The furore in Canada over the cancellation of the long-form census and the subsequent elation over its reinstatement in 2016 illustrates that – well – that Canadians get excited about odd things, such as being counted for statistical purposes. Of course, not all Canadians are enthusiastic about the census. Each census period a few objectors refuse to complete the long-form census, and some are even prosecuted for their refusal. While some opposition has been based on the past involvement of defense contractor Lockheed Martin in conducting the census (this involvement apparently ended for the 2016 census), other objections have been linked to privacy concerns. Perhaps because of the extensive measures in place to protect census privacy, these concerns have gained little traction either publicly or in the courts, although they did provide the former conservative government with an excuse to cancel the long-from census.
A recent Federal Court decision considers issues of privacy and the census in a somewhat different context. In O’Grady v. Canada (Attorney General), the objection was not to the census itself, but rather to the secondary use of census data for medical research. The applicant, Kelly O’Grady, objected to an agreement that had been entered into between Statistics Canada and McGill University’s Faculty of Medicine in 2011. This agreement, like others of its kind, provided the legal framework by which medical researchers could use Stats Canada data in population health research. The McGill project seeks to assess infant mortality and newborn health in Canada by linking perinatal outcomes with risk factors related to socioeconomic status, ethno-cultural background, and environmental conditions. The researchers needed to link a sample of births from the national birth record database with data from the 1996 and 2006 national censuses.
The collection and maintenance of census data is governed by the Statistics Act, which also establishes Statistics Canada. Stats Canada does not simply hand over data of this kind to researchers. Under the terms of the agreement with McGill, Stats Canada would make the linkages between the records, and then would provide researchers with access only to de-identified information. Further, only those researchers who were either employees or deemed employees of Stats Canada would have access to the data. Under the Statistics Act, “deemed employees” are individuals who are brought under the umbrella of the Act, who must swear oaths of office that affirm that they will comply with the Act and maintain confidentiality, and who are subject to penalties under the Act for any breaches of their obligations.
The applicant objected to the use of the census data under the terms of the Agreement. She argued that it violated of the Statistics Act and the federal Privacy Act. She argued that census data could only be shared with express consent of those who had shared their personal information, and this had not been obtained. Further, she maintained that under the Privacy Act government institutions can only share information without consent in narrowly limited circumstances, and only where the disclosure is consistent with the purposes for which the information had been collected. She argued that the census information had not been collected for medical or public health research, and therefore could not be disclosed for these purposes.
The applicant had complained to the Office of the Privacy Commissioner in 2012, arguing that her personal information had been improperly used in the study. In a 2014 decision, the Privacy Commissioner agreed that the applicant’s census data constituted her personal information, and also found that census information was being used in the study for purposes that went beyond those for which it was collected. However, the Commissioner had noted that the Statistics Act expressly permitted Stats Canada to use its data in this way. Perhaps more importantly, the Commissioner found that the applicant’s own personal information had not been used in the study. The Applicant had given birth within a period that would have been captured by the study, but she did so in Ontario, and the Ontario data had been excluded from the study because of concerns regarding its quality. The Commissioner concluded that the applicant’s complaint was not well-founded.
The fact that the applicant’s personal information had been excluded from the study was an important factor. The Federal Court found that the exclusion of her data meant that she had not been – nor could she ever be – personally affected by the study, and ruled that she did not have standing to bring this application. Further, Justice Russell noted that “[t]he issues she raises and argues can only really be decided on a set of facts that includes an applicant or applicants who were directly affected, or who may be directly affected by the Study when it is eventually released” (at para 52). He noted that there was, as yet, simply no indication that any personal information had been or would be improperly disclosed as a result of the study. He also observed that there was “no indication that the Applicant’s position is anything more than her own personal position, born of her academic interests and her social activism” (at para 52).
Despite ruling that the applicant had no standing in the matter, Justice Russell nevertheless considered the merits of the application. He found that it was clear that Stats Canada had not disclosed any personal information – whether of the applicant or any other person. Only employees and deemed employees of Stats Canada had access to the raw data for the purposes of creating the data linkages. The linked data was accessible only to employees or deemed employees of Stats Canada. Other members of the McGill research team only saw non-confidential aggregate data. Justice Russell noted that the applicant had provided no evidence to show how the aggregate data could be linked to specific individuals. Although the applicant had argued that postal code data was going to be provided to the researchers in order to enable them to assess environmental factors, Justice Russell ruled that the applicant’s claim that the postal code data could be used to re-identify individuals was nothing more than an assertion. Further, he noted that there was no evidence that any postal code data had been revealed to anyone who was not an employee or deemed employee of Stats Canada.
Justice Russell also considered the argument that the disclosure of the data violated the Privacy Act because it was not for a purpose for which it had been collected. He agreed that the census data was personal information. However, he found that while the specific purpose of using the data for this study was not formed at the time of its collection during the 1996 or 2006 censuses, the purpose of the study “is to compile and analyse statistics related to the health and welfare of Canadians”, and this was a consistent with both the purpose of the census and the mandate of Stats Canada. There was therefore no inconsistency with the terms of the Privacy Act.
Although he dismissed the application, Justice Russell cautioned that this was primarily because it both involved an applicant with no standing and was premature. It was premature in the sense that it was too early to know if any personal information might be improperly disclosed. He stated that his decision “should not prevent anyone whose personal information is inappropriately used or disclosed from bringing the matter before the Court in the future” (at para 86). The bottom line, therefore, is that individuals whose interests are directly affected by inappropriate actions by Stats Canada or by researchers will have recourse to the courts. However, there is little room to raise broader privacy arguments about the use in principle of Stats Canada data in appropriate research.