Methods
Modelling
Linear regression modelling is a statistical procedure that allows continuous data to be modelled according to a line of best fit, which is then used to extrapolate the trajectory of the data. The adequacy of the fit of the models are considered with relevancy to the underlying data, including procedures used to assess whether the modelling assumptions have been satisfied.
For this report, linear regression modelling has been used to establish a baseline annual trend for the rate of a selected characteristic or outcome between 2015 and 2019. This was to ensure that there were a sufficient number of years for the modelling and that it was based on the most recent trend.
The baseline annual trend was used to predict the proportion of a characteristic or outcome for 2020 and 2021. The predicted proportion was then compared to the observed proportion for 2020 and 2021, that is, the proportion that actually occurred in the population.
The observed proportions in 2020 and 2021 have been described as higher or lower than the predicted proportions where the relative difference is 5% or greater of the predicted value. This means that small absolute differences between observed and predicted proportions with low values may meet this threshold, while changes for larger absolute differences with higher values may not meet this threshold. For example, an observed value of 3.4% and predicted value of 3.2% would represent a relative difference of 6%, whilst an observed value of 83.6% and predicted value of 84.8% would represent a relative difference of 1%.
It was not possible to undertake linear regression modelling for some characteristics or outcomes due to the following:
- alcohol consumption during pregnancy did not have enough years of data to establish a baseline trend
- active resuscitation method did not have enough years of data to establish a baseline trend
- diabetes status had high variability between years, which did not meet the assumption of linearity
- stillbirths had high variability between years, which did not meet the assumption of linearity.
There was a potential for biased estimations if ‘Not stated’ values were included in the linear regression model. Therefore, these values were assumed to be missing completely at random and were excluded.
Perinatal mortality rates
Calculation of stillbirth rate
The stillbirth rate is presented as the number of stillbirths per 1,000 births (that is, all live births and stillbirths) in a specified population.
Stillbirth rate = 1,000 x Number of stillbirths / Total number of births
Calculation of neonatal mortality rate
The neonatal mortality rate is presented as the number of neonatal deaths per 1,000 live births in a specified population. Neonatal deaths are those which are live born and subsequently die within 28 days of birth.
Neonatal mortality rate = 1,000 x Number of neonatal deaths / Number of live births
Calculation of perinatal mortality rate
The perinatal mortality rate is presented as the number of stillbirths or neonatal deaths (perinatal deaths) per 1,000 births (that is, all live births and stillbirths) in a specified population.
Perinatal mortality rate = 1,000 x Number of perinatal deaths / Total number of births
Geography
Geographic data are based on the usual residence of the mother. Between 2017 and 2021, the usual residence of the mother is based on Statistical Area Level 2 (SA2) of the Australian Bureau of Statistics Australian Statistical Geography Standard Edition 2016 for all states and territories.
Prior to 2017, a different correspondence was used for SA2, meaning that differences between pre-2017 and 2017-2021 may be due to changes in the geographic borders rather than changes in the data. As a result, geography is reported from 2017 onwards in this report.
Note that disaggregating by geography was not possible for outcomes where there were a high proportion of small numbers requiring suppression (see Confidentiality). These outcomes include alcohol consumption during pregnancy, diabetes and hypertension status, home births, type of analgesia, active resuscitation method and perinatal deaths.
Primary Health Network
Primary Health Networks (PHNs) have been established by the Department of Health to increase the efficiency and effectiveness of medical services and improve the coordination of care for patients.
Perinatal data at Statistical Area Level 2 (SA2) were linked to 2017 PHNs using Australian Bureau of Statistics correspondence files.
The relevant proportion for each PHN was then calculated, and PHNs were categorised based on the median, interquartile ranges and 10th and 90th percentiles for the proportions at the PHN level. The categories were then adjusted to account for natural breaks in the distribution of the data and for easier interpretation (for example, a range with a maximum of 52.1% of mothers receiving antenatal care in the first trimester would be revised to a maximum of 50%). PHNs were allocated to categories based on unrounded proportions.
Remoteness
This report uses the Australian Statistical Geography Standard Remoteness Structure, which groups geographic areas into six classes of Remoteness Area based on their relative access to services using the Accessibility/Remoteness Index of Australia.
The six classes are: Major cities, Inner regional, Outer regional, Remote, Very remote and Migratory, see the Australian Statistical Geography Standard (ASGS): Volume 5 – Remoteness Structure, July 2016 (ABS 2018a).
Remoteness data used in this report are derived by applying this classification to the mother’s usual area of residence in the NPDC. Remoteness area was calculated where geographic area of usual residence was provided.
Statistical Area Level 3
Statistical Areas Level 3 (SA3) are geographical areas built from whole Statistical Areas Level 2 (SA2) and are designed for the output of regional data. SA3s create a standard framework for the analysis of ABS data at the regional level through clustering groups of SA2s that have similar regional characteristics. Whole SA3s aggregate to form Statistical Areas Level 4 (SA4). There are 358 spatial SA3 regions covering the whole of Australia without gaps or overlaps (ABS 2018b).
Perinatal data at Statistical Area Level 2 (SA2) were linked to Statistical Area Level 3 (SA3) using Australian Bureau of Statistics correspondence files.
Confidentiality
To maintain privacy and confidentiality of individuals, cells in the data tables are suppressed if there is a risk of disclosure of an attribute of an individual that was not already known. A cell in a table is considered identifiable if, as well as being able to identify the entity, other details are also revealed. It is AIHW policy that these cells need to be confidentialised, unless the attribute that would be disclosed is deemed to be non-sensitive in the context of the data being published.
Small numbers
Numbers of less than 5 have not been published (n.p.), in line with guidelines for protecting the privacy of individuals. Exceptions to this are small numbers in ‘Other’ and ‘Not stated’ categories. Consequential suppression of numbers has also been applied where required to prevent back-calculation of small numbers. However, all suppressed numbers have been included in the totals.
Per cents based on denominators of less than 100 have also been suppressed (n.p.) for reliability reasons.
Australian national birthweight percentiles by gestational age
Birthweight percentiles were calculated from data on all liveborn singleton babies born in Australia between 2004 and 2013 with a gestational age of 20–44 weeks.
Records with indeterminate sex were excluded from analysis. Records with missing or not stated data for sex, birthweight or gestational age were also excluded. Birthweight outliers were calculated and excluded using a method based on Tukey’s box and whisker plots.
Gestational age is reported in completed weeks of gestation, calculated from the first day of the last menstrual period (LMP) or estimated by prenatal and/or postnatal assessment if the LMP date was missing. Birthweight is reported to the nearest 5 grams.
Small for gestational age is defined as babies with birthweight below the 10th percentile according to the national birthweight percentiles for the period 2004 to 2013.
For more information on data used to assign percentiles see National Perinatal Data Collection annual update 2021 data table 6.1.
References
ABS (Australian Bureau of Statistics) (2018a). Australian Statistical Geography Standard (ASGS): Volume 5 – Remoteness Structure, ABS, Australian Government.
ABS (Australian Bureau of Statistics) (2018b), Australian Statistical Geography Standard (ASGS): Volume 1 – Main structure and Greater Capital City Statistical Areas, ABS, Australian Government,