The NHMRC Report

Australia's 2015 NHMRC report is one of the most instructive case studies available of what happens when a knowledge system tries to evaluate a form of knowledge its foundational assumptions prevent it from seeing. The report applied a strict definition of "reliable evidence," a hard sample-size threshold, and a contested review process -- and arrived at its conclusion before examining a single trial, because the conclusion was built into the assessment framework.

This page does two things. First, it dissects the report's methodology -- the threshold, the double-exclusion pipeline, the suppressed first review, and the sensitivity of meta-analytic conclusions to inclusion criteria. This internal critique is rigorous and stands on its own terms. Second, it asks the question the report never asked: whether the biomedical evidence-assessment framework is the right instrument for evaluating a participatory medicine. The methodological problems are not isolated errors. They are symptoms of a deeper structural condition.

The Objection

"The Australian NHMRC said there's no reliable evidence that homeopathy works."

The NHMRC's 2015 Information Paper: Evidence on the Effectiveness of Homeopathy for Treating Health Conditions concluded that "there are no health conditions for which there is reliable evidence that homeopathy is effective." That sentence has dominated the English-language conversation about homeopathy for a decade.

A fair-minded skeptic typically adds three implicit steps:

1. The NHMRC reviewed a broad body of research. Therefore it is unlikely that "hidden" high-quality evidence exists.
2. If reliable evidence existed, it would have appeared in such a review. Therefore the most plausible explanation is that the effect is not robust.
3. Public communication simplifies. "No reliable evidence" becomes "doesn't work," because in practice people care about what can be supported reliably.

This is a coherent position within the biomedical paradigm. It deserves a serious response -- both methodological and epistemological.

Why This Matters

When a national research body publishes a negative conclusion, the downstream effects are real: funding decisions, insurance coverage, professional standing, media framing, and the tone of the public conversation. For practitioners, it shapes regulation. For patients, it shapes whether they feel confident seeking homeopathic care. For researchers, it influences career viability.

But the deeper reason this objection demands careful handling is philosophical. Evidence reviews are not neutral summaries of data; they are machines that transform data into conclusions. Every machine has settings: what counts as "reliable," how heterogeneity is handled, whether small trials are treated as preliminary signals or filtered out, and how individualized interventions fit into condition-based frameworks.

The NHMRC report invites us to examine those settings -- and then to ask a harder question. Not just whether the machine's settings were correct, but whether this kind of machine is the right instrument for this kind of knowledge. A spectrophotometer is a precise instrument. It is the wrong instrument for evaluating a symphony. The limitation is not in the spectrophotometer's calibration; it is in the assumption that everything worth knowing can be measured in wavelengths. (For the full epistemological framework behind this claim, see How We Know What We Know.)

The Epistemological Context: Can the Biomedical Paradigm Evaluate Homeopathy?

Before examining the report's specific methodological choices, it is worth stepping back to ask why those choices were made -- and why they were considered natural and uncontroversial by the committee that made them.

The NHMRC operates within the biomedical evidence-assessment paradigm. That paradigm treats the randomized controlled trial as the gold standard of evidence, places meta-analyses at the apex of a fixed hierarchy, defines "reliable" as large, blinded, randomized, and statistically significant, and regards the practitioner's clinical judgment as a source of bias to be controlled for, not a source of knowledge to be cultivated. Every one of these commitments is a philosophical position, not a discovery about the nature of evidence. They follow logically from a specific epistemological tradition -- the Kantian inheritance that assumes human beings cannot know reality directly and must compensate through statistical inference. (For the full argument, see the epistemological framework.)

Homeopathy operates within a different epistemological tradition -- a participatory one. In this tradition, the trained practitioner's perception is the primary instrument of knowing. The patient is an unrepeatable totality -- what Hahnemann called the Inbegriff -- not an interchangeable data point. The totality of symptoms is a meaningful gestalt, not a list of variables to be averaged. And the remedy is selected for this patient by this practitioner based on this encounter, not administered as a standardized input to a diagnostically homogeneous population.

