SUPREME COURT OF CANADA
3 FEBRUARY 2012
(with whom McLachlin C.J. and Binnie, LeBel, Fish & Charron JJ joined)
Broad rights of access to government information serve important public purposes. They help to ensure accountability and ultimately, it is hoped, to strengthen democracy. “Sunlight”, as Louis Brandeis put it so well, “is said to be the best of disinfectants” (“What Publicity Can Do”, Harper’s Weekly, December 20, 1913, p. 10).
Providing access to government information, however, also engages other public and private interests. Government, for example, collects information from third parties for regulatory purposes, information which may include trade secrets and other confidential commercial matters. Such information may be valuable to competitors and disclosing it may cause financial or other harm to the third party who had to provide it. Routine disclosure of such information might even ultimately discourage research and innovation. Thus, too single-minded a commitment to access to this sort of government information risks ignoring these interests and has the potential to inflict a lot of collateral damage. There must, therefore, be a balance between granting access to information and protecting these other interests in relation to some types of third party information.
The need for this balance is well illustrated by these appeals. They arise out of requests for information which had been provided to government by a manufacturer as part of the new drug approval process. In order to get approval to market new drugs, innovator pharmaceutical companies, such as the appellant Merck Frosst Canada Ltd. (“Merck”), are required to disclose a great deal of information to the government regulator, the respondent Health Canada, including a lot of material that they, with good reason, do not want to fall into their competitors’ hands. But competitors, like everyone else in Canada, are entitled to the disclosure of government information under the Access to Information Act, R.S.C. 1985, c. A-1 (the “Act” or “ATI”).
The Act strikes a careful balance between the sometimes competing objectives of encouraging disclosure and protecting third party interests. While the Act requires government institutions to make broad disclosure of information, it also provides exemptions from disclosure for certain types of third party information, such as trade secrets or information the disclosure of which could cause economic harm to a third party. It also provides third parties with procedural protections. These appeals concern how the balance struck by the legislation between disclosure and protection of third parties should be reflected in the interpretation and administration of that legislation.
Health Canada received access to information requests relating to certain new drug submissions made to it by Merck. A series of disputes then arose between Merck, a third party to the requests, and the Minister of Health about what information had to be disclosed and what was exempt from disclosure. An avalanche of paperwork and court proceedings ensued. No fewer than five proceedings before the Federal Courts, generating a record of some 67 bound volumes of material, have brought the parties to this Court. At issue are the interpretation and application of several provisions of the Act that govern the disclosure or non-disclosure of third party confidential commercial information.
Merck says that the balance has swung too far in favour of disclosure, both in the way the Act was administered by Health Canada and in the way it was interpreted by the Federal Court of Appeal. Merck has three main complaints. First, it says that Health Canada failed to give it notice and an opportunity to make objections before disclosing some of its confidential information. This complaint raises issues about the threshold under the Act for giving third parties notice before disclosing their information. Second, Merck says that Health Canada failed to conduct an adequate review of the information before making its initial decision that the information was subject to disclosure. The effect of this, Merck claims, is that Health Canada effectively shifted its statutory obligations onto it, resulting in Merck having to expend extensive human and financial resources to deal with the access to information requests. In short, the process itself inflicted undue commercial injury. This point requires analysis of the nature of the government institution’s duties under the Act and the role of the third party when it claims exemption for the information sought. Third, Merck contends that both Health Canada and the Federal Court of Appeal held it to too onerous a standard of proof that the information was exempt. This contention requires an examination of the burden and standard of proof on a third party claiming exemptions from disclosure.
In addition to these main points, Merck also submits that the Federal Court of Appeal applied the wrong standard of appellate review and misapplied the provisions relating to the disclosure of information that can be reasonably severed from exempt material in the same record.
Although my view is that Merck’s appeals should be dismissed, there is nonetheless some merit to its complaints. I will take the opportunity the case provides to set out my understanding of when notice must be given to a third party, what the role of the government institution is in applying the third party exemptions and what are the applicable standards and burdens of proof in relation to them. I will address the standard of review on appeal and how the severance provisions should be applied. Finally I will deal with the specific rulings about the numerous pages of information still in contention. The main challenge of the appeals is to determine how to interpret and apply the Act so that it strikes the balance Parliament intended between broad rights of access and protection of third party information.
A good deal of background is required in order to understand the precise issues before the Court, which I will provide in the following section.
II. Facts, Proceedings and Issues
The case arises out of two access to information requests made with respect to information submitted by Merck to Health Canada in the course of seeking approval to market two products.
Merck applied to obtain approval to market Singulair®, an asthma medication, by filing a New Drug Submission (“NDS”) in early 1997. To obtain Health Canada’s approval, Merck had to make full and frank disclosure of all of its knowledge and information about the drug. Approval was granted approximately a year and a half later and, as a result, the drug was marketed and sold in Canada. In 1999, Merck applied for approval of Singulair® in a 4‑mg dose that would extend the permitted indications for the drug to patients two to five years of age. This required the submission of a Supplementary New Drug Submission (“SNDS”). A SNDS is submitted to request the authorization to market a drug that has already been approved and for which certain changes have been made, for instance and as in this case, proposing a new dosage. This process of approval, as with a NDS, required Merck to submit a great deal of information. The new dosage was approved and the drug marketed.
In due course, Health Canada received access to information requests relating to both Merck’s NDS and SNDS. With respect to the NDS, the requester sought access to the Notice of Compliance, the Comprehensive Summary, the Health Canada reviewers’ notes, and the correspondence between Health Canada and Merck Frosst. With respect to the SNDS, the requester asked for all releasable records.
As we shall see, these access to information requests led to lengthy exchanges between Merck and Health Canada about how Health Canada was processing them and what documents were or were not subject to disclosure, leading ultimately to extensive court proceedings.
These appeals engage two quite complex legislative and regulatory schemes, one relating to new drug approval and the other to access to information. I will, therefore, briefly outline these schemes. I will then set out a brief account of how Health Canada addressed the access to information requests, a brief summary of the ensuing court proceedings in the Federal Courts leading to the appeals to this Court and a statement of the precise issues that must be resolved.
A. The New Drug Approval Process
To seek approval to market a new drug in Canada, Merck was required to file a NDS which must comply with the Food and Drug Regulations, C.R.C., c. 870, s. C.08.002. This submission is a comprehensive disclosure of all of Merck’s information on the new drug. Amongst other things, it must submit a list of ingredients, the details of the methods of manufacture, details of the tests to be applied to control the potency, purity, stability and safety of the new drug, and detailed reports of the tests made to establish safety. Some of this information is made public upon approval of the new drug. Merck was also required to submit a statement of all representations to be made for the promotion of the new drug respecting the administration of the proposed dosage, the claims to be made and the contra-indication and side effects of the new drug.
Health Canada has issued quite detailed guidelines for the preparation of new drug submissions. The submission is to be in five main parts:
Part 1 – Master Volume;
Part 2 – Chemistry and Manufacturing, which sets out detailed information about the drug substance;
Part 3 – Comprehensive Summary, which sets out investigational studies relating to pharmacology, toxicology, microbiology, published and unpublished investigational articles, clinical studies and research and development of the drug. The comprehensive summary is the heart of the NDS, consisting of factual, concise descriptions of the methodology, results, conclusions and evaluations of the relevant investigational animal and clinical human studies;
Part 4 – Sectional Reports detailing investigational and clinical studies; and
Part 5 – Raw data from preclinical and clinical studies.
(Therapeutic Products Programme Guidelines – Preparation of Human New Drug Submissions (1991))
Once submitted, Health Canada reviews and evaluates this information. This produces what is referred to in the record as “reviewers’ notes”. During the review process, the reviewers of course comment on the information provided and frequently pose questions and seek additional information from the manufacturer. These requests, along with other communications passing between Health Canada and the manufacturer constitute what has been referred to in the record as correspondence. Before this Court, information in three types of documents is at issue: the comprehensive summary, the reviewers’ notes and the correspondence.
When all this information has been reviewed by Health Canada, a publicly available Product Monograph will be approved. This is a scientific document which contains the information for safe and effective use of the drug. It is based on data summarized in the comprehensive summary and is drafted and redrafted as Health Canada and the manufacturer discuss the product and exchange information. The final Product Monograph may not include all of the information exchanged between the parties. Rather, it is the result of discussions and compromise between them. It is published as part of the Notice of Compliance issued by Health Canada.
A SNDS follows a similar process.
B. Access to Information Legislation and Process
It is useful now to turn to a brief review of the legislative provisions that governed Health Canada’s response to the access to information requests relating to Merck’s NDS and SNDS. I have set out the most relevant provisions of the Act in the Appendix to these reasons.
The purpose of the Act is to provide a right of access to information in records under the control of a government institution. The Act has three guiding principles: first, that government information should be available to the public; second, that necessary exceptions to the right of access should be limited and specific; and third, that decisions on the disclosure of government information should be reviewed independently of government (s. 2(1)).
In Dagg v Canada (Minister of Finance),  2 S.C.R. 403, at para. 61, La Forest J. (dissenting but not on this point) underlined that the overarching purpose of the Act is to facilitate democracy and that it does this in two related ways: by helping to ensure that citizens have the information required to participate meaningfully in the democratic process and that politicians and officials may be held meaningfully to account to the public. This purpose was reiterated by the Court very recently, in the context of Ontario’s access to information legislation, in Ontario (Public Safety and Security) v Criminal Lawyers’ Association, 2010 SCC 23,  1 S.C.R. 815. The Court noted, at para. 1, that access to information legislation “can increase transparency in government, contribute to an informed public, and enhance an open and democratic society”. Thus, access to information legislation is intended to facilitate one of the foundations of our society, democracy. The legislation must be given a broad and purposive interpretation, and due account must be taken of s. 4(1), that the Act is to apply notwithstanding the provision of any other Act of Parliament: Canada Post Corp. v Canada (Minister of Public Works),  2 F.C. 110, at p. 128; Canada (Privacy Commissioner) v Canada (Labour Relations Board),  3 F.C. 609, at para. 49, aff’d (2000), 25 Admin. L.R. (3d) 305.
Nonetheless, when the information at stake is third party, confidential commercial and related information, the important goal of broad disclosure must be balanced with the legitimate private interests of third parties and the public interest in promoting innovation and development. The Act strikes this balance between the demands of openness and commercial confidentiality in two main ways. First, it affords substantive protection of the information by specifying that certain categories of third party information are exempt from disclosure. Second, it provides procedural protection. The third party whose information is being sought has the opportunity, before disclosure, to persuade the institution that exemptions to disclosure apply and to seek judicial review of the institution’s decision to release information which the third party thinks falls within the protected sphere. These appeals raise significant issues about the interpretation of the substantive protections as well as about how the procedural protections should operate.
I turn now to a brief overview of the most directly relevant provisions of the Act. Section 4 (as extended by the Access to Information Act Extension Order, No. 1, SOR/89-207) sets out the right of persons and corporations present in Canada to have, on request, “access to any record [defined to mean any documentary material regardless of medium or form] under the control of a government institution” (s. 4(1)). This right is accorded “[s]ubject to this Act” and, for present purposes, the important qualification of the right is found in s. 20. It sets out the exemptions relating to third party information. (A “third party” is defined to be a person, group of persons or organization other than the requester or a government institution (s. 3).) Subsection 20(1) provides that the government institution has a duty to refuse to disclose certain categories of third party information. The subsection, as material to these appeals, read as follows at the relevant time:
The duty not to disclose these sorts of third party information must be read with s. 25 of the Act, which may be called the severance provision. It requires the institution to disclose any part of a record that does not contain material which the institution is authorized not to disclose and which can reasonably be severed from any part that does contain exempted material. Section 25 provides:
Thus, we see that the general right of access is subject to a duty on government institutions not to disclose these types of third party information, including information that would normally be subject to disclosure, but cannot reasonably be severed from the exempted third party information. These are what I have called the substantive protections.
I turn now to the procedural protections for third parties. The Act, as noted, establishes a process of notification and judicial review. This process permits the third party to mount objections and have them considered before the information is disclosed. Section 27(1) of the Act details the circumstances in which a government institution must make every reasonable effort to give notice of its intention to disclose the third party’s information. At the time of the applications it read:
When a third party receives such a notice, it must be given the opportunity to make representations pursuant to s. 28 of the Act and the institution must then make a decision whether or not to disclose all or part of the record. Once again, the third party is given written notice of this decision and is accorded 20 days to request a review of it in the Federal Court, as provided for in s. 44. The text of ss. 28 and 44(1) is as follows:
(1) Health Canada’s Response to the Access to Information Requests
Health Canada identified about 550 pages in response to the NDS access to information request. It reviewed those pages and concluded that approximately 30 of them contained confidential information that could not be disclosed under s. 20(1) of the Act. Health Canada redacted those pages in part. It also concluded that 15 pages did not contain confidential information, with the exception of some information on one page that it redacted, and disclosed those pages without first notifying or consulting Merck.
Health Canada then notified Merck of the access to information request and of its intent to disclose part of the NDS record. It provided Merck with a copy of the over 500 still‑undisclosed pages that it sought to disclose to the requester, some of which were partially redacted. By letter dated August 16, 2000, Health Canada specified that some of those pages had already been redacted pursuant to s. 20(1) of the Act, and that others may also be subject to s. 20(1), however they were unable to determine this at the time. It sought Merck’s representations on the proposed disclosure pursuant to s. 27 of the Act. In particular, it asked Merck to explain which portions of the remaining record it considered to be confidential under s. 20(1), if any, and why. Merck responded on September 25, 2000. It took the position that, with the exception of the Product Monograph and some published studies, all of the information covered by the ATI request – including the already-disclosed pages – was exempt from disclosure under s. 20(1) of the Act.
Health Canada considered Merck’s response and redacted additional information from approximately 300 pages. Most of those pages were redacted in part, though some were withheld completely. Following these further redactions, approximately 490 pages were still at issue. On January 2, 2001, Health Canada sent Merck a second notice informing it of the additional redactions and enclosing the remaining 490 pages for Merck’s review. Health Canada informed Merck that, if Merck continued to object to the redactions, it could file a request for judicial review before the Federal Court in accordance with s. 44 of the Act. Merck filed such a request for judicial review on January 19, 2001.
With respect to the SNDS, Health Canada identified over 300 pages of information that were responsive to the access to information request. It concluded that about 60 of those pages contained confidential information that could not be disclosed under s. 20(1) of the Act. Those pages were redacted in part, or in a few cases deleted entirely. In addition, Health Canada concluded that eight pages contained no confidential information and could be disclosed to the requester directly. Health Canada disclosed those pages without advance notice to Merck.
Health Canada notified Merck of the access to information request, provided a copy of about 300 pages and solicited Merck’s submissions concerning their disclosure. Merck, as it had with respect to the NDS request, took the position that none of the pages could be disclosed, except for the Product Monograph and published studies. Health Canada replied by agreeing to some additional, partial redactions on about 45 pages and rejected the balance of Merck’s objections. Merck then sent a further reply based on a review prepared by outside consultants. The review identified as exempt from disclosure all of the information that was not already publicly available and which had not been redacted by Health Canada. In particular, the consultants identified information which Merck had requested Health Canada to withhold (i.e. everything except the Product Monograph or a published study – which were otherwise publicly available), and which was not already published on the U.S. Food and Drug Administration (“F.D.A.”) website. Merck maintained that none of this unpublished information could be disclosed. It did, however, agree to the partial disclosure of a number of pages.
In its second and final notice to Merck, Health Canada agreed to withhold additional details from about 10 more pages, but rejected the balance of Merck’s objections. Health Canada informed Merck of its right to seek judicial review before the Federal Court in accordance with s. 44 of the Act. Merck filed a request for judicial review in the SNDS file on January 8, 2002.
Merck maintained throughout the proceedings that Health Canada did not conduct a sufficiently detailed review of the documents before giving it the notices, while Health Canada maintained that Merck’s submissions did not address the exemptions it claimed specifically enough.
(2) Proceedings in the Federal Courts
The initial NDS judicial review was heard in the Federal Court before Harrington J., but the Federal Court of Appeal set aside his decision and directed a new hearing. The new hearing of that judicial review was heard in the Federal Court at the same time as the SNDS judicial review application. Both decisions were appealed to the Federal Court of Appeal. There are thus five decisions leading to the appeals now before the Court and I will briefly summarize them.
