Health behaviour modelling for prenatal diagnosis in Australia: a geodemographic framework for health service utilisation and policy development

Datasets

Study population

All women residing in Victoria who had amniocentesis or CVS in 1998 or 2002 and whose expected date of delivery was also in that year were included in the study. This was done to ensure that the births and prenatal tests occurred within the study period and to facilitate record linkage. The years were chosen because second trimester maternal serum screening was made available to all pregnant women in the public health sector in Victoria in 1996/1997 and became well established in 1998. The 2002 data were the most current available and incorporate diagnostic tests prompted by combined first trimester screening for women in the private health care sector, introduced in Victoria in 2001. Denominator data were obtained from all births (to calculate a Down syndrome live birth ratio) or confinements (to estimate uptake of prenatal diagnosis) recorded at the PDCU in 1998 and 2002. Data on births of babies with Down syndrome were obtained from the Birth Defects Register for the years of 1997 to 2002 and included all babies born 20 weeks gestation and later (live births, still births and neonatal deaths), excluding all terminations of pregnancies. Ethics approval for this study was obtained from the Victorian Government Department of Human Services Human Research Ethics Committee.

Geodemographic segments

The geographic identifier used in this study was the mother's postcode of residence. The allocation of postcodes in Victoria underwent a major change in 1996 followed by the creation of new Local Government Areas (LGAs) which have remained stable since. Where a number of postcodes are divided between two or more different LGAs, allocation to a specific LGA was assigned proportionally by the PDCU on the basis of Australian Bureau of Statistics census data. There are currently 79 LGAs in Victoria, each is made up of a number of Census Collection Districts. The Census Collection District (CCD) is the smallest geographic area as defined by the Australian Standard Geographical Classification and correct allocation to a CCD is only achieved by using a street address. As this variable is not routinely collected by the PDCU, it was necessary to interpolate the LGA data and re-distribute this data to the component CCDs in each LGA. The modelling for this involved first taking the rate of testing that applied in each LGA and applying this standard rate to each component CCD. The modelled information was then aggregated into 58 geodemographic lifestyle segments across the State. These segments are derived from comprehensive household data tabulated by the Australian Census of Population and Housing in 2001 and extensive consumer surveys and were used under the licence of Pathfinder Solutions (Australia) Pty Ltd (Pathfinder Segments of Australia©). Basic segment descriptions used to showcase customer behaviour profiles for commercial purposes, are available as online content (see Additional file 1).

Data analysis

Expected numbers of live births with DS were calculated by assigning a maternal age-specific risk at term to each birth in 2002, according to the mother's age. Maternal age-specific risks were taken as published by Reynolds in 1994 [19]. Modelled and aggregated expected and observed numbers of DS were then tabulated for each geodemographic segment and expressed as a proportion of all DS births. Modelling of data resulted in artificially high numbers of observations and it was not feasible to undertake a valid statistical analysis. Consequently, a descriptive approach was taken.

Modelled observed rates of prenatal diagnosis were compared to State average rates (expected) for both maternal age groups and visualised in Figures 1 and 2.

Modelled rates of prenatal diagnosis and confinements (all ages) for each geodemographic segment were expressed as a proportion of all confinements. A ratio for DS for each segment was created, where the observed rate being equal to the expected rate was defined as 1. A ratio above 1 indicated a higher than expected proportion of DS and a ratio below 1 indicated a lower than expected proportion of DS. This ratio was defined as the live birth ratio of DS. Similarly, a ratio for prenatal diagnosis was derived by comparing observed rates of uptake with the State average (all ages). Uptake of prenatal diagnostic testing was plotted against live birth ratio of DS as depicted in Figure 3.

