Plan at SA1, Not LGA: Why Population-Health Averages Hide Where the Need Really Is

An LGA average tells you who lives in a catchment.

An SA1 number tells you who needs care.

For Primary Health Networks doing needs assessment, commissioning, performance reporting, and board updates, that distinction is the difference between a plan that lands funding in the right postcodes and a plan that's already wrong on the day it's signed.

This post is the case for planning population-health work at SA1 — the smallest unit the ABS publishes — and treating LGA and catchment-level averages as a presentation layer, not a planning layer.

The averaging problem

Every PHN catchment is a composite. Most cover between five and twenty LGAs. Inside each LGA sit dozens of SA2s. Inside each SA2 sit dozens of SA1s. By the time you stack the averages all the way up to a catchment-wide number, you've smoothed away everything that makes the catchment interesting.

A concrete example. Take a roughly average urban PHN catchment of 10 LGAs and ~900,000 residents. The catchment-wide SEIFA decile might be 8 — comfortable, above-average advantage. The catchment-wide proportion of residents aged 65+ might be 16% — roughly the national figure.

Both numbers are true. Both numbers are useless for commissioning.

Inside that catchment:

The most advantaged SA1 sits at decile 10 with 22% of residents 65+ in a leafy suburb that needs less primary care, not more.
The least advantaged SA1 sits at decile 2 with 31% of residents born overseas, 18% with no English at home, and 24% in single-parent households. It does not need the same intervention as the leafy SA1, and it will not be reached by a catchment-wide program.

These two SA1s are 12 kilometres apart. The catchment-wide average is a lie that flatters everyone.

This isn't a hypothetical. Open the ABS DataPacks for any urban PHN catchment in Australia and you'll see the same shape: a long-tailed distribution at SA1 that compresses into a calm-looking median at SA2, a comfortable mean at LGA, and a useless single number at the catchment level.

Why "plan at SA1" sounds harder than it is

The most common objection is "we don't have the capacity to operate at SA1." It's worth unpacking what that actually means.

It does not mean delivering services SA1 by SA1. No PHN commissions a GP per SA1. The serving unit will always be larger — an LGA, a sub-region, sometimes a postcode cluster.

It means using SA1 as the analytical unit. When you're asking "where is the unmet need?", you want the smallest available grain. When you're asking "who do we fund to address it?", you aggregate back up to the unit of service delivery.

This is the inversion most teams get wrong. They start with the service unit (the LGA), pull data at that grain, and then never see the variance inside it. Better order of operations:

Pull every relevant indicator at SA1.
Identify the SA1s in the top decile of combined need (e.g. SEIFA × age × chronic-condition prevalence).
Roll those SA1s up to the service unit that contains them (LGA, SA2, postcode cluster — whichever fits the commissioned program).
Commission against the rolled-up units, but evaluate against the SA1-level indicators that triggered the program in the first place.

Step 4 is the part that matters. Programs commissioned at the LGA level can still be evaluated at SA1 — and that's how you find out if the program is reaching the people the data said needed it.

What SA1 actually is

A reminder of the geography for analysts new to ABS Australian Statistical Geography Standard (ASGS) territory:

SA1 — Statistical Area 1. Smallest unit ABS publishes most indicators at. Roughly 400 people on average; built so most variables have enough sample to be released without confidentiality suppression.
SA2 — Statistical Area 2. Aggregates 5–10 SA1s. Designed to represent a "community of interest" (a suburb or small town). 3,000–25,000 people. Most health and demographic indicators are released here too.
SA3 — Statistical Area 3. Sub-regional units, 30,000–130,000 people.
SA4 — Statistical Area 4. Labour-force regions, 100,000–500,000 people.
LGA — Local Government Area. Defined by state government boundaries, not ABS. Approximately stable but not perfectly nested with SA2s.

PHN catchments are defined as a list of LGAs (occasionally with sub-LGA carve-outs). They are not an ABS-published unit, which means every catchment-mapped analysis is a derived product.

For a fuller walkthrough of how to map PHN catchments to each ABS level, see How to Map a PHN Catchment to ABS Boundaries.

