LEE J:
1 Listed today was an interlocutory application filed by the applicant, Mr McNickle, whereby permission was sought to communicate with an independent barrister who has convened a number of expert conclaves in this matter for the purpose of requesting that the independent barrister facilitate the provision of a data transfer agreement, and the data underlying two epidemiology studies.
2 Although the interlocutory application was not moved upon today, it has been necessary to resolve the substantive issue raised by that application, which involves a number of novel issues.
3 There is no need for me to explain the issues in this proceeding in any detail. There have been several judgments of this Court that set out what might be described as the superstructure of the proceeding and the matters requiring ultimate resolution: McNickle v Huntsman Chemical Company Australia Pty Ltd (Hearing Vacation) [2022] FCA 133; McNickle v Huntsman Chemical Company Australia Pty Ltd (Assessors) [2021] FCA 780; (2021) 285 FCR 244; McNickle v Huntsman Chemical Company Australia Pty Ltd (Expert Evidence) [2021] FCA 370.
4 In short, it calls for the Court to receive complicated scientific and expert evidence. To assist in this process, ten joint expert conclaves were ordered in respect of several scientific disciplines. Relevantly, these conclaves included what was described, somewhat clumsily, as the "Epidemiological Biostatistics (The Correct Allocation of Person Years in the Cohort Studies which have been conducted in a Glyphosate) Conclave", which was conducted between Professor Ian Gordon, an expert retained by the applicant, and Dr Shelley Harris, an expert retained by the respondents (Conclave D).
5 The first (and only) question asked in Conclave D related to the methodologies and conclusions in two epidemiology studies arising from the Agricultural Health Study (AHS):
1. How reliable are the statistical methodologies used and the conclusions drawn in:
(a) G Andreotti, S Koutros, JN Hofmann, et al., Glyphosate use and cancer incidence in the Agricultural Health Study J Natl Cancer Inst, 110 (2018), pp. 509-516 [Andreotti]; and
(b) AJ De Roos, A Blair, JA Rusiecki, et al., Cancer incidence among glyphosate-exposed pesticide applicators in the Agricultural Health Study, Environ Health Perspect, 113 (2005), pp. 49-54 [De Roos].
6 It is unnecessary to go into the background of why this is a relevant subject of the report, but both parties consider it necessary for this material to be put before the Court. Accordingly, following the conclave, a joint report was finalised on 24 November 2022, and provided to the Court and the parties.
7 What emerged following Conclave D was that the main disagreement between the experts in analysing both De Roos and Andreotti relates to what is described as the "the correct allocation of person years and consequent calculation of rate ratios." It is Professor Gordon's opinion that the allocation of person years to levels of cumulative exposure to glyphosate was incorrectly carried out, and led to incorrect calculations of rate ratios. Dr Harris' opinion, however, was that the allocation of person years was correctly carried out, and led to accurate calculations of rate ratios.
8 Hence, the parties are at issue as to what Professor Gordon describes as a "serious analytic error". The consequence of this error is said to be that it could result in a "downward bias in the risk of NHL in the highest category of glyphosate exposure". It is said by Dr Harris that Professor Gordon has provided "no direct evidence of this error", and that this belief could only be substantiated "if the underlying AHS data were obtained and reanalysed".
9 No doubt I am missing something, but it seems to me it is a tad curious that the applicant, Mr McNickle, is the party seeking to obtain the AHS data. It is the respondents who seek to deploy opinions contained in articles from both Dr Roos and Andreotti in their case. If there is some prima facie, rational basis for suggesting that this material might have a serious analytical error, which can only be ascertained upon the checking of the underlying data, the failure to produce the data by the party seeking to adduce and rely upon the evidence, might have been relied upon to ground an application by the other party for rejection, or discretionary exclusion, or limitation of the material or, alternatively, support submissions going to the weight of the representations contained in the articles. But as it happens, both parties have reached a consensus that it is appropriate that the AHS data be obtained and reanalysed, and further questions be put to the experts along the following lines:
1. Having regard to the data provided under the Data Transfer Agreement, do you agree or disagree that the error described and summarised in paragraph [173] of Professor Gordon's report of 29 September 2021 as having been made in:
(a) AJ De Roos, A Blair, JA Rusiecki, et al., Cancer incidence among glyphosate exposed pesticide applicators in the Agricultural Health Study, Environ Health Perspect, 113 (2005), pp. 49-54. (De Roos 2005); and
(b) G Andreotti, S Koutros, JN Hofmann, et al., Glyphosate (a) use and cancer incidence in the Agricultural Health Study J Natl Cancer Inst, 110 (2018), pp. 509-516 (Andreotti 2018),
was made or not?
