Cumulative disadvantage? Exploring relationships between neighbourhood deprivation trends (1991 to 2006) and mortality in New Zealand

BACKGROUND Area-level socioeconomic deprivation has been shown to exert an independent effect on both individual and population health outcomes and health-related behaviours. Evidence also suggests that health and economic inequalities in many countries are increasing in some areas but may be on the decline in others. While area-level deprivation at a single point in time is known to influence health, the literature relating to longitudinal deprivation of communities and associated health impacts is sparse. This research makes a methodological contribution to this literature. METHODS Using a Latent Class Growth Model, we identified 12 deprivation trends (1991-2006) for small areas (n = 1621) in New Zealand. We then fitted regression models to assess the effects of trends of relative deprivation on a) all-cause mortality, and b) cardiovascular mortality (2005-2007) by census area unit. For comparison, we also fitted regression models to assess the effect of deprivation deciles (in 2006) on outcomes a) and b). RESULTS Using trends, we found a positive association between deprivation and mortality, except for two trends for both all-cause and CVD-related mortality. When comparing trends and deciles of deprivation, we observed similar patterns. However, we found that AIC values were slightly lower for the model including deciles, indicating better model fit. CONCLUSION While we found that current deprivation was a slightly better predictor of mortality, the approach used here offers a potentially useful alternative. Future deprivation research must consider the possible loss of information about health benefits of living in areas where relative deprivation has improved in cross-sectional analyses.