FOI 24/25-2110 DOCUMENT 1

From: To:

Subject: Date:

Hi Gary

Thank you for the opportunity to review your ‘Early Intervention’ and ‘Cost Drivers’ presentation packs. Please find attached feedback from a policy and research perspective, and from a data and analytical perspective.

Overall:

Early Intervention
- The key insights appear reasonable, with the exception of new entrants of participants with autism (Slide 4). The pathway from developmental delay to autism is not fully captured, with the analysis implying a decreasing number of participants with autism entering the Scheme which is incorrect.
- The definition of ‘Early Intervention’ is not as intended, nor consistent with NDIA’s definition, and needs refinement and further verification.
Cost Drivers
- This appears to contain three separate topics of investigation — access (for those with secondary disabilities), ‘over-servicing’ and support co-ordinators, rather than a complete picture of the cost drivers of the Scheme. There is no mention of other significant cost drivers, such as high numbers of children joining the Scheme, sustained plan and payment inflation, and increased demand for Supported Independent Living supports.
- The information collected by the NDIA on a participant’s secondary disabilities and/or conditions is limited. As such, inferences should not be drawn given the lack of confidence in the data.
- The attribution of ‘over-servicing’ to unregistered providers and support coordinators needs further verification.

Note that we have conducted a high-level review of the packs. If to be used further, or made public, a full technical review is recommended.

Thanks

Regards Kirsten

Page 1 of 53

redacted: s22(1)(a)(ii) - irrelevant material NDIS Review Liaison Office National Disability Insurance Agency E: redacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au NDIA logo.png

The NDIA acknowledges the Traditional Custodians of Country throughout Australia and their continuing connection to land, sea and community. We pay our respects to them and their cultures and to Elders past, present and emerging.

From: redacted: s22(1)(a)(ii) - irrelevant material Gary <Garyredacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au> Sent: Thursday, June 1, 2023 10:24 PM To: redacted: s22(1)(a)(ii) - irrelevant material Meng <Meng.redacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material Kim <Kimredacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au> Cc: redacted: s22(1)(a)(ii) - irrelevant material Mabruk <Mabruk.redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material Janet <Janetredacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material Shahar <Shahar.redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material Sidesh <Sideshredacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material Timothy <Timothyredacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material Jack <Jackredacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material Clare <Clareredacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material Kennedy <Kennedy.redacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au> Subject: RE: CAUTION: Email may contain unverified link [be careful if proceeding]Presentations for review [SEC=OFFICIAL:Sensitive]

                 OFFICIAL: Sensitive

Hi Meng. Just sending through the additional context on how these slides will be used, as discussed in the call earlier today.

Currently, the slide and analysis are just for discussion purposes, with the end goal of their current form to be used in preliminary discussions with Bruce and other teams in the Review. Ultimately, we expect some of it to work its way into the final report, but it’s unlikely it’d be the slides as they are currently, but rather pulling out specific facts and a key graph or two. This would only be a minority of the content of what’s in there now, and at that point, we’d come back to your team stating what we intend to use.

To us, what we think is worthwhile reviewing is: 1. Checking for any obvious data errors for any of the headline numbers (e.g. the main figure/trend on each slide). 2. Discussing the inferences we’re drawing out. Obviously we’d appreciate calling out if any of the inferences we’ve made are wrong/poorly interpreted, but it would also be really valuable to confirm/discuss the main ideas across the report. a. For the early intervention slides, this is the 5 points and summary provided on Slide 2 b. For the cost drivers slide, it’s the specific list of questions that we sent through in the first email

                         Page 2 of 53

While more thorough checking of the figures and commentary outside the main messages are helpful for us, they should be weighed up against the fact that we’re using these slides to generate discussion, and that what we’ll be taking forward is primarily the inferences we’re drawing from them rather than the slides themselves.

Happy to provide anything else that’s helpful. I’m also aware we’re yet to send through the underlying resources for the cost drivers work, which we’ll send over once we’ve organised it.

@Our team — do correct me if I’ve mischaracterised any of this.

Regards, Gary

From: redacted: s22(1)(a)(ii) - irrelevant material Sent: Wednesday, 31 May 2023 11:59 AM Cc: redacted: s22(1)(a)(ii) - irrelevant material

Subject: RE: CAUTION: Email may contain unverified link [be careful if proceeding]RE: Presentations for review [SEC=OFFICIAL:Sensitive]

Thanks very much Gary.

From: redacted: s22(1)(a)(ii) - irrelevant material

Sent: Wednesday, May 31, 2023 redacted: s22(1)(a)(ii) - irrelevant material

To: redacted: s22(1)(a)(ii) - irrelevant material

Subject: CAUTION: Email may contain unverified link [be careful if proceeding]RE: Presentations for review [SEC=OFFICIAL:Sensitive]

OFFICIAL: Sensitive Hi Meng. Attached are scripts and worksheets for the early intervention slides:

Early intervention worksheet (QA sheet).xlsx — workbook which contains the raw data and charts used for the early intervention slides. Chart data should be correctly named to the slide, and the raw data to the chart data. I haven’t kept intermediate data manipulation steps from the raw data to the chart data; these are usually either direct copying or taking summaries using pivot tables.
B4-early intervention analysis QA.R — R script which produces the raw data. Script is mostly but not completely documented. It loads datasets produced by ‘B1 early intervention data prep.R’
B1-early intervention data prep.R — dataset which is preparing the datasets and

Page 3 of 53

applying the filters used for this analysis. It draws from pre-processed datasets which we’ve coded up as the basis for our general use. Uses datasets produced by the following three pre-process scripts 01 Processing payments dataset.R – produces our preprocessed payments dataset. Loads and saves the payments dataset in a more efficient format, renaming columns and formatting date objects 02 Processing other datasets.R – produces our preprocessed participant demographics and plans dataset. Collects and formats columns/column names which we find useful 03 Effective plans method.R – script used to proportionally distribute plans across their effective time period

The first two are the only two which are directly relevant – the remaining 4 R scripts are all data cleaning. The last 3 especially reflect general scripts that we are using on our end for processing. I’m sending all of them for completeness, but would be surprised if you were reviewing back through all of the preprocessing. The first two I’ve tried to document so it should be relatively clear what’s going where, but the other ones I’ve not documented beyond what I ordinarily do. Consequently, do reach out if there is anything not documented or explained. The time we have to produce these scripts is limited, so excuse the fact things aren’t as edited, clear and standardised as best practice would be.

Regards, Gary

From: redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au> Sent: Tuesday, 30 May 2023 1:51 PM To: redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au> Cc: redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au> Subject: RE: Presentations for review [SEC=OFFICIAL]

Hi Gary,

Yes please, any codes/scripts and working spreadsheets.

Thanks,

Meng

From: redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au> Sent: Tuesday, May 30, 2023 1:28 PM To: redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au> Cc: redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au>

                         Page 4 of 53

Subject: RE: Presentations for review [SEC=OFFICIAL]

                    OFFICIAL

Thanks Meng. By underlying workings, do you mean scripts, underlying spreadsheets or something else?

Gary

From: redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au> Sent: Tuesday, 30 May 2023 1:09 PM To: redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au> Cc: redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au> Subject: RE: Presentations for review [SEC=OFFICIAL]

Thanks Gary.

We will start reviewing these. If you could please share the underlying workings behind the presentations, that would be helpful.

Thanks,

Meng

From: redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au> Sent: Tuesday, May 30, 2023 11:01 AM To: redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au> Cc: redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au>; redacted: s22(1)(a)(ii) - irrelevant material <redacted: s22(1)(a)(ii) - irrelevant material@ndis.gov.au> Subject: Presentations for review [SEC=OFFICIAL]

                    OFFICIAL

Hello Kim and Meng,

Attached are two presentations, which are preliminary versions of two projects listed on our key analysis projects dashboard. We would appreciate if you could review these, as a check on both the correctness of the results and the inferences that we draw out. For the moment, the purpose of this analysis is for discussion purposes. Down the line some of this analysis may be used in the final report, but at that point we’ll come back with the specific slides/analysis that will be published.

