects and incorporates clustering information using random effects. These models assume fixed effects have linear associations with the outcome and that the model's residual error is normally distributed conditional on fixed and random effects.…”

A Case for Synthetic Data in Regulatory Decision‐Making in Europe

“…Despite previous evidence and validation of synthetic data in research studies, 15,29,33 there is a current lack of methodological standards and procedural regulatory guidelines required by regulatory bodies. The European Network of Centers for Pharmacoepidemiology and Pharmacovigilance (ENCePP) sets the standards for methods used in RWE generation in the regulatory arena.…”

“…Some study results show that synthetic data have high validity and can be used as a proxy in clinical trials, 29 in neuroimaging studies, 30 or RWD 27,31–35 and do not lead to an increased bias, however, systematic and scoping reviews are still lacking. Utility metrics serve as a way to assess the consistency between RWD and synthetic data, as well as validate and measure utility 36 .…”

“…26 Importantly, risk evaluations of synthetic data use reveal that no direct connections between a generated data point and the real person are possible and that synthetic data (SD) are regarded as low-risk in practice. 27,28 Some study results show that synthetic data have high validity and can be used as a proxy in clinical trials, 29 in neuroimaging studies, 30 or RWD 27,[31][32][33][34][35] and do not lead to an increased bias, however, systematic and scoping reviews are still lacking. Utility metrics serve as a way to assess the consistency between RWD and synthetic data, as well as validate and measure utility.…”

“…One advantage of synthetic data is that diverse and copious data can be generated, however, methodological standards and validation studies must establish the degree to which synthetic data are an appropriate proxy for original data. 29,44,61,62 Despite previous evidence and validation of synthetic data in research studies, 15,29,33 there is a current lack of methodological standards and procedural regulatory guidelines required by regulatory bodies. The European Network of Centers for Pharmacoepidemiology and Pharmacovigilance (ENCePP) sets the standards for methods used in RWE generation in the regulatory arena.…”