(a) First Federal Court Decision, 2004 FC 959,  1 F.C.R. 587
The first decision pertains solely to Merck’s application for judicial review in relation to the NDS disclosure. Harrington J. allowed the application in part. He was of the opinion that Health Canada could not disclose any of the NDS record without prior notice to Merck. Further, apart from the one document called the Notice of Compliance, which is a public document published upon approval of the drug, Harrington J. found that although some of the information contained in the record was available in the public domain, it was not available “as such” and therefore remained confidential and should be exempted from disclosure (paras. 53 and 58). In addition, he held that this case was not a case where severance of the confidential information was reasonable. Accordingly, he ordered that no part of the record apart from the Notice of Compliance could be disclosed as, in his view, it was exempt from disclosure pursuant to s. 20(1)(b) of the Act. He did not include in his reasons any analysis of ss. 20(1)(a) or 20(1)(c).
(b) First Appeal, 2005 FCA 215,  1 F.C.R. 379
The Minister of Health appealed and a unanimous Federal Court of Appeal overturned Harrington J.’s decision. Desjardins J.A. found that Harrington J. erred in law in his interpretation of s. 20(1)(b). The Court of Appeal decided that rather than undertaking its own analysis of the records, the interests of justice would be better served by remitting the matter to the Federal Court.
(c) Rehearing of NDS Judicial Review and SNDS Judicial Review, 2006 FC 1201(CanLII) and 2006 FC 1200, 301 F.T.R. 241
The reviewing judge, Beaudry J., heard both the rehearing relating to the NDS (2006 FC 1201) and the SNDS application (2006 FC 1200). Merck sought two remedies: a declaratory order with regard to the lawfulness of the procedure followed by Health Canada in processing the request for access to information and an order prohibiting the disclosure of the NDS and SNDS records.
Turning first to the lawfulness of the process, Merck took issue with the disclosure of some of the record without being notified and objected to the fact that Health Canada had imposed on it the onus of showing why disclosure should be refused without having conducted its own genuine and thorough review. Health Canada argued that Merck could not ask for a declaratory order regarding the decision to disclose without notice because that decision was not properly before the court. The reviewing judge disagreed and held that the court should rule on the matter because not only were the issues serious, but also it would avoid the multiplication of decisions pertaining to the same access to information request. He also held that it was unrealistic to separate the process followed by Health Canada from the substance of the final decision. He held that the disclosure of some of the record without prior notice to the third party contravened the spirit or scheme of s. 20(1) of the Act. Given the potentially irreparable harm to third parties, disclosure without prior notice should not have occurred. The reviewing judge made this finding and concluded that Merck was entitled to declaratory orders in both cases.
The reviewing judge then turned to consider disclosure of the records. By the time he heard the NDS and SNDS matters, Health Canada had agreed to further redactions via affidavits so that the number of pages in issue was reduced to approximately 235 for the NDS request and 135 pages for the SNDS request. For the NDS request, the reviewing judge found that over 170 pages were exempted from disclosure pursuant to s. 20(1), while approximately 65 pages could be disclosed. For the SNDS request, he found that almost 60 pages were exempted pursuant to s. 20(1), and that the remaining pages could be disclosed.
With respect to the NDS records, the reviewing judge found that all three sub-paragraphs of s. 20(1) were implicated. He found that some of the records were exempted from disclosure because they contained trade secrets (s. 20(1)(a)), confidential information (s. 20(1)(b)) or information that if disclosed could reasonably be expected to result in material financial loss or gain to Merck or prejudice its competitive position (s. 20(1)(c)). The reviewing judge was of the view that, where the information contained in the record is more detailed than what is available in the public domain, it may be possible to resist disclosure based on the s. 20(1)(c) exemption. He held that in several instances Health Canada had wrongly applied the severance provision in s. 25; he was of the view that the material that was not exempt could not reasonably be severed from the material that was exempt.
In the SNDS file, the reviewing judge found that some information should be exempted pursuant to s. 20(1)(b) and (c), but found no trade secrets in these records.
(d) Second and Third Appeals and Cross-appeals, 2009 FCA 166, 400 N.R. 1
For both the NDS and SNDS judgments, Health Canada appealed and Merck cross‑appealed. The Federal Court of Appeal heard the two appeals and cross-appeals concurrently and delivered one judgment for all of them. Desjardins J.A., writing for a unanimous court, found that the reviewing judge made several legal errors. The Court of Appeal allowed the appeals and dismissed the cross-appeals, holding that all of the remaining pages at issue for both the NDS and SNDS should be disclosed.
With respect to the requirement to give notice, the Court of Appeal held that the obligation only arises if a record contains, or the head of the government institution has reason to believe that it might contain, information described in s. 20(1) of the Act. Contrary to the opinion of the reviewing judge, the Court of Appeal held that disclosure of records without prior notice to the third party does not contravene the text or spirit of the Act.
With respect to the s. 20(1)(a) exemption for trade secrets, the Court of Appeal was of the opinion that the term “trade secrets” should be interpreted narrowly and that when determining whether information constitutes a trade secret, a high threshold applies. The Court of Appeal found that the reviewing judge had failed to present any analysis in support of his decision to exclude some pages based on this exemption.
With respect to the s. 20(1)(b) and (c) exemptions for confidential information, the Court of Appeal held that there must be direct and objective evidence that the information is confidential in order for either exemption to apply. Merck, in the view of the Court of Appeal, did not provide sufficient evidence to meet its “heavy” burden (para. 62). Accordingly, the reviewing judge erred in refusing to order disclosure of the requested information pursuant to s. 20(1)(b). The Court of Appeal also found that Merck’s evidence relating to s. 20(1)(c) “remain[ed] vague, speculative and silent as to specifically how and why the disclosure of the requested information would be likely to bring about the harm alleged by Merck Frosst” (para. 93; see also para. 99). Thus, the Court of Appeal held that the reviewing judge erred in fact and law when he refused to order the disclosure of information pursuant to the s. 20(1)(c) exemption.
Finally, the Court of Appeal concluded that the reviewing judge had failed in his obligation to ensure compliance with s. 25 of the Act and to explain why severance was not reasonable. The court also concluded that the reviewing judge erred in law when he substituted his own discretion for that exercised by the head of the government institution where there was no evidence that the head of the institution’s assessment was incorrect.
I will first address the general issues of principle and then turn to the issues relating to particular claims for exemption.
The general issues are these:
what is the standard of appellate review and did the Federal Court of Appeal err in this regard?
what is the threshold for triggering the institutional head’s duty to give a third party notice of the access to information request and what sort of review of the record is required of the head of the institution in deciding whether or not to give notice?
What are the applicable burden and standard of proof on a third party claiming a s. 20(1) exemption?
After considering these issues I will turn to the principles relating specifically to the s. 20(1)(a), (b) and (c) exemptions and to the severance provision in s. 25.
A. General Issues
(1) What Is the Standard of Appellate Review and Did the Federal Court of Appeal Err in This Regard?
Merck submits that the Federal Court of Appeal erred in the standard of review it applied in these cases. The Court of Appeal, it argues, intervened based on its own re-assessment of the evidence and in the absence of any reversible error on the part of the reviewing judge. The respondent, Health Canada, accepts that the Federal Court of Appeal had to apply the usual standards of appellate review but it contests Merck’s position that the Court of Appeal failed to do so.
There are no discretionary decisions by the institutional head at issue in this case. Under s. 51 of the Act, the judge on review is to determine whether “the head of a government institution is required to refuse to disclose a record” and, if so, the judge must order the head not to disclose it. It follows that when a third party, such as Merck in this case, requests a “review” under s. 44 of the Act by the Federal Court of a decision by a head of a government institution to disclose all or part of a record, the Federal Court judge is to determine whether the institutional head has correctly applied the exemptions to the records in issue: Canada (Information Commissioner) v Canada (Commissioner of the Royal Canadian Mounted Police), 2003 SCC 8,  1 S.C.R. 66, at para. 19; Canada (Information Commissioner) v Canada (Minister of National Defence), 2011 SCC 25,  2 S.C.R. 306, at para. 22. This review has sometimes been referred to as de novo assessment of whether the record is exempt from disclosure: see, e.g., Air Atonabee Ltd. v Canada (Minister of Transport) (1989), 37 Admin. L.R. 245 (F.C.), at pp. 265-66; Merck Frosst Canada & Co. v Canada (Minister of Health), 2003 FC 1422 (CanLII), at para. 3; Dagg, at para. 107. The term “de novo” may not, strictly speaking, be apt; there is, however, no disagreement in the cases that the role of the judge on review in these types of cases is to determine whether the exemptions have been applied correctly to the contested records. Sections 44, 46 and 51 are the most relevant statutory provisions governing this review.
The decision of the judge conducting a review under the Act, which will often have a significant factual component, is subject to appellate review in accordance with the principles set out in Housen v Nikolaisen, 2002 SCC 33,  2 S.C.R. 235, and Canada (Information Commissioner) v Canada (Minister of National Defence), at para. 23.
The Federal Court of Appeal correctly set out the standard of review (para 25). Did it err in applying that standard? In my view, it did not. As I will explain in more detail in my analysis of each exemption provision, the reviewing judge did not make findings of fact and failed either to state the applicable legal principles or to explain how the legal principles applied to the facts before him or, in some cases, both. Generally, he gave no indication of the legal and factual findings that took him to his conclusions. His conclusions are not explicable when the documents and the evidence are reviewed. The Court of Appeal was therefore entitled to intervene and to carry out its own assessment of whether the reviewing judge had correctly applied the exemptions to the records. It would have been open to the Court of Appeal to remit the matter to the Federal Court for reconsideration by a judge of first instance. However, in light of the fact that this had already been done once in the NDS file, my view is the Court of Appeal was right to conduct its own assessment.
The Federal Court of Appeal did not simply fault the reviewing judge for failing to provide a detailed explanation of every conclusion or for failing to make his reasoning more explicit. The Federal Court of Appeal intervened because the reviewing judge made no findings of fact in the face of conflicting evidence, and generally provided no explanation of the applicable legal principles or how or why they applied to the disputed documents. The Court of Appeal did not err in doing so.
(2) What Is the Threshold for Triggering the Institutional Head’s Duty to Give a Third Party Notice of the Access Request and What Sort of Review of the Record Is Required in Deciding to Give Notice?
I briefly reviewed the notice provisions earlier. Before disclosing certain types of third party information, the head of a government institution must make every reasonable effort to give that third party written notice of the request for disclosure, except where the third party has waived the notice requirement. Unless the third party consents to disclosure, the head must also give the third party an opportunity to make representations as to why the record or part of it should not be disclosed: ss. 27(1), 27(2) and 28.
These appeals, strictly speaking, relate to judicial review applications of the institutional head’s decisions to release information in response to two access to information requests. It follows that the focus is on the decisions to disclose. However, the parties have made extensive submissions about how the notice provisions in ss. 27 and 28 of the Act ought to be applied. In light of the importance of the issues and the fact that both parties have made extensive submissions on the notice provisions, I will address them.
There are two main issues about this notice scheme. The first relates to the threshold for triggering the head’s obligation to give notice to the third party and the second to the nature of the head’s obligation to examine the record before deciding whether or not notice is required.
(a) The Threshold for Notice Under Section 27(1)
As noted earlier, s. 27(1) of the Act specifies when the head of the government institution must make reasonable efforts to give notice to a third party. (I will simply refer to this as the notice requirement.) For convenience, the text of the provision as it read at the time of the applications is as follows:
In this case, the Health Canada head disclosed some documentation without giving notice to Merck. Merck complains that it should have been given notice before any disclosure was made. In the Federal Court, the reviewing judge (after dealing with a number of procedural arguments that are not in issue before this Court) found that this disclosure without prior notice contravened the spirit of the legislation. Since disclosure without notice could result in irreparable harm to the third party concerned, such disclosure should not have taken place (2006 FC 1201, at para. 64). The Federal Court of Appeal disagreed. It found that s. 27(1) requires notice only if the record contains or might contain information the disclosure of which is prohibited by s. 20(1). In the Court of Appeal’s view, both the object of the Act as articulated in s. 2 and the contextual and grammatical analysis of s. 27(1) favour this conclusion.
Before this Court, Merck argues that the Federal Court of Appeal’s decision has the effect of unduly limiting the scope of the s. 20(1) exemptions by narrowing the procedural right conferred on third parties by s. 27. Merck suggests that the test for giving notice and the test for actually applying the exemption must be different. To have procedural fairness in this legislative scheme, s. 27(1) must set a low threshold for notice to affected parties. Merck therefore maintains that certain categories of records, because of their nature, should automatically trigger a right to notice. In its view, NDS and SNDS records, in light of the confidentiality and competitive value of the information they contain, fall within such a category where notice is required.
In my view, the text of the statute and the considerations identified by the reviewing judge and by Merck in its submissions support a fairly low threshold to trigger the obligation to give notice. However, I do not accept Merck’s submission that there is any “automatic” right to notice with respect to certain categories of records. Such a right to automatic notice is not supported by the text or purpose of the provisions or by the jurisprudence that has interpreted them.
Following the modern approach to statutory interpretation, the words of a provision are to be read in their entire context and in their grammatical and ordinary sense, harmoniously with the scheme of the Act, the object of the Act, and the intention of Parliament: Rizzo & Rizzo Shoes Ltd. (Re),  1 S.C.R. 27, at para. 21. The grammatical and ordinary sense of s. 27(1) makes plain that notice is required only if certain conditions are met in the particular circumstances. The section does not refer to particular categories of documents but rather to particular types of information that are or may be contained in records otherwise subject to disclosure. The subsection sets out specific conditions precedent for engaging the notice requirement. As the Federal Court Trial Division put it in words that were endorsed by the Federal Court of Appeal: “The essential condition precedent to the issuance of the notice is that the respondent has reason to believe the disclosure of the record might be contrary to his obligation under section 20 not to disclose records”: Twinn v Canada (Minister of Indian Affairs and Northern Development),  3 F.C. 368, at p. 373; aff’d (1987), 80 N.R. 263 (F.C.A.). To the same effect, MacKay J. put it this way in Air Atonabee, at p. 257: “the Act does not require notice to a third party before disclosure of information relating to that party, except in the circumstances set out in [s. 27(1)].”
While this precise issue has not been decided by this Court, the approach taken in Twinn and Air Atonabee is consistent with comments on this subject by both the majority and dissenting judges in H.J. Heinz Co. of Canada Ltd. v Canada (Attorney General), 2006 SCC 13,  1 S.C.R. 441, at paras. 41 and 66.
Merck’s submission that there is always a right to notice with respect to particular categories of records is thus not supported by the grammatical and ordinary meaning of the words of s. 27(1).
Neither is Merck’s position consistent with one of the Act’s animating principles, the principle that exceptions to the right of access should be limited and specific (s. 2(1)). The creation of classes of documents as proposed by Merck which would presumptively trigger the notice requirement and be presumptively exempt from disclosure would be inconsistent with this principle.
Finally, Merck’s proposed approach is not consistent with the scheme of the Act. It makes provision for giving effect to restrictions on rights of disclosure contained in other statutes. Section 24 provides that disclosure must be refused if disclosure is restricted by any provisions set out in Schedule II of the Act. As the respondent Health Canada points out, Parliament has decided not to establish such a regime for information of the type in issue here; nothing listed in Schedule II restrains the disclosure of information submitted to the Minister with a view to approval of a new medication. There is no statutory indication that the records in issue here – NDS and SNDS records – are intended to be approached on a categorical basis.
I therefore reject Merck’s contention that the proposed disclosure of any part of a NDS or SNDS automatically triggers the duty to give notice. I turn next to the circumstances that do engage the notice requirement.
The institutional head has a general duty, subject to the other provisions of the Act, to provide access to the record requested (s. 4(1)). This is the duty that Health Canada purported to carry out when it disclosed some documents without giving notice to Merck of its intention to do so. There is also a duty not to disclose information falling within the s. 20(1) exemptions. The notice provisions relate to how the institutional head carries out that duty.
In considering a request for disclosure of third party information under the Act, the institutional head has four main possible courses of action (aside from the exercise of discretion under s. 20(6)), two of which engage the notice provisions. He or she may decide to:
disclose the requested information without notice;
refuse disclosure without notice;
form an intention to disclose severed material with notice; or
give notice because there is reason to believe that the record requested might contain exempted material.
I will review each option briefly.