Attached are: 230525 Early intervention presentation.pdf – This is the extended version of what we presented to Bruce and NDIA on 25 May, and the corresponding output for #1 on our key

                         Page 5 of 53

analytical projects 230523 Early intervention presentation – Method slides.pptx – Accompanying slides outlining some method points for the early intervention presentation. Refer to this on the specific things we wanted reviewed. 230522 Cost drivers – Key takeaways.pdf – Main slides from the cost drivers work. Parts of this was presented to Bruce on 16 May. This is the corresponding output for #2 on our key analytical projects Key takeaways and specific data points/inferences we wanted reviewed are provided at the bottom of this email. 230530 Draft Key Analytical Projects.pptx – attached for reference; the analytic projects in our timeline

Apologies for the delay in sending this through. Ideally we have this back by Friday 9 June for us to work anything we need into the next Panel meeting – but happy to discuss timeframes and best way to proceed though.

Regards, Gary

Key takeaways and points of review for Cost drivers work

Please find the following attached analysis of cost drivers. These are the following key takeaways:

1. Participant complexity among new entrants is declining based on occurrence of co-
  occurring conditions
2. Participants may be using NDIS funds to access mainstream supports as evidenced by
  patterns in use of psychology for participants with co-occurring conditions
3. Over-servicing is on the increase, and is at least partially driven by providers (particularly
  unregistered providers) and support coordinators

We would particularly appreciate any checks/views the NDIS has on: Decrease in co-occurring conditions (slide 4 to 5): Is this data that is typically collected by the NDIS or submitted by participants at access? Does the agency have a view on how much of the decline is associated with record keeping? Accessing mainstream supports (slides 6 to 8): To what extent do co-occurring conditions affect planning decisions? (e.g. does the NDIS explicitly fund supports for co-occurring anxiety/depression, and if so does it require a threshold/ diagnostic assessment for co-occurring conditions to affect planning) Over-servicing analysis: Any commentary on the definition of over-servicing? And if there are issues with the term, would excess spend be preferable? Noting, we avoid over-utilisation or intra-plan inflation as this measure relates to spend rather than plan values (slide 17)

                         Page 6 of 53

The data implies that only ~350,000 ended in 2022 – is this consistent with expectations/longer typical plans on the increase (slide 15) Does the NDIA have any existing hypotheses on what drives over-servicing/ over- utilisation?

redacted: s22(1)(a)(ii) - irrelevant material (he/him)

Advisor | NDIS Review Secretariat Department of the Prime Minister and Cabinet Phone: redacted: s22(1)(a)(ii) - irrelevant material

Ngunnawal Country, One National Circuit, Barton ACT 2600 | PO Box 6500 CANBERRA ACT 2600 e. redacted: s22(1)(a)(ii) - irrelevant material@pmc.gov.au w. pmc.gov.au

         The Department acknowledges and pays respect to the past, present and emerging Elders and
             Traditional Custodians of Country, and the continuation of cultural, spiritual and educational
            practices of Aboriginal and Torres Strait Islander peoples.

IMPORTANT: This message, and any attachments to it, contains information that is confidential and may also be the subject of legal professional or other privilege. If you are not the intended recipient of this message, you must not review, copy, disseminate or disclose its contents to any other party or take action in reliance of any material contained within it. If you have received this message in error, please notify the sender immediately by return email informing them of the mistake and delete all copies of the message from your computer system.

IMPORTANT: This e-mail is for the use of the intended recipient only and may contain information that is confidential, commercially valuable and/or subject to legal or parliamentary privilege. If you are not the intended recipient you are notified that any review, re-transmission, disclosure, dissemination or other use of, or taking of any action in reliance upon, this information is prohibited and may result in severe penalties. If you have received this e-mail in error please notify the sender immediately and delete all electronic and hard copies of this transmission together with any attachments. Please consider the environment before printing this e-mail

FOI 24/25-2110 DOCUMENT 1.1

Early Intervention

Definitions

Participants who entered the Scheme through an existing Commonwealth program (RprtngAcsEntryTyp = Commonwealth) should also be classified as Legacy State (assuming this is to identify participants who were receiving disability supports prior to NDIS).
Legacy State (or Commonwealth) participants should not be excluded from the early intervention definition. Transferring from an existing program does not mean that early intervention is not relevant once they are in the NDIS (particularly for children).
Early intervention participants should be identified using the ‘Access Request Decision Reason’ variable in the access requests dataset (AcsRqstDcsnRsn = Benefit from Early Intervention), rather than using the ECEI gateway dataset. Some children do not enter the Scheme through the gateway, either because they live in an area without an Early Childhood partner, or they’ve applied directly to the Scheme.
The language of ‘autistic participants’ or ‘autism participants’ may be offensive to the disability community. The NDIA’s preferred language is participants with autism.

Exit rate (Slide 3)

The exit rates shown are not comparable across joining years, as these are unadjusted for exposure. The lower proportions from the more recent joining years reflect the shorter period of time participants have been in the Scheme, rather than a trend of the rate of exit.
Given demographic information is based on current data (March 2023?), the early intervention cohort is missing those who entered via early intervention but later received a permanent disability diagnosis who would now be identified as participants with an access request decision reason of permanent disability in the current data.

Entry type (Slide 4)

Given disability type is based on current data (March 2023?), the chart includes participants who entered the Scheme under Developmental Delay (DD) and then changed disability type to Autism to meet ongoing eligibility requirements. However, for recent years, the number of new entrants will be materially understated by the number of participants who have yet to convert from DD to Autism. The implied trend in this chart of reducing number of new entrants with Autism is misleading as numbers are increasing over time.
The entry type reflects the current status (early intervention versus permanent disability), rather than the entry type upon entering the Scheme.

Line item spend (Slide 5)

The methodology appears to use plan data combined with an average utilisation percentage to arrive at total spend. The preferred approach is to use the payments data directly.
The line items “Assessment Recommendation Therapy And/or Training” cover the provision of therapeutic supports by Allied Health Practitioners to participants aged 7 or older. This includes delivery of therapy, assessment of need for therapeutic supports, and assessments for AT, Home Mods
```
                         Page 9 of 53
```

and Housing. These items cannot be separately identified. The same applies to line items “Capacity Building Supports for Early Childhood Interventions” for children aged 6 and under.

Average funding by age (Slide 6)

While the results appear reasonable, the methodology for calculating average funding needs further verification. The preferred approach is to use annualised average funding based on the participant’s latest plan.

Growth in spend (Slide 7)

The first year in each chart is an incomplete year in terms of capturing payments, as participants can join throughout the year. An alternative is to look at the average annualised funding over time.

Cost Drivers

General

The ‘Cost Drivers’ presentation appears to contain three separate topics of investigation – access (for those with secondary disabilities), ‘over-servicing’ and support co-ordinators, rather than a complete picture of the cost drivers of the Scheme. There is no mention of other significant cost drivers, such as high numbers of children joining the Scheme, sustained plan and payment inflation, and increased demand for Supported Independent Living supports.

Access

The information collected by the NDIA on a participant’s secondary disabilities and/or conditions is limited. At 31 March 2023, secondary disability information was recorded for 31% of active participants. Of participants with any secondary disability recorded, the majority (67%) have only one condition listed. File reviews on participant samples indicate there is likely to be a significant under- reporting of secondary disabilities in the data. As such, inferences should not be drawn given the lack of confidence in the data.
The trend showing a declining rate of participants with ‘co-occuring’ conditions may reflect the phasing pattern of the Scheme, with those transitioning from existing State/Territory or Commonwealth programs having more severe or complex conditions than more recent entrants (many of whom are children).
The presentation focuses on mental health conditions and psychology supports only, with no mention of children, ageing participants or participants with chronic health conditions. Consideration should also be given to intersection with other mainstream services, including education, health, housing, aged care and the justice system.

