New Zealand's first specialist assessment target is defined as a share of people waiting and its published definition does not mention declined referrals, which were 13.1 percent of 2,918,557 first referrals analysed for 2018–2022

What is happening?

One of the five government health targets in New Zealand is that 95 percent of patients wait less than four months for a first specialist assessment, to be reached by 2030. The Health New Zealand fact sheet on the targets defines the measure as the proportion of people waiting less than four months for that assessment, counted from the date of referral. For October to December 2024 the same sheet reported 197,179 people waiting, of whom 119,439 had waited less than four months, giving a national result of 60.6 percent. Dividing the second number by the first reproduces that result, which confirms that the published measure takes the people on the waiting list as its base. The most recent ministerial release, dated 2026-09-23, reported 66.1 percent for April to June 2026, against 62 percent a year earlier.

There is a step before anyone is counted as waiting. When a general practitioner refers a patient, the specialist service assesses the referral and records one of four prioritisation outcomes: accepted, redirected, not yet decided, or declined. A 2016 press report on the first national figures described it the same way: a patient whose referral is accepted may then go onto a waiting list.

That step has been measured nationally. A peer-reviewed study of National Patient Flow data published on 2025-09-02 found that 13.1 percent of 2,918,557 first referrals analysed for 2018–2022 were declined at prioritisation. The rate rose from 11.6 per 100 in 2018 to 14.1 in 2021 and 13.9 in 2022. The published one-line definition of the target does not mention this outcome. In January 2026 Health New Zealand told the press that it can see declined patients within a hospital department but cannot easily capture them at a national level.

Whose problem is this?

A person whose referral is declined leaves the specialist system at the point of prioritisation and returns to general practice, and no source opened here follows what happens to that person next.

Where does this problem end?

The question here is narrower than whether too many referrals are declined, because no source opened here judges whether any particular decline was clinically right. What this document tracks is whether the decline outcome sits anywhere in the way the target is defined and reported.

What is the state now, and what should it be?

Now

Needs a new measurementthe target state for declined referrals: no institutional document opened here sets a target, a reporting rule or a deadline for measuring declined referrals at national level. The 95 percent target concerns people waiting, not declines. Health New Zealand has described its intentions only in statements reported by the press, set out below as dated attributions rather than as a target.

How big is it?

Four different quantities appear in the sources, and they measure different things.

The waiting stock was 197,179 people at the end of 2024. That is the base of the target measure.

The flow of first referrals analysed in the study was 563,222 in 2018, 564,073 in 2019, 585,683 in 2020, 610,707 in 2021 and 594,872 in 2022, a total of 2,918,557 after one region with 115,782 referrals was excluded for data quality reasons. Over the same period the population grew by about 7 percent while referral numbers stayed broadly stable.

Multiplying each year of referrals by the decline rate for that same year gives roughly 65,000 to 86,000 declined first referrals a year across 2018–2022. That range is derived here, not stated by the study, and it counts referrals rather than people.

Other counts of declines exist but cannot be set beside the national rate. A 2016 press report of the first National Patient Flow figures gave 7,229 referrals declined as below threshold in July to September 2015, about 5 percent, under a narrower definition at an early stage of data collection. The union representing salaried senior doctors reported 101,962 declines in 2023 and 110,916 in 2024 across seven districts, from official information responses. A 2026 press report gave South Island counts of 36,814, 39,498 and 37,662 patients for the financial years 2022/23 to 2024/25. These figures differ in year basis, unit, definition and coverage, so none of them is a continuation of the national series.

Under what conditions does it arise?

1. The target measure is built on the waiting list. The published definition takes the people who are waiting as its base, and a declined referral does not add a person to that list, so the measure moves with how long accepted patients wait and is silent on how many were not accepted.

2. The threshold for acceptance depends on resources as well as need. The Auditor-General wrote in 2011, for what were then called scheduled services, that thresholds may need to alter in response to changes in the capacity and funding of district health boards. In 2025, writing about treatment thresholds, the Auditor-General stated that where a threshold is set depends on both clinical need and the resources available to meet demand.

3. Thresholds differ between districts. Health New Zealand said in January 2026, four years after twenty district health boards were merged into a single national body, that acceptance thresholds are not a level playing field across the country.

4. The data exists but is not routinely assembled nationally. National Patient Flow collects the prioritisation outcome of every referral, the Ministry of Health published a first national figure in 2016, and researchers produced a five-year national rate in 2025. Health New Zealand nonetheless said in 2026 that it cannot easily capture declines at national level.

What has been tried?

The collection that would answer the question has existed for about a decade. What the sources show has been tried is one national publication, one academic analysis and a stated intention to report, and none of these has been attached to the target that is published every quarter.

What was found?

Why is it still unsolved?

Measurement absent — the outcome exists in a national data collection, but it is not part of the measure the target reports.

A target defined on the waiting list can only see people who have been placed on it. The prioritisation decision that comes first is recorded referral by referral, yet the published definition of the target does not mention it, and the body responsible for both the list and the decision has said it cannot easily count declines nationally. The result is that the quarterly figure and the decline rate are two separate pieces of information, and only one of them is published on a schedule.

