The Need for Speed (In Breach Protection)

Posted on April 26, 2016 I Written By

The following is a guest blog post by Robert Lord, Co-founder and CEO of Protenus.
Robert Protenus
The speed at which a hospital can detect a privacy breach could mean the difference between a brief, no-penalty notification and a multi-million dollar lawsuit.  This month it was reported that health information from 2,000 patients was exposed when a Texas hospital took four months to identify a data breach caused by an independent healthcare provider.  A health system in New York similarly took two months to determine that 2,500 patient records may have been exposed as a result of a phishing scam and potential breach reported two months prior.

The rise in reported breaches this year, from phishing scams to stolen patient information, only underscores the risk of lag times between breach detection and resolution. Why are lags of months and even years so common? And what can hospitals do to better prepare against threats that may reach the EHR layer?

Traditional compliance and breach detection tools are not nearly as effective as they need to be. The most widely used methods of detection involve either infrequent random audits or extensive manual searches through records following a patient complaint. For example, if a patient suspects that his medical record has been inappropriately accessed, a compliance officer must first review EMR data from the various systems involved.  Armed with a highlighter (or a large excel spreadsheet), the officer must then analyze thousands of rows of access data, and cross-reference this information with the officer’s implicit knowledge about the types of people who have permission to view that patient’s records. Finding an inconsistency – a person who accessed the records without permission – can take dozens of hours of menial work per case.  Another issue with investigating breaches based on complaints is that there is often no evidence that the breach actually occurred. Nonetheless, the hospital is legally required to investigate all claims in a timely manner, and such investigations are costly and time-consuming.

According to a study by the Ponemon Institute, it takes an average of 87 days from the time a breach occurs to the time the officer becomes aware of the problem, and, given the arduous task at hand, it then takes another 105 days for the officer to resolve the issue. In total, it takes approximately 6 months from the time a breach occurs to the time the issue is resolved. Additionally, if a data breach occurs but a patient does not notice, it could take months – or even years – for someone to discover the problem. And of course, the longer it takes the hospital to identify a problem, the higher the cost of identifying how the breach occurred and remediating the situation.

In 2013, Rouge Valley Centenary Hospital in Scarborough, Canada, revealed that the contact information of approximately 8,300 new mothers had been inappropriately accessed by two employees. Since 2009, the two employees had been selling the contact information of new mothers to a private company specializing in Registered Education Savings Plans (RESPs). Some of the patients later reported that days after coming home from the hospital with their newborn child, they started receiving calls from sales representatives at the private RESP company. Marketing representatives were extremely aggressive, and seemed to know the exact date of when their child had been born.

The most terrifying aspect of this story is how the hospital was able to find out about the data breach: remorse and human error! One employee voluntarily turned himself in, while the other accidentally left patient records on a printer. Had these two events not happened, the scam could have continued for much longer than the four years it did before it was finally discovered.

Rouge Valley Hospital is currently facing a $412 million dollar lawsuit over this breach of privacy. Arguably even more damaging, is that they have lost the trust of their patients who relied on the hospital for care and confidentiality of their medical treatments.

As exemplified by the ramifications of the Rouge Valley Hospital breach and the new breaches discovered almost weekly in hospitals around the world, the current tools used to detect privacy breaches in electronic health records are not sufficient. A system needs to have the ability to detect when employees are accessing information outside their clinical and administrative responsibilities. Had the Scarborough hospital known about the inappropriately viewed records the first time they had been accessed, they could have investigated earlier and protected the privacy of thousands of new mothers.

Every person seeks a hospital’s care has the right to privacy and the protection of their medical information. However, due to the sheer volume of patient records accessed each day, it is impossible for compliance officers to efficiently detect breaches without new and practical tools. Current rule-based analytical systems often overburden the officers with alerts, and are only a minor improvement from manual detection methods.

We are in the midst of a paradigm shift with hospitals taking a more proactive and layered approach to health data security. New technology that uses machine learning and big data science to review each access to medical records will replace traditional compliance technology and streamline threat detection and resolution cycles from months to a matter of minutes. Making identifying a privacy breach or violation as simple and fast as the action that may have caused it in the first place.  Understanding how to select and implement these next-generation tools will be a new and important challenge for the compliance officers of the future, but one that they can no longer afford to delay.

Protenus is a health data security platform that protects patient data in electronic medical records for some of the nation’s top-ranked hospitals. Using data science and machine learning, Protenus technology uniquely understands the clinical behavior and context of each user that is accessing patient data to determine the appropriateness of each action, elevating only true threats to patient privacy and health data security.