Over-servicing and Support coordinators

The logic and calculation of ‘over-servicing’ makes sense, noting however the below considerations: a. ‘Over-servicing’ is not a complete measure of inflation as a cost driver. It does not give a quantum of growth (like plan or payment inflation does), and it does not take into account the growth in funding from plan to plan, focusing on the pattern of utilisation only.
```
                        Page 10 of 53
```

b. The underlying assumption that supports are used uniformly throughout the plan is not always true. For example, capital supports are one-off and lumpy by nature. It’s unclear how this is allowed for in the calculation. c. There is a lag between when supports are provided and when payments for those supports are made. For example, payments can still occur in 2023 in relation to plans that ended in 2022. Any measure of utilisation should allow for an appropriate lag for payments to fully develop. d. The language of ‘over-servicing’ implies that over utilisation is provider driven. While this may be partly true, there may be other reasons including higher support needs required due to decrease in functional capacity or decrease in informal supports, plans not sufficient to provide required supports for the participant, or funding appears to be used in line with support needs participant chooses to access. e. While ‘over-servicing’ focuses on over utilisation of plans that have been shortened, the concept is similar to ‘intra-plan inflation’. The latter is a concept that is published in the Quarterly Report to Disability Ministers. There may be risks and confusion involved in introducing new terms that are similar to published concepts. 18. The attribution of ‘over-servicing’ to unregistered providers and support coordinators needs further verification. Without details of the underlying workings and data sources used, our ability to verify the inferences drawn is limited.

                            Page 11 of 53

FOI 24/25-2110 DOCUMENT 1.2

NDIA Feedback on the NDIS Review’s ‘Early Intervention Presentation’ report.

Current State Review

The Early Intervention Current State review found that 47,320 participants (or 30% of s25 participants) have a diagnosis of autism. This is across all age cohorts, not just aged 14 and under.
The current state review does not touch on the number of new s25 participants with autism receiving access each year.
Whilst the current state review doesn’t touch on exit rates quantitatively, it found that operational guidance for the NDIA does not articulate the process for transitioning an early intervention participant back into mainstream and community settings. Further, the Applied Principles and Tables of Support (APTOS) does not cover the criteria required to support participants to exit the Scheme.
Average annualised committed support funding capacity building for s25 participants is very close to that of s24 participants ($17,600 compared to $17,200), however core support funding is significantly lower for s25 participants ($9,100 compared to $69,100). This was not broken down further by age or autism diagnosis in the review.
For s25 participants with autism, average annualised committed supports peaked in the 19-24 age cohort, at $44,300 (this includes both core and capacity building support funding).

Issues with eligibility reassessment (ER) as a contributing factor

Suggest consideration be given to the difficulty NDIA experiences in implementing eligibility reassessment as a contributing factor to the volume of children continuing within the Scheme. a. There is no power to require (compel) information for the purpose of eligibility reassessment, that would initiate an administrative consequence where the participant or child representative refuses to comply with the request (and after reasonable support to comply was provided). b. Operational experience has shown that information for ER is provided reluctantly, and sometimes not at all. While new operational guidance has worked to clarify the information needed, participant/child representative refusal to provide information, particularly in instances where revocation is likely to be explored, provides a barrier to the evidence needed for any section 30 revocation action to be taken.

How to ensure funding spent on evidence based Early Intervention?

Insights into current use of funds: Suggest an analysis using ABN data of the top providers for key cohorts in Early Intervention (i.e. children with autism, under 7 and 7-14) would complement this pack. a. Particularly to highlight the extent of funds being spent at businesses /suppliers such as Kogan, K-mart, Woolworths and Amazon. b. Note access to this data to be confirmed with ADA, we understand that ABN information is not routinely captured but can be obtained through the NDIS app, and that this analysis has been previously conducted.

Rise in core as children in EI cohort develop, with concurrent Capacity Building funding

Another relevant factor in the increase in core over time while Capacity Building remains steady is likely to be the interaction with the planning construct (line by line negotiation) and the interaction between what is reasonable parental responsibility in provision of core supports.
```
                         Page 12 of 53
```

Core supports may be provided when the child is young, where these are R&N. A potential contributor to the rise in core supports as the child ages is that consideration of what is reasonable for a parent to provide shifts over time. Providing less core supports when the child is young/or attending school may be a more defensible position for the Agency.

Impact of Access Lists on EI and continued eligibility / participation within the Scheme

Where a participant has accessed the scheme through List A or particularly for the EI cohort List D, they have not been required to provide information about their functioning. The detail and reliability of functional information in these circumstances can be variable. This information is critical for understanding if early intervention is working, and to understand the person’s support needs in planning.
Early intervention is not expected to change a permanent impairment, rather to moderate the impact and / or improve functional capacity (that is, influence the persons capacity and social and environmental context). As such, information about functioning and outcomes is critical.
In relation to children continuing within the Scheme, the Lists may have further impact. A diagnosis of one of the health conditions on List A will in most cases be a reliable proxy for an applicant meeting the Disability Requirements. However, some health conditions result in different levels of impairment for different people, meaning that a diagnosis may not be a reliable proxy for some legislative criteria, in particular substantially reduced functional capacity.
Autism is included on List A with the caveat that it be assessed as level two or three using the current Diagnostic and Statistical Manual of Mental Disorders (DSM-V)1 by a specialist multi-disciplinary team, paediatrician, psychiatrist or clinical psychologist experienced in the assessment of Pervasive Developmental Disorders.
The NDIA is concerned that Autism level two on the DSM-V is not a reliable proxy for substantially reduced function. Further, there is evidence that its inclusion on List A has influenced the behaviour of treating professionals and driven an increase in level two diagnoses.
During the independent assessment pilot, ten of the 79 pilot participants with a level two or three diagnosis scored within assumed normal levels of functioning on two assessment tools (Vineland and PEDI-CAT). This suggests not all participants currently being diagnosed with level two or three autism have substantially reduced functional capacity.

It is also worth noting that the NDIA has also completed a piece of work which sought to understand current practice around early intervention and how this could be applied across age and disability cohorts, and that a lot of colour can be found within that current state snapshot (CSS) report. This is particularly true for issues around escalation, assessment, planning, reviews and ‘off-boarding’.

1 https://ourguidelines.ndis.gov.au/home/becoming-participant/applying-ndis/list-conditions-are-likely-meet- disability-requirements

                            Page 13 of 53

NDIA Feedback on the NDIS Review’s ‘Cost Drivers’ report

Recording disabilities on a participant’s CRM record

When an applicant accesses the NDIS, access delegates are required to assign a ‘primary disability.’ a. If a participant meets access based on a diagnosis on an Access List, that condition is recorded as the participant’s primary disability. b. For all other participants, the Access Request Form (ARF) requests treating health professionals designate a primary disability, being the disability “that has the greatest impacts most on your daily life”2
As well as primary disability, the ARF asks applicants ‘Do you have any other disabilities that substantially affect your everyday life?’ Access delegates are required to assess disabilities against the NDIS access criteria, and add up to four ‘secondary disabilities.’
While the National Access and Reassessments Branch has rigorous processes in place for adding secondary disabilities to a participant’s CRM record, other NDIA business areas (i.e. planners) are able to add or change participants’ disabilities, and there is anecdotal evidence of disabilities being added to a participant’s record with lower evidentiary standards.

Changes in the data over time

The NDIS Review notes that there has been a decrease in additional disabilities being recorded over time. There are several hypotheses for this trend: a. Changes in NDIS cohorts over time – as observed by the Review, during the Scheme launch and transition years NDIS entry was focussed on those with higher needs, generally these were people who were transitioned from existing state schemes. During this transition delegates may have simply copied the disabilities that were listed in the records of those from State and Territory schemes into the NDIS system. b. Changes in agency processes over time – It is likely that over time access delegates have become better at assessing and then only listing conditions which met the access criteria.
CRM records are not intended to provide a full summary of a participant’s disability and health circumstances. Listing conditions on a participant’s CRM record is primarily for data collection purposes and to record the disabilities for which a participant may receive funded supports.
We expect that CRM data significantly underestimates the rate of co-occurring health and mental health conditions. For instance, it has been observed in cases at the Administrative Appeals Tribunal that many participants request supports in relation to conditions and impairments which are not listed on their CRM records.