Description of data

The 2002 State average uptake of prenatal diagnosis in women aged less than 37 years was 3.8% (Figure 1). There were marked differences in the rates of uptake between metropolitan and rural geodemographic segments, ranging from 3.1% (Young Single Parent Families) to 5.7% (Asset Rich, Income Rich) and from 2.0% (Dairy Farming) to 3.2% (Affluent Coastal Lifestyle) respectively (Figure 1). The 2002 State average uptake of prenatal diagnosis in women of advanced maternal age (37 years and over) was 42.2% (Figure 2). Uptake in metropolitan Victoria ranged from 38.9% (Young Single Parent Families) to 52.5% (Inner Suburban Lifestyle Seekers) and in rural areas it ranged from 15.5% (Wheat Farmers) to 39.4% (Affluent Coastal Lifestyle). Most point estimates differed markedly from State average and there was no apparent association between increasing SES and levels of uptake, although the four highest uptake rates in younger mothers were seen amongst the eight highest SES segments of metropolitan Victoria. The level of prenatal diagnosis in younger mothers reflects uptake of prenatal screening or diagnostic testing paid for through the private health care sector [22].

Figure 3 depicts uptake of prenatal diagnosis across all ages compared to State average (8.4%) plotted against the modelled live birth ratio of DS. In the metropolitan segments, the live birth ratio of DS mostly lies within 20% of the expected rate with only two outliers, one of which was 40% lower than expected (Asset Rich, Income Rich). The other (Older Money & Asset Rich) was 30% lower than expected. Ten of the 16 rural segments however, show a live birth ratio of DS of >20% higher than expected. Figure 3 also shows an almost universal inverse relationship between rates of uptake of prenatal testing and live birth ratio of DS.

Uptake of prenatal diagnosis (Figures 1 and 2)

Given that the results were based on modelled data, extrapolated from actual, we discuss only those segments with rates greater or less than 20% variation from State average, (above or below the line in figures) for each region.

Typical residents of the four metropolitan segments with the highest uptake in younger women (Asset Rich, Income Rich, Rising Wealth, Inner Suburban Lifestyle Seekers and Older Money & Asset Rich) are high income couples with or without children who are at least twice as likely to have a household income of A$100,000 per year compared to the rest of the population. They live in the inner suburbs, either own their home outright or are paying high mortgages and drive late model luxury cars. They have private health insurance and spend around 20% more on specialist physician's fees than the average Australian household. For women aged 37 years and over who have access to testing through the public system, increasing SES had no additional bearing on uptake rates. The "Inner Suburban Lifestyle Seekers" was the only segment of the four with a considerably higher than average uptake of prenatal diagnosis in women of advanced maternal age (52.5%).

In contrast, there were four high income segments in younger mothers that did not show above average uptake rates of prenatal diagnosis. Reasons for this were that the segments were defined by an over representation of professionals, associate professionals, advanced clerical and service workers (Moving Up), families with teenage children (Comfortable and Owned Outright, Mature Families) and couples aged 40–44 years old with young children aged around 5 to 14 years (Professionals with Young Families). In addition, people living in these segments also subscribe to private health insurance but do not spend extra money on specialist fees when compared to all of Australia.

There was a marked peak in the uptake rate by younger women in one lower socioeconomic rank (High Rise Rentals, 4.9%). As high rise rentals in metropolitan Melbourne are interspersed throughout the wealthy inner suburbs, the finding of a high uptake rate in women living in this segment is most likely related to an artefact of the data modelling, rather than an actual characteristic of this group. The segment is defined by a large proportion of residents born overseas (40%), with over 60% having arrived in Australia in the last ten years and a strong skew towards the 20 to 34 year old age groups. They are 30% less likely to have private health insurance than the average Australian.

There were also relatively high prenatal diagnosis uptake rates for both age groups in the "Vietnamese Migrant Enclaves". Although these did not vary more than 20% from State average, the peaks may be explained by the fact that there is a well established metropolitan community based antenatal clinic for Vietnamese women [20] and by a tendency to accept prenatal testing as advised by their doctors [21].

Uptake of prenatal diagnosis in both maternal age groups in rural areas was universally low and most segments were well below 20% less than the State average in the younger women. Segments such as "Affluent Coastal Lifestyle", "Thriving Regional Living", "Rising Country" and "Regional Mix" are predominantly located in regional centres and access to services may have played a role in their slightly higher rates of uptake, particularly in the older women. The "Regional Battlers" segment, while also defined as located in regional centres, mainly consists of one parent families, separated or widowed people and residents aged over 60. Household income and level of education are well below average and a high proportion of households do not have a car. Residents in this segment are 31% less likely to spend money on specialist physician's fees than average. The issue of access to testing for women in this segment more likely relates to a lack of individual opportunity, than their physical location.