I turn first to disclosure without notice. The practical realities as well as the text of the notice provision in s. 27(1) suggest a high threshold for disclosure without notice. Such disclosure is only justified in clear cases, that is where the head, reviewing all the relevant evidence before him or her, concludes that there is no reason to believe that the record might contain material referred to in s. 20(1). The institutional head cannot repent after the fact from an ill-advised decision to disclose. Disclosure without notice and any harm that might follow are irreversible. Giving notice in all but clear cases reduces the risk of irremediable harm to the third party through inappropriate disclosure. Moreover, the institutional head may not have enough information to make a correct judgment about whether the information is exempt; the input of the third party may be required in order for the institutional head’s decision to be properly informed. It is, therefore, both prudent and consistent with the text of the Act for the institutional head to disclose without notice only where the exemptions clearly cannot apply.
I turn to the second option, refusal to disclose without notice. It is important to recognize that the institutional head has a duty both to disclose non-exempt material and to refuse to disclose exempted material. Just as the institutional head must not deny access without due consideration, he or she also must give due consideration to whether access must clearly be refused. This latter point was well put by MacKay J. in SNC-Lavalin Inc. v Canada (Minister of Public Works) (1994), 79 F.T.R. 113 (T.D.). He noted, at para. 47, that the institutional head’s duty under s. 20 to refuse to disclose the information described in that section is not discharged by simply noting the possibility that the information may fall within the duty to refuse disclosure, but leaving it solely up to the third party to prove to the head’s satisfaction that it ought not to be disclosed.
Institutional heads must have some reason to believe that access cannot be refused without notice to the third party. They must apply their minds to the record in light of the known circumstances. They should be able to articulate a rational basis, emerging from this initial review, on which the exemptions from disclosure may not apply. To put it simply, institutional heads must take their duty not to disclose exempt third party information as seriously as their duty to disclose information that the Act requires to be disclosed.
That brings us to the two situations relevant to this case in which notice must be given under s. 27(1): first, when the head has reason to believe that the record might contain information described in s. 20(1); and, second, when the head proposes to disclose information severed from other information as required by s. 25. An element of each of these conditions is that the head “intends to disclose [a] record” and this phrase needs careful consideration. (I put to the side the situation in which the head proposes to use the s. 20(6) public interest override.)
I turn first to the situation in which the head “intends to disclose any record .... that the head of the institution has reason to believe might contain” exempted third party information. How, it may be asked, can the head “intend to disclose” something that he or she has reason to believe is exempt from disclosure? The answer, in my view, is that the phrase “intends to disclose” must be understood in the context of the scheme of the Act. There need not be an actual, present intention to disclose in the sense of a decision taken subject only to being talked out of it by the third party. For the purposes of s. 27(1), the institutional head “intends to disclose” a record that might contain exempt information if the head concludes that he or she cannot direct either refusal or disclosure without notice according to the principles I have just outlined.
As discussed earlier, in order to disclose third party information without giving notice, the head must have no reason to believe that the information might fall within the exemptions under s. 20(1). Conversely, in order to refuse disclosure without notice, the head must have no reason to believe that the record could be subject to disclosure. If the information does not fall within one of these clear categories, notice must be given. I would therefore interpret the phrase “intends to disclose” as referring to situations which fall between those in which the head concludes that neither disclosure nor refusal of disclosure without notice is required. In other words, the head “intends to disclose” a record “that the head .... has reason to believe might contain” exempted information unless the head concludes either
that there is no reason to believe that it might contain exempted information (in which case disclosure without notice is required) or
that he or she has no reason to believe that disclosure could be required by the Act (in which case refusal of disclosure without notice is required).
To the extent that the reasons of the Court of Appeal, at para. 34, suggest the head must have actually formed an opinion on the matter as opposed to simply having no “reason to believe”, I respectfully disagree.
The approach I propose sets quite a low threshold for the requirement of giving notice. This is not only consistent with the text of the Act, but properly reflects the balance the Act strikes between disclosure and protection of third parties.
Given the nature of the exemptions in issue – trade secrets, financial and other confidential information etc. – the third party whose information is being considered is generally in a better position than the head of the institution to identify information that falls within one of the s. 20(1) exemptions. The third party knows and understands the industry in which it participates and has an intimate knowledge of the specific information, how it has been treated and the possible harm that could come from its disclosure. As Deschamps J., writing for the majority of the Court in H.J. Heinz put it (para. 51):
The unique notice given to third parties is tied to the specific nature of the exemption. [A] government institution would not have any specific knowledge of the business or scientific dealings of a third party .... In the case of confidential business information, .... the assistance of the third party is necessary for the government institution to know how, or if, the third party treated the information as confidential. Indeed, the third party’s information management practices may be an important means of determining whether the information actually meets the definition of “confidential”. Whether the information is confidential cannot be determined without representations from the third party.
Moreover, observing a low threshold for third party notice ensures procedural fairness and reduces the risk that exempted information may be disclosed by mistake. In addition, because the giving of notice opens the way to judicial review of a decision to disclose, observing a low threshold for third party notice also accords with one of the Act’s animating principles – that decisions on the disclosure of government information should be reviewed independently of government – while also being consistent with the principles that government information should be available to the public and that necessary exceptions to the right of access should be limited and specific (s. 2(1)).
The approach to notice I have just outlined makes sense of the statutory direction to give notice when the institutional head “intends to disclose any record .... that the head of the institution has reason to believe might contain” exempt third party information. But what about the part of the provision that requires the institutional head to give notice when he or she “intends to disclose any record .... that contains” such information? How, it might be asked, could the head form an intention to disclose any record that falls within the s. 20 exemptions and which he or she therefore has a duty not to disclose? The head obviously cannot form an intention to do what the Act prohibits. In my view, there are two answers. The first lies in s. 20(6), which allows disclosure of some otherwise exempt information in the public interest. This so-called public interest override is not relevant here.
The second answer lies in the severance provision in s. 25 of the Act. As noted, s. 25 requires the institutional head to disclose information in a record that contains exempted information where the disclosable information can reasonably be severed from the exempted information. Where the institutional head proposes to take the course required by s. 25, he or she “intends to disclose any record .... or any part thereof, that contains” exempted material. It follows, in my view, that notice is required whenever the head intends to disclose a record containing third party information by severing the non-exempt information and disclosing only that as required by s. 25.
In this case, a document disclosed by the head to the requester without notice to Merck contained severed information pursuant to s. 25. In my respectful view, this was not in accordance with the requirements of the Act.
To sum up my conclusions on s. 27(1):
(b) The Nature of the Review by the Head of a Government Institution
Having established the threshold for notice and for disclosure without notice, the next question is what sort of review the institutional head should conduct in order to determine whether these thresholds have been met. Merck argues that the head of the institution must conduct a “genuine and thorough” analysis of whether a s. 20(1) exemption applies before forming any intention to disclose the record and sending a notice of intent to disclose to a third party (Merck factum; para. 40). Forming an intention to disclose without such an analysis, Merck submits, would result in placing the onus on the third party to prove page by page, line by line, that the information falls within one of the s. 20(1) exemptions. Merck maintains that where, as is the case where NDS and SNDS records are involved, the head of the institution already knows that the record contains confidential information which is of value to competitors, it is unreasonable for the head of the institution to form the intention to disclose any part of the record without giving notice.
Both the reviewing judge and the Court of Appeal disagreed with Merck on this point. They concluded that the Act did not require the head of the institution to undertake a genuine and thorough examination of the record before forming the intention to disclose part or all of it: see Beaudry J., at para. 91 (2006 FC 1200); C.A. reasons, at paras. 112-14.
There are important policy and practical considerations that must be balanced in order to decide what sort of review is required of the head when deciding to give notice.
First, information should be disclosed whenever required by the Act.
Second, third party confidential commercial information must receive the protection which the Act intends for it.
Third, it is the duty of the institutional head to make the disclosure decision and respect the rights of third parties without simply shifting that responsibility onto the third party.
While the head will often require the assistance of the third party in order to reach a decision about how the Act ought to apply, the duty to decide whether to disclose or not remains with the head. The head does not discharge that duty by simply giving notice at the first sign of potentially exempted information and leaving it to the third party to do all the work. The head is not entitled to simply put the entire onus of review on the third party. Finally, the practical constraints on the head must be considered. The head may not be well informed about the subject matter of the information and may therefore be disadvantaged in assessing it. The head is also bound by the time limits under the Act; one of the responsibilities of the head is to provide timely access to the record.
In my view, the head must conduct a sufficient review of the requested material in order to decide if the threshold for notice, as I have discussed it above, has been met. The federal government’s Access to Information Policy, Chapter 1-1, published in the Treasury Board Manual at the relevant time, specified that institutions must review each individual record to determine which portions, if any, may be excluded or exempted. This statement in my view correctly describes the nature of the review required before the decision is made to give notice to the third party. The institutional head must make a serious attempt to apply the exemptions within the constraints I have noted. The same principle applies, in my view, to the head’s severance of material under s. 25. I will discuss that question more fully in the part of my reasons dealing with s. 25. However, my view is that applying s. 25 is part and parcel of the head’s initial review, subject of course to the constraints I have mentioned.
Once notice has been given, the third party has an opportunity to make its representations. At this point in the process, I am not convinced that it is useful to speak of the third party having an onus. The material filed on behalf of the respondent insisted that the third party has an onus to persuade it that the exemptions apply. Indeed, in a document entitled Access to Information Act – Third Party Information – Operational Guidelines, Health Canada describes what it requires as representations from a third party once a notice has been issued pursuant to s. 27(1). Health Canada refers to the Federal Court jurisprudence which holds that the third party bears the onus of establishing that the information falls under an exemption (p. 3 of the Guidelines). I do not think this reference is apt. There is no doubt that, once the head has given notice of a decision to disclose, the third party has the onus to show why this decision was wrong on judicial review under s. 44: see, e.g., Maislin Industries Ltd. v Minister of Industry, Trade and Commerce,  1 F.C. 939, at pp. 942-43; Canada Packers Inc. v Canada (Minister of Agriculture),  1 F.C. 47, at p. 65; Rubin v Canada (Canada Mortgage and Housing Corp.),  1 F.C. 265, at p. 276; Air Atonabee, at p. 263. However, the responsibility to decide whether disclosure is required or prohibited by the Act rests initially with the institutional head.
From the third party’s perspective, it is, of course, prudent and in accordance with common sense to be as helpful as it can be in identifying precisely why disclosure is not permitted. Nonetheless, the head must make a serious attempt, with the available information and having regard to the practical constraints, to discharge the responsibility imposed by the Act to apply the requirements to disclose or not disclose. A cooperative approach is necessary in order for the system to work. The head cannot simply shift his or her responsibility on the third party and similarly the third party must provide reasonable assistance to the head in carrying out his or her duties under the Act.
At this stage of the proceedings, there is no point in retracing the interchanges between the parties to decide if the notice provisions and the appropriate level of review were correctly applied throughout. I have already indicated that notice must be given if the record is subject to s. 25 and I have described the threshold for giving of notice and the obligations of the institutional head to review the record. However, it may be useful to observe that the impression I have is that both Merck and Health Canada at times took rather extreme positions that were not in accordance with the purpose, letter or spirit of the Act. The record suggests both that the institutional head emphasized the duty to disclose rather than the equally important duty not to disclose and that Merck was not as helpful as it could have been in making clear and targeted submissions in relation to its various objections. It is to be hoped that the clarifications that I have set out above will lead to more constructive and cooperative approaches to these issues in the future.
(3) The Burden and Standard of Proof on Judicial Review
(a) The Burden of Proof
Who bears the burden is not controversial. The third party bears the burden of showing why disclosure should not be made when it seeks judicial review (under s. 44 of the Act) of the head’s decision to disclose material which has been the subject of a notice under s. 27. This has been clear since the early case law construing the Act: see, e.g., Maislin Industries.
(b) The Standard of Proof
The applicable standard of proof is less clear. Merck argues that the Federal Court of Appeal erred in applying a heavier standard of proof than that of the balance of probabilities. For example, at para. 62, in the context of her analysis of s. 20(1)(b), Desjardins J.A. spoke of there being a “heavy” burden on the objecting party. Similarly, in relation to s. 20(1)(a), she referred, at para. 54, to a “high threshold”.
This notion of a “heavy burden” appears in many places in the jurisprudence relating to the exemptions: see, e.g., AstraZeneca Canada Inc. v Canada (Minister of Health), 2005 FC 189 (CanLII) (with supplementary reasons at 2005 FC 648 (CanLII)), at para. 52, aff’d 2006 FCA 241, 353 N.R. 84, and Canada (Information Commissioner) v Canada (Prime Minister),  1 F.C. 427 (T.D.) (“Canada v Canada”), at p. 441. However, it is important to differentiate between the standard of proof and how readily that standard may be attained in a given case. It is now settled law that there is only one civil standard of proof at common law and that standard is proof on the balance of probabilities: F.H. v McDougall, 2008 SCC 53,  3 S.C.R. 41, at para. 40. Nothing in the Act suggests that we should depart from this standard. However, as noted in McDougall, “context is all important and a judge should not be unmindful, where appropriate, of inherent probabilities or improbabilities or the seriousness of the allegations or consequences” (para. 40). Proof of risk of future harm, for example, is often not easy. Rothstein J. (then of the Federal Court) captured this point in Canada v Canada where he noted that there is a “heavy onus” on a party attempting to prove future harm while underlining that the obligation to do so requires proof on a balance of probabilities (p. 476). Therefore, I conclude that a third party must establish that the statutory exemption applies on the balance of probabilities. However, what evidence will be required to reach that standard will be affected by the nature of the proposition the third party seeks to establish and the particular context of the case.
Turning to the Court of Appeal’s reasons in the present case, I am of the opinion that they applied a higher burden than the civil standard of the balance of probabilities in relation to the s. 20(1)(a) and (b) exemptions. As noted, the court called for a “high threshold” in relation to s. 20(1)(a) (para. 54) and applied a “heavy” burden in relation to s. 20(1)(b) (para. 62). While exemptions are the exception and disclosure the general rule, with any doubt being resolved in favour of disclosure, the applicable standard of proof is still the civil standard of the balance of probabilities.
B. The Section 20(1)(a), (b), or (c) Exemptions
(1) Overview of the Exemptions
The Act contains a number of exemptions from the general rule of disclosure. The ones relevant to these appeals relate to third party confidential commercial information as set out in s. 20(1). The appeals raise a number of issues about the interpretation of these exemptions. Before turning to those issues, however, it will be helpful to put the s. 20(1) exemptions in the context of the other exemptions in the Act.
The Act sets out a series of exemptions listed from ss. 13 to 24. They may be categorized according to whether they are class- or harm-based exemptions and according to whether they are mandatory or discretionary. Where there is a class exemption, the exemption applies to all records determined to fall into that class of record. However, a harm-based exemption applies only if the specified harm or risk of harm is present. Some exemptions are mandatory: once the record has been shown to fall within the exemption, the head of the institution has no discretion and must refuse to disclose it, subject only to any applicable override, such as the one found in s. 20(6), a topic not in issue here. Other exemptions are discretionary: once there has been an initial determination that the record falls within the statutory exemption, the head has discretion as to whether or not disclosure will be refused or granted.
Turning specifically to the s. 20(1) exemptions for third party confidential commercial information, it is clear that all of these third party exemptions are mandatory: if the record falls within the exemption, the head must refuse to disclose it (putting aside the s. 20(6) public interest override):
The trade secrets (s. 20(1)(a)) and the confidential information (s. 20(1)(b)) exemptions are class-based: once information in the record corresponds to the statutory provision, that information is exempted and the head must refuse to disclose it. The s. 20(1)(c) exemption is harm-based and applies only if disclosure could reasonably be expected to result in material financial loss or gain or to prejudice the competitive position of a third party.
(2) Section 20(1)(a): the Trade Secrets Exemption
Section 20(1)(a) of the Act provides an exemption from disclosure for “any record .... that contains trade secrets of a third party”. The issues for decision here are first, how should “trade secrets” be defined and, second, did the Federal Court of Appeal err in its approach to s. 20(1)(a) in this case?
In the NDS matter, the reviewing judge found that some parts of the record contained trade secrets and that they should be exempted from disclosure. However, he did not set out any definition of the term “trade secrets” or any chain of reasoning that led him to this conclusion. In the SNDS file, he did not find any documents to which the trade secrets exemption applied.
