Cyber, secrecy and the public body

Here’s a copy of a presentation I gave yesterday at the High Technology Crime Investigation Association virtual conference. It adresses the cyber security pressures on public bodies that arise out of access-to-information legislation, with a segment on how public sector incident response differs from incident response in the private sector

IPC/Ontario – Appropriate for hospital to notify of breach because it maintained a shared EMR

The IPC/Ontario has issued a significant decision about information governance under the Personal Health Information Protection Act. Specifically, it held that a hospital that gives a physician access to an electronic medical record for use in private practice is a health information custodian together with the physician, but that it can retain a duty to notify of a breach arising out of the private practice.

Background

The hospital maintained an EMR system and gave access to its credentialed physicians and their employees for use in private practice. Employees in two such private practices accessed EMRs without authorization. The hospital notified affected patients and reported the breach to the IPC, which led the IPC to investigate.

In the course of investigation it came to light that some of the employees had shared their login credentials with others outside of the hospital, but apparently to enable health care. The employees also apparently accessed some records (for non-health care purposes) with the consent of friends of family members. Both of these actions violated hospital policy.

Decision

The IPC held that the access enabled by credential sharing and the access made with the consent of family members was made in breach of PHIPA. Although a more benign form of unauthorized access, the IPC found a breach based on section 10(2) of PHIPA, which states, “A health information custodian shall comply with its information practices.”

Regarding the identity of the custodian, the IPC held that both the hospital and the two private practice physicians were custodians in the circumstances – the physicians being custodians “when they access patient information in [the EMR] for the purpose of privatizing health care to their private practice patients.” Such access, the IPC explained, invites a disclosure by the hospital and a collection by the physicians; in this context the physicians were not the hospitals’ agents.

Despite the physicians’ custodianship, the IPC held it was appropriate for the hospital to notify in the circumstances. It said:

[122]   In the cases under review, THP and the private practice physicians also treated THP as the health information custodian responsible for notifying affected individuals of the private practice employees’ unauthorized accesses in THP’s EMR. In these circumstances, I agree that THP was the appropriate party to give notice under section 12(2) of PHIPA. As the health information custodian who maintains the EMR, THP was best placed to discover and investigate the extent of the employees’ activity in the EMR, identify all the parties whose personal health information had been accessed without authority, and initiate contact with these individuals, all of whom are THP patients, but some of whom may not have any relationship with the particular private practice physician for whom the employee worked. In these cases, notification by THP was appropriate, taking into account not only the language of section 12(2)[29] but also the interests of the affected individuals.

[123]   I also agree with THP that in some circumstances, notification by the collecting custodian may be more appropriate, and a reasonable approach to fulfilling the notice obligation in section 12(2). For example, in a case where the private practice physician has a more significant relationship with the patient whose privacy was breached, notice from that physician (rather than from the custodian who disclosed the information) may be prudent. So long as the notice is given as required upon the events described in section 12(2) (and complies with the other requirements of that section), I agree with THP that circumstances such as the patient’s interests and the relationships between the patients and the various custodians involved may be relevant factors in deciding how best to fulfil the notification obligation. I am not persuaded that applying such an approach to notification in future cases would have the consequences of discouraging hospitals from adopting EMR technologies, or from participating in broader initiatives like a provincial electronic health record system.

Implications

The kind of shared accountability invited by this decision can cause confusion and risk. It will behoove hospitals and other custodians who provide shared access to their EMR systems to be very clear and detailed in establishing who is responsible for what. The hospital in this case, for example, decided post-incident to make more clear that physicians who are given outside access are responsible for training and supervising their employees. It also expressly obligated physicians to participate in privacy investigations arising from the actions of an employee.

The IPC’s finding on who provides notification is very qualified, and rests partly on the fact that the hospital in this case voluntarily provided notification to affected individuals. While taking control of notification may be beneficial to hospitals who maintain and provide third-party access to EMR systems, providing notification may also signal responsibility for a breach and for the related risks for which hospitals have little or no ability to control. The hospital in this case dealt with this tension by stipulating to its physicians that they may be named in hospital notification letters “as being responsible for the breach.” Other hospitals, may wish to require physicians to notify themselves in certain circumstances. The IPC’s decision does not appear to preclude such alternatives.

Trillium Health Partners (Re), 2020 CanLII 15333 (ON IPC).

Legal Privilege and Data Security Incident Response – Law and Practice

I’m off to a cyber conference in Montreal this week to sit on a panel about threat exchanges. My role will be to address the legal risks associated with sharing threat information and a university’s ability to effectively assert a confidentiality interest in the same information. I’m genuinely interested in the topic and have prepared not just one, but two papers!

