Time-series analysis of routine health data: challenges and opportunities

Date: Wednesday 27 October 2021, 4.30PM
Location: Online
Online - joining instructions will be sent to those registered
Section Group Meeting


Share this event

Routinely collected healthcare time-series data is often modelled to provide evidence needed for policymaking on preventative measures or to take action to change the course of specific public health issues or disease episodes. However, considering that routine healthcare data has not been collected for a particular research purpose or according to a specific study design, challenges arise when using these data for research purposes, especially when it derives from multiple sources or multiple geographical locations with different national health systems. In this session, our speakers will address the challenges and opportunities when analysing routine health time-series data while exploring classical and novel approaches.

 

Routinely collected healthcare time-series data is often modelled to provide evidence needed for policymaking on preventative measures or to take action to change the course of specific public health issues or disease episodes. However, considering that routine healthcare data has not been collected for a particular research purpose or according to a specific study design, challenges arise when using these data for research purposes, especially when it derives from multiple sources or multiple geographical locations with different national health systems. In this session, our speakers will address the challenges and opportunities when analysing routine health time-series data while exploring classical and novel approaches.

Dr Laurence Watier (Inserm & Institut Pasteur):
“Challenges in analysing time series”

Through various examples, we will focus on the steps to be undertaken when interested in evaluating and quantifying the effect of an intervention or estimating the attributable fraction from time series. The source and quality of data used will be discussed as well as the impact of using very long time series.

Professor Antonio Gasparrini (London School of Hygiene and Tropical Medicine):
“The case time series design”

We will illustrate a new study design that embeds the longitudinal structure of time series within an individual-level self-matched setting. The modelling framework is highly adaptable and based on efficient estimation and computational methods that make it suitable for the analysis of routinely collected longitudinal data.

 
Dr Laurence Watier (Inserm & Institut Pasteur)
Professor Antonio Gasparrini (London School of Hygiene and Tropical Medicine)