The Court of Appeal took issue with the reviewing judge’s conclusion about the legal test for trade secrets and the sort of evidence that is required to bring a record within the exemption. It noted that the reviewing judge had failed to set out the applicable legal test or how it applied to the documents which he found to be exempt. The Court of Appeal found that the proper test was that set out in the reasons of Strayer J. (as he then was) in Société Gamma Inc. v Canada (Department of the Secretary of State) (1994), 56 C.P.R. (3d) 58 (F.C.T.D.), at para. 7, and as elaborated on by Phelan J. in AstraZeneca, at paras. 62-65. The Court of Appeal also concluded that the evidence to support such an exemption would have to meet a “high threshold” and that “[a]nyone who relies on that provision must necessarily furnish specific, objective and detailed evidence that the information constitutes a trade secret” (para. 54). Finding that Merck’s statements on trade secrets were very broad and entangled with its s. 20(1)(b) claims, the Court of Appeal concluded that Merck had not met its burden of providing objective and specific evidence that any of the records contained information that constituted a trade secret.
(b) The Definition of “Trade Secrets”
I agree with the Federal Court of Appeal that the approach to trade secrets set out by Phelan J. in AstraZeneca is correct, although I do not find it helpful to characterize this as a restrictive definition or as setting a high threshold. My reasons follow.
Under the modern approach to statutory interpretation, the interpretation of “trade secrets” must take into account the text, purpose and scheme of the legislation (Rizzo, at para. 21). Turning first to the text, there is no definition of “trade secrets” in the Act. Given that fact and that the term is a familiar legal term which has only a technical meaning, I infer that Parliament intended that the technical legal definition should apply: see Ruth Sullivan, Sullivan on the Construction of Statutes (5th ed. 2008), at pp. 57-58. However, although “trade secrets” is a technical legal term, it does not have a comprehensive definition: Roger T. Hughes, Dino P. Clarizio and Neil Armstrong, Hughes and Woodley on Patents (2nd ed. (loose-leaf)), at §. 102; Roger T. Hughes and Dino P. Clarizio, Halsbury’s Laws of Canada – Patents, Trade Secrets and Industrial Designs (2007), “Trade Secret”, at para. HPT-180.
I turn next to the broad legal context of the term as understood in the civil and common law. In Quebec civil law, two expressions are used to convey the notion of trade secret: “secret industriel” and “secret commercial”. While these are technical legal terms, as they are in the common law, they do not have comprehensive definitions. Raymond Doray and François Charette, in Accès à l’information: Loi annotée, jurisprudence, analyse et commentaires (loose-leaf), at p. II/22-4, in fact suggest that “secret industriel” is a common law notion. I would note, however, that the French-language phrase “secrets industriels”, which is the phrase used in the French version of the Act, suggests that the information referred to must relate to a technical matter capable of commercial or industrial application: see Serge Parisien, Les secrets commerciaux et la loi sur l’accès à l’information du Québec (1993), at pp. 22-25 on the meaning of “secret industriel” in the equivalent Quebec legislation. At common law, it is clear that a trade secret is a subset of confidential commercial information, but other than in the employment setting, the common law has tended not to make a clear distinction between trade secrets and the broader category of confidential commercial information: see, e.g., R. v Stewart,  1 S.C.R. 963, at pp. 974‑75; David Vaver, “Civil Liability for Taking or Using Trade Secrets in Canada” (1981), 5 Can. Bus. L. J. 253, at p. 258.
Turning next to the scheme of the Act, this distinction between trade secrets and confidential commercial information finds expression in ss. 20(1)(a) and (1)(b). The former provision provides for an exemption for trade secrets, while the latter provision provides separately for an exemption for confidential financial, commercial, scientific or technical information. This suggests that “trade secrets” in s. 20(1)(a) was intended to be a narrower concept than the more general class of confidential, financial, commercial, scientific and technical information set out in s. 20(1)(b). That a narrower ambit for trade secrets must have been intended is reinforced by the fact that the trade secrets exemption is not subject to the public interest override in s. 20(6), while the confidential information exemption in s. 20(1)(b) is subject to it. This approach also accords with the principle that exceptions to the right of access should be limited and specific (s. 2(1)). In this way, the Act’s purpose of providing broad access rights is protected.
I turn to discuss a few of the leading authorities. One often-cited case is the decision of Chevrier J. in R. I. Crain Ltd. v Ashton,  O.R. 303 (H.C.J.), aff’d  O.R. 62 (C.A.). In the context of an action against a former employee to restrain disclosure of the former employer’s trade secrets, several characteristics of a trade secret are set out. These include that it is a plan or process, tool, mechanism or compound known only to its owner and his employees to whom it is necessary to confide it and that it usually is understood to mean a secret formula or process not patented but known only to certain individuals using it in compounding some article of trade having a commercial value (pp. 308-9).
I should refer as well to two other influential decisions concerning the definition of a “trade secret”. In Société Gamma, Strayer J. considered Société’s claim that tender submissions it made in connection with its bid to obtain a contract for translation services constituted a trade secret and were therefore exempt from disclosure. He noted that there is no need to demonstrate harm in order to fall within either the trade secret or the confidential commercial information exemption and that there must be some difference between a “trade secret” and something which is merely “confidential” and supplied to a government institution. He then defined “trade secret” as follows, at pp. 62-63:
I am of the view that a trade secret must be something, probably of a technical nature,* which is guarded very closely and is of such peculiar value to the owner of the trade secret that harm to him would be presumed by its mere disclosure.
*This impression is strengthened by the French version which uses the term “secrets industriels” as the equivalent of “trade secret”.
Another influential decision is that of Phelan J. in AstraZeneca, which was a review of a decision to release records related to a NDS. Phelan J. held that Parliament’s intention was to protect genuine trade secrets based on the common law definition of the term. He cited the Société Gamma definition, but noted that the question is not whether the interpretation of “trade secrets” should be broad or narrow but whether the record falls within the legal definition of “trade secrets” (paras. 62-63). He referred with apparent approval to Health Canada’s Access to Information Act – Third Party Information – Operational Guidelines, which sets out four criteria to be met by a trade secret (para. 64). These elements are the same as in the Guidelines in evidence before us, which read:
Phelan J. concluded, at para. 65:
The type of information which could potentially fall into this class includes the chemical composition of a product and the manufacturing processes used. However, it is not every process or test which would fall into this class particularly where such process or test is common in a particular industry.
Health Canada argues that this is the appropriate definition of “trade secret”. I agree. I particularly underline Phelan J.’s comment that the point is not whether the term is to receive a “broad” or a “narrow” definition (para. 63), but rather that the term should be given its traditional legal meaning.
Phelan J.’s reasons, along with the portion of the Guidelines which he adopts, appropriately capture that traditional legal meaning. A “trade secret” for the purposes of s. 20(1) of the Act should be understood as being a plan or process, tool, mechanism or compound which possesses each of the four characteristics set out in the Guidelines which I have quoted above. This approach is consistent with the common law definition of “trade secrets” and takes account of the clear legislative intent that a trade secret is something different from the broader category of confidential commercial information which is separately and specifically protected under the Act. This approach is also consistent with the use of “secrets industriels” in the French version of the Act, as discussed above.
Merck suggests that the Security of Information Act, R.S.C. 1985, c. O-5, s. 19(4), and the NAFTA definitions of “trade secret” should colour the definition to be given to “trade secrets” under the Act. Merck, however, stops short of advancing a particular definition, submitting simply that the term must be defined in a way that is consistent with other federal statutes and Canada’s international treaty obligations.
Turning to the former point first, Merck argues that “trade secrets” in the Act should be interpreted consistently with the definition of that term in s. 19(4) of the Security of Information Act:
Merck also argues that the Act’s notion of a “trade secret” should be construed according to the NAFTA which provides, at article 1711:
(North American Free Trade Agreement, Can. T.S. 1994 No. 2)
I am not persuaded by these submissions. I am not sure there is much difference between the definition in AstraZeneca and the definition in the Security of Information Act. Moreover, we cannot simply incorporate into the Access to Information Act, which contains no definition of the term “trade secrets”, definitions adopted in different contexts. The Official Secrets Act was amended in the wake of the September 11, 2001 attacks as part of the Anti-Terrorism Act, S.C. 2001, c. 41, where it was renamed the Security of Information Act (s. 25). The definition of “trade secret” in s. 19(4) was part of that initiative. The purposes of the Access to Information Act and the Security of Information Act are significantly different. While maintaining national security is not incompatible with ensuring government accountability and democracy, it seems clear that access to information may be limited where issues of national security come into play: see ss. 15 and 16 of the Act. Therefore, it would not be appropriate to import into the access to information context the definition of “trade secret” set out under the heading Economic Espionage in the Security of Information Act.
As for the appellant’s reliance on article 1711 of the NAFTA, it does not support the conclusion for which Merck contends. Before discussing the significance of the NAFTA definition I would note that article 39 of the Agreement on Trade-Related Aspects of Intellectual Property Rights, 1869 U.N.T.S. 299 (“TRIPS”) is also potentially relevant. This article calls for the protection of secret, commercially valuable “information”. I accept, of course, that to the extent possible domestic legislation should be interpreted so that it is consistent with Canada’s international obligations: see R. v Hape, 2007 SCC 26,  2 S.C.R. 292, at para. 53; see also, e.g., Zingre v The Queen,  2 S.C.R. 392, at pp. 409-10; Ordon Estate v Grail,  3 S.C.R. 437, at para. 137; Baker v Canada (Minister of Citizenship and Immigration),  2 S.C.R. 817, at para. 70; and Schreiber v Canada (Attorney General), 2002 SCC 62,  3 S.C.R. 269, at para. 50. However, Canada is not necessarily required to adopt the treaty definition of “trade secrets” into its access to information law in order to fulfill its treaty obligations. These obligations could be fulfilled in other ways. As the respondent notes, Canada has opted to address these obligations in the pharmaceutical context by focussing on protecting parties against commercial use of their trade secrets by others. The amendments to the Food and Drug Regulations in 1995 and 2006 reflect this approach: Food and Drug Regulations, amendment, SOR/95-411, and Regulations Amending the Food and Drug Regulations (Data Protection), SOR/2006-241. It would not be appropriate to decide in this case whether this complies with Canada’s treaty obligations. This choice of approach, however, undermines Merck’s argument that Parliament intended the definition of “trade secret” in s. 20(1)(a) to mirror the NAFTA definition.
Similarly, protection of a broader class of confidential commercial information under s. 20(1)(b) of the Act and the protection of personal information under s. 19 further weaken Merck’s argument in relation to s. 20(1)(a). These provisions cover information that may be caught by the purportedly wider article 1711 of the NAFTA and article 39 of the TRIPS. This suggests that Parliament has intended to fulfill its international obligations by means that go beyond protection of “trade secrets” under s. 20(1)(a). I conclude that consideration of the NAFTA and the TRIPS does not indicate that Parliament intended a definition of “trade secrets” that is broader than the definition I have endorsed above.
The Court of Appeal, while citing AstraZeneca and the Guidelines, insisted that “trade secrets” should be interpreted restrictively and that the jurisprudential threshold for invoking this exemption is high. These comments, in my respectful view, do not reflect the correct approach to this exemption. The question on review is simply whether the party claiming the exemption has established on the balance of probabilities that the record falls within the definition I have set out above.
I agree with the Court of Appeal’s definition of “trade secrets”, but in my respectful view it placed an unduly heavy burden on Merck to establish that the definition applied as I have just outlined. By imposing the burden it did, the Court of Appeal erred in law. However, my view is that this error did not result in the Court of Appeal reaching the wrong conclusion about how s. 20(1)(a) applies here. In my respectful view, it did not err in finding that Merck’s evidence simply was not capable of establishing that the documents, which the reviewing judge found to be exempted, contained trade secrets or revealed trade secrets.
Before this Court, the parties filed a Joint Record of Pages in Issue indicating that Merck relies on the s. 20(1)(a) exemption for over 150 pages in the NDS and 9 pages in the SNDS, all of which are contested by Health Canada. However, Merck’s submissions in this Court focus exclusively on 37 pages in the NDS which the reviewing judge concluded were exempt from disclosure because they contained trade secrets: see Beaudry J., at para. 105 (2006 FC 1201). Of these, the Joint Record indicates that 7 pages are no longer in issue. Merck contends that the records in issue contain details of the specific manufacturing process used for the drug and that such information is quintessentially a trade secret. Health Canada submits that to the extent that portions of these records reveal trade secrets, such information has been redacted.
Health Canada does not seriously contest the proposition that confidential information about the specific manufacturing process used for the drug may be a “trade secret” under s. 20(1)(a), provided that the other elements of the definition I have outlined above are satisfied. The difficulty here is not about the definition of “trade secrets” or about how elevated the threshold is, but that Merck’s evidence is not responsive to the records as currently redacted. It does not explain how what remains in the records constitutes trade secrets within the meaning of the exemption.
One example will demonstrate the difficulty with Merck’s position and the reviewing judge’s decision. In its factum, Merck reproduces an excerpt from the affidavit of Robert Sarrazin (sworn June 1, 2001) which sets out Merck’s objections to the disclosure of pages 469 and 470 of the records at issue in the NDS file: see C.A. reasons, at para. 49. Merck advances this as “an example of the supporting evidence” which Merck submits shows that “it has met any evidentiary burden upon it in relation to s. 20(1)(a) of the ATIA” (factum, para. 155; see also C.A. reasons, at para. 49). The reviewing judge found that both pages were exempt from disclosure, saying simply that they contained information which constituted trade secrets: see Beaudry J., at para. 105 (2006 FC 1201). Although Merck set out this evidence in its factum, it advised after the hearing in this Court that it is no longer seeking exemption for page 470. However, the evidence in relation to these two pages illustrates the deficiencies in the reviewing judge’s decision.
In his affidavit, Mr. Sarrazin objects to disclosing pages 469 and 470 because they contain certain information which in his view constitutes trade secrets. However, in the version of pages 469 and 470 which Health Canada proposed to disclose and which was before the reviewing judge, both have been heavily redacted with the entire table on page 470 having been removed: see C.A. reasons, at paras. 49 and 51. Thus the evidence before the reviewing judge from Merck was that these pages contained trade secrets while Health Canada’s evidence was that it had agreed to delete all such content and indeed that the content of the table on page 470 had been deleted more than four years before the hearing before the reviewing judge. On this record, the reviewing judge’s conclusion that virtually blank pages constituted trade secrets is a palpable and overriding error.
As noted in Housen, the failure of a judge at first instance to discuss a relevant factor in depth, or even at all, is not itself a sufficient basis for an appellate court to reconsider the evidence (para 39). However, the judge’s failure to address relevant matters may constitute a material error justifying appellate intervention if the omission gives rise to the reasoned belief that the trial judge must have forgotten, ignored or misconceived the evidence in a way that affected his or her conclusion (Housen, at para. 39, citing Van de Perre v Edwards, 2001 SCC 60,  2 S.C.R. 1014, at para. 15). In this case, the reviewing judge ruled that an essentially blank page is a trade secret. He made this ruling without referring to either the applicable legal test or to the relevant evidence. This, in my respectful view, was a material error as discussed in Housen and justified the Federal Court of Appeal’s intervention. While it could have decided to remit the matter for further review by a judge at first instance, the fact that the record had already been the subject of two review processes in the Federal Court supports its decision to review the record and apply the exemption itself.
I agree with my colleague Deschamps J. that a reviewing judge should not be required to provide a word by word, line by line, or even page by page explanation for his or her decision and that, if appropriate, the judge can address the case on the basis of categories of records. However, there is some considerable distance between a judge not providing a word by word explanation and, as here, a judge providing no explanation at all in circumstances that give rise to a reasoned belief that the judge must have forgotten, ignored or misconceived the evidence in a way that affected his conclusion. It is well within the bounds of deference as described in Housen for an appellate court to reconsider the record in these circumstances.
I would therefore uphold the decision of the Court of Appeal which set aside the reviewing judge’s decision that these documents fall within the exemption.
(3) The Section 20(1)(b) Exemption
Section 20(1)(b) provides an exemption for a third party’s confidential financial, commercial, scientific or technical information:
The reviewing judge upheld the exemptions Merck claimed with respect to 6 documents or portions of documents in the NDS file, and 49 documents in the SNDS file of which 44 remain in dispute. In both files, the reviewing judge noted that information found in the public domain, even if in a different form, is no longer confidential and then simply stated, without referring to the evidence or offering any further explanation, that certain pages were exempt from disclosure under s. 20(1)(b) because they contained confidential information that had been treated as such by Merck and was not in the public domain: see Beaudry J., at paras. 95 and 106 (2006 FC 1201); and at paras. 103 and 113 (2006 FC 1200).
