What are some of the most urgent cybersecurity issues facing the IT leaders at patient care organizations and health plans right now, in the wake of some recent, massive attacks on organizations worldwide? And what must healthcare IT leaders do in order to ramp up their cybersecurity strategies and implementations? Those were some of the questions pondered by industry leaders during a session entitled “Practical Tips for Creating a Cybersecurity Framework that Meets Your Privacy Standards,” on Friday, August 11. The session was held on the second day of the Healthcare Informatics Health IT Summit Series-Philadelphia, being held at the Warwick Hotel in center-city Philadelphia.
The panel was moderated by Sriram (Sri) Bharadwaj, chief information security officer and director, information services, at UC Irvine Health (Irvine, Calif.). Bharadwaj was joined by Leo Scanlon, HHS senior advisor for healthcare and public health and deputy chief information security officer in the Department of Health and Human Services; and J. Mark Eggleston, vice president and chief information security officer and privacy officer at Health Partners Plans (Philadelphia).
Early on in the discussion, Bharadwaj asked his fellow discussants, “Why do we need to make use of a framework for data security? What do we need to get done, from a framework perspective, Leo?”
“I was listening to the last panel with a great deal of interest, because the description of what’s being done” in patient care organizations “mirrors what’s been done in the federal government for the past 12 years,” Scanlon told Bharadwaj. “It’s very difficult work; we have a lot of scars. And the hardest thing is persuading people of the risks involved. We have a framework based on NIST [the cybersecurity framework from the National Institute of Standards and Technology], based on FISMA [the Federal Information Security Modernization Act of 2014]. The guideline,” in short, he said, is, “use commercial software and manage the risk. That’s basically what the FISMA statute said. NIST was tasked with developing the framework. Congress asked them to develop what became 800-53 [the Security and Privacy Controls for Federal Information Systems and Organizations, from the Joint Task Force Transformation initiative], the control set. And we wasted probably five to seven years, spending a tremendous amount of money, demonstrating that we were compliant with controls, but showing very little that we had developed the craft of risk management,” Scanlon recalled.” And then there was a revolt among federal CISOs, and at the same time, NIST was busy developing the control set. One could argue that a framework should have been developed first, but they were under pressure to develop a control set,” he said, so that’s what happened.
“The reality,” Scanlon continued, “is that we’re going to use software that is not secure and was never designed to be secure. Andin the outside world, outside segments of the military, the reality is that nothing will ever be fully secure. The key is to suggest approaches.” In the areas of the Department of Health and Human Services (HHS) focused on data security, he reported, “We’ve developed a whole array of tools and self-assessments. We use audits, and we have a staff of people who organize audits and do nothing but respond to audits. And we use those audits as the cross-check against the self-assessments in various areas, to measure maturity. The cybersecurity framework was developed to be directly connected to the NIST framework. It involves a maturity model assessment. So our big lesson,” he said, “was how to get out of a compliance mode and get out of checking boxes and writing reports, but rather, moving towards targets.”
(l. to r.) Panelists Scanlon,Eggleston, and Bharadwaj discuss cybersecurity issues
In addition, Scanlon said, “The other thing out of the federal government was an executive order, which said that cybersecurity is the responsibility of the individual agency, full stop. The executive order said, stop this, pay attention, and lead it.” Meanwhile, he said, “The third leg of what the federal government has done to support the framework is workforce development. To implement the framework, you need skill sets at every level—people who can understand these things, lead, and translate this into strategies at the upper management level. And need to evaluate tools and technologies, and apply them in the appropriate way. That’s what all of this is about, and what drives the cybersecurity framework in the federal government.”
Bharadwaj turned to Eggleston and asked him, “Mark, how do you use frameworks like this?” “We kind of stay away from the checkbox compliance piece. Studies have shown time and time again that simply checking boxes and being compliant doesn’t make you more secure,” Eggleston responded. “I’ve long been a proponent of HIPAA [the Health Insurance Portability and Accountability Act of 1996]; but HIPAA is the floor; it’s a dated protocol or law. We augment HIAPA by a series of vulnerability audits. We do quarterly tests. And we do a full external audit every two years. In actuality, I do do it every year. And I’m a big advocate of the NIST 800 series. And the CSC20—there are free protocols out there,” he said, referring to the Critical Security Controls for Cyber Defense framework. This is a framework that doesn’t require you to conduct a gap assessment. You hit the ground running with these protocols, and you know automatically that you’ll reduce your risk. And also in healthcare, you’ve got HITRUST”—the HITRUST CSF Framework from the Frisco, Tex.-based, not-for-profit HITRUST Alliance—“which is actually free.”
Moving towards an enterprise-wide risk management framework
“Here is my viewpoint of the concept of frameworks,” Bharadwaj stated. “Everybody understands the management of risk, at the board level. But the minute you talk about managing cybersecurity risk, people kind of run away. So you have a conversation with the board to help them understand that you have to have an enterprise-wide risk management framework, or ERM. The folks at the top need to understand that this is something we abide by and do. The concern we have is that the security industry has used a check-the-box approach for many years—‘I passed HIPAA’—that kind of thing. We put the cart before the horse. We didn’t talk about how we managed risk, but rather, we focused on check boxes and completing tasks.”