The concepts of ‘additional disabilities’ and ‘co-morbid health and mental health conditions’

The NDIS Review’s slide pack describes ‘secondary disabilities’ as ‘co-occurring conditions,’ and the analysis implies that often co-occurring conditions require supports which could best be described as treatments and which are the responsibility of mainstream services.
This analysis has correctly identified a dynamic which contributes to cost-shifting to the NDIS. However, there are nuances to be aware of:

2 The Providing evidence of your disability, Operational Guideline.

                            Page 14 of 53

a. There are conceptual challenges with distinguishing between ‘disabilities’ and ‘co-morbid health and mental health conditions.’ In general: i. Disability arises from permanent impairments; ii. Comorbid health and mental health conditions are impairments which are in their nature responsive to and most appropriately addressed by treatment – and therefore are not ‘permanent’ according to NDIS legislation. b. Many of the conditions listed as secondary disabilities on participants’ records will have been assessed as permanent, and may be appropriately considered to be part of the participant’s disability. 9. In addition, many conditions listed as secondary disabilities are in fact indistinguishable from the primary disability in a clinical sense. For example, its noted that anxiety/depressive disorders are a common secondary disability. However, where these co-occur with a primary psychosocial disability, the extent to which they can be regarded as separate and distinguishable in nature is questionable. For example, treatment for the mental health condition and disability support for the psychosocial disability may in fact take place through the same intervention, i.e. a session with a psychologist.

Implications of research

The co-occurring conditions section of the pack focusses on the implications of the distinction between disability and mental health by focussing on the provision of psychology supports to NDIS participants with co-occurring conditions such as depression and anxiety. This focus is appropriate, though there are broader implications.
There is a large cohort of participants who experience a range of chronic health conditions (e.g., backpain, obesity, arthritis, diabetes, COPD) which are highly prevalent in Australian society, and not most appropriately addressed by specialist disability supports through the NDIS.
Disparities in the service footprints of the NDIS and mainstream service systems incentivises participants and their health professionals to define health and mental health conditions in disability terms to have them recognised as part of the participant’s disability in the NDIS.
Other suggested areas of focus for this kind of analysis include: a. The broad range of chronic health conditions in the NDIS (e.g., backpain, obesity, arthritis, diabetes, COPD) b. People applying to the NDIS with life-limiting terminal conditions, where their support needs are most appropriately addressed through the palliative care system
```
                        Page 15 of 53
```

FOI 24/25-2110 OFFICIAL DOCUMENT 2

Scheme cost drivers — Access, over-servicing and support coordanators analysis

Page 16 of 53 NDIS Review - OFFICIAL 30 May, 2023 | 1

Summary of Key findings:

New entrants into the Scheme appear to have less complex needs than existing participants
- Co-occurring conditions have been on the decline since 2017 – from 50% to 12% in 2022
- The decline in co-occurring conditions may reflect declining complexity of disabilities or limitations in collections of data
- Mental health conditions have become an increasing share of co-occurring conditions where they are recorded
Data on co-occurring conditions suggests that NDIS there is potential to use NDIS for services that overlap with mainstream supports
- Adults with anxieties/depressive disorders are 2x more likely to use a psychologist than other participants, and spend 3x as much on psychologists
- Overall, psychology expenditure among adults makes up $170 million in supports, but it is one example supports that could be funded by mainstream systems (e.g. medicare) as well as NDIS. A number of other therapies accessible through the NDIS may also overlap with mainstream services.
Over-servicing has become an increasing share of Scheme expenditure and is at least partially driven by providers and support coordinators
- Over-servicing has increased from 9% in 2017 to 16% of annualised expenditure by 2022
- An unusual number of providers have elevated rates of over-servicing for their participants, particularly unregistered providers
- Participants using support coordinators are much more likely to experience over-servicing, particularly unregistered support coordinators
```
                                                    Page 17 of 53
    NDIS Review | [Presentation Title]                                                                         30 May, 2023    | 2
```

FOI 24/25-2110

Analysis section 1. Access requests and mainstream supports

Page 18 of 53

OFFICIAL

NDIS Review

The overall rate of co-occurrence among Scheme entrants has declined from above 50% in 2017 to 12% by 2022

The rate of co-occurrence has declined sharply since 2018

Total number of entrants (#) vs. rate of entrants with co-occurring conditions (%), 2013-2022

Since the Scheme trial begun, the rate of participants with co-occurring conditions has been declining. This is likely consistent with expectation, as Scheme entry in the initial years was more focussed on higher needs participants, who are likely to present with more complex disabilities (with the exception of SA where the initial NDIS trial was focussed on children).

This decline has largely coincided with the decline in participant entry as the Scheme approaches maturity, though in 2022 entry trends have reversed, with a slight uptick over 2021. This suggests that in 2022 some share of applicants with lower needs than past entrants have been granted access to the Scheme.

However, it is also likely that record-keeping/accounting of co-occurring conditions is a factor for this trend. Further information from the NDIA will be required to what extent the decline represents change in record keeping vs. real change

OFFICIAL

Page 19 of 53 NDIS Review | Cost Drivers Analysis — Access distortions 30 May, 2023 | 4

OFFICIAL NDIS Review

Mental health conditions are an increasing share of co-occurring conditions

% of participants with a mental health conditions as share of all participants with co-occurring conditions, 2013-2022

The chart on the left examines the incidence of mental health conditions for all participants with reported co-occurring conditions.

Mental health conditions are making up an increasing share of participants, in particular anxiety disorders and Depression/Bipolar Affective Disorder.

OFFICIAL

Page 20 of 53 NDIS Review | Cost Drivers Analysis — Access distortions 30 May, 2023 | 5

OFFICIAL

Participants with anxiety/depression are twice as likely as others to use the NDIS to access psychologists and spend 3x more

More than 1 in 3 adults in the Scheme with anxiety and depression use a psychologist, compared to less than 1 in 5 for other participants

% using psychology supports, 2022	Avg. $ spent on psychology supports, 2022	Total $ spent on psychology supports, 2022
Do not have Anxiety and or Depression: 18%	Do not have Anxiety and or Depression: $452	Do not have Anxiety and or Depression: $131 Million
Have Anxiety and or Depression: 37%	Have Anxiety and or Depression: $1,183	Have Anxiety and or Depression: $39 Million

OFFICIAL Source: NDIA Unit record data, participants over the age of 15 with expenditure in 2022

Page 21 of 53 NDIS Review | Cost Drivers Analysis – Access distortions 30 May, 2023 | 6

OFFICIAL

NDIS Review

Both participants and mainstream service providers have incentives to seek NDIS access for participants to use mainstream services

Participants / Providers	Incentive
Participants seeking clinical psychology services	NDIS provides more generous funding (sessions fully covered, capped by budget not quantity p.a.)
Mainstream Health networks	Opportunity to reduce load on system by redirecting participants towards NDIS
Private clinical psychologists	Opportunity to grow demand for services and provide clients an additional avenue to pay.

Important to note that these participants are likely to have a genuine need for these services, which at times may have a direct relationship to their primary disability.

However, the likely use of NDIS services to pay for psychologists to treat depression or anxiety raises clear questions regarding equity. Many Australians without access to the NDIS likely suffer from anxiety/depression and do not benefit from the same ease of access to psychologists as NDIS participants may have.

OFFICIAL

Page 22 of 53 NDIS Review | Cost Drivers Analysis — Access distortions 30 May, 2023 | 7

OFFICIAL

NDIS Review

The impact of anxiety/depression on psychology use is likely to be underestimated due to imperfect data records

Source of bias	Impact	Impact
Under-recording of anxiety and depression among participants	Based on relatively low prevalence rate within NDIS participant population, anxiety or depression is likely under-recorded in the “control” group – this suggests that a portion of participants in this group may be using psychologists to treat anxiety/depression	Typical use of, and spend on psychology by control group is elevated by presence of participants with anxiety /depression
Infrequent updates of co-occurring conditions for participants with anxiety/depression	Infrequent updates of co-occurring conditions for participants with anxiety/depression suggests that in this group, anxiety/depression may be over-recorded – meaning some of these participants spend less on psychologists than they would if they were still impacted by these conditions	Typical use of, and spend on psychology by control group is dragged down by presence of participants who no longer experience anxiety /depression within this group

Both of these effects are likely occurring in the data, leading to underestimation of the impact of the use of NDIS supports to fund mainstream needs

OFFICIAL

Page 23 of 53 NDIS Review | Cost Drivers Analysis — Access distortions 30 May, 2023 | 8

FOI 24/25-2110

Analysis section 2. Over-servicing

Page 24 of 53

FOI 24/25-2110 NDIS Review

To analyse over-servicing, plan expenditure is divided into over-servicing and regular servicing (for plans that have ended) in 3 steps

Calculate modified utilisation $$\text{Modified utilisation} = \frac{\text{Total plan spend}}{\text{Total plan values}} \times \left( \frac{\text{Original plan duration}}{\text{Most recent plan duration}} \right)$$
Calculate regular servicing and over-servicing
- Regular servicing: The proportion of spending that is within the pro-rated annualised budget (based on original plan duration):
  - Minimum of:
    1. Annualised budget OR
    2. Annualised budget $\times$ modified utilisation
- Over-servicing: The proportion of spending that is greater than the pro-rated annualised budget (based on original plan duration):
  - Maximum of:
    1. 0 OR
    2. Annualised budget $\times$ (modified utilisation — 1)
Calculate over-servicing share $$\text{Over-servicing share} = \frac{\text{Over-servicing}}{\text{Regular servicing} + \text{Over-servicing}}$$

Page 25 of 53 NDIS Review | Cost Drivers Analysis — Over-servicing 30 May, 2023 | 10

OFFICIAL

Over-servicing has been increasing since July 2019, from 9% to 16% of spend as at February 2023

What is over-servicing?