The Federal Court of Appeal set aside these findings. In the NDS file, it found that Merck’s evidence was not responsive to the most recent redacted version of the documents in issue and that Merck had not provided any direct and objective evidence bringing those documents within the s. 20(1)(b) exemption (paras. 67-77). With respect to the SNDS file, the Court of Appeal similarly found an absence of evidence from Merck that the documents in issue were confidential or treated as such. If the Court of Appeal was correct that there was no evidence to support the reviewing judge’s conclusion, appellate intervention was justified (Housen, at para. 1).
Merck asks us to conclude that all of the documents still in issue for which it claims a s. 20(1)(b) exemption should not be disclosed. Thus it seeks not only exemption with respect to those pages for which the reviewing judge upheld Merck’s objections, but also a finding that all of its other claimed s. 20(1)(b) exemptions as set out in the Joint Record of Pages in Issue should be upheld.
Merck challenges the Court of Appeal’s conclusion that there was no evidence to show the records in issue were exempt. I will address its submissions on that point at the end of this section. Three other issues also arise with respect to s. 20(1)(b). Two relate to whether the records are confidential financial, commercial, scientific or technical information. The third issue relates to what constitutes information “supplied to a government institution by a third party”. It will be helpful to explain how these issues arise before turning to a more detailed discussion of them.
In order to qualify for the exemption, the information must be
financial, commercial, scientific or technical information;
confidential and consistently treated in a confidential manner by the third party; and
supplied to a government institution by a third party.
The parties accept the factors identified by MacKay J. in Air Atonabee as being appropriate to consider the question of whether information is confidential within the meaning of s. 20(1)(b).
The first of those factors considers whether the information in the record has already been made publicly available (Air Atonabee, at p. 272). In particular, the question is whether disclosure by the government institution of a compilation of publicly available scientific articles in a NDS or SNDS to a competitor of the third party discloses more information than is already available in the public domain. Merck’s position is that the records contain confidential “financial, commercial, scientific or technical information” of two different sorts.
The first type of confidential information is the substantive content of the record, that is, the information directly relayed by its contents. An example is a record that refers to the findings of a confidential study. The confidential information is the substantive content of the study itself. This aspect of Merck’s objection was the focus of the Court of Appeal’s reasoning. I will come back to this question; for now, it is worth noting that its resolution turns mostly on the evidence about whether Health Canada’s redactions had removed this sort of confidential information.
Merck’s submission that there is a second type of confidential information depends on a more subtle argument. It contends that, quite apart from whether the substantive contents of the record are confidential, the manner in which the information is presented and the fact that Merck listed a particular study or relied on it in a particular way are both confidential scientific, or technical information within the meaning of s. 20(1)(b). I will address these points in turn.
(b) Formatting and Structure of the Submission
The Court of Appeal noted that, before it, the scientific or technical nature of the information was not contested (para. 64). In the submissions in this Court, that is also the case with respect to the substantive content of much of the information. However, the parties are divided about whether Merck’s formatting and structure of the submission are confidential commercial, financial, scientific or technical information for the purposes of the s. 20(1)(b) exemption. Merck argues that they are. Disclosure of this information, it contends, allows competitors to use Merck’s methods of assembling a NDS or SNDS but without incurring the significant research and development time and costs that Merck expended.
To address this submission, I begin by accepting three propositions that are well‑established in the Federal Court’s jurisprudence.
First, the terms “financial, commercial, scientific or technical” should be given their ordinary dictionary meanings. As MacKay J. in Air Atonabee stated, at p. 268:
… dictionary meanings provide the best guide and that it is sufficient for purposes of subs. 20(1)(b) that the information relate or pertain to matters of finance, commerce, science or technical matters as those terms are commonly understood.
Second, the case law also holds that in order to constitute financial, commercial, scientific or technical information, the information at issue need not have an inherent value, such as a client list might have, for example. The value of information ultimately “depends upon the use that may be made of it, and its market value will depend upon the market place, who may want it, and for what purposes, a value that may fluctuate widely over time” (Air Atonabee, at p. 267).
Finally, I agree that administrative details such as page and volume numbering, dates and location of information within the records are not scientific, technical, financial or commercial information (AstraZeneca, at para. 73).
In general, the same can be said about the formatting and structure of submissions such as the choice to use a graph or table to present information or the precise organization and ordering of sections of a document the general contents of which are the subject of publicly available guidelines as is the case here: see, e.g., Société Gamma, at pp. 63-64. Of course, whether or not the exemption applies must be considered in light of the nature of the information and the evidence in the particular case.
(c) Listing of Publicly Available Studies
Merck further says that the disclosure of a compilation of publicly available scientific articles in a NDS or SNDS discloses more information than is already in the public domain; the fact that an innovator pharmaceutical company has relied on specific studies to support approval of a specific drug is competitively valuable information that is not otherwise publicly available. Relying on Janssen-Ortho Inc. v Canada (Minister of Health), 2007 FCA 252, 367 N.R. 134, aff’g 2005 FC 1633 (CanLII), it submits that the fact it has chosen to rely on particular studies is confidential information which is not disclosed by the publication of the studies.
Merck submits that the Federal Court’s jurisprudence on this point is conflicting. It notes that in the first Federal Court of Appeal decision in these matters (2005 FCA 215, a decision that is not before this Court on these appeals) the court held that s. 20(1)(b) did not apply to references to publicly available research articles or other public documents in the NDS and SNDS, while in Janssen-Ortho the company’s evaluation of the studies it had relied on in its submission was found to be confidential.
I can only accept Merck’s submissions in part.
As set out earlier, information is not confidential if it is in the public domain, including being publicly available through another source. As MacKay J. put it in Air Atonabee, at p. 272, to be confidential, the information must not be available from sources otherwise accessible by the public or obtainable by observation or independent study by a member of the public acting on his or her own. It follows that information that has been published is not confidential. Moreover, information which merely reveals the existence of publicly available information cannot generally be confidential: knowledge of the existence of the information can be obtained through independent study by a member of the public. To the extent that Merck submits that its compilations of such studies are confidential for the purposes of s. 20(1)(b) because the compilations might help a competitor to learn of the existence of the studies, I do not agree.
Merck’s main point, however, is that while the content of published studies may not be confidential information for s. 20(1)(b) purposes, the fact that it relied on certain published studies is not publicly available and is confidential. Merck cites what Simpson J. said in Janssen-Ortho, at para. 39 (2005 FC 1633): although a description of information in published studies would normally be disclosed, the fact that the party opposing disclosure considers those findings accurate and trustworthy has not been publicized and therefore may fall within the realm of confidential information. The Federal Court of Appeal upheld this conclusion. Sexton J.A. writing for the court noted that the applicant’s opinions relating to public documents were confidential and these opinions were not publicly available; the fact that the applicant may have relied upon certain public information was not public knowledge (para. 6).
I do not accept Merck’s submission that Janssen-Ortho is inconsistent in principle with other jurisprudence of the Federal Courts in this area. The point is that the content of published studies will generally not be confidential because that content is available from another source in the public domain. The decision in Janssen-Ortho, as I read Simpson J.’s reasons, accepts this as a general proposition when it concludes that a description of findings in published studies would normally be disclosed. However, the third party’s reliance on or evaluation of those studies may be shown by the evidence to fall within the definition of confidential information. What took the Janssen-Ortho case out of the general proposition was that the record also disclosed that the applicant considered the findings to be accurate and trustworthy and that this information was not publicly available. Simpson J. found that such material, to the extent that it included “expert advice, opinions, conclusions and information about the studies the Applicant considered reliable” was confidential commercial information (para. 40).
In my view, therefore, the simple reference to a publicly available study or a description of its contents in a submission generally is not confidential information; a mere reference simply notes the existence of the study and a description of its contents simply summarizes information which is publicly available. Knowledge both of the existence of the study and of its contents will generally be obtainable by a member of the public, albeit with more effort, through independent study. However, much will depend on the evidence in a particular case.
I underline this last point. Once the relevant legal principles are established, whether or not a record is confidential is primarily a question of fact. Care must be taken, therefore, not to over-generalize the holdings of particular cases, by failing to give due regard to the evidence which was before the court in those cases. It may be, for example, that the relevance of a particular study to a particular line of inquiry might in some cases be shown to be confidential. Similarly, as in Janssen-Ortho, express or implicit statements of the applicant’s evaluation of the reliability of a study will generally meet the definition of confidential information. Of course, where the existence or contents of studies themselves meet the definition of confidential information in s. 20(1)(b), references to such studies will also generally be confidential for the purposes of the exemption. Similarly, if the fact that the applicant has evaluated or relied on the study is publicly available, that fact will not be confidential. The key point is that these principles are not self-applying and must be considered in light of the evidence in each case.
It seems to me that the dispute between the parties on this point turns more on a question of fact rather than on a question of legal principle. Merck’s affiants acknowledged that numerous articles pertaining to asthma and its treatment are found in the public domain. However, the assertion is that the fact that information in a particular publication was used in a Canadian NDS, in this case Merck’s, is not public knowledge. However, Merck’s affiants do not assist the Court in understanding whether the information the institutional head proposes to disclose shows Merck’s assessment or evaluation of the studies. Health Canada specifically disavows any intent to disclose the unpublished opinions or evaluations by Merck’s experts of any study. Moreover, Health Canada maintains that the fact Merck relied on the studies as listed is public information from the Product Monograph and documents published by the U.S. F.D.A.
(d) “Supplied to a Government Institution by a Third Party”
One of the requirements of the s. 20(1)(b) exemption is that the information be supplied by the third party to the government institution. There remains an issue about whether this requirement was met by certain documents in the record. The documents in issue are reviewers’ notes prepared by scientists retained by Health Canada to evaluate the drug and correspondence between Merck and Health Canada. While these records contain information supplied by Merck, they also contain other information, such as the analysis and observations of the reviewers, their conclusions and recommendations, as well as information from scientific literature. Health Canada says that following the approach set out in Canada Packers, it is obliged to release these reviewers’ notes after having redacted the material provided by Merck.
The parties join issue on what qualifies as information supplied to a government institution by a third party. Health Canada contends that while documents prepared by scientists employed or retained by it to evaluate the proposed medication contain some information that was supplied by Merck, that is not the case with all of the information contained in these sorts of records. They also contain, for example, the reviewers’ observations, analysis, conclusions and recommendations, as well as information coming from the scientific literature and the reviewers’ questions arising from all of this material. This sort of information is not supplied by Merck. Health Canada also submits that much of the correspondence that passed between it and Merck is similarly not information provided by a third party.
What, then, are the governing legal principles?
The first is that a third party claiming the s. 20(1)(b) exemption must show that the information was supplied to a government institution by the third party.
A second principle is that where government officials collect information by their own observation, as in the case of an inspection for instance, the information they obtain in that way will not be considered as having been supplied by the third party. As MacKay J. said in Air Atonabee [p. 275]:
In my view, where the record consists of the comments or observations of public inspectors based on their review of the records maintained by the third party at least in part for inspection purposes, the principle established by Can. Packers Inc., supra, applies and the information is not to be considered as provided by the third party.
See also Canada Packers, at pp. 54-55; Les viandes du Breton Inc. v Canada (Canadian Food Inspection Agency), 2006 FC 335 (CanLII), at paras. 44-49.
A third principle is that whether or not information was supplied by a third party will often be primarily a question of fact. For example, if government officials correspond with a third party regarding certain information, it is possible that the officials have prior knowledge of the information gained by their own observation or other sources. But it is also possible that they are aware of this information because it was communicated to them beforehand by the third party. The mere fact that the document in issue originates from a government official is not sufficient to bar the claim for exemption. But in each case, the third party objecting to disclosure on judicial review will have to prove that the information originated with it and that it is confidential.
To summarize, whether confidential information has been “supplied to a government institution by a third party” is a question of fact. The content rather than the form of the information must be considered: the mere fact that the information appears in a government document does not, on its own, resolve the issue. The exemption must be applied to information that reveals the confidential information supplied by the third party, as well as to that information itself. Judgments or conclusions expressed by officials based on their own observations generally cannot be said to be information supplied by a third party.
As noted earlier, Merck’s submissions on s. 20(1)(b) raise three points. In my respectful view, it fails on all of them.
Turning to Merck’s first submission, it contends that the Court of Appeal erred in finding that it had not discharged its burden of proof that the documents, as redacted, continued to contain confidential information. This argument is concerned with the substantive content of the information referred to and turns on whether, as Health Canada contends, all information that is not in the public domain has been redacted.
The Court of Appeal’s decision about the s. 20(1)(b) exemption turned on this point (paras. 62 and 67). It concluded that Merck had failed to discharge its burden with respect to whether the information was confidential and had been consistently treated as such. The court found that the affidavits submitted by Merck failed to provide “direct and objective evidence” and had not shown that the s. 20(1)(b) exemption applied.
I agree with Merck that the Court of Appeal applied an unduly onerous standard of proof. The Court of Appeal stated that the third party opposing disclosure has a “heavy” burden to establish the s. 20(1)(b) exemption (para. 62). For reasons I have set out earlier, this is an error of law. The burden is to show on the civil standard that the exemption applies. However, I do not think the result reached by the Court of Appeal turns on its description of the standard of proof. The court’s decision rested on what it concluded to be an absence of evidence responsive to the claimed exemptions in light of the extensive redactions made by Health Canada. I will explain.
With respect to the NDS, the Court of Appeal’s holding turned mainly on the way Health Canada came to concede that extensive redaction was necessary, coupled with Merck’s failure to respond specifically to the portions remaining after these final redactions had been made.
In September of 2001, Health Canada filed a new affidavit to which were attached more heavily redacted versions of documents that it proposed to release: see C.A. reasons, at para. 16. Merck did not respond specifically to the newly edited versions of the documents. The Court of Appeal found that the affidavit evidence submitted by Merck before that date was of limited use because it was impossible to tell if an objection continued to apply in light of the new editing. The Court of Appeal also found that the affidavits submitted by Merck after that date did not provide any direct and objective proof that the exemption applied to the documents in their newly redacted form (paras. 75-76). For that reason, the Court of Appeal set aside all of the reviewing judge’s conclusions in relation to the s. 20(1)(b) exemption for the NDS.
With respect to the SNDS, the Court of Appeal simply concluded that there was no direct and objective evidence from Merck to show that the information remaining in the records was confidential (para. 79).
Both of these conclusions focussed on the question of whether the substance of the information was publicly available.
Merck’s factum in this Court addresses these specific conclusions in three short paragraphs. It submits that Merck consistently treats the information in issue confidentially and that, contrary to the findings of the Court of Appeal, there was “ample evidence in the record that Merck treats the information in issue as confidential” (para. 180). As an example, it refers to evidence that access to Merck’s facilities is restricted, that its employees and consultants must sign confidentiality agreements, that access to paper and computer discs is restricted on a “need to know” basis and in particular that information pertaining to a NDS is accessible to its Regulatory Affairs personnel, and to a restricted number of the company’s officers on a “need to know” basis.
Respectfully, these submissions are of no assistance with respect to the issue that concerned the Court of Appeal; the submissions do not explain how what is left on the often heavily redacted pages is confidential in the face of Health Canada’s evidence that the unredacted material is in the public domain and therefore not confidential. Merck’s submissions and my perusal of the record have not persuaded me that the Court of Appeal erred in its conclusion on this point. The consideration of two specific examples will explain why.
In the NDS file, consider by way of example the evidence in relation to page 33 of the records in issue. That page is part of the Comprehensive Summary; specifically it is from the section setting out certain investigational studies. The page refers to studies by number, the particulars of which are listed elsewhere in the submission.
On Health Canada’s final redacted version of page 33, there are redactions made to the list of studies. On page 137, all details have also been redacted from one of the studies referred to on page 33. By way of contrast, the earlier January 2, 2001 version had no redactions from page 33, nor were there any redactions to the details of the studies on page 137: see C.A. reasons, at paras. 68-69.