2. Having regard to your answer to question 1, how reliable are the statistical methodologies used and the conclusions drawn in:
(a) De Roos 2005; and
(b) Andreotti 2018?
10 So far, so good. One would have thought this would be a relatively straightforward process of obtaining the data, providing it to the experts and obtaining their further opinions on it. But like so many matters in this case, it is not quite as simple as that. The AHS data is maintained by the National Cancer Institute (NCI), a United States government body that is part of the Department of Health and Human Services, being part of the executive government of the United States of America. For well over a year, efforts have been made by the solicitors for the applicant to obtain a copy of this data.
11 Remarkably, since August 2021, around 60 communications have been exchanged concerning the appropriate facilitation of the data to experts. I had indicated at a case management hearing in September 2022 that I was content for the parties to continue to make efforts to obtain material to be provided to the facilitator in the same way as other documents in order for the experts, if they thought the material relevant, to express their opinions as to the questions posed to them. What has become apparent, however, following communications by the solicitors for both parties and representatives of the NCI, is that the position of the NCI is as follows (as explained in the Affidavit of Mr Lee Taylor sworn 7 November 2022 (at [18])):
(a) The NCI explained that the raw data are not contained in simple spreadsheets that are legible or reviewable by non-specialists. Rather, the data - which comprise the records of over 50,000 participants in the AHS - are raw text which can only be analysed within statistical software platforms. The NCI maintains the data in a software suite known as SAS (Statistical Analysis System) which, I understand, was developed in part by the National Institute of Health, of which the NCI is a part. Dr Beane-Freeman explained that they are able to export the data in a form that can be used in comparable analytic platforms (such as MATLAB).
(b) Because the data are in the form described above, there would be no utility to providing the AHS Data in its raw form to the parties' lawyers or indeed to the Court. The data cannot be interpreted or "interrogated" by non-specialists without appropriate statistical qualifications, training, and experience, as well as the necessary statistical analysis software. To the extent, therefore, that the parties' lawyers need to challenge any expert's analysis of the AHS Data, they will likely need to rely on their own expert(s) (who will have the data).
(c) The NCI's policy and practice of providing AHS Data to expert researchers is critical to their overarching mission as a public health body. The AHS Data comprise personal medical information of rural farmers enrolled in the AHS, and if the NCI is unable to guarantee that its subjects' privacy is fully protected, the integrity of the AHS (and similar projects) would be fundamentally at risk. That imperative is served by providing data only to experts in relevant fields who are properly trained in managing and using health data, and who are experienced in deploying that data in a manner consistent with the privacy requirements (such as in publications using the data), and whose institutions typically use such data and therefore have appropriate systems in place to manage and protect the data.
(d) The NCI is aware that the AHS Data are sought in connection with this proceeding and that they would be used by expert witnesses.
(e) The Proposed DTA does not restrict the experts or the parties from using the experts' analyses of the data (as distinct from the raw data themselves) in open court, nor would it preclude those analyses being incorporated into the Court's analysis.
(f) Under the Proposed DTA, the lawyers and/or the Court can request the experts to generate data outputs from the raw AHS Data. There is no restriction on how many or what type of outputs can be generated by the experts, other than that those outputs cannot be based on fewer than five individual study participants (of over 50,000). This is to ensure that study participants' privacy is protected. A data output is generated by aggregating the raw data of individual AHS participants to allow this data to be analysed and studied under different circumstances. Data outputs would reflect the results of particular analytic queries or calculations performed on the raw data; for example, the number of AHS participants were exposed to glyphosate in a particular year, might be extracted and then rendered in a form that would be intelligible to laypersons, such as a graph or chart, a table, or a textual description. Similarly, more complex outputs might be generated such as a re-calculation of participants' intensity-weighted cumulative lifetime days exposure (which is a focus of Professor Gordon's analysis), which would also be rendered in a form intelligible to the parties' legal representatives and the Court.