Over-servicing represents spend that is above the budgeted amount a participant receives in their plan. For example, if a participant receives a $20K budget over 12 months, and spends the entirety in 3 months, then 75% of their expenditure would be over-servicing ($5,000 of expenditure is within budget, and $15,000 is above budget). Only plans that have been shortened can are assumed to have potential over-servicing.

Some over-spending can reflect inadequate plan values

For the purpose of this analysis it has been assumed that plan values reflect limits on what reasonable and necessary supports are, and that delivery of supports beyond budgets represents over-servicing – some share of over-servicing will thus represent inadequate plan budgets, however some share of over-servicing is likely driven by excessive provider charging (though providers who do not know participants’ budgets may be doing so unintentionally).

Over-servicing has substantially increased in 2022

Prior to 2022 over-spending has generally been in decline, with an uptick coinciding with the COVID-19 pandemic (potentially due to additional NDIS support funding associated with COVID-19). However the beginning of 2022 over-servicing has increased rapidly. This is consistent with NDIA published reporting of higher intra-plan inflation in the June and September 2022 quarters (intra-plan inflation is a similar measure to over-servicing as used in the report).

Page 26 of 53 NDIS Review | Cost Drivers Analysis – Over-servicing 30 May, 2023 | 11

Of the 5,000 largest providers delivering services to participants in 2022:

The 100 most over-servicing providers derived nearly 60% of revenue from over-servicing (57%).
The top 500 most over-servicing providers derived at least 33% of their revenue from over-servicing.

The distribution suggests that providers are in part be driving over-servicing – if over-servicing was explained purely by under-funding, we would expect the distribution to be flatter (See appendix A).

This is caveated by the fact that providers may typically not know participants’ budgets and may be engaging in over-servicing unintentionally.

Page 27 of 53 NDIS Review | Cost Drivers Analysis – Over-servicing 30 May, 2023 | 12

FOI 24/25-2110 NDIS Review

Unregistered providers are more likely to over-service participants compared to registered providers

Of the 5,000 largest providers delivering services to participants in 2022, 4,000 were registered and 1,000 were unregistered:

Unregistered providers are consistently deriving a greater share of revenue from over-servicing compared to registered providers.

Note: Registered providers make up most service delivery in this sample, as unregistered providers are typically smaller.

Unregistered providers make up 10% of total over-servicing, and approximately 7% of regular servicing.

Page 28 of 53 NDIS Review | Cost Drivers Analysis — Over-servicing 30 May, 2023 | 13

FOI 24/25-2110

Analysis section 3. Support co-ordinators

Page 29 of 53

Most participants in the Scheme use registered support coordinators

About 10x as many participants use registered support coordinators as opposed to unregistered support coordinators.

Number of ended plans by type of support coordinator used, 2022:

No support coordinator: 168,419
Unregistered Support Coordinator: 11,807
Registered support coordinator: 157,967

NOTE: Due to the NDIS shifting to longer plans, the number of ended plans in 2022 is lower than the number of expected ended plans if all plans were annual.

Further, this data only includes participants were the actual support coordinator could be identified, meaning all self-managed participants are excluded.

Page 30 of 53 NDIS Review | Cost Drivers Analysis – Support coordinators 30 May, 2023 | 15

NDIS Review

Unregistered support coordinators are more likely to have over-serviced participants than registered support coordinators

Participants using a support coordinator are consistently more likely to have an early plan review (these are highly correlated to over-servicing) and consistently receive a greater share of expenditure from over-servicing.

In particular, participants using unregistered support-coordinators have higher rates of over-servicing and higher rates of unscheduled plan reviews.

Page 31 of 53 NDIS Review | Cost Drivers Analysis — Support coordinators 30 May, 2023 | 16

Participants using unregistered & conflicted support coordinators are more likely to experience over-servicing

Participants using registered support coordinators for other services are not more likely to have over-serviced plans.
Participants using unregistered support coordinators for other services display greater levels of over-servicing.

Note: plan-management has been excluded from the data, such that support-coordinators delivering support coordination and plan management are not considered conflicted.

Page 32 of 53 NDIS Review | Cost Drivers Analysis – Support coordinators 30 May, 2023 | 17

FOI 24/25-2110

Appendix A

Page 33 of 53

OFFICIAL NDIS Review

Across almost all major disability cohorts, the presence of anxiety/depression is associated with greater use of NDIS funds on psychology services

The impact of anxiety/depression on psychologist use is persistent when controlling for a large range of factors.

These cohorts capture approximately 2/3 of all participants with anxiety or depression listed as a co-occurring condition.
Across 201 of 207 cohorts within the major adult disability groups (97%) (psychosocial, Autism and Intellectual disability) use of psychologists is higher for participants with anxiety/depression.

Page 34 of 53 NDIS Review | Cost Drivers Analysis — Access distortions 30 May, 2023 | 19

OFFICIAL NDIS Review

Core accounts for 79% of Scheme over-servicing, with Capacity Building — Daily Activities accounting for a further 10%

Core currently accounts for 79% of Over-servicing, and Capacity Building Daily Activities accounts for 10% of over-servicing (Up from 5% in July 2018) — This distribution is only slightly different to the overall spend distribution (dotted lines).
Note: Capital over-servicing is likely an inflated estimate, as participants are expected to exhaust capital budgets early in their plan.

Page 35 of 53 NDIS Review | Cost Drivers Analysis — Over-servicing 30 May, 2023 | 20

OFFICIAL

Observed over-servicing across providers does not appear to be randomly distributed

Some providers receive systematically higher shares of revenue from over-servicing than expected.

At the most extreme end, actual over-servicing is slightly higher than the maximum result in simulations. This reflects that under natural variation it is expected that there will always be some outliers.
However, outside the most extreme band, the actual distribution is consistently “fatter tailed” than expected if over-servicing was randomly distributed across participants, but uncorrelated to providers. This suggests that some providers are systematically over-servicing/delivering supports that are not budgeted for.

Page 36 of 53 NDIS Review | Cost Drivers Analysis – Over-servicing 30 May, 2023 | 21

FOI 24/25-2110 DOCUMENT 3

Early intervention deep dive

Page 37 of 53 26 May, 2023 | 1

OFFICIAL — PRELIMINARY NDIS Review

Report summary

Purpose of this report

This report is the first of the key analytical projects, and the first deep dive conducted by the Review using unit record data. This provides a focus on early intervention and autism, allowing for analysis of multiple disaggregations of demographics, more granular data reporting and longitudinal trends.

Method notes

Results are all preliminary. Report focuses on data until the end of 2022 and excludes SIL participants from the analysis.

Key evidence points of this report

Exit rates from early intervention are low, with the majority transitioning to an autism diagnosis
Early intervention and List A have contributed to at least 25,000 new autistic participants every year since 2019
Line item analysis shows autistic children use significant funding for capacity building therapies including OT, psychology, speech and ‘other’
By age 20, effectively all autistic participants’ plans comprise of a majority of core funding, noting capacity building does not decrease significantly
Preliminary evidence suggests early intervention is not leading to low or declining future spend

“So what”

Given the above, there is a significant cost pressure from these participants aging in and not leaving the NDIS. This comes despite capacity building making up the majority of supports for autistic children, including over $500 million on therapies in 2022 (do we need a more stringent mechanism for funding capacity building supports?).

