Summary
Traditionally, child mortality rates are estimated using cross-sectional or point in time analysis. In this method, deaths (numerator) and population at risk (denominator) are derived from two different data sources resulting in lack of comparability between the two sources in terms of Indigenous identification, and inconsistency between the numerator and the population at risk.
In the cohort-specific method, birth cohorts are followed over time and child mortality can be estimated for this cohort. This method provides a more accurate measure of infant and child mortality because of the consistency in Indigenous identification of both deaths (numerator) and population at risk (denominator) as both these data are derived from the same data collection—the Perinatal Data Collection (PDC).
Modelling risk factors for adverse pregnancy and birth outcomes
This report set out to examine the following:
- The feasibility of using linked perinatal, birth and national death data to estimate cohort specific deaths rates and how these compare with cross-sectional or point in time death rates of infant and under-5 child mortality
- the methodological and data quality issues associated with linking perinatal records to birth registration and death registration records to track deaths and the cause of death of Indigenous infants and children aged less than 5 years to their originating birth cohorts
- data quality issues relating to modelling the risk factors associated with adverse pregnancy and birth outcomes among Indigenous and non-Indigenous women.
To undertake these analyses, a Linked Perinatal, Birth Death Data set was created by linking jurisdictional perinatal and birth registration records to the National Death Index (NDI) to identify Indigenous under-5 deaths occurring in specified birth cohorts within jurisdictional Perinatal Data Collections (PDCs).
The results from these examinations are described in this Summary.
Feasibility of linking perinatal and birth records to the NDI
The analysis showed that it is feasible to link jurisdictional perinatal and birth registration records to the National Death Index (NDI) to identify Indigenous under-5 deaths occurring in specified birth cohorts within jurisdictional Perinatal Data Collections (PDCs).
The quality of this linkage depends on the quality of linkage variables on the three named data sets. The quality of these variables has been improving over time.
Comparing cross-sectional and cohort-specific under-5 deaths
- The National Mortality Database (NMD) contains records of all deaths that occurred in Australia and were registered in Australia. Earlier evidence shows that the NMD underestimates the number of under-5 deaths of Indigenous children as it has not been enhanced for Indigenous misclassification.
- Cross-sectional or ‘point-in-time’ deaths are derived from the NMD. The NMD was not linked to the jurisdictional PDCs.
- Cohort specific analysis of under-5 mortality using the Linked Perinatal, Birth, Death Dataset was able to identify about 90% of all cross-sectional under-5 deaths and about 84% of cross-sectional under-5 deaths of Indigenous children nationally.
- Excluding Tasmania and the Australian Capital Territory (both having extremely small number of deaths and small Indigenous populations), the percentage of cross-sectional Indigenous under-5 deaths on the NMD that was identified in the linked perinatal dataset ranged between 65% in Queensland and 97% in Western Australia.
- The lower percentage of Indigenous under-5 deaths identified in the linked perinatal data set (84%) compared to the number of Indigenous under-5 deaths estimated from cross-sectional data is due to known factors.
- the change from a paper-based to an electronic platform for the recording and collating of perinatal data in Victoria, particularly during 2008 and 2009.
- the absence of personal identifiers on the Queensland perinatal data until 2008
- the absence of personal identifiers on the Northern Territory perinatal data for all records in scope.
Modelling risk factors for adverse pregnancy and birth outcomes
Linking perinatal to death records made it possible to model the risk factors associated with adverse pregnancy and birth outcomes, such as perinatal mortality (comprising stillbirth and neonatal mortality), pre-term birth, low birthweight, as well as infant and child mortality rates.
While the NDI contained information on date of death, age at death and cause of death, the perinatal data contained information on the maternal contextual, demographic and socioeconomic characteristics, health risk factors and pregnancy complications associated with adverse birth and pregnancy outcomes (such as stillbirth, pre-term birth, low birthweight, infant death and child death).
Modelling the risk factors associated with cause of death
Cause of death information is not available on perinatal records except for a few early age at death records.
However, linking perinatal records to the NDI enabled nearly all under-5 deaths to be analysed in terms of cause of death and age at death, in conjunction with the maternal contextual, demographic and socioeconomic characteristics, health risk factors and pregnancy complications associated with the under-5 deaths.
Improvements in data quality
- There has been continuous improvement in the quality of the perinatal data since 2010, particularly in the availability and quality of linkage variables to enable jurisdictional PDCs to be linked to jurisdictional birth registration records and death records.
- Since 2010, there have also been improvements in the availability of key risk factors on jurisdictional PDCs for modelling the differentials in adverse pregnancy and birth outcomes between Indigenous and non-Indigenous women.
Summary
1. Linked Perinatal, Birth, Death Data set
- Objective of the Dataset
- Purpose of this report
- Structure of this report
2. Data and methods
- Sources of data
- Jurisdictional Perinatal Data Collections
- Jurisdictional birth registrations
- National Death Index and National Mortality Database
- Enhanced Mortality Database
- Justification for 2-step process
- Ethics approval
- Data custodian approval
- Protection of data privacy
- Protection of data confidentiality
3. Linkage of perinatal and birth records
- Overview
- Data linkage process
- Quality of linkage variables
- Linkage rate: perinatal to birth registration records
- Quality of linkage of perinatal to birth registration records
4. Linkage of perinatal to death records
- Overview
- Quality of data linkage
- Consistency between cohort-specific and cross-sectional deaths
- Age at death
- All under–5 deaths
- Indigenous under–5 deaths
- Birth cohorts
- All under-5 deaths
- Indigenous under-5 deaths
5. Indigenous identification
- Sources of information on Indigenous identification
6. Cause of death
- Records with missing cause of death information
- Age at death
7. Pregnancy and birth outcomes
- Data availability
8. Cohort-specific under-5 mortality
- Trends in cohort-specific under-5 mortality rates
- Trends for Australia
- Trends for each jurisdiction
- Trends for each jurisdiction
- Conclusion
Appendix A: Estimating cohort-specific deaths with or without data linkage
- Estimating cohort-specific child mortality using data linkage
- Estimating cohort-specific child mortality without data linkage
Appendix B: Cross-sectional and cohort-specific under-5 mortality rates
- Current approach to estimating the under-5 mortality rate
-
Issues with cross-sectional estimates of the infant mortality rate
-
End matter: References; Acknowledgments; Abbreviations; Symbols; List of tables; List of figures