Merck’s affidavit evidence relating to the substantive content of this page is to the effect that its contents reveal confidential results which are not in the public domain. Health Canada’s evidence accepts that the page in issue contains specific confidential commercial, financial, scientific and technical third party information that is properly exempted from disclosure, but maintains that all such information has been redacted and the remaining information either is not confidential by nature (such as the format of the page) or is in the public domain. Merck’s evidence does not respond to Health Canada’s evidence. Faced with the unanswered evidence that all confidential information had been redacted, and with no explanation from the reviewing judge as to why he had rejected this potentially decisive and unanswered evidence, the Federal Court of Appeal appropriately intervened and made its own assessment: see C.A. reasons, at paras. 72-76.
I turn to the s. 20(1)(b) exemptions claimed in the SNDS. Consider by way of example the evidence in relation to page 115 of the records in issue. This page is also part of the Comprehensive Summary. Throughout the process, Health Canada did not redact any part of this page. The reviewing judge found that it contained confidential information and that it was to be exempted from disclosure in its entirety (see Beaudry J., at para. 113 (2006 FC 1200)), a holding which the Court of Appeal reversed.
In its letter dated July 20, 2001, Merck responded to Health Canada’s notice and submitted its objections to disclosure. Merck’s objections address generally the nature and the type of information found in the Comprehensive Summary pages. Merck asserted the confidential nature of this information by noting the limited distribution of the Comprehensive Summary within Merck and that it was submitted to Heath Canada with a confidentiality notice. There is no specific comment about the confidentiality of the substantive content of what is found at page 115.
Health Canada responded to Merck on October 2, 2001. It agreed to some additional, partial redactions in other documents, but it rejected the balance of Merck’s objections: see Beaudry J., at para. 19 (2006 FC 1200). Health Canada asserted that there could be no blanket exemption on the Comprehensive Summary and that some of its information was already available in the public domain. It provided for no redaction of the information contained at page 115.
Merck sent a further reply on October 31, 2001, based on a review prepared by outside consultants. The review identified all of the information that was not already publicly available, and which had not been redacted by Health Canada: see Beaudry J., at para. 23 (2006 FC 1200). At this time, Merck and the consultants suggested redacting only certain paragraphs and several references to other sections of the SNDS. Merck argued that such information remained confidential. It also argued that its methodology in preparing the SNDS was confidential and therefore the references to other parts of the SNDS should be redacted.
Health Canada sent Merck its final notice on December 19, 2001. It agreed to redact some additional details in other documents, but rejected the balance of Merck’s objections: see Beaudry J., at paras. 24-26 (2006 FC 1200). As mentioned above, there was no redaction to page 115, and so Merck applied to the Federal Court for review.
Before the Federal Court, Merck submitted a detailed table listing the parts of the record in dispute and its representations regarding the reasons why they should be exempted from disclosure. With regard to the confidentiality of page 115, Merck referred to an affidavit which says that pharmaceutical companies generally consider this information confidential. It also refers to another affidavit, which similarly states that pharmaceutical companies usually treat this type of information as confidential and that Merck does the same: see C.A. reasons, at paras. 96 and 98.
Once again, I see no error in the Court of Appeal’s conclusion that Merck’s evidence was not, in law, capable of discharging its onus of showing how the substantive information on page 115 contained confidential information: see C.A. reasons, at para. 79.
I therefore agree with the Court of Appeal to the extent that it held that Merck had not shown that the substantive contents of the records in dispute contained confidential financial, commercial, scientific or technical information. However, the Court of Appeal did not address the other two aspects of Merck’s submissions made in this Court in relation to s. 20(1)(b).
Merck’s second point is that the formatting and structure of the submissions qualifies for exemption. As noted earlier, I do not agree with this submission, for the reasons I set out above.
Merck’s third point is that the very fact it listed studies and therefore would be understood to have relied on them in its new drug submissions was confidential information. Even if the studies themselves are in the public domain, Merck maintains that it assembled the list of articles and studies and that what it chose to include is not in the public domain. The very fact that a particular publication was used in a Canadian NDS is not public knowledge. The list also provides a screening of articles and links them to the product and the NDS/SNDS. Merck also emphasizes that this information confers a competitive advantage on a competitor with concomitant harm to it. This last point, of course, is relevant to the s. 20(1)(c) exemption, not the s. 20(1)(b) exemption under consideration here.
The problem with Merck’s arguments about the listing of studies is that Merck itself proposed that copies of all published articles referred to in the submissions should be provided to the requester: letter from Merck to Health Canada dated September 25, 2000 (see reasons of Harrington J., at para. 38 (2004 FC 959); see also Beaudry J., at para. 16 (2006 FC 1201)). In addition, Health Canada underlines the point that the fact Merck had referred to many studies was already in the public domain as a result of the publication of the Product Monograph and documents published by the U.S. F.D.A. (Health Canada factum, at para. 134). I cannot see any difference in principle between published articles and studies otherwise in the public domain. Releasing these articles in response to an access to information request relating to a NDS or SNDS of course shows that Merck selected the studies and draws the link between the articles and the product and the NDS or the SNDS. This, in my view, completely undermines Merck’s claim for confidentiality of the fact that it selected the studies for inclusion or the existence of a link between them to the product and the NDS/SNDS. Having reviewed the relevant evidence, my view is that Merck did not show that the listing of the studies was itself confidential financial, commercial, scientific or technical information. While I would not foreclose the possibility of a claim of this nature being established in some cases in which the evidence supported it, the evidence does not support it here. I would add that the simple listing of studies does not engage the Janssen‑Ortho principle which relates to information revealing the third party’s evaluation of those studies.
I would therefore uphold the decision of the Federal Court of Appeal to reject Merck’s claims for s. 20(1)(b) exemptions on this record.
(4) Section 20(1)(c): Disclosure of Information that Could Reasonably Be Expected to Harm the Third Party
I turn now to the harm-based exemption in s. 20(1)(c). The exemption applies if the third party establishes that the disclosure “could reasonably be expected to result in material financial loss or gain to, or .... prejudice the competitive position of, a third party”:
Once again, there are submissions from both parties about the applicable principles as well as how those principles apply. The debate in this Court centred on three aspects of this exemption:
the degree of likelihood required by this provision that harm will result from disclosure;
whether disclosing information already in the public domain can cause harm; and
the types of harm contemplated by the provision.
After a brief review of the decisions below, I will address these issues and deal with the more specific submissions about how the relevant principles apply in this case.
(a) Decisions in the Federal Court and the Federal Court of Appeal
Turning first to the NDS file, it is unclear what threshold of probability of harm the reviewing judge applied. At para. 101 (2006 FC 1201), he referred to the burden of “establishing probability of harm” while, at para. 107, he tracked the statutory language of s. 20(1)(c) by referring to whether disclosure “could reasonably be expected to .... prejudice” the third party. He ruled that only one page of the contested reviewers’ notes and correspondence between Merck and Health Canada fell within the s. 20(1)(c) exemption (para. 93). He also rejected in general terms Merck’s position that it would probably suffer prejudice from disclosure of information that was already in the public domain (para. 101). Nonetheless, he found the exemption applied to about 130 documents, mostly in the Comprehensive Summary, because they contained information that was more precise or more detailed than that which was publicly available (para. 104).
With respect to the SNDS file, the reviewing judge applied the same legal framework and found that five documents fell within the s. 20(1)(c) exemption on the basis that they contained information that was more precise or more detailed than that available in the public domain and that their disclosure “would likely cause the Applicant significant loss of profit or undermine its competitiveness” (para. 112, 2006 FC 1200).
The Federal Court of Appeal overturned the reviewing judge’s findings in both files and found none of the records was exempt under s. 20(1)(c). Desjardins J.A. for the Court held that for the third party to benefit from this exemption, the information could not be in the public domain and that the third party had to establish a “reasonable expectation of probable harm”: at para. 81, citing AstraZeneca. In her view, a mere possibility of harm was not sufficient (para. 84). Desjardins J.A. found that the record did not support the reviewing judge’s conclusion that the exemption under s. 20(1)(c) applied to any of the documents as redacted. She found that Merck’s evidence was vague, speculative and silent as to specifically how and why the disclosure of the requested information would be likely to bring about the harm alleged by Merck (para. 93).
Other than four documents from the NDS that are no longer in issue, Merck seeks restoration of all of the s. 20(1)(c) exemptions, found by the reviewing judge, as well as exemptions for the other documents for which it unsuccessfully claimed exemptions before the reviewing judge. These latter records amount to over 100 documents in each of the NDS and the SNDS appeals.
With respect to the s. 20(1)(c) exemptions claimed by Merck but rejected by the reviewing judge, Merck’s submissions in this Court have not shown how the reviewing judge erred in his rejection of its claims or how the Federal Court of Appeal erred by upholding his findings. The focus of my analysis therefore will be on the s. 20(1)(c) exemptions upheld by the reviewing judge but set aside by the Court of Appeal; that is, about 130 documents in the NDS and 5 in the SNDS.
I return to the three issues identified earlier.
(b) The Degree of Likelihood that Harm Will Occur
For about 20 years, the Federal Courts have read s. 20(1)(c) as requiring the third party to demonstrate “a reasonable expectation of probable harm”: see, e.g., Canada Packers, at pp. 58-60; Canada v Canada, at pp. 440 ff.; Air Atonabee, at pp. 277-78 and 280; Ottawa Football Club v Canada (Minister of Fitness and Amateur Sports),  2 F.C. 480 (T.D.), at pp. 487-88; Saint John Shipbuilding Ltd. v Canada (Minister of Supply and Services) (1990), 67 D.L.R. (4th) 315 (F.C.A.); Brookfield Lepage Johnson Controls Facility Management Services v Canada (Minister of Public Works and Government Services), 2004 FCA 214, 322 N.R. 388, at paras. 11 ff.
Merck proposes that this line of jurisprudence should be abandoned and that the word “probable” should be omitted. Merck submits that the proper test was articulated by the Nova Scotia Court of Appeal in Chesal v Nova Scotia (Attorney General), 2003 NSCA 124, 219 N.S.R. (2d) 139, where the court held that the introduction of “probable” into the language of the test was incorrect. Hence, all that is required is a “reasonable expectation of harm” (para. 37). Merck’s proposed test would therefore require the third party to show a “reasonable expectation of harm” resulting from disclosure.
Health Canada maintains that the well-established standard applied by the Federal Court of Appeal in this case should be maintained. As the case law clearly indicates that a mere possibility is insufficient, the proper test is a reasonable expectation of probable harm. Any test in between cannot be conceptualized.
I am not persuaded that we should change the way this test has been expressed by the Federal Courts for such an extended period of time. Such a change would also affect other provisions because similar language to that in s. 20(1)(c) is employed in several other exemptions under the Act, including those relating to federal-provincial affairs (s. 14), international affairs and defence (s. 15), law enforcement and investigations (s. 16), safety of individuals (s. 17), and economic interests of Canada (s. 18). In addition, as the respondent points out, the “reasonable expectation of probable harm” test has been followed with respect to a number of similarly worded provincial access to information statutes. Accordingly, the legislative interpretation of this expression is of importance both to the application of many exemptions in the federal Act and to similarly worded provisions in various provincial statutes.
It may be questioned what the word “probable” adds to the test. At first reading, the “reasonable expectation of probable harm” test is perhaps somewhat opaque because it compounds levels of uncertainty. Something that is “probable” is more likely than not to occur. A “reasonable expectation” is something that is at least foreseen and perhaps likely to occur, but not necessarily probable. When the two expressions are used in combination – “a reasonable expectation of probable harm” – the resulting standard is perhaps not immediately apparent. However, I conclude that this long-accepted formulation is intended to capture an important point: while the third party need not show on a balance of probabilities that the harm will in fact come to pass if the records are disclosed, the third party must nonetheless do more than show that such harm is simply possible. Understood in that way, I see no reason to reformulate the way the test has been expressed.
I note that in Lavigne v Canada (Office of the Commissioner of Official Languages), 2002 SCC 53,  2 S.C.R. 773, at para. 58, the Court referred with apparent approval to the “reasonable expectation of probable harm” formulation and to the statement by Richard J. (as he then was) in Information Commissioner (Can.) v Immigration and Refugee Board (Can.) (1997), 140 F.T.R. 140 (T.D.), that this standard implies a “confident belief” (para. 43). In applying the standard, the Court concluded that the evidence did not “provide a reasonable basis for concluding that disclosure .... could reasonably be expected to be injurious” so as to result in the harm alleged (para. 61). This comment, while not requiring proof that harm will occur on the balance of probabilities, nonetheless underlines the point that something well beyond a mere possibility of harm must be shown. As for the causal link between disclosure and harm, the Court indicated that there need not be a causal relationship as in tort law, but that there must be proof of a “clear and direct connection between the disclosure and the injury that is alleged” (para. 58; see also Canada Packers, at pp. 58-59).
The Court also recently interpreted a similar phrase which occurs in another statutory context. In Hilewitz v Canada (Minister of Citizenship and Immigration), 2005 SCC 57,  2 S.C.R. 706, the majority referred, at para. 60, to language specifying that a foreign national is inadmissible under s. 38(1)(c) of the Immigration and Refugee Protection Act, S.C. 2001, c. 27, if that individual’s health condition “might reasonably be expected to cause excessive demand on health or social services.” Abella J. noted that the wording establishes a requirement that “any anticipated burdens on the public purse be tethered to the realities, not the possibilities, of [the] applicants’ circumstances” (para. 60; emphasis added).
I would affirm the Canada Packers formulation. A third party claiming an exemption under s. 20(1)(c) of the Act must show that the risk of harm is considerably above a mere possibility, although not having to establish on the balance of probabilities that the harm will in fact occur. This approach, in my view, is faithful to the text of the provision as well as to its purpose.
As with any question of statutory interpretation, the Court must interpret the words of this statute in their entire context, in their grammatical and ordinary sense harmoniously with the scheme of the Act, the object of the Act and the intention of Parliament.
I begin with the English text of the provision. The words “could reasonably be expected to result” seem to avoid either the standard of mere possibility or the standard of probability. We must assume, I think, that both of those standards would clearly have been known to the drafters. This suggests that some middle ground was intended: something cannot reasonably be expected to occur if it is a mere possibility; but something may be reasonably expected even if it is not more likely than not to occur. The word “expected” derives from the verb “to expect”, a primary meaning of which is to “regard as likely” (The Canadian Oxford Dictionary (2nd ed. 2004)). The word “likely” is more difficult to pin down. While it can mean “probable” it may also mean “such as well might happen” (ibid). In legal usage, the standard of proof on the balance of probabilities is often expressed by saying that something must be shown to be more likely than not. I conclude that the English text of the statute suggests a middle ground between that which is probable and that which is merely possible. The intended threshold appears to be considerably higher than a mere possibility of harm, but somewhat lower than harm that is more likely than not to occur.
Turning to the French text of s. 20(1)(c), the phrase “risquerait vraisemblablement de causer” is a challenging one to interpret. The conditional “risquerait de causer” might be rendered into English by either “could” or “would” cause. The drafter here chose the less definite “could.” The word “vraisemblablement” is capable of meaning “probably” or “likely”: see, for example, Kwiatkowsky v Minister of Employment and Immigration,  2 S.C.R. 856, at pp. 863-64. However, it is often used in federal statutes as the equivalent of the English words “likely” or “reasonably” or to convey the sense of risk of something happening or not happening. Some examples follow. In the Competition Act, R.S.C. 1985, c. C-34, s. 11(1), “qu’une personne détient ou détient vraisemblablement des renseignements pertinents à l’enquête en question” was drafted in English as “that a person has or is likely to have information that is relevant to the inquiry”, and in s. 74.11(4) “s’il est convaincu que le paragraphe (3) ne peut vraisemblablement pas être observé” is drafted in English as “where it is satisfied that subsection (3) cannot reasonably be complied with”. In the Criminal Code, R.S.C., 1985, c. C-46, s. 25.1(9), “qui entraînerait vraisemblablement la perte de biens ou des dommages importants à ceux-ci” is drafted in English as “that would be likely to result in loss of or serious damage to property”, and in s. 382.1(2) “sachant qu’ils seront vraisemblablement utilisés pour acheter ou vendre, même indirectement, les valeurs mobilières en cause ou qu’elle les communiquera vraisemblablement à d’autres personnes qui pourront en acheter ou en vendre” was drafted in English as “knowing that there is a risk that the person will use the information to buy or sell, directly or indirectly, a security to which the information relates, or that they may convey the information to another person who may buy or sell such a security”. In the Insurance Companies Act, S.C. 1991, c. 47, s. 294(6), “provoquerait vraisemblablement une modification sensible du prix des valeurs mobilières de la société” was drafted in English as “might reasonably be expected to materially affect the value of any of the securities of the company”.