Page 38 of 53 26 May, 2023 | 2

OFFICIAL — PRELIMINARY NDIS Review

The exit rate for early intervention is around 10%, while the majority transition from delay to an autism diagnosis

Where early intervention participants have ended up: Count of early intervention participants by select disability cohorts and exit status, for annual new entrants from pre-2016 to 2022.

For participants joining after 2016, the exit rate of early intervention participants has been 10% or lower — noting that exit rates for cohorts will be increasing over time. This has been significantly higher than the rest of the NDIS — more than 3x above in the closest year — but means 90% of early intervention participants are remaining on the scheme.
Excluding participants who have exited or are still on a (non-global) delay diagnosis, most early intervention entries are leading to an autism diagnosis. This holds for each year except 2022 — which is still too early to tell.

Page 39 of 53 26 May, 2023 | 3

OFFICIAL — PRELIMINARY NDIS Review

Since 2019 a consistent number of autistic participants have entered as New participants through Section 24 or Early Intervention

How each of the 201,000 autistic participants on the NDIS entered: Count of autistic participants, by join year and entry type, as of December 2022.

Despite the association of autism with young children and young children with early intervention, the highest proportion (43%) of autistic participants entered as new participants in Section 24 — likely via List A.
While a material proportion of current autism participants arrived from legacy state schemes — 50,000 or 25% — the majority of this inflow was through years 2016-2019.
Since 2018, a steady stream of autism entrants have been “new” (i.e. non-legacy) participants through Section 24, and through early intervention.

Page 40 of 53 26 May, 2023 | 4

OFFICIAL — PRELIMINARY

NDIS Review

Line item spend shows therapies are the main supports used by autistic children

What supports autistic children are using

Top 20 line items by total spend for plan- and agency-managed autistic 0-14 children in 2022.1

Support type: Core Capacity building

Support type	Line item	Spend ($m)	% of total spend
Capacity Building	Assessment Recommendation Therapy And/or Training (Incl. AT) - Other Therapy	$169	16.4%
Capacity Building	Capacity Building Supports For Early Childhood Interventions - Other Therapy	$111	10.8%
Core	Assistance With Self-Care Activities - Standard - Weekday Daytime	$69	6.7%
Capacity Building	Plan Management - Financial Administration	$61	5.9%
Capacity Building	Assessment Recommendation Therapy And/or Training (Incl. AT) - Psychology	$58	5.7%
Capacity Building	Assessment Recommendation Therapy or Training - Occupational Therapist	$55	5.3%
Core	Access Community Social And Rec Activities - Weekday Daytime	$51	5.0%
Capacity Building	Assessment Recommendation Therapy And/or Training (Incl. AT) - Speech Pathologist	$42	4.1%
Core	Support Coordination Level 2: Coordination of Supports	$34	3.3%
Capacity Building	Specialist Behavioural Intervention Support	$34	3.3%
Capacity Building	Individual Skill Development And Training Including Public Transport Training	$19	1.8%
Core	Access Community Social And Rec Activities - Standard - Saturday	$15	1.4%
Capacity Building	Capacity Building Supports For Early Childhood - Allied Health Assistant Level 2	$14	1.4%
Capacity Building	STA And Assistance (Inc. Respite) - 1:1 - Weekday	$14	1.4%
Capacity Building	Therapy Assistant - Level 2	$13	1.3%
Capacity Building	Behaviour Management Plan Incl. Training In Behaviour Management Strategies	$12	1.2%
Capacity Building	Assessment Recommendation Therapy And/or Training (Incl. AT) - Physiotherapy	$12	1.1%
Core	Assistance With Self-Care Activities - Standard - Saturday	$12	1.1%
Capacity building	Plan Management And Financial Capacity Building - Set Up Costs	$11	1.1%
Core	Assistance With Self-Care Activities - Standard - Sunday	$10	1.0%
-	All other line items	$212	20.6%
	Total	$1,030	100%

Most spend for autistic children is towards therapy — either under “Assessment Recommendation Therapy” or “Capacity Building Supports For Early Childhood Interventions”. By comparison, otherwise common core supports in the NDIS of assistance with self-care and accessing community are relatively low in spend.

The largest individual line item is “Other Therapy”, reflecting the fact that line items until July 2022 did not break out key therapies such as occupational therapy or speech pathology.

Notes: 1. This data reports at the line item level — the most granular spending data available. The only aggregation is removing line items which differ by TTP i.e. temporary transition payments. Source: Panel estimates based on unpublished data supplied by the National Disability Insurance Agency

26 May, 2023 | 5

OFFICIAL — PRELIMINARY NDIS Review

Average funding for core supports begins rising at age 11, and by age 20 overtakes capacity building supports for most autistic participants

Shift in plan nature across ages for autistic participants

Average plan value for autistic participants across support types, by age, in 2022.

Support type: Core Capacity building

Separating plans by support type shows the drivers of plan value over time. While capacity building funding remains steady across all ages, core funding increases rapidly from around age 11.

This shift differs by the level of function, which follows given the higher needs of low function participants. For these participants, plans are core-dominant by age 12. This change still occurs but at a slower pace for moderate and high functioning participants who have a majority core plan by age 20.

Page 42 of 53 Source: Panel estimates based on unpublished data supplied by the National Disability Insurance Agency 26 May, 2023 | 6

OFFICIAL — PRELIMINARY NDIS Review

Across the most common autism cohorts, there is no evidence that early intervention has led to low or declining spend

How spend growth for early intervention participants compares

Average annualised spend for select autism cohorts, quarterly from various starting dates to end of 2022.1

Entry type: New Section 24 Legacy State Early intervention

To ensure like-for-like comparisons, each graph represents a cohort of autistic participants: participants with the same year of entry, age range at entry and level of functioning.

Across 6 of the most common cohorts there is no evidence early intervention participants show low or declining spend.

Each of the 0 to 6 cohorts show largely similar spend patterns. At 7+ there is more variation in spend, with those transitioning from legacy state schemes unsurprisingly the highest spend. However, where new Section 24 and 25 participants do differ, it is early intervention participants that show a higher spend growth. This pattern is the same across other cohorts tested.2

Notes: 1. Data shows quarterly spend multiplied by 4 for an annualised figure. Averages are calculated to include participants who have since exited, to remove any survivorship bias. Sample sizes for New, State and El participants respectively for each graph moving left to right are: 814, 956, 1950; 2937, 1540, 752; 2793, 839, 305; 1372, 820, 163; 2308, 1540, 1364; 2969, 221, 1619. 2. More cohorts have been analysed - these 6 cohorts are large cohorts which represent a spread across age and functional types. Source: Panel estimates based on unpublished data supplied by the National Disability Insurance Agency 26 May, 2023 | 7

| > ba ef tf tf Ff 7 7 3

FO} 24/25-2110

Appendix

Page 44 of 53

OFFICIAL — PRELIMINARY NDIS Review

In 2022 autism participants spent $4.3 billion while participants who entered under Section 25 (early intervention) spent $2.4 billion

Current state spending for participants by autism, delay and main entry types

Total (left) and average (right) spend by disability group in 2022. Non-SIL plans only.

Disability: Autism Developmental delay All other disabilities

Total (left) and average (right) spend by entry type in 2022.

Entry type: New Section 24 Legacy State Early intervention

Page 45 of 53

Source: Panel estimates based on unpublished data supplied by the National Disability Insurance Agency 26 May, 2023 | 9

OFFICIAL — PRELIMINARY NDIS Review

Out of 592,000 NDIS participants, 229,000 (39%) are new entering via Section 24, while 184,000 (31%) are new entering via Section 25, i.e. Early Intervention

How 592,000 participants entered the NDIS

Count of all participants, by entry type and join year, as of December 2022.1

Entry type: New Section 24 Legacy State Early intervention

The NDIS can be divided into three types of participants based on how they entered:

New participants entering through Section 24

Participants transitioning from legacy state systems

New participants entering through Section 25, i.e. early intervention

Data by join year shows the difference in timing for each entry type. ‘New’ participants account for the majority and peaked in 2019-2020. Legacy state participants was the majority of entering participants until 2018.

By contrast, early intervention has increased each year, and reached it’s highest in 2022.