Thomas Kuhn and Paul Feyerabend independently developed the concept of incommensurability to describe precisely this situation: two paradigms that cannot evaluate each other on neutral ground because they define their fundamental terms differently. In the biomedical paradigm, "evidence" means statistically significant results from blinded, randomized trials. In homeopathic practice, "evidence" means the totality of the individual patient's symptoms as perceived by a trained observer, the response to the remedy, and the accumulated clinical knowledge of two centuries of practice. These are not two measures of the same thing. They are two different concepts of what "evidence" means.

The structural parallel with Galileo is precise. Feyerabend's meticulous reconstruction of the Galileo case showed that the demand to prove Copernicanism by Aristotelian physics was the demand that a new paradigm validate itself using the old paradigm's criteria -- criteria designed so that the new paradigm must fail. Galileo was right, but for reasons that no methodology of his time could have sanctioned. The demand that homeopathy prove itself by RCT is structurally identical: it requires a participatory medicine to demonstrate its knowledge using instruments calibrated to detect something else entirely.

This is not conspiracy theory. It does not claim that the NHMRC scientists acted in bad faith. It claims something more fundamental: that the biomedical paradigm lacks the conceptual resources to perceive what homeopathy knows. The NHMRC's inability to find "reliable evidence" tells us as much about the limitations of its assessment framework as about the limitations of homeopathy.

With this context established, the methodological critique that follows becomes not merely a catalog of questionable choices but a diagnostic: each problematic decision traces back to the structural inability of one epistemological system to accommodate another.

What the Evidence Shows

1) What the NHMRC actually concluded -- and what it did not

The NHMRC's key line -- "no health conditions for which there is reliable evidence" -- is often read as a direct claim about reality ("homeopathy is ineffective"). It is more precise than that.

In evidence assessment, there is an important distinction between:

• No reliable evidence of effectiveness (the evidence does not meet the review's reliability criteria), and
• Reliable evidence of no effectiveness (good evidence that the effect is absent or clinically negligible).

The NHMRC's phrasing belongs to the first category. That is not a loophole or rhetorical trick. It is simply what the words mean in the logic of evidence synthesis. The NHMRC did not conclude that homeopathy does not work. It concluded that by its criteria, no reliable evidence of effectiveness could be identified.

But there is a deeper distinction to make. "Reliable" is itself a term of art defined by the biomedical paradigm. What counts as reliable evidence depends on what you consider reliable instruments of knowing. If the trained practitioner's perception is a reliable instrument -- as Goethe, Steiner, Hahnemann, and the Chinese medical tradition all maintain -- then two centuries of consistent clinical observation, systematic provings, and a materia medica verified across cultures and continents constitute reliable evidence of a kind the NHMRC was not equipped to recognize. The NHMRC did not find what it was not looking for. This is not a failure of the evidence. It is a limitation of the search.

2) The sample-size threshold and its consequences

The most disputed methodological choice in the NHMRC process is the use of a minimum sample-size threshold -- n >= 150 -- as a gating criterion for "reliability." (NHMRC, 2015; Homeopathy Research Institute, 2019).

This matters because many homeopathy RCTs have sample sizes below that cutoff -- typically between 30 and 120 participants. A hard threshold excludes a large fraction of the evidence base before other quality dimensions (randomization, blinding, attrition, outcome reporting) can be weighed.

From a statistical perspective, the argument for a threshold is understandable: small trials are noisy, small-study effects and publication bias are real, and a national body wants to avoid over-interpreting weak signals. The critique is not that sample size is irrelevant. The critique is that a single hard cutoff substitutes for a fuller treatment of imprecision.

Frameworks commonly used in evidence assessment (for example, GRADE) treat imprecision as one factor among several, with downgrading based on confidence intervals and decision thresholds -- not as a blanket exclusion rule. This threshold does not appear in GRADE, Cochrane methodology, or standard evidence assessment frameworks. If applied uniformly across medicine, it would exclude early-stage evidence for many conventional treatments. Phase II clinical trials routinely enroll fewer than 150 participants.

The practical consequence: the NHMRC report is, by construction, a review that treats much of the homeopathy trial literature as not meeting its definition of reliability. That is a defensible position within a specific paradigm, but it is not the same thing as "the trials consistently show no effect."

3) The double-exclusion problem

Another technical feature of the NHMRC approach is that it primarily assessed systematic reviews rather than re-extracting primary trial data. (NHMRC, 2015).

