A live audit of 158 PTAC decisions, sorted by what was wrong with the evidence and what the committee decided. One pattern keeps showing up.
When PTAC raised a cost-effectiveness concern, 21 of 24 times the outcome was conditional access — not rejection. CE uncertainty is managed by attaching criteria, not by saying no.
29 of 31 not-recommended decisions had no cost-effectiveness concern flagged at all. Evidence design — not price — drives rejection.
The dominant rejection driver is evidence quality — insufficient, absent, or answering the wrong clinical question. Not cost.
NOT_RECOMMENDED · N=31 · RAW CSV
High clinical need does not protect against rejection. Need alone is not a shield — evidence design is what carries the decision over the line.
Need score is an A²I™ classification applied to PTAC minutes text — not PTAC's own terminology.
NEED_SCORE 4–5 · RAW CSV
CE concern is a conditionality signal, not a rejection signal. The committee manages cost uncertainty by attaching criteria — not by saying no. Evidence design drives rejection, not price.
Hover or tap a concern type to see how PTAC ruled.
When the evidence doesn't answer the question PTAC is asking — wrong endpoint, wrong comparator, wrong population — the application is not recommended.
Indirectness means the evidence submitted did not answer PTAC's specific question. This is not about evidence quality — the studies may have been well-conducted. The problem is that the trial measured the wrong thing:
When evidence is indirect, PTAC cannot draw conclusions about the benefit in the funded setting. In this corpus, indirectness was fatal every time — 7 of 7 cases resulted in rejection, with zero recommendations or conditional approvals.
Representative paraphrase of recurring PTAC language — not verbatim from a single decision.
CNS — 43% rejection rate. Esketamine, memantine, and modafinil account for repeated not-recommended outcomes; low certainty evidence is the dominant failure mode.
Oncology — all 7 indirectness rejections in the corpus. PFS without OS demonstrated = rejection every time.
Rare disease — 31% rejection, zero conditional approvals. Binary outcomes: recommended or not recommended.
Immunology — best cohort: 69% recommended, 19% conditional, 11% rejected.
Therapy area classifications are derived from indication and drug_name fields in ptac_corpus_v9_final.csv. Outcome counts verified against PTAC minutes (spot check n=11, May 2026). 2 decisions unclassified (excluded from breakdown).
Drugs reviewed more than once split into two distinct patterns.
Multiple reviews where PTAC eventually recommended or broadened access as evidence matured — the committee waited for the right data, then funded.
Repeated submissions that never crossed the line — same evidence gaps, same not-recommended or deferred outcomes across reviews.
Empagliflozin (4 reviews · 2 rec / 2 not rec) excluded — insufficient pattern signal.
MULTI-REVIEW DRUGS · RAW CSV
Cost-effectiveness concerns lead to conditional access — not rejection. When CE was flagged, 21 of 24 times PTAC managed uncertainty by attaching criteria.
REJECTION BREAKDOWN (31 total)
Zero rejections on CE alone. When CE was flagged, the committee attached criteria — it did not say no. Deferral is a different signal: evidence genuinely absent and expected, not insufficient dossier evidence.
Key insight: Evidence existed but quality insufficient. PTAC could not draw reliable conclusions on magnitude of effect.
Examples: esketamine (3 rejections TRD, MDSI, post-augmentation), empagliflozin (HFrEF and HFpEF), secukinumab (AS and HS)
Key insight: Evidence did not answer PTAC's question — surrogate endpoint (PFS not OS), wrong comparator, population mismatch. Fatal every time.
Examples: atezolizumab (TNBC — 2), pembrolizumab combinations (cervical, mesothelioma), bevacizumab (ovarian), eribulin (breast), nivolumab (urothelial)
Key insight: Required data simply not in the submission. No RCT. No comparative data. No long-term outcomes.
Examples: clodronate, octreotide, deflazacort, ferric carboxymaltose, secukinumab (HS)
5 total · all evidence gap absent, CE concern none
PTAC defers when evidence is genuinely absent and expected. It rejects when evidence is submitted but insufficient or doesn't answer the question. These are fundamentally different signals.
Source: ptac_corpus_v9_final.csv — raw extraction, no enrichment.
In your Phase II protocol. In your comparator choice. In your endpoint selection. A²I™ maps 158 PTAC decisions to show what evidence determined access — by indication, by claim type, by pattern. Know what PTAC needs before you build it.
What funding evidence claims have historically determined access — by indication, by therapy area, by claim type.
Which claims PTAC has historically challenged — ranked by how load-bearing they are to the overall recommendation.
What evidence has historically closed the gap when PTAC raised a specific concern type.
75 DRUG DECISIONS · 23 MEETINGS · 2020–2025 · VERBATIM EQUITY FIELD EXTRACTION
PTAC documents equity on 89% of decisions. It acts on it in 17%. This section surfaces what five years of minutes reveal about equity evidence, research gaps, and the distance between documentation and outcome. All figures are drawn directly from raw equity field extraction across PTAC meeting minutes 2020–2025. Zero enrichment applied.
