Impact of weighting on the association between sociodemographic characteristics, health behaviours and cancer, cardiovascular and all-cause mortality in the Australian 45 and Up Study

Sarsha Yap; Qingwei Luo; Stephen Wade; Preston Ngo; David Goldsbury; Peter Sarich; Emily Banks; Marianne Weber; Karen Canfell; Michael David; Julia Steinberg

doi:10.1016/j.canep.2024.102567

Impact of weighting on the association between sociodemographic characteristics, health behaviours and cancer, cardiovascular and all-cause mortality in the Australian 45 and Up Study

Cancer Epidemiol. 2024 Jun:90:102567. doi: 10.1016/j.canep.2024.102567. Epub 2024 Apr 10.

Authors

Affiliations

¹ The Daffodil Centre, The University of Sydney, a Joint Venture with Cancer Council NSW, Sydney, New South Wales, Australia. Electronic address: sarshay@nswcc.org.au.
² The Daffodil Centre, The University of Sydney, a Joint Venture with Cancer Council NSW, Sydney, New South Wales, Australia.
³ National Centre for Epidemiology and Population Health, Australian National University, Canberra, Australian Capital Territory, Australia.
⁴ The Daffodil Centre, The University of Sydney, a Joint Venture with Cancer Council NSW, Sydney, New South Wales, Australia; School of Medicine and Dentistry, Griffith University, Gold Coast, Queensland, Australia.

PMID: 38603997
DOI: 10.1016/j.canep.2024.102567

Abstract

Background: Weighting can improve study estimate representativeness. We examined the impact of weighting on associations between participants' characteristics and cancer, cardiovascular and all-cause mortality in the Australian 45 and Up Study cohort.

Methods: Raking weighted cohort data to the 2006 Australian population for seven sociodemographic characteristics. Deaths were ascertained via linkage to routinely collected data. Cox's proportional hazards regression quantified associations between 11 sociodemographic and health characteristics and cancer, cardiovascular and all-cause mortality. The ratios of hazard ratios (RHRs) compared unweighted and weighted estimates.

Results: Among 195,052 included participants (median follow-up 11.4 years), there were 7200 cancer, 5912 cardiovascular and 21,840 all-cause deaths. Overall, 102/111 (91.9%) weighted HRs did not differ significantly from unweighted HRs (100%, 86.5% and 89.2% of 37 HRs for cancer, cardiovascular and all-cause mortality, respectively). Significant differences included a somewhat stronger association between single/widowed/divorced (versus married/de-facto) and cardiovascular mortality (unweighted HR=1.25 (95%CI:1.18-1.32), weighted HR=1.33 (95%CI:1.24-1.42), RHR=1.06 (95%CI:1.02-1.11)); and between no school certificate/qualification (versus university degree) and all-cause mortality (unweighted HR=1.21 (95%CI:1.15-1.27), weighted HR=1.28 (95%CI:1.19-1.38), RHR=1.06 (95%CI:1.03-1.10)).

Conclusion: Our results support the generalisability of most estimates of associations in the 45 and Up Study, particularly in relation to cancer mortality. Slight distortion of a few associations with cardiovascular or all-cause mortality were observed.

Keywords: 45 and Up Study; Australia; Cancer mortality; Cardiovascular mortality; Mortality; Raking weighting; Risk factors.

MeSH terms

Aged
Australia / epidemiology
Cardiovascular Diseases* / epidemiology
Cardiovascular Diseases* / mortality
Cause of Death*
Cohort Studies
Female
Follow-Up Studies
Health Behavior*
Humans
Male
Middle Aged
Neoplasms* / epidemiology
Neoplasms* / mortality
Sociodemographic Factors
Socioeconomic Factors