This is not inherently wrong -- reviews-of-reviews can be efficient. But they introduce a compounding effect: if a systematic review already used restrictive inclusion criteria, and the NHMRC then applied additional reliability filters on top, the end result is a double-exclusion pipeline that leaves very little evidence in the top tier.

The Homeopathy Research Institute's critique argues that this compounding restrictiveness materially shaped the final conclusion. Their analysis also identified inconsistent quality assessment -- some trials were downgraded for features intrinsic to homeopathic research design, such as individualized prescribing protocols -- and the absence of any homeopathy research expertise on the expert committee. (Homeopathy Research Institute, 2019).

The downgrading of individualized prescribing deserves particular attention. Individualization is not a methodological inconvenience that reviews should try harder to accommodate. It is the foundational epistemological commitment of homeopathic practice -- the claim that each patient is an unrepeatable totality, and that the remedy must be selected for this specific patient. The RCT's inability to handle individualization is not a limitation that can be fixed with better study design; it is a structural incompatibility between two epistemological systems. When a review penalizes individualized prescribing as a methodological weakness, it is judging homeopathy by the criteria of a paradigm whose foundational assumptions contradict homeopathy's own. (For the Organon's account of why individualization is epistemologically necessary, see Hahnemann, Organon, Aphorisms 7, 83, and 153.)

4) The suppressed 2012 draft and the question of process

The point that cannot be ignored is procedural: an earlier draft review, completed in 2012 by a team led by OptumInsight, reached more nuanced conclusions and was not published. The NHMRC instead commissioned a second review with different criteria and a different methodology. The existence of the earlier work became publicly established through Freedom of Information requests and Australian Senate Estimates hearings in 2016.

What this implies: transparency matters in evidence synthesis. When an earlier draft exists, non-publication raises questions about what changed and why. If the evidence base is broadly similar but the conclusion shifts, the likely driver is the assessment framework -- which reinforces the central point: in a method-sensitive literature, the framework determines the conclusion.

In 2017, the Australian Commonwealth Ombudsman investigated complaints about the NHMRC review process. The Ombudsman did not find evidence of deliberate misconduct but noted procedural concerns about transparency. The investigation confirmed the existence of the first review and its non-publication. Process problems do not prove efficacy. But they are evidence that a single institutional conclusion should be treated as one assessment within one paradigm, not as a final verdict on a form of knowledge that paradigm was not designed to evaluate.

5) The sensitivity analysis: why inclusion rules are not a minor detail

If you want to see why trial-selection criteria matter, the broader homeopathy meta-analysis literature provides an instructive case.

• Linde et al. (1997, The Lancet) pooled placebo-controlled trials and reported an overall result favoring homeopathy, while emphasizing that trial quality limitations made strong conclusions difficult.
• Shang et al. (2005, The Lancet) reached a much more negative conclusion, heavily influenced by a restricted subset of larger, higher-quality trials.
• Ludtke & Rutten (2008, Journal of Clinical Epidemiology) showed that Shang's conclusion depends strongly on which trials are included -- different, defensible definitions of "large" and "high quality" produce different outcomes.
• Rutten & Stolper (2008, Homeopathy) emphasized how post-publication scrutiny and data handling can materially affect conclusions in this field.

These are not cited to win the NHMRC argument by proxy. They demonstrate a methodological reality: when a literature is heterogeneous and many studies are small, conclusions can pivot on selection thresholds. That is exactly the kind of environment in which a hard n >= 150 cutoff becomes outcome-determinative.

6) What more recent systematic reviews suggest

To make this concrete, consider two systematic reviews by Mathie and colleagues:

• Mathie et al. (2014, Systematic Reviews) analyzed randomized placebo-controlled trials of individualized homeopathic treatment and reported a pooled effect favoring homeopathy (OR 1.53, 95% CI 1.22--1.91), while rating the overall evidence quality as low/unclear and calling for better trials.
• Mathie et al. (2017, Systematic Reviews) performed a meta-analysis on non-individualized homeopathic treatment and again reported results that, depending on sub-analysis, favored homeopathy but remained constrained by trial quality and heterogeneity.

These papers do not settle the question. But they are hard to reconcile with the claim that every rigorous synthesis shows nothing. The picture is mixed: some syntheses find small effects with quality caveats; some find results compatible with placebo; and the disagreement tracks methodology.