Women residing in the heartland of farming in Victoria (Dairy Farmers, Wheat Farmers, Cattle Country and Sheep Runs) had the lowest rates of uptake of testing. Most residents in these areas were born in Australia and married couples with children are over represented. Although there is low unemployment, generally household incomes are below average. While private health insurance cover is common, people living in farming areas spend little on specialist physician's fees compared to the rest of Australia.

Interestingly, patterns of utilisation of prenatal diagnosis in rural women were not uniform in both maternal age groups. "Dairy farmers" had the lowest rates of prenatal diagnosis in the younger women, whereas older women in "Wheat farming" areas had exceptionally low rates of uptake. A similar discordance was observed for the "Mixed Farming" segment, which was very low in the younger women. Households with more than five people are over represented in farming areas and low uptake of testing in older mothers and, to an extent, in the younger mothers may be explained by a previous finding that women of higher parity are less likely to have testing [8]. A combination of this and access to testing, possibly further contribute to low uptake in younger women in the "Dairy Farming", "Mixed Farming" and "Cattle Country".

Down syndrome live birth ratio (Figure 3)

We observed an almost universal inverse relationship between utilisation of prenatal testing and live birth ratio of DS. This is not surprising as approximately 75% of cases with DS are diagnosed prenatally in Victoria [22] and studies have shown that most pregnancies are terminated following diagnosis [23, 24]. However, two of the four highest rates of testing in metropolitan Victoria did not result in equally low rates of babies born with DS. Women in the metropolitan "Rising Wealth" and "Inner Suburban Lifestyle Seekers" segments showed a high overall uptake of testing in Figure 3, whereas the live birth ratio of DS was less than 20% lower than expected. This finding is difficult to explain within the constraints of the data available, but an underlaying skewed maternal age distribution and the associated risk for DS may be a contributing factor. Table 2 shows that by 2002, one third of all babies with DS were born to the high risk but relatively small group of women of advanced maternal age and it is possible that testing is not reaching this particular high risk population in the most effective way. It has certainly been shown that prenatal screening is a more effective filter for prenatal diagnosis than advanced maternal age alone [22]. Consequently, the relationship between uptake of prenatal diagnosis and live birth rate of Down syndrome may be affected by the extent to which women use prenatal diagnosis with or without consideration of screening results [25]. In addition, a contribution to the departure from the general inverse relationship between uptake of prenatal diagnosis and live birth ratio of DS may be that some women in these segments are more likely than others to continue with their pregnancy in the event of a positive test result.

Table 3 shows that, in 2002, there were proportionally more babies born with DS in rural areas when compared to the overall birth rate. This prevalence appears to be directly related to levels of uptake of prenatal diagnosis (Figure 3), in particular in the farming segments. The live birth ratio of DS was highest in wheat and mixed farming areas with a rate 70% higher than expected. Only 26% of people in wheat and mixed farming areas and 31% in sheep farming areas live in inner regional districts and lack of access to services may partly contribute to these findings. The "Low income and ageing coastal" and the "Affluent coastal lifestyle" segments also showed a live birth ratio of DS of 70% higher than expected. These segments are at either end of the socioeconomic scale but largely co-exist in similar geographic locations and without a more specific geographic identifier, it is difficult to clarify how the characteristics of the segments may contribute to these findings.

In summary, the lack of statewide geodemographic consistency in uptake of prenatal diagnosis implies that there is a need to target health professionals and pregnant women in specific areas to ensure there increased equity of access to services and that all pregnant women can make informed choices that are best for them. Equally as important is the increased opportunity for reproductive choice and provision of appropriate health services for families of children with Down syndrome. Our findings show that these potential interventions are particularly relevant in rural areas.