Here is the first one – a nuts and bots presentation on privilege and data security incident response. I hope it is useful to you. Feedback welcome through PMs.

Experts, privilege and security incident response

I’d encourage you to read David Fraser’s blog post from last weekend – The value of legal privilege: Your diligent privacy consultant may become your worst enemy.

David’s basic point is sound: structuring a security or privacy expert retainer to support a privilege claim can prevent your own expert’s advice from being used against you. Most often this is done by having legal counsel retain an expert in anticipation of litigation and for the dominant purpose of litigation, with instructions and conclusions going strictly between counsel and expert.

David explains a scenario in which an organization retained an expert to advise on some form of due diligence connected to a subsequent security incident. The expert was apparently quite candid in its written advice, outlining a security problem that amounted to what David compares to a “dumpster fire.” The organization responded partly but not wholly to the expert’s recommendations. That expert’s report will therefore become, as David says, the plaintiff’s Exhibit A.

Being faced with your own expert’s advice is very bad, hence the soundness of David’s point. My additional point: legal privilege is no solution to a bad client-counsel-expert relationship.

The views on what is a reasonable investigation or remediation in the data security context can vary widely between equally qualified experts. Too often, perhaps driven by conflicting interests, security experts recommend what’s possible and rather than what is “due.” A breach coach can help address this problem, identifying trusted experts and working with them to reach a shared and acceptable understanding of the due diligence required in responding to a security incident. With such a relationship, departing from an expert’s recommendations (even though they are privileged) represents a real and meaningful risk. The facts – i.e., the things done based on an expert’s recommendations – are never privileged. If litigation ensues those facts will be picked apart by other experts, and you want the good ones to view the facts the same way as you and your trusted advisor.

Experts that are prone to floating long lists of options need to be retained under privilege because they are dangerous, but even under privilege their advice is worth little. The prescription: do everything you can to build a great client-counsel-expert relationship. Use a breach coach. Keep a roster of trusted experts on retainer. Don’t use experts retained for due diligence advice to do the very remedial work they recommend.

In snooping investigations, disclose the logs

When an employer confronts an employee with an allegation of improper access to personal information, it is important to give the employee the event log data that proves the allegation. It may often be voluminous and difficult to interpret, but presenting a general allegation or summarizing events without particulars will give the employee a good reason to deny the allegation.

This is what happened in this very illustrative British Columbia case in which an arbitrator held he could not infer dishonesty from the grievor’s initial failure to admit wrongdoing because the grievor had not been given log data. Also, if an employee continues to deny responsibility, log data can be difficult to rely upon; even if it can be established to be authentic, there are issues about presenting log data in a meaningful and privacy-protective way. An early admission can go a long way.

Fraser Health Authority (Royal Columbian Hospital) v British Columbia Nurses’ Union, 2017 CanLII 72384 (BC LA).

Cyber insurance and incident response practice

Here’s a deck from a Monday panel presentation that I participated in with some colleagues from the sector.  It features a cyber incident scenario and some questions. See if you can answer them, and if you’d like to have a discussion, please comment or get in touch.

Alberta CA demands greater scrutiny of privilege claim re internal investigation

On July 4th the Court of Appeal of Alberta held that a chambers judge erred by accepting a claim that all documents created or collected in the course of an internal investigation were privilege without conducting a record-by-record analysis.

Legal counsel for the company initiated the investigation after a workplace fatality and directed the investigation team to segregate the investigation documents and to endorse all material as privileged and confidential. Legal counsel later swore that the dominant purpose of the investigation was the contemplation of litigation, which the chambers judge said, “invariably and logically leads to the collateral finding that, within the context of Suncor’s internal investigation that was carried out in anticipation of litigation, the information and documents created and/or collected during the internal investigation with the dominant purpose that they would assist in the contemplated litigation, are integrally covered by litigation privilege.”

The Court of Appeal held that the chambers judge erred by not conducting an analysis about the reason for the creation of each record (or bundle of records). It explained that statements may have been taken, for example, under a standing workplace protocol or that surveillance video or business records may have been collected – and that neither kind of record would be the subject of a proper privilege claim.

Alberta v Suncor Inc, 2017 ABCA 221 (CanLII).

Court approves settlement, limits recovery of class counsel fees

On August 29th, Justice Perell of the Ontario Superior Court of Justice approved settlement of an action brought against Home Depot following a significant 2014 payment card system intrusion. The Court approved a settlement that featured a $250,000 non-reversionary settlement fund for documented claims of “compromise” and an agreement to pay up to $250,000 in credit monitoring. It also denied payment of approximately $407,000 in (docketed) legal fees to class counsel as unjustified, approving instead, payment of $120,000 in fees.