As noted earlier, the word “likely” is a good fit with the statute’s text of “could reasonably be expected to”. The shared meaning rule for the interpretation of bilingual legislation dictates that the common meaning between the English and French legislative texts should be accepted (Sullivan, at pp. 99 ff., and Michel Bastarache et al., The Law of Bilingual Interpretation (2008), at pp. 32 ff). By resorting to the shared meaning rule, I would interpret “could reasonably be expected to” in the English version and “risquerait vraisemblablement” in the French version as meaning “likely”, a standard considerably higher than mere possibility, but somewhat lower than “more likely than not”. This sense is captured by the long-standing test enunciated by the Federal Courts: “reasonable expectation of probable harm”.
This interpretation also serves the purposes of the Act. A balance must be struck between the important goals of disclosure and avoiding harm to third parties resulting from disclosure. The important objective of access to information would be thwarted by a mere possibility of harm standard. Exemption from disclosure should not be granted on the basis of fear of harm that is fanciful, imaginary or contrived. Such fears of harm are not reasonable because they are not based on reason: see Air Atonabee, at p. 277, quoting Re Actors’ Equity Assn. of Australia and Australian Broadcasting Tribunal, Re (No. 2) (1985), 7 A.L.D. 584 (Admin. App. Trib.), at p. 590. The words “could reasonably be expected” “refer to an expectation for which real and substantial grounds exist when looked at objectively”: Watt v Forests,  NSWADT 197, at para. 120. On the other hand, what is at issue is risk of future harm that depends on how future uncertain events unfold. Thus, requiring a third party (or, in other provisions, the government) to prove that harm is more likely than not to occur would impose in many cases an impossible standard of proof.
Health Canada applied an unduly onerous test of probability of harm. For example, an officer at Health Canada at the relevant time deposed that, in deciding whether disclosure could be expected to be prejudicial to a third party, the financial loss or the prejudice to a third party’s competitive position must be “immediate” and “clear”. This approach is not, in my respectful view, consistent with the language of s. 20(1)(c).
To conclude, the accepted formulation of “reasonable expectation of probable harm” captures the need to demonstrate that disclosure will result in a risk of harm that is well beyond the merely possible or speculative, but also that it need not be proved on the balance of probabilities that disclosure will in fact result in such harm.
(c) Could Disclosure of Publicly Available Information Cause Harm?
Merck notes in its submissions that there is no reference to confidentiality in s. 20(1)(c). Its position therefore is that in some cases information that is not confidential within the meaning of s. 20(1)(b) may still be exempted under s. 20(1)(c). It submits, for example, that a compilation of material may fall within s. 20(1)(c) even though each item in the compilation is in the public domain. Another example offered is of information which has had very limited previous disclosure. In both of these types of situations, Merck submits, the evidence may establish that harm could reasonably be expected to result from disclosure.
As the respondent points out, it is very hard to show that harm can reasonably be expected to result from disclosure of publicly available information: AstraZeneca, at paras. 81 and 109; Cyanamid Canada Inc. v Canada (Minister of Health and Social Welfare) (1992), 9 Admin L.R. (2d) 161 (F.C.A.), at paras. 28-29. As Phelan J. put it in AstraZeneca, at para. 81, “[a]s a general proposition, publicly available information is not exempt information under section 20 either as a class of documents or under the [harm] test. It requires compelling evidence to dislodge the logical conclusion that information in the public domain will be used, particularly by knowledgeable users”.
I accept this general principle. In this case, however, the difference between the parties is more concerned with what the information really consists of than with whether it is publicly available. For example, Merck submits that the compilation of publicly available studies is a separate work from the studies themselves and one which is created by Merck’s employees with a considerable investment of time and resources. Thus, the information at issue is not the publicly available information that the studies exist or what their content includes. What is not publicly available, says Merck, is the way a group of publicly available studies was compiled for a particular purpose.
I do not think this submission fails as a matter of principle. It may be possible in some cases to show that the way in which publicly available information has been assembled in a particular situation is not, itself, publicly known. Once that is done, the question becomes whether disclosing it has been shown to give rise to the risk of harm required under s. 20(1)(c).
(d) Types of Harm to the Third Party
I now turn to address the parties’ submissions about the type of harm on which a third party may rely in claiming the s. 20(1)(c) exemption. It is for the reviewing judge to decide whether the evidence shows that disclosure could reasonably be expected to result in harm of the nature specified in s. 20(1)(c). I mention this to underline the point that while the case law can set out general principles governing the provision’s application, at the end of the day, there is a significant factual component to the inquiry which will turn on the particular circumstances and evidence in each case.
To begin, it is worth noting that the list of types of harm in s. 20(1)(c) is disjunctive. It is sufficient for a third party to show that disclosure could reasonably be expected to result in any one of a financial loss or gain or in prejudice to the third party’s competitive position. In other words, it is not necessary for the third party to show that the “prejudice” to his or her competitive position also results in “harm”: see Brookfield Lepage, at paras. 9-10.
That brings us to the types of harm alleged by Merck. I will discuss each one briefly. The submissions here focus on material that is not otherwise exempted because it is a trade secret or confidential commercial information which ought to be exempted under ss. 20(1)(a) and (1)(b). Putting aside what may be described as bald assertions, Merck has raised three types of harm:
facilitating a competitor’s drug development with concomitant losses to Merck,
facilitating a competitor’s preparation of a NDS or SNDS with concomitant losses to Merck, and
giving an incorrect impression concerning Singulair’s safety.
I will briefly comment on each type of harm.
I turn first to the evidence about how release of the pages in issue could cause harm to Merck by facilitating a competitor’s new drug development process. In essence, the allegation is that the information could help a competitor bring its drug to market more quickly than it otherwise could and that this in turn would cause prejudice to Merck’s competitive position. This point was advanced with respect to a number of types of information:
Lists of references and cross-referencing: Merck maintains that this information provides a first screening of all articles available in the public domain and is therefore likely to facilitate the competitor’s drug development program by indentifying key elements investigated, developed or used in the NDS or SNDS submission.
Manufacturing information: information about how the product is manufactured will facilitate and accelerate the competitor’s own drug production.
Merck’s responses to questions raised by Health Canada during the review process: this information alerts competitors as to what the issues are in registering this class of product. A competitor’s advance knowledge of these issues would expedite its drug development and submission review, jeopardizing Merck’s competitive position.
Disclosure of some information would give a misleading impression of the product’s safety and be exploited by competitors with resulting financial loss to Merck.
I do not understand Health Canada to contest the proposition that manufacturing and other scientific information about the product which is not in the public domain may well be exempt from disclosure under either s. 20(1)(a) or (b). Health Canada maintains, rather, that information of that nature has been redacted from the records which it intends to disclose. For the reasons set out earlier, my view is that Merck’s evidence does not effectively answer Health Canada’s evidence on this point. I will therefore put to the side Merck’s submissions that any such confidential manufacturing or scientific information which is not in the public domain remains in issue under s. 20(1)(c). The points to be resolved under s. 20(1)(c), therefore, are whether lists of references and cross-referenced published sources, or information about how the approval process unfolded, are exempted under s. 20(1)(c).
I turn then to Merck’s submission that its compilations of published studies could allow a competitor to copy its work for the purposes of the competitor’s own drug development or approval process, thereby prejudicing Merck’s competitive position and causing it financial loss. There are three relevant general principles that need to be borne in mind in this regard.
The first is that disclosure of general information such as dates, numbering and location of information within a NDS or the manner of its presentation generally does not give rise to the necessary expectation of harm or competitive prejudice. The same may be said about lists of studies or acknowledgement that certain studies which have been released to the public have been consulted. Of course, everything will turn on the evidence in the particular case.
Second, knowledge that may be gleaned from the records about how the regulatory process works is, as Phelan J. succinctly put in AstraZeneca, “not the type of information which section 20 is designed to exempt from disclosure” (para. 94). A purpose of access to information legislation is to make the workings of government more transparent. It would be contrary to this purpose to hold that disclosure of information about how the regulatory process works in general, or how it worked in a particular case, confers a competitive advantage or disadvantage.
Third, disclosure of information, not already public, that is shown to give competitors a head start in developing competing products, or to give them a competitive advantage in future transactions may, in principle, meet the requirements of s. 20(1)(c). The evidence would have to convince the reviewing court that there is a direct link between the disclosure and the apprehended harm and that the harm could reasonably be expected to ensue from disclosure: see, e.g., AB Hassle v Canada (Minister of National Health and Welfare) (1998), 161 F.T.R. 15, at para. 42, aff’d  3 F.C. 360 (C.A.); Wells v Canada (Minister of Transport) (1995), 103 F.T.R. 17, at para. 9; Culver v Canada (Minister of Public Works and Government Services),  F.C.J. No. 1641 (QL) (T.D)., at para. 17; Bitove Corp. v Canada (Minister of Transport) (1996), 119 F.T.R. 278, at para. 10; Coradix Technology Consulting Ltd. v Canada (Minister of Public Works and Government Services), 2006 FC 1030, 307 F.T.R. 116, at para. 31; Canada Post Corp. v National Capital Commission, 2002 FCT 700, 221 F.T.R. 56, at paras. 16-17; Aventis Pasteur Ltd. v Canada (Attorney General), 2004 FC 1371, 262 F.T.R. 73, at paras. 32-33; and Prud’homme v Agence canadienne de développement international (1994), 85 F.T.R. 302, at para. 7. Even if information taken in isolation may not seem to fall within the exemption, the information should nonetheless be examined in its entirety in order to determine the likely impact of its disclosure.
I conclude that as a matter of principle, the disclosure of information that is not already in the public domain and that could give competitors a head start in product development, or which they could use to their competitive advantage, may be shown to give rise to a reasonable expectation of probable harm or prejudice to the third party’s competitive position. The question here is whether Merck’s evidence did so.
The reviewing judge ruled against Merck on many of its claims for exemption under s. 20(1)(c) in both the NDS and the SNDS. Merck did not persuade the Federal Court of Appeal that he had erred in those rulings and it advances no document-specific submissions in this Court in relation to any of them. The only document-specific submissions by Merck before us relate to about 25 pages in the NDS to which the reviewing judge found that the s. 20(1)(c) exemption applied. That ruling was set aside by the Court of Appeal.
Having reviewed Merck’s submissions and the evidence referred to, my view is that the Federal Court of Appeal’s appellate intervention was justified and that, in making its own assessment, it did not err in its disposition of these claims for exemption. Health Canada’s evidence was to the effect that virtually all of the unredacted information on the pages in issue was in the public domain and it gave extensive and precise references to where the information could be publicly obtained. This evidence was largely unanswered by Merck and it did not provide evidence showing how the disclosure of the redacted form of the information could reasonably be expected to give rise to the harm and prejudice it claimed. I also reiterate here my conclusions in relation to s. 20(1)(b) as it applies to lists of studies.
That leaves for consideration Merck’s submission that release of some of the pages could give an inaccurate perception of the product’s safety. Merck says that refusal to disclose this sort of information under s. 20(1)(c) is not problematic because proper information in proper context is provided in the Product Monograph. Moreover, there are reporting requirements relating to information where public safety is concerned and, in an appropriate case, the public interest override could be invoked to release such information even if it is found to be exempt under s. 20(1)(c), providing disclosure is in the public interest.
I do not accept the principles inherent in these submissions. The courts have often – and rightly – been sceptical about claims that the public misunderstanding of disclosed information will inflict harm on the third party: see, e.g., Air Atonabee, at pp. 280-81; Canada Packers, at pp. 64-65; Coopérative fédérée du Québec v Canada (Ministre de l’Agriculture et de l’Agroalimentaire) (2000), 180 F.T.R. 205, at paras. 9-15. If taken too far, refusing to disclose for fear of public misunderstanding would undermine the fundamental purpose of access to information legislation. The point is to give the public access to information so that they can evaluate it for themselves, not to protect them from having it. In my view, it would be quite an unusual case in which this sort of claim for exemption could succeed.
It is particularly important to allow broad access to this sort of information in the context of the pharmaceutical industry. As the respondent points out, Health Canada systematically posts on its website reports about undesirable effects of all drugs sold in Canada. In addition, the Food and Drug Regulations require pharmaceutical companies to report adverse reactions of their drugs to Health Canada (s. C.01.017). Information about those reactions is publicly available. It is therefore difficult to see how release of such reports through an access to information request could result in harm to the third party.
Merck has not persuaded me that it established the grounds for a s. 20(1)(c) exemption for documents of this nature.
To conclude on s. 20(1)(c), Merck has not shown that the Federal Court of Appeal erred in the principles it applied or how it applied them.
(5) Conclusion with Respect to Section 20(1) Exemptions
In my view, Merck has not shown that any of the pages in issue, as redacted, contain any information exempted under s. 20(1)(a), (b) or (c).
C. Severance of the Record Under Section 25 of the Act
When heads of government institutions determine that a requested record contains exempted information in respect of which they are authorized to refuse disclosure, they must go on to consider the issue of severance. By virtue of s. 25 of the Act, they are required to disclose any part of the record that does not contain such exempted information and which can reasonably be severed from any part that does contain exempted information. Section 25 reads:
In this case, the reviewing judge found entire pages were exempt that the institutional head had decided could be disclosed with exempted material redacted. The Court of Appeal found the judge had erred in this regard and Merck challenges that conclusion.
In both the NDS and SNDS files, the reviewing judge found that, apart from a few instances in which he noted specific passages could be redacted, the entirety of all the other pages contained exempted information and should not be disclosed. He held in both the NDS and SNDS files that “it would be extremely difficult to isolate the information that should not be disclosed”: see Beaudry J., para. 114 (2006 FC 1200); see also, Beaudry J., at para. 108 (2006 FC 1201). With respect to the NDS decision (2006 FC 1201), I believe the judge’s reference to examples in para. 108 was intended to be to para. 106 rather than to para. 107. The judge did not set out the applicable legal principles or indicate how they applied in the circumstances.
The Court of Appeal found fault with the reviewing judge’s decision on two grounds. The Court of Appeal faulted the reviewing judge for substituting his discretion for that of the institutional head with respect to s. 25 (para. 104). I respectfully do not agree with the Court of Appeal on this point. As Merck submits, on the s. 44 review, it was the role of the reviewing judge to review the disclosure decision of the institutional head and to determine whether that decision was in accordance with the Act. It follows that the Court of Appeal was in error to the extent it faulted the judge for having “substituted” his view for that of the institutional head. The reviewing judge was required to consider whether the institutional head had properly applied s. 25.
The second error identified by the Court of Appeal was that the reviewing judge failed to explain why the non-exempt material could not reasonably be severed and disclosed. Here, in my view, the Court of Appeal was on firm ground. The reviewing judge did not explain why it would be “extremely difficult” to sever and disclose the non-exempt information. In the absence of any explanation from the reviewing judge (and none being apparent from his reasons read in the context of the whole record), the Court of Appeal was obliged to intervene.
Additionally, in this Court, Merck did not provide any submissions as to why the non-exempt material could not reasonably be severed. In other words, it did not offer submissions in defence of the substance of the reviewing judge’s decision. In the absence of such submissions, and in light of my conclusion that appellate intervention was justified, I would uphold the Court of Appeal’s decision setting aside the reviewing judge’s decision in relation to s. 25.
It will be helpful, however, to reiterate some of the key principles in relation to s. 25.
To begin, it is important to recognize that applying s. 25 is mandatory, not discretionary. The section directs that the institutional head “shall [not “may”] disclose any part of the record that does not contain” exempted information, provided it can reasonably be severed: see Dagg, at para. 80. Thus, the institutional head has a duty to ensure compliance with s. 25 and to undertake a severance analysis wherever information is found to be exempt from disclosure.