Notes: 1. Includes participants who have since exited. 2. Participants who were marked as both ‘Legacy State’ and ‘Early Intervention’ (n = 11,000) were counted in the ‘Legacy State’ category. Source: Panel estimates based on unpublished data supplied by the National Disability Insurance Agency 26 May, 2023 | 10

OFFICIAL — PRELIMINARY

NDIS Review

Participants vary by entry type, including lower level of function from legacy state schemes and higher function levels from Early Intervention

How participants by different entry types vary

Count of active participants across entry types, by age and level of function, as of December 2022.

Entry type: New Section 24 Legacy state Early intervention

New Section 24	High func.	Mod func.	Low func.	Total
0 to 6	701	1,005	994	2,700
7 to 14	5,450	39,631	7,957	53,038
15 to 18	2,464	15,250	3,331	21,045
19 to 29	4,453	17,878	4,526	26,857
30 to 44	5,911	21,261	7,248	34,420
45+	12,039	39,819	27,266	79,124
Total	31,018	134,844	51,322	217,184

Legacy state	High func.	Mod func.	Low func.	Total
0 to 6	698	236	212	1,146
7 to 14	11,841	14,608	10,939	37,388
15 to 18	3,516	6,058	7,528	17,102
19 to 29	5,245	18,131	14,099	37,475
30 to 44	2,559	13,004	10,909	26,472
45+	4,953	20,145	20,637	45,735
Total	28,812	72,182	64,324	165,318

Early intervention	High func.	Mod func.	Low func.	Total
0 to 6	61,412	12,013	6,241	79,666
7 to 14	32,651	22,258	5,979	60,888
15 to 18	5,360	4,389	1,545	11,294
19 to 29	4,354	4,799	976	10,129
30 to 44	853	2,094	567	3,514
45+	1,080	4,310	1,867	7,257
Total	105,710	49,863	17,175	172,748

Across the three entry types, three distinct profiles of participants are evident — each in line with expectations.

Participants from legacy state schemes have the highest support needs, with 64,300 or 39% of the cohort listed with low level of function.

In comparison, new participants entering via Section 24 are highly concentrated at the moderate level of function with 134,800 or 62% of participants. The age profile appears bimodal, split between the largest cohort at the 45+ range and the second largest at 7 to 14.

Finally, early intervention is clearly focused around high functioning and 0 to 6 age participants, driven by early childhood early intervention gateways.

Notes: 1. Figures here show active participants as of December 2022, which differs from the previous slide which includes participants who have since exited the Scheme. Source: Panel estimates based on unpublished data supplied by the National Disability Insurance Agency

26 May, 2023 | 11

OFFICIAL — PRELIMINARY NDIS Review

Autistic participants are much younger and much more likely to have moderate level of function, compared to all other disabilities

How participants with autism and developmental delay vary

Count of active participants across disability group, by age and level of function, as of December 2022.

Disability: Autism Developmental delay All other disabilities

Autism	High func.	Mod func.	Low func.	Total
0 to 6	6,035	7,547	5,618	19,200
7 to 14	18,746	67,873	18,168	104,787
15 to 18	5,131	20,762	6,744	32,637
19 to 29	4,000	19,700	7,230	30,930
30 to 44	781	5,605	1,982	8,368
45+	268	1,940	664	2,872
Total	34,961	123,427	40,406	198,794

Developmental delay	High func.	Mod func.	Low func.	Total
0 to 6	40,743	3,428	656	44,827
7 to 14	8,363	578	181	9,122
15 to 18	17			26
19 to 29				10
30 to 44				3
45+				3
Total	49,129	4,020	842	53,991

All other disabilities	High func.	Mod func.	Low func.	Total
0 to 6	16,033	2,279	1,173	19,485
7 to 14	22,833	8,046	6,526	37,405
15 to 18	6,192	4,930	5,656	16,778
19 to 29	10,048	21,102	12,371	43,521
30 to 44	8,542	30,752	16,741	56,035
45+	17,802	62,333	49,106	129,241
Total	81,450	129,442	91,573	302,465

Three clearly different distributions of participants are evident across autism, developmental delay and all other disabilities.

Autism is highly concentrated in the 7 to 14 age range, moderate level of function cohort — comprising 67,900 or 34% of all autistic participants. The next highest age-severity groups are all either in the same 7 to 14 age range or with the same moderate level of function.

Conversely, the profile for all other disabilities is split much more towards older participants and a more even spread across moderate and low functioning. The two largest cohorts for all other disabilities is 62,300 and 49,100 for 45+ participants with moderate and low functioning — together comprising 37% of the total.

Notes: Some participants for developmental delay were removed due to counts below privacy threshold (i.e. cohorts of less than 10). Source: Panel estimates based on unpublished data supplied by the National Disability Insurance Agency 26 May, 2023 | 12

OFFICIAL — PRELIMINARY NDIS Review

Autistic participants from legacy schemes are slightly older with higher support needs, while El participants are younger with lower support needs

How participants from different entry types vary

Count of active autistic participants across entry type, by age and level of function, as of December 2022.

Entry type: New Section 24 Legacy state Early intervention

New Section 24	High func.	Mod func.	Low func.	Total
0 to 6	344	909	880	2,133
7 to 14	3,108	38,147	7,010	48,265
15 to 18	1,099	13,679	2,158	16,936
19 to 29	1,076	9,839	1,813	12,728
30 to 44	373	3,408	526	4,307
45+	129	1,347	236	1,712
Total	6,129	67,329	12,623	86,081

Legacy state	High func.	Mod func.	Low func.	Total
0 to 6	46	78	73	197
7 to 14	4,209	11,080	6,746	22,035
15 to 18	1,417	3,913	3,753	9,083
19 to 29	1,516	7,147	4,950	13,613
30 to 44	177	1,664	1,396	3,237
45+	54	421	389	864
Total	7,419	24,303	17,307	49,029

Early intervention	High func.	Mod func.	Low func.	Total
0 to 6	5,645	6,560	4,665	16,870
7 to 14	11,429	18,646	4,412	34,487
15 to 18	2,615	3,170	833	6,618
19 to 29	1,408	2,714	467	4,589
30 to 44	231	533	60	824
45+	85	172	39	296
Total	21,413	31,795	10,476	63,684

Overall, the profiles of autistic participants based on entry types are not too dissimilar. Each have a concentration in the 7 to 14 age band and moderate level of function profile.

As expected, those transitioning to the NDIS from legacy state systems are more likely to be lower functioning and older — although still mostly 29 or younger. New participants entering via Section 24 are overwhelmingly in the moderate level of function (78%) — suggestive of entry via an Autism 2 diagnosis and List A.

Those under 14 are the majority (81%) of early intervention autistic participants.

Page 49 of 53 Source: Panel estimates based on unpublished data supplied by the National Disability Insurance Agency 26 May, 2023 | 13

OFFICIAL — PRELIMINARY

NDIS Review

From 19,300 exits, the most common exit is early intervention delay at 7-14, followed by high and moderate functioning autistic participants

Profile of non-mortality exits from the NDIS

Count of non-mortality exits by disability, age, entry type and level of function, as of December 2022.

Disability	0-6	7 to 14	15 to 18	19 to 29	30 to 44	45+	Early int.	New	State	High func.	Mod func.	Low func.	Total
Autism	30	1,426	496	792	99	30	842	953	1,078	1,144	1,493	236	2,873
Developmental delay	552	8,263	73	21			6,955	239	1,715	8,456	368	85	8,909
Global delay	46	1,000	21				692	62	316	952	92	26	1,070
Hearing Impairment	50	142	33	77	31	159	236	193	63	423	63		492
Intellectual Dis.		448	143	539	207	263	312	420	872	823	603	178	1,604
Other Neurological		105	23	49	26	374	81	296	204	161	147	273	581
Other Physical		100	30	37	25	231	78	153	198	172	174	83	429
Other Sensory		1,127	183	59			827	138	408	1,215	129	29	1,373
Psychosocial		63	30	83	221	548	60	444	441	245	527	173	945
All other	12	306	47	93	90	473	194	352	475	528	268	225	1,021
Total	705	12,980	1,079	1,753	701	2,079	10,277	3,250	5,770	14,119	3,864	1,314	19,297

Source: Panel estimates based on unpublished data supplied by the National Disability Insurance Agency

26 May, 2023 | 14

OFFICIAL — PRELIMINARY NDIS Review

Similar with all other disabilities, there is a significant increase in supports from the age of 15

Comparison of plan value and spend for autism and all other disabilities in 2022

Average plan value and spend for autistic and all other disabilities, by age, in 2022.