This is precisely why the NHMRC report should be read as: "under our reliability framework, we do not consider the evidence base sufficient to claim effectiveness for any condition." That is a statement about what cleared a specific reliability bar within a specific paradigm. It is not the same statement as "the clinical effects have been reliably shown to be absent."

What We Acknowledge

A grounded analysis is only credible if it states its own constraints. Here are the ones that matter within the framework this article defends.

1. The NHMRC is a reputable institution within the biomedical paradigm. Its reviews carry legitimate weight within that paradigm. But institutional authority does not extend to evaluating knowledge systems whose foundational commitments differ from its own. The NHMRC can assess whether homeopathic RCTs meet the biomedical standard of reliability. It cannot assess whether homeopathic knowledge is valid, because that question requires engagement with an epistemological framework the NHMRC does not possess. Feyerabend's point is precisely that institutional authority is not epistemological authority.

2. The evidence base is vast -- but most of it exists in forms the biomedical paradigm does not recognize as evidence. Provings, systematic case documentation, the materia medica verified across two centuries and multiple continents, clinical observation refined through generations of practice -- this constitutes an enormous body of knowledge. The relative scarcity of large RCTs reflects not a failure of homeopathic research but the epistemological mismatch between the RCT and an individualized medicine. The RCT demands standardization; homeopathy requires individuality. The mismatch is structural, not a gap to be filled by more funding.

3. Procedural concerns reveal paradigm assumptions, not conspiracies. The suppressed first review and the unusual sample-size threshold raise legitimate questions about process. But the deeper lesson is not that the NHMRC was biased in the ordinary sense. It is that the entire evidence-assessment apparatus -- threshold selection, quality grading, the decision to assess reviews-of-reviews -- expresses the philosophical commitments of the paradigm in which the assessors were trained. Different commitments produce different conclusions. This is exactly what one would predict from the philosophy of science.

4. Other governmental reviews have reached different conclusions from overlapping evidence. The Swiss Health Technology Assessment (Bornhoft & Matthiessen, 2011) evaluated both clinical trial data and real-world effectiveness evidence and concluded that homeopathy could be considered effective, contributing to its inclusion in Swiss mandatory health insurance in 2017. India's Central Council for Research in Homoeopathy has published Standard Treatment Guidelines for multiple conditions based on clinical verification studies. These are not fringe institutions. The divergence across jurisdictions is precisely what the concept of incommensurability predicts: the conclusion depends on the paradigm within which the assessment is conducted.

5. Communicating across paradigms is genuinely difficult. The hardest limitation to acknowledge is not a lack of evidence but the challenge of making homeopathic knowledge intelligible to audiences trained in a different epistemological tradition. The evidence becomes visible when one develops the capacity to perceive the individual case as a meaningful whole. This capacity can be developed -- it requires training, not faith -- but it cannot be conveyed by a meta-analysis. The invitation is not "believe us despite the evidence" but "learn to see differently, and the evidence becomes self-evident."

The Broader Context

The NHMRC report did not emerge in a vacuum. It was published during a period of active policy debate about the regulation of homeopathy in Australia, and it was subsequently used to justify decisions about insurance coverage and public health recommendations. The political and institutional context does not invalidate the assessment, but it is relevant to understanding why a national research body chose to commission a second review after the first one reached different conclusions.

The broader pattern is diagnostic. Across different countries, different assessment frameworks have produced different conclusions from substantially overlapping bodies of evidence. This is not unusual in health technology assessment -- reasonable methodological choices can shift conclusions, especially when the evidence base is heterogeneous. What is unusual is the degree to which a single review has dominated the English-language conversation, while equally credible reviews from other jurisdictions have been largely ignored.

But the deepest lesson of the NHMRC report is epistemological. The report reveals, with unusual clarity, what happens when a paradigm that has blinded itself by design -- that has institutionalized the conviction that human beings cannot know directly and must therefore rely on statistical proxies -- tries to evaluate a form of knowledge that requires a different kind of seeing. The result is not a neutral finding of absence. It is the predictable outcome of applying an instrument to a phenomenon it was not designed to detect.

