Participation in cervical screening by Indigenous women in the Northern Territory: a longitudinal study

Participation rates for NT women (Indigenous and non-Indigenous combined) were calculated for four biennial periods between 1997 and 2004 for the seven Aboriginal and Torres Strait Islander Commission (ATSIC) regions in the NT. These comprise two urban regions (Alice Springs and Darwin) and five rural/remote regions (Box 1). We used ATSIC regions, which closely approximate NT Government Administrative Health Districts, because estimated resident populations were available from the Australian Bureau of Statistics (ABS) for the ATSIC regions but not for the health districts.

Previously, the NTPSR allocated women to administrative health districts on the basis of their postal address postcode. However, mail to some border communities in South Australia and Western Australia is delivered via the NT, and, consequently, women in these communities were incorrectly coded as NT residents (eg, Ernabella, SA, has the postal address “via Alice Springs PO Bags, NT 0872”). In our study, we allocated regions of residence using information from all address fields (not just postcode) in the NTPSR database. As a result, about a quarter of women previously coded as residents of the Alice Springs Rural Administrative Health District (equivalent to the Apatula region shown in Box 1) were excluded from our study as residents of other states. This reallocation also corrected previous regional misclassifications of women within the NT.

Consistent with national performance indicators, the rate of participation in cervical screening was calculated as the proportion of women in the target population who had at least one Pap smear in each 2-year period. We divided the number of women aged 20–69 years who had at least one smear in a 2-year period by the number of eligible women in the same time period (estimated as the mean of the two annual ABS population estimates for the period).

Although the NTPSR is permitted by NT Public Health Regulations to include data on Indigenous status, it does not do so, as these data are not recorded by federally funded pathology services.9 Therefore, we could not directly calculate Indigenous participation rates in cervical screening. Instead, we estimated rates indirectly for the nine rural/remote areas in which Indigenous women comprised 70% or more of the total female population (Box 1). We chose these areas because indirect estimates would have been less accurate for populations with lower proportions of Indigenous women. These nine areas encompass 55% of the NT female Indigenous population.

As with regional participation, area of residence was allocated using information from all address fields in the NTPSR database. We used Statistical Local Areas (part of the Australian Geographic Classification System10) to define rural/remote area boundaries, as Indigenous and non-Indigenous population estimates were available for these areas for the year 2001 from the ABS. We derived population estimates for other years by assuming that the rate of population change (within 5-year age groups, separately for Indigenous and non-Indigenous populations) was the same for each area as for the corresponding ATSIC region.

We calculated participation rates for Indigenous women in these areas, assuming that rates for non-Indigenous women were equivalent to those for the Australian population (ie, 61% in 2001). Thus, the number of Indigenous women having smears in each area in each biennial period was estimated by subtracting 61% of the eligible non-Indigenous women from the number of eligible women who had at least one smear. This result was divided by the mean of the two mid-year estimates of the eligible Indigenous population to give an indirect estimate of Indigenous participation rates.

We performed sensitivity analyses by varying the assumption of a 61% non-Indigenous participation rate by 10% (ie, 51% and 71% participation). The effect of the different assumptions was examined using absolute rate differences.

All participation rates were age-adjusted to the 2001 Australian population age distribution,11 and all population figures were adjusted to remove the proportion of women estimated to have undergone a hysterectomy (based on the 2001 National Health Survey hysterectomy fractions12). Results were compared with total NT and Australian rates obtained from national reports8 over the corresponding time periods.

Tests for trend were not performed, as trend assessment is difficult when there are four or fewer data points, and as there appeared to be more than one trend within the 8-year period for many of the comparisons (eg, a large increase during one period followed by a gradual decline thereafter).

Stata 8.0 statistical software (StataCorp, College Station, Tex, USA) was used for statistical analysis.

Our study was approved by the Human Research Ethics Committee of the NT Department of Health and Community Services and the Menzies School of Health Research.

Participation rates in NT urban regions were higher than national rates from at least 1997 onwards in Darwin, and from 2000 in Alice Springs (Box 2). Rates in all rural/remote regions were lower than Australian rates. The three Top End rural/remote regions (Jabiru, Katherine and Nhulunbuy) showed improvements in participation rates between 1997–1998 and 1999–2000, with no overlapping confidence intervals. These improvements were maintained above the 1997–1998 baseline despite a downward trend in subsequent years. Since 1999–2000, the Apatula region in Central Australia has consistently had the lowest rates.

In 1997–1998, the overall estimated participation rate for Indigenous women living in the “> 70% Indigenous areas” was about half the Australian rate (33.9% [95% CI, 32.6%–35.2%] versus 63.9% [95% CI, 63.8%–63.9%]). In 1999–2000, the rate for Indigenous women increased to 44.0% (95% CI, 42.7%–45.4%). There was little change from 1999–2000 to 2003–2004, with Indigenous participation rates remaining 18–19 percentage points lower than Australian rates.

Since 1999–2000, Indigenous participation in these “> 70% Indigenous areas” has remained higher in the Top End than in Central Australia, where there has been no overall improvement (Box 3). Within Central Australia, trends varied between these rural/remote areas (Box 4): in Area H, rates increased in the last time period, but, in Area G, rates fell after 2000. Participation in Area I improved, but overall this area had the lowest participation. Notably, Area D in the Top End had participation rates that were higher than total NT and Australian rates from 1999. This area also had the highest proportion of Indigenous residents.

Non-Indigenous women comprised only 11.1% of the population in the nine rural/remote areas studied. Varying the assumption of the rate of non-Indigenous participation made little difference to estimated Indigenous participation rates: if non-Indigenous participation was 51%, the Indigenous participation rate would have been, on average, 1.1% higher (range, 0.3%–1.6%) in areas with an Indigenous population greater than 90% and, on average, 3.0% higher (range 1.5%–4.8%) in areas with Indigenous populations between 70% and 89%. If non-Indigenous participation was 71%, Indigenous rates would have been, on average, 1.3% lower (range, 0.1%–2.7%) and 3.2% lower (range, 1.7%–4.6%), respectively.