Abstract
Background
Stroke is the third leading cause of mortality and the fourth most common cause of disability globally. Unlike other World Bank high-income countries, available estimates show no decrease in stroke incidence in New Zealand over the last decade, however, these estimates are regional and periodic. Understanding the predisposing factors for stroke is important in its prevention and management and currently there is insufficient evidence about how emerging factors contribute to stroke risk. This doctoral thesis was aimed at validating administrative health data in New Zealand for its use in stroke monitoring and epidemiological studies and estimating stroke incidence. In addition, it aimed to assess associations with emerging stroke risk and protective factors, specifically socioeconomic factors and green space, and investigate possible underlying factors that may explain some of these associations.
Methods
Literature synthesis on factors associated with stroke incidence were conducted through narrative and scoping reviews. Next, a validation study was carried out on administrative health data in Statistics New Zealand’s Integrated Data Infrastructure using the fifth Auckland Regional Community Stroke study (ARCOS V) as a reference, and the validated data was used to calculate national and regional age-standardised stroke incidence (first-ever-in-a-lifetime strokes).
Subsequently, two whole population, case-control studies using unconditional regression analyses controlled for possible confounders, were conducted; first with 7,155 incident stroke cases between September 2020 and August 2021, to assess the associations with socioeconomic deprivation (using New Zealand’s Index of Deprivation (NZDep) and Indices of Multiple Deprivation (IMD)). Secondly, with 6,213 incident stroke cases in 2016 to assess the association with green space (measured as normalized difference vegetation index (NDVI)), including stratified analyses by stroke types and geographical location (urban-rural). All cases in both studies were frequency matched by age and sex to stroke-free controls in a 1:5 ratio. SAS Enterprise Guide 8.5 (SAS Institute) was used for statistical analyses.
Results
The literature reviews revealed well-documented evidence of ethnic disparities in stroke incidence in multicultural settings. There was also convincing evidence of an inverse relationship between socioeconomic status and stroke. On the other hand, existing literature on associations between incident stroke and geographical location or green space were sparce and inconclusive. Using ARCOS V data as reference, health administrative data had overall sensitivity of 87% [95%CI: 85%-89%] and PPV of 91% [95%CI: 89%-92%] for first event strokes. Similar values were noticed in subgroup analysis by age, sex, ethnicity, socioeconomic deprivation and stroke types. Age-standardised annual stroke incidence in New Zealand was 110.6 per 100,000 person-years. Sub-analyses showed higher incidence in males, Pacific Peoples and Māori.
For both case-control studies, stroke cases had significantly higher proportions of well-established stroke risk factors (hypertension, diabetes, atrial fibrillation, dyslipidaemia, and smoking) and lived in areas of higher socioeconomic deprivation. Using those living in the least deprived area (decile 1) as reference, there were increased odds of stroke as deprivation increased. Adjusting for all covariates, persons living in the most deprived areas (decile 10) had the highest odds of stroke [comparing decile 10 vs 1: NZDep aOR=1.33, 95%CI: 1.16-1.52; IMD aOR=1.41, 95%CI: 1.22-1.61]. Assessment of the association between green space and stroke showed that higher levels of residential green space were inversely associated with stroke in the unadjusted models, but this association disappeared in fully adjusted analyses. Stroke type stratification showed similar results for ischaemic, but not haemorrhagic stroke.
Conclusion
These findings show health administrative data in New Zealand has good validity and can be used for stroke epidemiological research. The age-standardised incidence of stroke is higher than some similar World Bank high-income countries, with evident sex and ethnic disparities. There was a clear association between incident stroke and socioeconomic deprivation, but not with green space. Continuous monitoring of stroke and targeted risk factor prevention are recommended.