Disability: Autism Developmental delay All other disabilities

For both autism and all other disabilities, there is a significant increase in both plan value and spend across the 11-21 age range. The sharpest increase is from the 15-19 age range.

For autism, this is a doubling of average plan value from $30,000 to $60,000 from 15 to 19. This is higher but comparable to the increase from $44,000 to $76,000 for all other participants — noting that it is the same magnitude of increase but off a higher base value.

Page 51 of 53 Source: Panel estimates based on unpublished data supplied by the National Disability Insurance Agency 26 May, 2023 | 15

OFFICIAL — PRELIMINARY NDIS Review

While 15-18 year old autistic participants are still using therapies, the majority of the top line items are core supports

What supports 15-18 year old autistic participants are using

Top 20 line items by total spend for plan- and agency-managed autistic 15-18 children in 2022.

Support type: Core Capacity building

Support class	Line item	Spend ($m)	% of total spend
Core	Assistance With Self-Care Activities - Standard - Weekday Daytime	$55	11.4%
Core	Access Community Social And Rec Activities - Weekday Daytime	$48	10.0%
Capacity Building	Assessment Recommendation Therapy And/or Training (Incl. AT) - Other Therapy	$35	7.3%
Capacity Building	Assessment Recommendation Therapy And/or Training (Incl. AT) - Psychology	$20	4.1%
Core	Support Coordination Level 2: Coordination of Supports	$19	4.0%
Capacity Building	Plan Management - Financial Administration	$18	3.7%
Capacity Building	Specialist Behavioural Intervention Support	$16	3.4%
Core	STA And Assistance (Inc. Respite) - 1:1 - Weekday	$16	3.3%
Capacity Building	Assessment Recommendation Therapy or Training - Occupational Therapist	$13	2.6%
Core	Access Community Social And Rec Activities - Standard - Saturday	$13	2.7%
Core	Assistance With Self-Care Activities - Standard - Saturday	$10	2.0%
Core	Assistance With Self-Care Activities - Standard - Sunday	$9	2.0%
Core	Assistance With Self-Care Activities - Level 2 - Weekday Daytime	$9	1.9%
Core	Access Community Social And Rec Activities - Standard - Sunday	$8	1.7%
Core	STA And Assistance (Inc. Respite) - 1:1 - Saturday	$7	1.5%
Core	Assistance With Self-Care Activities - Night-Time Sleepover	$7	1.4%
Capacity Building	Individual Skill Development And Training Including Public Transport Training	$7	1.4%
Capacity Building	Assessment Recommendation Therapy or Training - Speech Pathologist	$6	1.3%
Core	STA And Assistance (Inc. Respite) - 1:1 - Sunday	$6	1.2%
Capacity Building	Behaviour Management Plan Incl. Training In Behaviour Management Strategies	$6	1.1%
-	All other	$153	32.0%
	Total	$479	100%

The top two line items for the 15-18 autistic cohort in 2022 are the two main core line items in the NDIS: assistance with self-care and access community social and rec activities — both for standard supports in weekday daytime.

Assessment recommendation therapies are still being utilised, but there are a number of new line items relative to the 0-14 cohort. This includes more variations of the two main core supports and STA Assistance.

Page 52 of 53 Source: Panel estimates based on unpublished data supplied by the National Disability Insurance Agency 26 May, 2023 | 16

A > bf ef tf tf Ff 7 7 2

FO} 24/25-2110

Page 53 of 53

RE: Presentations for review [SEC=OFFICIAL:Sensitive] — s47F - personal privacy on behalf of NDIS Review Liaison Office to s47F - personal privacy (2023-06-01)

Early Intervention

Definitions

Exit rate (Slide 3)

Entry type (Slide 4)

Line item spend (Slide 5)

Average funding by age (Slide 6)

Growth in spend (Slide 7)

Cost Drivers

General

Access

Over-servicing and Support coordinators

NDIA Feedback on the NDIS Review’s ‘Early Intervention Presentation’ report.

Current State Review

Issues with eligibility reassessment (ER) as a contributing factor

How to ensure funding spent on evidence based Early Intervention?

Rise in core as children in EI cohort develop, with concurrent Capacity Building funding

Impact of Access Lists on EI and continued eligibility / participation within the Scheme

NDIA Feedback on the NDIS Review’s ‘Cost Drivers’ report

Recording disabilities on a participant’s CRM record

Changes in the data over time

The concepts of ‘additional disabilities’ and ‘co-morbid health and mental health conditions’

Implications of research

Scheme cost drivers — Access, over-servicing and support coordanators analysis

Summary of Key findings:

Analysis section 1. Access requests and mainstream supports

The overall rate of co-occurrence among Scheme entrants has declined from above 50% in 2017 to 12% by 2022

Across participants with co-occurring conditions, mental health conditions are representing an increasing share of conditions, particularly anxiety disorders

Participants with anxiety/depression are twice as likely as others to use the NDIS to access psychologists and spend 3x more

Both participants and mainstream service providers have incentives to seek NDIS access for participants to use mainstream services

The impact of anxiety/depression on psychology use is likely to be underestimated due to imperfect data records

Analysis section 2. Over-servicing

To analyse over-servicing, plan expenditure is divided into over-servicing and regular servicing (for plans that have ended) in 3 steps

Over-servicing has been increasing since July 2019, from 9% to 16% of spend as at February 2023

What is over-servicing?

Some over-spending can reflect inadequate plan values

Over-servicing has substantially increased in 2022

A small number of providers have a disproportionate share of their NDIS revenue originating from over-servicing

Unregistered providers are more likely to over-service participants compared to registered providers

Analysis section 3. Support co-ordinators

Most participants in the Scheme use registered support coordinators

Unregistered support coordinators are more likely to have over-serviced participants than registered support coordinators

Participants using unregistered & conflicted support coordinators are more likely to experience over-servicing

Appendix A

Across almost all major disability cohorts, the presence of anxiety/depression is associated with greater use of NDIS funds on psychology services

Core accounts for 79% of Scheme over-servicing, with Capacity Building — Daily Activities accounting for a further 10%

Observed over-servicing across providers does not appear to be randomly distributed

Early intervention deep dive

Report summary

Purpose of this report

Method notes

Key evidence points of this report

“So what”

The exit rate for early intervention is around 10%, while the majority transition from delay to an autism diagnosis

Since 2019 a consistent number of autistic participants have entered as New participants through Section 24 or Early Intervention

Line item spend shows therapies are the main supports used by autistic children

What supports autistic children are using

Average funding for core supports begins rising at age 11, and by age 20 overtakes capacity building supports for most autistic participants

Shift in plan nature across ages for autistic participants

Across the most common autism cohorts, there is no evidence that early intervention has led to low or declining spend

How spend growth for early intervention participants compares

In 2022 autism participants spent $4.3 billion while participants who entered under Section 25 (early intervention) spent $2.4 billion

Current state spending for participants by autism, delay and main entry types

Out of 592,000 NDIS participants, 229,000 (39%) are new entering via Section 24, while 184,000 (31%) are new entering via Section 25, i.e. Early Intervention

How 592,000 participants entered the NDIS

Participants vary by entry type, including lower level of function from legacy state schemes and higher function levels from Early Intervention

How participants by different entry types vary

Autistic participants are much younger and much more likely to have moderate level of function, compared to all other disabilities

How participants with autism and developmental delay vary

Autistic participants from legacy schemes are slightly older with higher support needs, while El participants are younger with lower support needs

How participants from different entry types vary

From 19,300 exits, the most common exit is early intervention delay at 7-14, followed by high and moderate functioning autistic participants

Profile of non-mortality exits from the NDIS

Similar with all other disabilities, there is a significant increase in supports from the age of 15

Comparison of plan value and spend for autism and all other disabilities in 2022

While 15-18 year old autistic participants are still using therapies, the majority of the top line items are core supports

What supports 15-18 year old autistic participants are using