The heart of the s. 25 exercise is determining when material subject to the disclosure obligation “can reasonably be severed” from exempt material. In my view, this involves both a semantic and a cost-benefit analysis. The semantic analysis is concerned with whether what is left after excising exempted material has any meaning. If it does not, then the severance is not reasonable. As the Federal Court of Appeal put it in Blank v Canada (Minister of the Environment), 2007 FCA 289, 368 N.R. 279, at para. 7, “those parts which are not exempt continue to be subject to disclosure if disclosure is meaningful”. The cost-benefit analysis considers whether the effort of redaction by the government institution is justified by the benefits of severing and disclosing the remaining information. Even where the severed text is not completely devoid of meaning, severance will be reasonable only if disclosure of the unexcised portions of the record would reasonably fulfill the purposes of the Act. Where severance leaves only “[d]isconnected snippets of releasable information”, disclosure of that type of information does not fulfill the purpose of the Act and severance is not reasonable: Canada (Information Commissioner) v Canada (Solicitor General),  3 F.C. 551 (T.D.), at pp. 558-59; SNC-Lavalin Inc., at para. 48. As Jerome A.C.J. put it in Montana Band of Indians v Canada (Minister of Indians and Northern Affairs),  1 F.C. 143 (T.D.) [pp.160‑61]:
To attempt to comply with section 25 would result in the release of an entirely blacked-out document with, at most, two or three lines showing. Without the context of the rest of the statement, such information would be worthless. The effort such severance would require on the part of the Department is not reasonably proportionate to the quality of access it would provide.
That said, one must not lose sight of the purpose of s. 25. It aims to facilitate access to the most information reasonably possible while giving effect to the limited and specific exemptions set out in the Act (Ontario (Public Safety and Security), at para. 67).
Section 25 must also be considered in relation to the question of giving notice to a third party under s. 27. As I discussed earlier, notice is required whenever the institutional head forms the intention to release information that he or she believes might be exempt, including severed material under s. 25. I am also of the view that the institutional head is obliged to do his or her best to apply s. 25 before giving notice and should indicate to the third party what redactions will be made as a result of the institutional head’s initial review. I note that the notice provision in ss. 27(3)(a) and (b) refers to the institutional head giving notice of the intention to release “a record or a part thereof” and providing the third party with “a description of the contents of the record or part thereof” that may relate to the third party. This suggests that the institutional head is to make an effort to apply s. 25 at the notice stage.
Of course, the institutional head can only proceed on the basis of the information reasonably available. But consistent with the thresholds for notice which I set out earlier, the institutional head should identify the records he or she has determined clearly fall within the s. 20(1) exemptions and, having done so, go on to determine under s. 25 what information must be disclosed because it can reasonably be severed from the exempt material.
For the reasons set out earlier, I would affirm the Federal Court of Appeal’s disposition of the s. 25 issue.
I would dismiss the appeals with costs.
(with whom Abella and Rothstein JJ joined, dissenting)
I have read the reasons of my colleague Cromwell J. I agree with his approach to the issue of notice. The head of the institution must review all the relevant material before him or her and can disclose information without notice only if there is clearly no reason to believe that the record might contain exempt information. I also agree with my colleague that, on the issue of the standard of proof, the Court of Appeal erred in imposing a standard higher than that of proof on a balance of probabilities.
However, in my view, the Federal Court’s judgments (2006 FC 1200, 301 F.T.R. 241 and 2006 FC 1201 (CanLII)) do not contain a palpable and overriding error that would justify this Court’s intervention. I would restore the findings of the Federal Court, subject to any agreements the parties may have concluded since its judgments were rendered.
A. Appellate Review
Although my colleague indicates at para. 54 that appellate review is governed by Housen v Nikolaisen, 2002 SCC 33,  2 S.C.R. 235, his subsequent analysis does not, in my opinion, comport with the principles established in that case. My colleague agrees with the approach of the Federal Court of Appeal, which found that the reviewing judge had not explained in sufficient detail how he came to his conclusions (2009 FCA 166, 400 N.R. 1). Cromwell J. endorses that court’s conclusions despite finding that it both applied the wrong standard of proof and inappropriately characterized the definition of “trade secrets” as a restrictive one.
Turning to the Federal Court’s judgment, my colleague faults the reviewing judge for failing “either to state the applicable legal principles or to explain how the legal principles applied to the facts before him or, in some cases, both” (para. 55). I cannot accept the requirements my colleague’s approach imposes on trial judges or the message it sends to the legal community. The rule from Housen is that an appellate court must defer to a trial judge’s findings on questions of fact as well as on questions of mixed fact and law. The standard to be applied on such questions is that of a “palpable and overriding error”. Deferring to trial judges’ findings where it is appropriate to do so ensures that judicial resources are used efficiently, enhances access to justice and is consistent with the institutional role of the appellate court.
It must be noted that although the Federal Court is being asked to review an administrative decision, one made by Health Canada in this case, the process is atypical in the sense that it differs from the one that applies to the review of most administrative decisions. The latter process – and the question of which standards ought to govern it – has occupied the forefront of administrative law in the past decade. In Dunsmuir v New-Brunswick, 2008 SCC 9,  1 S.C.R. 190, this Court sought to bring clarity to this issue in the context of the first level of review. In the “classic” process, appellate review consists in verifying whether the court at the first level of review has correctly applied the standard in reviewing the administrative decision. What this means in practice is that in “step[ping] into the shoes” of the lower court, an appellate court’s focus is, in effect, on the administrative decision (Prairie Acid Rain Coalition v Canada (Minister of Fisheries and Ocean), 2006 FCA 31,  3 F.C.R. 610, at para. 14; Zenner v Prince Edward Island College of Optometrists, 2005 SCC 77,  3 S.C.R. 645, at para. 30).
There are exceptions to this classic process. Under s. 44 of the Access to Information Act, R.S.C. 1985, c. A-1 (“ATIA”), the appeal court’s focus is on the reviewing judge’s findings, and the rule from Housen applies to that court’s decision (Canadian Imperial Bank of Commerce v Canada (Chief Commissioner, Human Rights Commission), 2007 FCA 272,  2 F.C.R. 509, at paras. 8, 9 and 72; Rubin v Canada (Minister of Health), 2003 FCA 37, 300 N.R. 179, at paras. 4-5; Merck Frosst Canada Inc. v Canada (Minister of National Health), 2002 FCA 35 (CanLII); SNC Lavalin Inc. v Canada (Minister for International Co-operation), 2007 FCA 397, 77 Admin. L.R. (4th) 1, at paras. 2, 3 and 7).
The peculiarities of the review process provided for in s. 44 ATIA explain this distinctiveness. The scheme of the ATIA reveals that Parliament intended to set up an independent review process – a function which is not fulfilled by the head of the institution: 3430901 Canada Inc. v Canada (Minister of Industry), 2001 FCA 254,  1 F.C. 421, at para. 36. Furthermore, the federal Information Commissioner, unlike his or her Ontario and Quebec counterparts (Freedom of Information and Protection of Privacy Act, R.S.O. 1990, c. F.31; Act respecting access to documents held by public bodies and the Protection of personal information, R.S.Q., c. A-2.1), has no adjudicatory powers and can only make recommendations. The role of the head of the federal institution is as much that of a party as that of a decision maker. The institution’s opinion on the obligation to disclose or refuse to disclose is no more authoritative than that of other interested parties (Canadian Imperial Bank of Commerce, at para. 63).
The Federal Court judge is thus the first impartial gatekeeper a party seeking disclosure (s. 41 or 42 ATIA) or objecting to it (s. 44 ATIA) can turn to. By the time the matter reaches that court, the content of the file will often have evolved: Air Atonabee Ltd. v Canada (Minister of Transport) (1989), 37 Admin. L.R. 245 (F.C.T.D.). The Federal Court reviews the evidence, which can be extensive. New evidence may be filed and cross-examinations may be conducted, as in the case at bar (ss. 45-46 ATIA; Air Atonabee, at pp. 264-66). The Federal Court may hear new arguments if necessary. The judge makes his or her own findings and draws inferences on the basis of the information in the court’s record at that time. The judge may order any remedies he or she considers appropriate (ss. 50-51 ATIA).
In sum, the judge does not conduct the kind of review that is usually conducted in an administrative law context. The Federal Court’s task, in essence, is to start afresh and assess the issue de novo. This is akin to the role of a trial court. For this reason, the appellate court’s role is to review the reviewing judge’s decision, not that of the Commissioner or the head of the institution. The appellate court’s role may be different in instances in which the decision of the head of the institution is discretionary by law, but such instances are irrelevant to the case at bar.
B. Did the Federal Court Make a Palpable and Overriding Error in this Case?
In Housen, at para. 1, Iacobucci and Major JJ. described the “palpable and overriding error” standard as follows:
A proposition that should be unnecessary to state is that a court of appeal should not interfere with a trial judge’s reasons unless there is a palpable and overriding error. The same proposition is sometimes stated as prohibiting an appellate court from reviewing a trial judge’s decision if there was some evidence upon which he or she could have relied to reach that conclusion.
They aptly explained at para. 3 what this means for an appellate court by quoting from Underwood v Ocean City Realty Ltd. (1987), 12 B.C.L.R. (2d) 199 (C.A.), at p. 204:
The appellate court must not retry a case and must not substitute its views for the views of the trial judge according to what the appellate court thinks the evidence establishes on its view of the balance of probabilities.
And, as they soundly observed at para. 4:
While the theory [just described] has acceptance, consistency in its application is missing.
In the case at bar, Beaudry J.’s findings on the exemptions are fact-based or bear on questions of mixed fact and law, so deference is owed to them. There was “some evidence upon which he … could have relied” to reach his conclusions. No palpable and overriding error can be found in his judgments. Consequently, the Federal Court of Appeal erred in retrying the case.
Unlike the Federal Court of Appeal, Beaudry J. applied the correct standard, that of proof on a balance of probabilities. He mentioned it once in each of his sets of reasons. He did not need to refer to it each time he considered a different argument. It is clear that he was mindful of all the submissions made by the parties, including Health Canada’s arguments that Merck Frosst Canada Ltd.’s representations were insufficiently specific and that Merck had not discharged its burden of showing that disclosure should be refused. Though his reasons could have been more explicit, a judge is not required to explain every conclusion in detail for reasons to be considered sufficient. Moreover, he referred in his judgments to the evidence and to tables filed by the parties. The tables summarized the parties’ evidence for each page or group of pages. Beaudry J. even appended the tables to his judgments. For each exemption category, he reached a decision that he supported by referring to the tables. It is clear from other tables the parties filed jointly with this Court (Joint Record of Pages in Issue) in which the pages at issue are indicated that he did not mechanically endorse the position of either Merck or Health Canada. While one may disagree with the result, Beaudry J.’s conclusions can easily be explained by referring both to his reasons and to the parties’ submissions.
There is, at first glance, a difficulty related to the evidence: prior to the hearing before Beaudry J., Health Canada responded to Merck’s last affidavit by redacting additional information, but Merck chose to proceed without responding to these redactions. Merck took the position that the information in the court’s record was sufficient and that it was not obliged to respond further. Its stance opens the door to the argument that the evidence it adduced did not respond to changes to the court’s record.
My colleague accepts Health Canada’s submissions and enters into the fray, arguing that one page or another illustrates the vacuity of Merck’s position and that Beaudry J. erred in this regard. Health Canada points to one page (p. 470 of the pages in issue in the New Drug Submission file) from which all confidential information was redacted, but that nevertheless found its way onto the list of hundreds of documents which Beaudry J. found to be exempt. In my view, this minor error does not justify this Court’s intervention, especially since it was rectified before we heard the case. The page in question is not at issue. With respect to other documents, still at issue, that were redacted to either a limited or a significant extent but were found to be exempt, the extent of the redaction should not be determinative in and of itself. Where information is highly technical – as is the case here – it may mean little to a non-expert but be of significance to a competitor who can “connect the dots” (in the words of Harrington J. in the first review conducted in this case (2004 FC 959,  1 F.C.R. 587)). Information that is superficially benign because its significance is lost on the person conducting the review can cause harm if disclosed. My colleague’s review lacks the insight the reviewing judge gained in the four days the latter spent hearing the case.
Beaudry J. heard all the parties’ arguments. More importantly, the same arguments that Merck’s representations were insufficiently specific and that its evidence did not respond to changes to the court’s record were presented to him. They did not carry the day. Health Canada’s statement that all confidential information had been redacted is just an argument. It is not proof that all such information has in fact been redacted. Indeed, at the beginning of the proceedings, Health Canada took the position that none of the information was confidential. The number of documents that either were subsequently found to be exempt in their entirety or were redacted extensively is a clear indication that Health Canada’s word cannot be taken as proof. Health Canada does not make a convincing case that Beaudry J. erred on this point, and I do not think we can dismiss his judgments.
The size of the record, the time allotted to the parties to argue their cases in this Court, and the Court’s institutional role are all factors that militate against reviewing the facts in such minute detail. The deferential approach dictated by Housen is more consistent with this Court’s role. Furthermore, I am not convinced that this Court ought to be conducting the kind of technical review which is required in order to determine whether information qualifies for an exemption from disclosure.
Health Canada and Merck fought tooth and nail for over 5 years before being heard by Beaudry J. Access to information may be becoming the favourite battleground of innovative and generic drug manufacturers. The quantity of resources, both public and private, expended as a consequence of the war between the parties in the case at bar is appalling. This may be a sign of a more wide-scale problem. If so, the message this Court sends will be particularly important. The message should be that Health Canada and third party applicants in cases such as this must take a responsible approach to disclosure and do the best they can earlier in the process. The redaction of documents could sometimes be simplified by establishing categories rather than reviewing every word. It is clear that the word-by-word approach is not working in cases such as the one at bar. A purposeful review of the file is more apposite.
If, once the parties have reviewed the file and – if possible – established categories, they do not agree, they may take the matter to a reviewing judge. An appellate court owes deference to the product of that judge’s review. The reviewing judge should not be required to provide a word-by-word, line-by-line or even page-by-page explanation for his or her decision. If appropriate, the judge can address the case on the basis of categories, provided that the judgment makes it clear which documents or categories of documents are exempt. Unless there is a palpable and overriding error, the Federal Court of Appeal, and this Court, should refrain from embarking on a review of the facts.
My concerns with having an appellate court reassess the evidence and with requiring detailed reasons are not limited to the message this sends to parties and the stringent requirements it imposes on reviewing judges when they draft their reasons. In my view, reviewing the evidence at the appeal level imports a high risk of error in a case such as this. The facts are typically reviewed in the first instance after the parties have argued on the record, and the reviewing judge will often have asked questions about specific documents. This is not, and should not be, done in this Court.
I will provide one example which illustrates the risk the Court runs in conducting a review in a case like this one. My colleague takes note of Merck’s argument that the manner in which the articles and studies are presented in the Comprehensive Summary and the fact that it relied on them at a particular stage of the development of the product would be of value to competitors. He endorses, as I do, the ratio of Janssen-Ortho Inc. v Canada (Minister of Health), 2007 FCA 252, 367 N.R. 134, which supports that argument. However, my colleague dismisses Merck’s position on the basis that Merck has agreed to the release of some of the articles and studies. Acquiescence in the publication of certain articles and studies differs from acquiescence in the publication of the Comprehensive Summary, as the latter shows how Merck relied on the articles and studies. This is what Janssen-Ortho protects. Unless the Court can show precisely where Merck consented to the disclosure of the excerpts from the Comprehensive Summary referring to the articles and studies, I do not think it is open to us to infer that it did.
Having reviewed Beaudry J.’s judgments, the record and the parties’ arguments, I am of the view that there was clearly “some evidence upon which he .... could have relied to reach [his] conclusion”. To find otherwise would be to fail to show proper deference.
For these reasons, I would allow the appeals, with costs in the Court of Appeal and in this Court, and would restore the judgments of the Federal Court, subject to any agreements entered into by the parties since those judgments were rendered.
A P P E N D I X
Access to Information Act, R.S.C. 1985, c. A-1
Loi sur l’accès à l’information, L.R.C. 1985, c. A-1
As in force at the time of the applications for judicial
Telle qu’en vigueur au moment où les demandes de contrôle judiciaire ont été présentées review
(as in force at the NDS application)
(telle qu’en vigueur au moment où la demande relative à la PDN a été présentée)
(as in force for the SNDS)
(telle qu’en vigueur au moment où la demande relative à la PSDN a été présentée)
Appeals dismissed, Deschamps, Abella and Rothstein dissenting.
Catherine Beagan Flood and Patrick Kergin (m/s Blake, Cassels & Graydon, Toronto), for the appellant.
Bernard Letarte and René LeBlanc (Attorney General of Canada, Ottawa), for the respondent.
Anthony G. Creber and John Norman (m/s Gowling Lafleur Henderson, Ottawa), for the intervener.
all rights reserved