(g) These outputs will enable the data to be digested and analysed so that the parties' lawyers and the Court can properly understand what the data show. These outputs from the raw data can be used by the parties' lawyers or the Court to test the experts' opinions, methodology and analyses. The outputs from the raw AHS Data can be referred to in open court.
(h) The lawyers and the Court would then be able to ask the experts questions regarding their analyses, opinions and methodology, including based upon any outputs generated by the experts from the raw AHS Data.
12 This seems to me to give rise to an important issue. It is uncontroversial that learned articles published in scientific journals can be admissible. In the recent decision of Karpik v Carnival plc (The Ruby Princess) (Evidential Ruling) [2022] FCA 1318, Stewart J set out the general principles as follows (at [1]-[7]):
1. An issue has arisen with regard to the admissibility of certain learned articles published in scientific journals during the course of the trial of this class action. The issue has arisen after all witnesses have completed their evidence and been excused, and there is no intention to apply to recall any witness.
2. There is a substantial volume of medical-related expert evidence in the trial. Amongst other issues, the evidence is directed at questions such as the incubation period of the SARS-CoV-2 virus, the prevalence of asymptomatic infections of the virus, the modes of transmission of the virus including what the state of scientific knowledge was in that regard in early March 2020, the extent of asymptomatic and pre-symptomatic transmission of the virus, the contagiousness of the virus including with reference to the basic reproductive number (R0) and the secondary attack rate (SAR), and the significance of a negative SARS-CoV-2 PCR test followed some time later by serology (blood) tests positive for SARS-CoV-2 antibodies. As might be expected, the expert reports are replete with references to learned articles published in scientific journals on which the expert opinions are based.
3. It is uncontroversial that those articles are admissible in evidence. As it was explained in Bodney v Bennell [2008] FCAFC 63; 167 FCR 84 at [93] by Finn, Sundberg and Mansfield JJ, there is nothing in the Evidence Act 1995 (Cth) that displaces the body of common law that provides that experts are entitled to rely upon reputable articles, publications and material produced by others in the area in which they have expertise as a basis for their opinions. Experts may not only base their opinions on such sources, but may give evidence of fact which is based on them. They may do this although the data on which they base their opinion or evidence of fact will usually be hearsay information, in the sense that they rely for such data not on their own knowledge but on the knowledge of someone else. That statement was approved in Dasreef Pty Ltd v Hawchar [2011] HCA 21; 243 CLR 588 at [110] by Heydon J.
4. Significant in the body of law referred to is Borowski v Quayle [1966] VR 382 at 386-387 where Gowans J quoted the following passage from Wigmore on Evidence (3rd ed) Vol 2, pp 784-785 para 665(b):
The data of every science are enormous in scope and variety. No one professional man can know from personal observation more than a minute fraction of the data which he must every day treat as working truths. Hence a reliance on the reported data of fellow-scientists, learned by perusing their reports in books and journals. The law must and does accept this kind of knowledge from scientific men. On the one hand, a mere layman, who comes to court and alleges a fact which he has learned only by reading a medical or a mathematical book, cannot be heard. But, on the other hand, to reject a professional physician or mathematician because the fact or some facts to which he testifies are known to him only upon the authority of others would be to ignore the accepted methods of professional work and to insist on finical and impossible standards. Yet it is not easy to express in usable form that element of professional competency which distinguishes the latter case from the former. In general, the considerations which define the latter are (a) a professional experience, giving the witness a knowledge of the trustworthy authorities and the proper source of information, (b) an extent of personal observation on the general subject, enabling him to estimate the general plausibility, or probability of soundness, of the views expressed, and (c) the impossibility of obtaining information on the particular technical detail except through reported data in part or entirely.
5. It can thus be seen that expert reliance on such learned publications is not excluded by the rule against hearsay (s 59) or the opinion rule (s 76).
