Geographical variation in alcohol access, risk and harms in selected Australian states

These distance analyses use known information on Australian road maps, and locations of alcohol outlets to explore measures of access for the population centre point (called the weighted population centroid) of a defined small area (also known as SA1) to the alcohol outlets. This information is used to determine:

• the distance to the closest outlet
• the number of alcohol outlets within an SA1
• the distance to the alcohol outlet closest to the population centroid for an SA1
• the accessibility to alcohol outlets in an area (the spatial access index).

The details of how these measures were calculated for SA1 and SA2 are under the Distance analysis heading below.

Alcohol outlet data source

Information on alcohol outlets for this report have been obtained from the liquor licensing authorities for each of the following Australian States and Territories: New South Wales, Victoria, Queensland, Western Australia, South Australia, Tasmania.

Data for the period of interest was not available for the Australian Capital Territory or Northern Territory.

The information also included whether the alcohol outlet was licensed for on premise drinking, off premise drinking or both. These data have not been included in the report.

Software

ESRI Arc Pro 3.0 network analyst was used to calculate travel distance between population centroids and alcohol outlets.

Population data

Population data for both the 18 and over population at the SA1 level and total population data for the mesh block level were sourced from the 2016 Census data using the Australian Bureau of Statistics Table Builder (ABS 2021).

Geographical data

SA1 boundary and mesh block boundary in 2016 were sourced from the ABS website (ABS 2016). G-NAF (Geocoded National Address File) was from Geoscape Australia. Geoscape G-NAF is the geocoded address database for Australian businesses and governments. It is Australia’s trusted source of geocoded address data, with over 50 million contributed addresses distilled into 15.4 million G-NAF addresses. This data is represented as points for the address locations. Commercial addresses are excluded for this analysis purpose.

Road network data

The road network data for distance analysis was sourced from ESRI Street map premium.

Calculating the weighted centre of population for areas (weighted population centroids)

Populations tend to be distributed throughout areas, not concentrated at one location. A centre point can be used in calculations to represent the population of an area in the same way a mean represents the average point within a data set. This ensures the population centre best represents the location of people within the area - particularly for more sparsely populated areas. Population centre points for SA1 were used (calculated from mesh blocks) because they are the smallest geographic level at which ABS population data are available and because of their relatively small size.

Using the data from G-NAF, each address in a mesh block was used to calculate the average centre of population for the mesh block using the calculation:

This was then merged with the table of population counts based on Mesh Blocks code using spatial join. 

The populations of each mesh block were used as weights within an SA1 to shift the population centre and create a weighted population centre for the area. This was done using the Mean Centre tool in ArcPro.

Data were also aggregated to SA2 level after analysis of drive time to alcohol outlet.

Distance analysis

1. Minimum drive distance (meters) to closest alcohol outlet (SA1/SA2)

   The minimum drive distance from the population weighted centroid of an SA1 to the closest alcohol outlet was computed using the closet facility solver in ArcGIS Pro. The closest facility solver measures the travel distance between each SA1 population-weighted centroid and alcohol outlets and determines which are nearest to one another.

   The Minimum drive distance to the closest alcohol outlet at the SA2 level is the average result of minimum drive distance to the closest alcohol outlet at the SA1 level. It sums up all SA1s’ minimum drive distance to the closest alcohol outlet within each SA2 boundary and then divided by the number of SA1s in each SA2 to get the result.

2. Number of alcohol outlets per SA1/SA2

   The number of alcohol outlets was computed using the spatial join method in ArcGIS Pro. It counts the number of alcohol outlets within each SA1/SA2 boundary.

3. Number of alcohol outlets per person (SA1/SA2)

   The number of alcohol outlets per person is calculated by dividing the number of alcohol outlets by the total population in each SA1/SA2.

4. Number of alcohol outlets per adult (SA1/SA2)

   The number of alcohol outlets per person is calculated by dividing the number of alcohol outlets by the population aged 18 and over in each SA1/SA2 (see population data section for more information).

