Efficient and transparent methods for linking and analysing longitudinal population studies and administrative data (360G-Wellcome-212953_Z_18_Z)

The Wellcome Trust LPS Strategy states that research is needed to i) underpin efficient linkage of multiple datasets, ii) quantify potential biases resulting from linkage, and iii) to handle linkage error in data analyses. We propose a programme of methodological research to develop efficient and transparent methods for linkage and analysis, to help maximise the joint potential of existing longitudinal population studies and multiple administrative datasets. Our proposed work packages will focus on linkage of LPS and multiple administrative datasets: WP1: Facilitate development of innovative linkage methods for multiple datasets, by developing shareable software to generate synthetic datasets that are generalisable to a range of research settings. WP2: Develop methods for efficient linkage of multiple datasets by expanding existing probabilistic linkage methods to the setting in which multiple administrative or LPS datasets are to be linked dynamically, and where there are multiple sets of identifiers (e.g. collected at different time-points). WP3: Provide appropriate tools for analysis of linked data by extending existing imputation methods for handling linkage uncertainty and avoiding bias within analysis. WP4: Maximise the value of approaches developed in WP2-3 through evaluation using exemplar linkages and dissemination of methods to the LPS community and other key stakeholders.