6. There are two categories of learned publications the admissibility of which the parties dispute. Common to both categories is that the articles are not referred to in any of the expert reports that have been tendered, confirmed on oath or affirmation and that have been the subject of cross-examination. In the first category are two articles sought to be tendered by the applicant which were put to one of the respondents' expert witnesses, Professor Catherine Bennett, in cross-examination. The four articles in the second category, which are also sought to be tendered by the applicant, were not referred to in the evidence of any of the experts.
7. The objection taken to both categories of articles is that admission of them would be contrary to the opinion rule in Pt 3.3 of the Evidence Act read together with Div 23.2 of the Federal Court Rules 2011, which deals with parties' expert witnesses and expert reports, and the Court's Expert Evidence Practice Note (GPN-EXPT). The short point that is made is that even if it were to be accepted that the opinions expressed in the articles were wholly or substantially based on the authors' specialised knowledge based on their training, study or experience (as required by s 79(1) of the Evidence Act), the requirements of the Court's rules and practice with regard to the admission and testing of expert reports have not been satisfied in relation to the articles.
13 Here, of course, we are not only dealing with the proposed tender of articles per se, but also with a different category of proposed opinion evidence, that is, two experts giving evidence on the reliability of other representations made in the learned articles. The experts propose to express their views as to the reliability of those learned articles by the application of their specialised knowledge. Again, for reasons not presently entirely clear to me, it is evident from the submissions put to me that this evidence is proposed to be adduced at trial and not on the voir dire (in support of an application by the applicant to exclude the tender of the articles).
14 Of course, in giving this evidence at a hearing, the starting point is s 76(1) of the Evidence Act 1995 (Cth) (Evidence Act): "Evidence of an opinion is not admissible to prove the existence of a fact about the existence of which the opinion was expressed." One of the several exceptions to the opinion rule is contained in s 76(1), and relates to opinions based on specialised knowledge. The term "opinion" is undefined in the Evidence Act, but it has come to be defined as "an inference from observed and communicable data": see Allstate Life Insurance Co v Australia & New Zealand Banking Group Ltd (No 5) (Allstate Judgment No 32) (1996) 64 FCR 73 (at 75 per Lindgren J).
15 Two conditions must be met for an opinion to be admissible under s 79. First, the witness must have "specialised knowledge based on a person's training, knowledge and experience"; and secondly, the opinion must be "wholly or substantially based on that knowledge". It is the second condition that is relevant here because, although it is sufficient if the opinion is substantially based on specialised knowledge based on training, study and experience, the opinion must be presented in such a way that it is possible to determine its basis: see Honeysett v The Queen [2014] HCA 29; (2014) 253 CLR 122. As J D Heydon KC explained in his article "Evidence of Forensic Scientific Opinion and the Rules of Admissibility" (2015) 36 Adelaide Law Review 101 (at 104):
There does remain quite a sharp division on the question of whether it is a precondition to the admissibility of expert evidence that there be evidence capable of supporting the assumptions of primary fact - evidence from either the expert or other sources, which either has been tendered or will be before the close of the tendering party's case. Decisions in State courts favour the strict view. Decisions of the Federal Court of Australia favour the laxer view.
16 Without commenting on the author's characterisation of the decisions of various courts, it is fair to say the question of how one approaches the perceived necessity there be evidence capable of supporting all the assumptions made by an expert prior to admitting opinion evidence has been somewhat controversial. For present purposes, however, it is sufficient to have regard to what the High Court said in Dasreef Pty Ltd v Hawchar [2011] HCA 21; (2011) 243 CLR 588 (at 605 [41]-[42] per French CJ, Gummow, Hayne, Crennan, Kiefel and Bell JJ), where it was explained that the real issue on the question of admissibility is being able to ascertain whether the opinion expressed is wholly or substantially founded on specialised knowledge based on training, study or experience and:
41. … [w]hat has been called the basis rule is a rule directed to the facts of the particular case about which an expert is asked to proffer an opinion and the facts upon which the expert relies to form the opinion expressed. The point which is now made is a point about connecting the opinion expressed by a witness with the witness's specialised knowledge based on training, study or experience.