The second part is that the threshold is not fixed. Where acceptance is set depends on resources as well as need, according to the Auditor-General, and Health New Zealand has said it differs between districts. A measure that does not carry the decline outcome therefore cannot show whether a change in the waiting figure came with a change in who was accepted. No source opened here says that it did; the point is that the published measure could not show it either way.

The third part is that the pieces have appeared only once each. A ministry table in 2016, an academic study in 2025, a union analysis of seven districts in 2026 and a stated plan to report validated data have never been joined into a regular national series, so each new count arrives on its own definition and cannot be compared with the last.

What observation would mean it is solved?

Candidates — (a) Health New Zealand publishes, on a regular schedule, the national number and rate of first specialist assessment referrals declined at prioritisation, alongside the target result (b) the technical definition of the target is published and states how declined referrals are treated (c) nationally consistent acceptance criteria are adopted and reported against by district.

(a) alone is weaker than it looks. A rate without a stable definition repeats the problem of the existing counts, which cannot be compared with each other. It also counts referrals, not people, so repeat referrals can inflate it while people deterred from being referred are not in it at all.

(b) alone only describes. Stating how declines are treated makes the measure honest about its base, but does not by itself put the decline outcome in front of anyone reading the quarterly figure.

(c) alone moves the threshold without measuring it. Consistent criteria could make declines more or less frequent, and without (a) nobody outside the system would see which. The three have to be read together.

What is it connected to?

Fills with researchtreatment-stage thresholds examined by the Auditor-General in 2025, the load on general practice when referrals return, and measures of unmet need in other health systems are neighbouring questions. Relation types and evidence grades were not established in this round.

What these sources do not say

• Whether national decline figures are already published. A Ministry of Health page titled as tables of prioritisation outcomes for first specialist assessment referrals, and the National Patient Flow statistics page, both returned 403 and could not be read. The sources opened here therefore do not establish whether routine national publication exists.
• How the technical definition treats declines. The target page links to performance resources that were not in the retrieved text. The gap described here is limited to the published one-line definition; the technical definition was not seen.
• Any link between the target figures and the decline rate. No source opened here connects a rise in declines with a movement in the target result, and the years they cover do not overlap. The 2025 Auditor-General report expressly left the first specialist assessment phase outside its scope.
• Whether declines were clinically appropriate. No source judges this. The study authors state that their data cannot separate differences in population need from differences in referral and triage practice.
• What happens to people after a decline. No opened source measures re-referral, private care, giving up, or health outcomes.
• National decline figures after 2022. The study stops at 2022. Health New Zealand said in May 2026 that validated data was expected by the end of the year; its publication was not found by 2026-09-27.
• The limits of the study, as stated by its authors. Data quality varied across regions, one region was excluded, complete-case analysis may introduce bias, triage criteria and referral practice may have changed over time, people deterred from seeking a referral are not captured, and without individual identifiers repeat referrals may be counted more than once. The study was funded by a primary care sector organisation, the side of the system to which declined referrals return; the authors state that they worked independently.
• One definition or two. The fact sheet describes the measure as the share of people waiting under four months, while a Ministry briefing describes it as the share of people on the list who were seen within the timeframe. The sources do not say which is the stock and which the flow. The arithmetic of the fact sheet matches the stock description.
• The number of districts. The 2025 Auditor-General report counts 19 districts; the fact sheet and the Ministry briefing count 20. The union estimate of more than 255,000 declines nationally uses 19 and is not recalculated here.
• A formal response from Health New Zealand to the 2025 Auditor-General report. It was not found in the pages opened.
• What the Ministry briefing activity figure refers to. The briefing says the target percentage can mask the volume of activity and cites about 12,000 more assessments completed in a quarter it labels Q4 2024/25, but the briefing is dated 2025-01-31, before that quarter ended, so the period of the figure is not settled.
• Whether the 2016 first publication covered all districts. The report does not say.

See the evidence

The central figures rest on documents read directly. The target definition and the October to December 2024 counts come from the Health New Zealand fact sheet, and the national decline rate comes from the full text of the peer-reviewed study. The Health New Zealand statements on national capture, threshold consistency and planned reporting are known only through press reports, and the union figures only through press reports and an editorial, because the union report itself returned 403. Nine rows above could not be opened and are kept with the reason rather than removed. This is a research-based definition, so observation_refs is empty and provenance_mode: press-derived.

People affected

Estimated range 65,334–86,110 As of annual range across calendar years 2018-2022, declined first referrals

Sensitivity The unit is declined referrals, not people. Without individual identifiers in the data, one person referred more than once can be counted more than once, which inflates the count. Two things deflate it: one region with 115,782 referrals over the five years was excluded for data quality, and people deterred from seeking a referral never enter the data. The yearly rates are published rounded to one decimal place, so each product carries a small rounding error. The range says nothing about years after 2022 or about whether any decline was clinically appropriate.

Regional breakdown The study reports regional rates for anonymised regions only and warns that data quality varies widely between regions, so a regional breakdown cannot be attached to named districts.