COMPASS seminars 2020
Our series is suspended until further notice, during social distancing and lockdown. We hope to be able to resume in due course. Stay safe!
Estimated Resident Population in the IDI: What happens if you want a denominator pre-2008?
Dr Anna Howe, School of Population Health
Tuesday 17 March @ 12 Grafton Road, 12PM
There is a well-established and widely used method for calculating an Estimated Resident Population (ERP) from IDI data. However, this method cannot be used for years prior to 2008 due to data limitations.
The original algorithm for creating an ERP retains individuals with recent activity in one of the following IDI data sets: accident claims, tax, health, or education; or for those under five years of age, birth or visa approval. However, some of these data sets are unreliable before 2008 and therefore cannot be used in constructing pre-2008 ERPs.
Population counts for 2008–2018 with these data sets removed were compared with the 2008–2018 ERP based on the full data sets, in order to gauge the likely effect on estimated population counts for years earlier than 2008. I will describe a modification that can be used for years prior to 2008, and illustrate the impact of this modification on population counts.
Anna Howe is a Research Fellow with the School of Population Health, and is interested in maternal and child health.
Trends in Antidepressant Exposure during Pregnancy
Dr Stephanie D'Souza
POSTPONED UNTIL LATER IN THE SEMESTER, STAY TUNED.
Globally, trends have shown an increase in antidepressant exposure during pregnancy. The current project was interested in examining whether similar trends were present in New Zealand’s pregnant population. Using community pharmaceutical data from the Integrated Data Infrastructure, we investigated antidepressant dispensing in all resident New Zealanders with at least one successful pregnancy, over the period 20007/08 to 2017/18. We further explored dispensing by medication type, trimester, ethnicity (total response), age, and area-level deprivation.
Gestational antidepressant dispensing prevalence increased over the study period from 2.6% to 4.4%, with SSRIs being the most common antidepressant. Across the study period, trimester-specific dispensing identified highest dispensing for trimester 3 and lowest dispensing for trimester 1. All subgroups also showed increased dispensing, though with considerable variation between categories. Specifically, the most deprived quintile had a notably lower dispensing prevalence relative to other quintiles. Ethnic differences in antidepressant dispensing were also apparent across the study period.
Stephanie D’Souza is a Research Fellow with COMPASS Research Centre. She completed her PhD in Psychology at the University of Auckland, using data from the Growing Up in New Zealand study to investigate behavioural difficulties in early childhood. Her research interests include child development, life course research, and working with large longitudinal and administrative data sets.
Data Quality of the 2018 New Zealand Census
Associate Professor Barry Milne
Tuesday 3 March @ 12 Grafton Road, 12PM
One in six New Zealanders did not complete the 2018 New Zealand Census, and completion in some areas of the country was less than fifty percent. The low – and patchy – completion rate, and use of alternative data sources when census data were not available, have raised concerns about the quality of data produced from the 2018 census.
I will present work undertaken to evaluate various aspects of the quality of the 2018 Census data set, highlight the topic areas – and populations subgroups – that have been most impacted, and provide some tips and cautions for data users. This work was undertaken while I was a member of the 2018 Census External Data Quality Panel.
Barry Milne is the Director of COMPASS Research Centre. He has a Masters degree in psychology from the University of Otago and a PhD in Psychiatric Epidemiology from Kings College London. His main interests are in lifecourse research, survey research, and the use of large administrative data sets to answer policy and research questions.