42. A failure to demonstrate that an opinion expressed by a witness is based on the witness's specialised knowledge based on training, study or experience is a matter that goes to the admissibility of the evidence, not its weight. …
17 Hence, opinion evidence is generally admissible even if proof of the factual basis for that opinion is controversial and issues relating to the factual basis cannot be resolved until the end of the trial. Put another way, opinion evidence is generally admissible if there is evidence which, if accepted, can establish the truth of the assumptions: Rhoden v Wingate [2002] NSWCA 165; (2002) 36 MVR 499 (at 525 [86] per Heydon JA).
18 Accordingly, my focus at trial in assessing whether this additional proposed expert material will be admissible is, in accordance with s 140(2) of the Evidence Act, whether it is established on the balance of probabilities that the opinions expressed in relation to these articles are wholly or substantially based on the expert's knowledge. Put simply, this means it will be necessary for the opinions to be presented in a form that makes it possible to determine whether they are based on the knowledge of the experts. The primary duty of the experts will be to identify the facts and reasoning process that justify the opinion expressed. This will allow me to evaluate the opinion expressed on the balance of probabilities.
19 But the present case will be one where, as I understand it, if the data is provided only to the experts, they will then be able to analyse it and then express the conclusions they have drawn from the data. Accordingly, this does not seem to be a case where there will be an insuperable admissibility problem by reason of the Court not having access to the underlying data. Even if this were to be the case, s 190(1) of the Evidence Act allows the Court, if the parties consent, to make an order dispensing with the application of the provisions of Pt 3.3, which deals with opinion evidence.
20 The problem here is not one of the parties' making. They have done their best to obtain the underlying material. It seems to me that given the parties have indicated they would consent to a waiver in respect of this material, then provided the supplementary report is prepared in a way that discloses the basis of the opinions expressed, there should be no difficulty with its admissibility. Accordingly, in these circumstances, I was prepared to make the orders proposed.
21 I should add two further points. The first is, for the purposes of analysing the AHS data, it will be necessary, as I understand it, for the experts to generate outputs of the raw data, which will be available to the Court. Although this material is in itself an opinion, it appears there will be sufficient material to enable the parties and the Court to test or interrogate the analyses and opinions of the experts effectively. Moreover, obtaining the outputs of the raw data by experts and putting it into a comprehensible form will assist the Court and the legal practitioners to understand the underlying data in a meaningful way.
22 The second is that, as noted above, the AHS data comprises highly confidential, personal medical information of rural farmers enrolled in the AHS, and the view taken by the NCA is that if it is unable to guarantee that the participant's privacy is protected fully, the integrity of the AHS (and similar projects) would be impaired.
23 Against this background, it is understandable that the NCI is solicitous as to the confidentiality of the material. This leads me in turn to consider the data transfer agreement proposed by the NCI. Ordinarily, it would be a very odd exercise of judicial power to require third party experts to enter into an agreement with another stranger to the proceeding for the purpose of providing the experts with the material needed by the experts for them to give evidence. Usually an order requiring production would issue, but the necessary material is located outside of the Court's jurisdiction and there is no practicable means by which it may be obtained through a compulsory process.
24 In this regard, I did raise with the parties during an earlier case management hearing the possibility of making an application for civil discovery in aid of a foreign proceeding under 28 USC s 1782 of the Federal Rules of Civil Procedure (US), but it is evident that this would likely be opposed by the NCI and conditions would be placed upon the provision of the material. Even if such an order were granted, I am satisfied it is likely that confidentiality restrictions would be placed on the provision of the material by the foreign Court in any event.
25 The solicitors for the parties have had experience in dealing with the NCI and hopefully a form of data transfer agreement can be provided to allow the data to be provided to the two experts participating in the conclave. It does not seem to me that would cause any difficulty for the independent barrister in accessing the material, as he presumably would suffer the same difficulties in understanding it as everyone else in the courtroom. It would be inappropriate for me, of course, to order what amounts to a mandatory injunction requiring the experts to enter into a particular form of agreement. It is ultimately a matter for them as to whether they wish to participate in the conclave. Hopefully, however, they are willing to do so, which will serve to facilitate a sensible solution to the unusual difficulty that has arisen.
I certify that the preceding twenty-five (25) numbered paragraphs are a true copy of the Reasons for Judgment of the Honourable Justice Lee.