Equity is assessed on almost every decision. It is cited in the recommendation in fewer than 1 in 5.
FIVE-YEAR TREND · DOCUMENTATION VS ACTION
Of 75 decisions, 67 contain a Māori equity section and 56 contain a Pacific equity section. Equity is cited in the actual recommendation text in 13. The gap is not the committee — it is the evidence that was submitted.
In 2020, equity was cited in zero recommendations despite appearing in 73% of decision records. By 2023, every decision had a Māori and Pacific equity section. Equity in recommendations dropped to 7% by 2025. The 2021 Terms of Reference change made documentation mandatory. It did not make evidence mandatory.
11 decisions have a Māori section but no Pacific section. In 2020, Pacific coverage was 47% versus 73% for Māori. Both reached 100% by 2023 — but the gap persisted for three years after mandatory sections were introduced. Equity documentation was not equal across populations even within the same mandatory framework.
PHARMAC's five Māori health priority areas — lung cancer, breast cancer, cardiovascular disease, diabetes, mental health — account for 55 of 75 decisions reviewed. Equity language reaches the recommendation in 13 of those 55. The Hauora Arotahi framework has been in place since 2016, confirmed by whānau Māori in 2018. The evidence to act on it has not been generated.
Five years of PTAC minutes show the same pattern. The committee documents equity burden. It notes absent evidence. It sets equity aside. Pivotal trials are designed years before a PHARMAC submission — in markets where Māori and Pacific peoples do not exist as a patient population. Underrepresentation is structural, not incidental. NZ-specific registry data, ethnicity-stratified real-world analyses, and post-market equity studies exist as a possibility. The minutes show they are almost never submitted. The committee notes the absence every time. Requiring documentation of a gap does not close it — from 2023, every decision has an equity section. Equity in recommendations dropped to 7–25%.
PHARMAC has documented Māori health priority areas since 2016. Eight years later, the minutes still show the same phrase: "no specific evidence available for this population." Equity burden is well characterised. Equity evidence — the kind that informs a funding decision — is not.
The funding system requires evidence to act. The research system has not generated it for these populations. Access inequity is upstream of the funding decision — it is a research design problem before it is a reimbursement problem.
This extraction captures whether equity was documented and whether it appeared in the recommendation. It cannot tell us whether equity evidence — had it existed — would have changed the outcome. That is the unanswered research question this dataset surfaces.
Methodology & Limitations
Data source: PTAC meeting minutes 2020–2025, publicly available at pharmac.govt.nz. 25 PDFs processed; 2 meetings absent from extraction (2021-11 and 2024-11) due to download failure during automated processing.
Extraction method: Automated verbatim field extraction via LLM-assisted parsing. Fields extracted per drug decision: drug name, meeting date, recommendation level, Māori impact section present (yes/no), Māori impact text (verbatim, max 100 words), Pacific equity section present (yes/no), Pacific equity text (verbatim, max 100 words), equity cited in recommendation (yes/no), Hauora Arotahi flag (yes/no). No manual curation of extracted text.
Field definitions: — equity_cited_in_recommendation: text-presence flag only — captures whether equity language appeared in the recommendation statement. Does not indicate that equity drove or determined the outcome. — hauora_arotahi_flag: relevance classifier — indicates whether the drug's indication falls within PHARMAC's five designated Māori health priority areas. Not a measure of equity evidence quality or decision weight. — Pre-2021 equity sections: informal and scattered. Mandatory equity sections introduced via PTAC Terms of Reference revision, 2021.
Hauora Arotahi: PHARMAC's five Māori health areas of focus (lung cancer, breast cancer, cardiovascular disease, diabetes, mental health). Framework in place since 2016. Confirmed by whānau Māori through consultation, 2018. Reference: pharmac.govt.nz/hauora-arotahi
Corpus: ptac_corpus_v9_final.csv — 158 rows, 2020–2025. equity_raw_extraction.csv — 75 rows, 23 meetings. Equity extraction is a net-new dataset, not present in the v9 corpus.
Limitations: This analysis does not establish causation. Equity being absent from the recommendation text does not confirm equity was not considered — it reflects what was documented in the published minutes. The extraction cannot determine whether equity evidence, had it been submitted, would have changed the recommendation level. Therapy area counts may shift when the full ptac_raw_extended table is built with validated therapy_area field. Outcome sub-splits within concern types have not been validated against the raw CSV. Directional use only. This report is produced for informational and research purposes. It does not constitute regulatory, clinical, or funding advice. Ascentra Medical makes no warranty as to the completeness or accuracy of automated extraction outputs.
© 2026 Ascentra Medical. Source data © PHARMAC New Zealand.
The evidence that determines your access outcome is being generated right now — in your Phase II protocol, your comparator choice, your endpoint selection. Evaluate it early. Fix it before it's locked.