A resampling approach for causal inference on novel two-point time-series with application to identify risk factors for type-2 diabetes and cardiovascular disease

Two-point time-series data, characterized by baseline and follow-up observations, are frequently encountered in health research. We study a novel two-point time series structure without a control group, which is driven by an observational routine clinical dataset collected to monitor key risk markers of type-$2$ diabetes (T2D) and cardiovascular disease (CVD). We propose a resampling approach called 'I-Rand' for independently sampling one of the two time points for each individual and making inference on the estimated causal effects based on matching methods. The proposed method is illustrated with data from a service-based dietary intervention to promote a low-carbohydrate diet (LCD), designed to impact risk of T2D and CVD. Baseline data contain a pre-intervention health record of study participants, and health data after LCD intervention are recorded at the follow-up visit, providing a two-point time-series pattern without a parallel control group. Using this approach we find that obesity is a significant risk factor of T2D and CVD, and an LCD approach can significantly mitigate the risks of T2D and CVD. We provide code that implements our method.