What data is actually available at SA1

A practical question for analysts: which indicators can you pull at SA1, and which can't you?

Available at SA1 (Census 2021):

Population by age, sex, country of birth, language spoken at home
Household composition, single-parent households, multi-generational households
Tenure type, dwelling structure, rent and mortgage stress
Highest year of school completed, post-school qualifications
Industry of employment, labour-force status, hours worked
SEIFA indices (IRSAD, IRSD, IER, IEO)
Unpaid care for a person with a disability, voluntary work

Available at SA2:

ABS Population Estimates and Projections
ABS Building Approvals (sometimes lower)
ABS Estimated Resident Population by single year of age

LGA or higher only:

Most AIHW indicators (METeOR-defined population-health metrics)
State crime statistics (typically LGA in NSW/VIC/QLD/SA/WA)
MBS and PBS service utilisation (usually SA3 or LGA)
Hospital admissions and ED activity (LGA or hospital-catchment)

The gap in the middle — health-service indicators are typically only published at LGA or above — is where the work of joining ABS social/demographic data at SA1 to health-service data at LGA happens. Done right, you can identify SA1s with high social vulnerability and then check the service-utilisation rates of the LGA they sit in. Mismatches are signals.

The "plan at SA1, present at SA2/LGA" rule of thumb

A useful pattern for PHN teams:

Plan at SA1. When you're identifying need, hotspotting, designing intervention areas — go to the finest available grain.
Present at SA2 or LGA. Board reports, council briefings, public dashboards — present at the unit your stakeholders recognise. The variance you found at SA1 should still be visible in your text ("Most of the catchment sits at decile 7–8; one cluster of SA1s in southern XYZ-LGA sits at decile 2"), even if the headline number is at LGA.
Evaluate at SA1. Programs commissioned at LGA can still be evaluated against the SA1 indicators that triggered them. If the program raised LGA-level service uptake but didn't move the SA1-level utilisation in the hotspot, that's worth knowing.

How to actually get SA1 data into a planning workflow

Three pragmatic options:

ABS TableBuilder / DataPacks. Free, official, but cumbersome. TableBuilder builds custom tables interactively; DataPacks ship pre-built SA1 zips per Census topic. Useful for one-off pulls; painful for ongoing planning work.
AURIN. The Australian Urban Research Infrastructure Network exposes a wide library of harmonised ABS + government datasets at SA1/SA2 via a portal and API. Free for academic use; institutional licences for non-academic. Good for ad-hoc analysis but not built for ongoing PHN reporting cadence.
A pre-joined data layer like Domato PHN. Pre-maps every PHN catchment in Australia to LGA, SA2, SA1 and postcode and pre-joins ABS, AIHW, ACARA and state datasets. Built specifically for PHN planning and reporting cadence; live in production with Sydney North Health Network.

The right choice depends on your team's existing analyst capacity. If you have an in-house team that already works in R or Python against ABS files, you have the foundations to do this yourself — it just takes time. If you don't, you'll spend more on stitching the data than on acting on it.

A short checklist for your next plan

Before you publish the next needs assessment, board pack, or commissioning paper, ask:

What's the smallest unit our indicators are available at? (Likely SA1 for ABS, LGA for AIHW.)
Did we calculate at that unit, or did we average up first?
Where in the catchment is the variance? (Show the distribution, not just the mean.)
If we're commissioning at a larger unit, can we still evaluate at SA1?
Are the SA1s with highest need actually in the LGA we're funding most?

If the answer to the last question is "we don't know", that's the gap your data layer needs to close.

Need a starting point? Start with one indicator at SA1 — SEIFA IRSAD is a good first pick — and look at the distribution across your catchment. The exercise is short. The shock is genuine. The plan that comes out of it is usually a different plan.

Domato PHN is the white-labelled population-health data layer for Australia's 31 Primary Health Networks. Every catchment pre-mapped to LGA, SA2, SA1 and postcode; ABS, AIHW, ACARA and state open data pre-joined. See the product page or reach out for a pilot.