This is a good outcome for organizations exposed to potential class action claims for data security incidents. It was driven by two factors: (1) the Court found the incident was associated with a limited risk of damage; and (2) the Court was impressed by Home Depot’s incident response.

Regarding damage, the Court assessed the risk of damage flowing from a compromise to payment card information and e-mail address information as minimal:

[46] Professor Archer outlined three heads of damage to consumers from a payment card breach:  (1) the risk of a fraudulent charge on one’s credit card; (2) the risk of identity theft; and (3) the inconvenience of checking one’s credit card statements. The so-called non-reversionary Settlement Fund of $250,000 is designed to provide compensation for these heads of damages.

[47] Of the three heads of damage, practically speaking, there is little risk of fraudulent charges because of sophisticated safeguards developed by credit card companies. Moreover, when there are frauds, the losses are almost always absorbed by the credit card company or the retailer. The credit card companies are not Class Members.

[48] In the immediate case, there is no evidence that a Class Member absorbed a fraudulent charge. Neither Merchant Law Group nor McPhadden Samac Tuovi LLP have been contacted by a putative Class Member who said that he or she suffered a financial loss attributable to the data breach.

[49] There is also little risk that the data breach, including the disclosure of email addresses, increased the risk of identity theft, because the stolen data would have been inadequate to allow a criminal to fake another’s identity.

[50] Mr. Hamel’s evidence was that for identity theft, the most important information to have is a government-issued identification number such as a driver’s licence number, social insurance number or passport number and preferably all three. In the immediate case, the data stolen from Home Depot did not include this information.

[51] As for inconvenience damages, in the immediate case, there are none, because credit card holders are already obliged to check their statements for fraudulent purchases.

(Note that the Office of the Information and Privacy Commissioner of Alberta has recognized that the loss of e-mail address is associated with a risk of spear phishing – a risk that is arguably remote.)

Regarding incident response, Home Depot had offered to pay for a number of fraud protection services following the incident – including credit monitoring, identity theft insurance and credit repair services. The Court commented that this reduced the need for behavior modification:

[100] The case for Home Depot being culpable was speculative at the outset and ultimately the case was proven to be very weak. The real villains in the piece were the computer hackers, who stole the data. After the data breach was discovered, there was no cover up, and Home Depot responded as a good corporate citizen to remedy the data breach. There is no reason to think that it needed or was deserving of behaviour modification. Home Depot’s voluntarily-offered package of benefits to its customers is superior to the package of benefits achieved in the class actions.

These two factors led the Court to place little value on the action or the settlement. Justice Perell (who is outspoken), commented, “I would have approved a discontinuance of Mr. Lozanski’s proposed class action with or without costs and without any benefits achieved by the putative Class Members.”

Lozanski v The Home Depot, Inc., 2016 ONSC 5447 (CanLII).

Saskatchewan health authority criticized for slow incident response

Good incident response involves nailing your timing – not going too fast or too slow. 

On August 17th the Saskstchewan Information and Privacy Commissioner held that a health authority breached the Saskatchewan Health Information Privacy Act by failing to respond to an incident in a timely manner. 

The Commissioner’s report does describe a dilatory response – with a discovery of “snooping” in mid October 2015, an investigation that led to a paid suspension at the end of January 2016, notification to the Commissioner at the end of February 2016, notification to the Commissioner towards the end of March that the breach was bigger than first reported and eventual notification to affected individuals in July 2016. 

Think and don’t react, and you can even pause to momentarily to gain confidence in a next critical step, but always keep the ball moving.

Investigation Report 030-2016 (17 August 2016, Sask OIPC).  

The Saskatchewan OIPC okays health authority’s incident response

On June 8th, the Office of the Saskatchewan Information and Privacy Commissioner issued an investigation report in which it held that a regional health authority responded appropriately to a privacy breach. Most notably, the OIPC reinforced a recommendation about notification included in its 2015 publication, Privacy Breach Guidelines. The recommendation:

Unless there is a compelling reason not to, [health information] trustees should always notify affected individuals.

This is a novel and conservative variation on the normal harms-related principle that guides notification. It is simply a recommendation – and one directed only at public agencies and health information trustees in Saskatchewan. It is notable nonetheless, however, in that it reflects an arguably developing public sector norm. Right or wrong, there is a unique pressure on public sector institutions to notify that should always be considered as part of a public sector institution’s careful response to a data handling incident.

Investigation Report 101-2016 (8 June 2016).