5. Weighted spatial accessibility index (SA1/SA2)

   The spatial accessibility index for SA1 is first calculated using the following formula:

   SAI (SA1) = sum of inverse distance of closest 5 outlets = (1/distance a) + (1/distance b) + (1/distance c) + (1/distance d) + (1/distance e)

   The weighted spatial accessibility index at the SA1 level is calculated with this formula:

   Weighted SAI (SA1) = SAI (SA1) * population aged 18 and over in each SA1

   The weighted spatial accessibility index at the SA2 level is calculated with this formula:

   SAI (SA2) = sum of each Weighted SAI (SA1) within SA2 /sum of population aged 18 and over in each SA1 within SA2

   For all the analyses, we excluded statistical areas in the Australian Capital Territory, the Northern Territory, Other territories, islands such as Norfolk Island and Lord Howe Island with no road network, or areas without geographical boundary (these SA2s are the ones with “No usual address or “Migratory - Offshore – Shipping). All the analyses include both on-premises and off-premises alcohol outlets.

Steps of the summarised result

1. National level

   The summarised results (minimum, maximum, average, standard deviation) at national level are calculated directly from the result table. Australian Capital Territory and North Territory are excluded from the summary as data on alcohol outlet locations was not available at the time of this report.

2. GCCSA level

   We firstly spatial join with SA1/SA2 layer with ABS GCCSA layer to capture the GCCSA information for each SA1/SA2. Then we summarised the results (minimum, maximum, average, median) for Greater Capital City areas (called Capital city areas in the report) and Rest of State areas (called Non-capital city areas in the report) separately.

Validation

For the results of minimum drive distance to the closest alcohol outlet at the SA1 level. We compared our drive distance results with that in Google Maps. We chose a couple of Origin-Destination pairs to compare the results and they are compared in different distance ranges. Our analytical results are very close to the drive distance in Google Maps, which indicates our results are reasonable. The differences between our results and Google Maps mainly result from two reasons:

• Google Maps considers real-time traffic and road conditions, so the road restrictions and conditions may be different from our road network database.
• The road network is very sparse in some remote areas, so the estimated drive distance both in our results and Google Maps are not as accurate as city areas. This could lead to a larger difference.

Methods for demonstration analyses of risks and harms by geographic areas

In this report, measures of alcohol-related risks and harms were derived for people aged 14 and over from the AIHW National Drug Strategy Household Survey (NDSHS) 2019 (AIHW 2021b). The NDSHS collects information on alcohol and tobacco consumption, and illicit drug use among the general population in Australia. The survey has been conducted every 2 to 3 years since 1985. 

The analysis of risky alcohol consumption and alcohol related harms is only presented in this report for Capital city areas. Using the same analysis for Non-capital city areas was limited by the variation across Non-capital city areas in distance to the closest alcohol outlet, due to the vast distances between populations and alcohol outlets in some small geographic areas of Australia. This meant it was difficult to create relevant distance categories to summarise the data for statistical analyses.

Measuring risky drinking

Measuring risky drinking according to the Australian NHMRC 2020 alcohol guidelines (NHMRC 2020) was calculated for people aged 14 and over from NDSHS 2019 questions as outlined in the AIHW report Measuring risky drinking according to the Australian alcohol guidelines (AIHW 2021a). 

Measuring alcohol-related incidents from someone under the influence of alcohol

Measuring harms experience due to someone under the influence of alcohol were derived from question Y1. in the 2019 NDSHS questionnaire (AIHW 2021b):

In the last 12 months, did any person under the influence of or affected by alcohol (Mark one response for each row)

• Verbally abuse you (Y/N)
• Physically abuse you (Y/N)
• Put you in fear (Y/N)

Distance to closest alcohol outlet

Distance to the nearest alcohol outlet was calculated from the population-weighted centroid of the SA1 in a Capital city area (called Greater capital city area using the ABS geography (ABS 2016)) that the survey respondent resided in. All results were weighted to the Australian population by age, sex and strata, and differences were deemed to be statistically significant if the chance of seeing the observed difference under the null hypothesis was less than 5% (p <0.05). Significance testing was done using the distance group of interest and comparing it against all other distance groups combined. See National Drug Strategy Household Survey 2019: Technical information (AIHW 2021b) for further information.