Abstract:
New Zealand has a strong history of longitudinal research, with studies such as the Multidisciplinary Health and Development Study, the Christchurch Health and Development Study, the Pacific Island Families Study, and Growing Up in New Zealand. However, all these studies involve single cohorts, and only Growing Up in New Zealand includes a large proportion of Māori in its cohort. New Zealand also has an important research resource in the Integrated Data Infrastructure (IDI), a database of administrative and survey datasets containing a wide range of variables linkable at the individual level. The existing longitudinal studies are not able to link to the range of information within the IDI. Therefore, such longitudinal studies are yet to demonstrate the ability to link official statistics with administrative data in the IDI resource. This thesis aims to extend the utility of linking an official statistics survey with administrative data and the ability to complete longitudinal analysis on this cohort. Practical examples using various individual, household and geographic variables from the IDI are conveyed, and their effects on two health outcomes for the Te Kupenga 2013 cohort: Ambulatory Sensitive Hospitalisations (ASH) and COVID-19 Vaccinations. Regression analysis displayed that housing and geographic factors do not affect ASH events. However, individual characteristics such as disability and medical discrimination impacted the odds of an ASH event. Further analysis uncovered that measures of trust in fair healthcare and whānau wellbeing impact COVID-19 Vaccinations. This thesis demonstrates that it is feasible to turn a sample survey into a cohort for longitudinal analysis in the IDI. The process for doing this is described in detail, including data management and analytic code that can be used with Te Kupenga or other datasets with some modifications. Developing this method has highlighted that changes to the structure and function of the IDI resources would simplify similar research in the future. This thesis concludes by outlining these issues and potential solutions and then provides recommendations for improving IDI’s capability for longitudinal research.