The NHMRC was not wrong within its own framework. But its framework is one framework among several, and it is the wrong framework for evaluating a participatory medicine. The report's conclusions are not definitive. They are paradigm-bound. And they should be read accordingly.

Where does that leave a reader who is skeptical but curious? Three questions deserve to be separated:

1. What did the NHMRC report demonstrate under its criteria? It did not find condition-level evidence that met its definition of reliable.
2. What does the broader landscape suggest under other frameworks? Mixed results in the biomedical frame; an extensive body of clinical knowledge in the participatory frame -- provings, case documentation, the materia medica -- that the biomedical frame does not recognize as evidence.
3. What is the most honest sentence we can say today? The NHMRC report is an assessment produced within a specific epistemological paradigm. Its conclusions are valid within that paradigm and inapplicable outside it. Different paradigms require different methods of evaluation.

For the epistemological framework behind this analysis, see How We Know What We Know. For a detailed critique of the Shang meta-analysis that the NHMRC report echoes, see The Lancet Study (Shang 2005). For the wider map of how we handle evidence across the site, see the Evidence Overview. To explore other skeptic objections, the Skeptic Hub is the index.

Summary

The NHMRC's 2015 report is best read as a strict evidence assessment within the biomedical paradigm -- not as a demonstration that homeopathy is ineffective and not as a neutral evaluation of homeopathic knowledge.

Its headline conclusion -- "no health conditions for which there is reliable evidence" -- is shaped by methodological settings that express the philosophical commitments of that paradigm: a hard sample-size threshold (n >= 150) that excludes many homeopathy RCTs before other quality dimensions are weighed; a double-exclusion pipeline that compounds the restrictiveness of layered reviews; the penalizing of individualized prescribing as a methodological weakness rather than recognizing it as an epistemological principle; and the absence of any homeopathy research expertise on the expert committee. A more favorable first review, completed in 2012 by OptumInsight, was not published -- its existence was later established through FOI requests and Senate Estimates questioning.

The broader evidence landscape is method-sensitive: meta-analyses and systematic reviews (Linde, 1997; Shang, 2005; Mathie, 2014; Mathie, 2017) have reached different conclusions depending on inclusion criteria and quality thresholds. Other government bodies -- in Switzerland and India -- examined overlapping evidence with different frameworks and reached different conclusions.

The NHMRC report is a case study of what happens when a knowledge system that has blinded itself by design tries to evaluate a form of knowledge that requires a different kind of seeing. Its conclusion was determined by its framework before a single trial was assessed. The framework is not the only framework. And the conclusion is not a verdict on homeopathy -- it is a diagnosis of the limits of biomedical evidence assessment when confronted with a participatory medicine.

Frequently Asked Questions

Did the NHMRC report prove that homeopathy is ineffective?

No. It concluded that there were no health conditions for which the NHMRC found reliable evidence of effectiveness under its criteria. That is a statement about what cleared a specific reliability bar within a specific paradigm, not a demonstration that effects are absent. The report used a sample-size threshold of 150 participants that excluded most published homeopathic trials from the "reliable" tier -- a threshold that does not appear in GRADE, Cochrane methodology, or standard evidence assessment frameworks.

Why does the n >= 150 threshold matter so much?

Because it functions as a gatekeeper. If many trials in a field have sample sizes below that cutoff, a review that treats sub-150 trials as "not reliable" will, by design, end up with very little evidence in the highest tier. That may reduce false positives, but it also ensures that preliminary signals are filtered out rather than weighed with appropriate uncertainty. The result: the conclusion is built into the inclusion criteria.

What is the significance of the suppressed 2012 draft review?

A first review was completed in 2012 by OptumInsight and reached more nuanced conclusions. The NHMRC did not release it and instead commissioned a second review with different methodology. The existence of the first review was confirmed through Freedom of Information requests and Australian Senate Estimates hearings in 2016. The Commonwealth Ombudsman subsequently investigated and noted procedural transparency concerns. The significance: when two reviews of overlapping evidence reach different conclusions, the driver is the assessment framework, not the evidence. This reinforces the central point that conclusions are paradigm-bound.

How should I interpret conflicting meta-analyses on homeopathy?

As confirmation that the literature is heterogeneous and that methodological choices determine pooled results. Linde et al. (1997) and Mathie et al. (2014, 2017) reported pooled effects favoring homeopathy with quality caveats; Shang et al. (2005) reached a more negative conclusion; Ludtke & Rutten (2008) showed how sensitive such conclusions can be to inclusion decisions. The disagreement is real -- and it tracks methodology. This is exactly what the philosophy of science predicts when one paradigm's tools are applied to another paradigm's knowledge.

Did other governments reach different conclusions?

Yes. The Swiss Health Technology Assessment concluded that homeopathy could be considered effective, contributing to its inclusion in Swiss mandatory health insurance. India's CCRH has published clinical treatment guidelines for multiple conditions based on institutional research. France reimbursed homeopathic treatments through its national health system until 2021. These divergent conclusions from government bodies examining overlapping evidence demonstrate that the answer depends on the assessment framework -- which is to say, on the epistemological paradigm within which the assessment is conducted. Different paradigms require different methods of evaluation.

Is this article saying the NHMRC acted in bad faith?

No. The argument is not about bad faith. It is about structural limitation. The NHMRC scientists operated within the biomedical paradigm with professional integrity. The claim is that the biomedical paradigm itself lacks the conceptual resources to evaluate a participatory medicine -- just as Aristotelian physics lacked the conceptual resources to evaluate Copernican astronomy. The limitation is in the framework, not in the individuals who applied it.

References

1. National Health and Medical Research Council (NHMRC). NHMRC Information Paper: Evidence on the Effectiveness of Homeopathy for Treating Health Conditions. Canberra: NHMRC; 2015.
2. Homeopathy Research Institute (HRI). Critique of the 2015 NHMRC Report on Homeopathy. Homeopathy Research Institute; 2019.
3. Rutten ALB, Stolper CF. The 2005 meta-analysis of homeopathy: the importance of post-publication data. Homeopathy. 2008;97(4):169--177. doi:10.1016/j.homp.2008.09.008.
4. Australian Senate Community Affairs Legislation Committee. Estimates Hearings, Budget Estimates 2016 (evidence relating to the NHMRC homeopathy review process). Commonwealth of Australia; 2016.
5. Commonwealth Ombudsman. Investigation into the NHMRC's Assessment of Homeopathy. Commonwealth of Australia; 2017.
6. Bornhoft G, Matthiessen PF, editors. Homeopathy in Healthcare: Effectiveness, Appropriateness, Safety, Costs. Berlin: Springer; 2011.
7. Central Council for Research in Homoeopathy (CCRH). CCRH Standard Treatment Guidelines. Ministry of AYUSH, Government of India.
8. Mathie RT, Lloyd SM, Legg LA, et al. Randomised placebo-controlled trials of individualised homeopathic treatment: systematic review and meta-analysis. Systematic Reviews. 2014;3:142. doi:10.1186/2046-4053-3-142.
9. Mathie RT, Ramparsad N, Legg LA, et al. Randomised, double-blind, placebo-controlled trials of non-individualised homeopathic treatment: systematic review and meta-analysis. Systematic Reviews. 2017;6:63.
10. Linde K, Clausius N, Ramirez G, et al. Are the clinical effects of homoeopathy placebo effects? A meta-analysis of placebo-controlled trials. The Lancet. 1997;350(9081):834--843. doi:10.1016/S0140-6736(97)02293-9.
11. Shang A, Huwiler-Muntener K, Nartey L, et al. Are the clinical effects of homoeopathy placebo effects? Comparative study of placebo-controlled trials of homoeopathy and allopathy. The Lancet. 2005;366(9487):726--732. doi:10.1016/S0140-6736(05)67177-2.
12. Ludtke R, Rutten ALB. The conclusions on the effectiveness of homeopathy highly depend on the set of analyzed trials. Journal of Clinical Epidemiology. 2008;61(12):1197--1204. doi:10.1016/j.jclinepi.2008.06.015.
13. Feyerabend P. Against Method: Outline of an Anarchistic Theory of Knowledge. 3rd ed. London: Verso; 1993.
14. Kuhn TS. The Structure of Scientific Revolutions. 2nd ed. Chicago: University of Chicago Press; 1970.
15. Hahnemann S. Organon of Medicine. 6th ed. O'Reilly critical edition. Redmond, WA: Birdcage Books; 1996.