Bharadwaj went on to say that “The feds have released the NIST framework, but I still say that it lacks a risk management perspective, though it has marketing and communication elements in it. But in our organization, we’ve developed our own risk management asset” to address cybersecurity issues more broadly. “We went ahead and took the NIST 800 document and started working on it. But we also asked, how easy is it to fix specific problems? It doesn’t take a ton of hours to do this; it does require some management.”
“We’ve done studies of that type,” Scanlon replied. “And the SANS controls—SANS [the Bethesda, Md.-based SANS Institute] took the NIST 853 Controls, and said, look, let’s boil these groups down into one form of control. Do you basically have identity management in place, to make sure that you know what you’re doing in that area? One big part of the federal task force on that was a framework for developing a healthcare version of this. Not everyone can manage use of the NIST 853 controls. And the government is developing a common set of security controls. Fundamentally,” he said, “you need to ask, what are your assets, do you know where they are, and do you know their vulnerabilities? You need to know where to patch. If you don’t know that, you can’t fix it. Mark described upper-level capabilities; most organizations don’t have that. There’s a capability matrix. A small organization can’t consume a highly sophisticated level of automated information. This is where framework comes in, with the ability to communicate to ourselves in terms of where we are and what we need. And there’s a value to getting this sort of assessment in place.”
Further, Scanlon said, “We did an interesting study where we took the whole control set, and looked at the tools we were using. Now, most security tool vendors will give you something that looks like the NIST security list,” he continued. “Most tools can do multiple things. We know that a particular tool can do a particular thing. We were able to map our risk state and our capability state, and we were able to put a plan in place. And we’ve found in that sense, that the framework helps us to mitigate security risk management.”
“Every vendor wants to do everything, and every vendor wants to become your sole service provider,” Bharadwaj noted. “With that scenario in mind, and with vendors trying to do anything all the time, if you use a framework or checklist, how do you do that?”
“In my world, I get three resources to do what I do, people, process and technology,” Eggleston responded. “And the order of this is important. Vendors are great. They’ like to come to you and say, we can make you HIPAA-compliant, and we are HIPAA-compliant. Neither can be true. They can assist you in making yourself more compliant. And for the next two decades, we’ll see a shortage of security staff. How do you get the people? Some organizations are using SOCs [security operations centers], hosted by people you can call on, to alert you to the bad guys when they try to get into your network. And patching work isn’t very sexy, right? But it’s necessary.”
In the wake of WannaCry and Petya/NotPetya
“Well, being fully patched just got sexy after WannaCry!” Bharadwaj exclaimed. “But it’s a very different approach from going to a vendor and thinking they will manage everything. The number of people who want to get into cybersecurity is very large, but the number who can really do it is very small.” Meanwhile, he asked, “How do you really deploy your framework? I come from an academic medical center, and academic medical centers have this other side, the university, where anything goes. So for us, putting in controls in a university-type scenario is even more difficult. I cannot just block GMail; it just can’t happen. So we have to rely on people. People are part of the fabric, and they have to speak security.”
Bharadwaj went on to note that, “If people understand that security is paramount, they’ll do a good job. And helping people understand is very important for us to do. So when WannaCry happened, we sent out a note to everybody. We were not worried about our systems; we were concerned about people [computing] at home. We weren’t concerned about work, but about home. And people will be connecting to your network from home.”
“WannaCry and Petya/NotPetya were very big events for us. And we run the Indian Health Service, so we were vulnerable” as a provider system, Scanlon noted.”We can patch a thousand servers in a day. But an interesting problem in the aftermath of that is—when you try to mobilize, patching out of cycle is very, very expensive. We have contracts with SLAs [service level agreements] that determine when patches will be employed. And we have contracts that allow for emergency patching; but it’s very, very expensive. And it took us about 24 hours from the time the WannaCry virus became a well-recognized threat, to figure out where we were. And the Secretary said, just patch or get offline. You can’t just go to CMS that’s running 50,000 data centers, and tell them, shut everything off or get offline. But the key question is, what’s my patch state today? What’s my average time for propagation? Is that time window acceptable to me without invoking the emergency clauses in contracts? And you look at high-value systems. And those systems get the most attention first. It sounds like a simple thing,” he said, but it’s difficult to do: getting people to think, can I report this? And what would that look like? In many cases, it’s not worth the effort and the cost to jump that last 5 percent without thinking about it. And you’re right: patching isn’t sexy.”
Scanlon continued,” What security is all about is good operating practice. We’re training our network operators to operate efficiently and correctly. And why is it hard to patch? Because sometimes a patch breaks the system. So we’re really thinking through, how do you drive good operational practice on a day-to-day basis?”
“For us, coming from a medical center perspective, if you want to cry, go buy a medical device,” Bharadwaj said. “We’ve got tons of devices that are plugged in across the industry. And by the time that those guys understand that they can’t be running on Windows OS—it’s a pain, it’s tough. If you really want to cry, go talk to the FDA [Food and Drug Administration]. We were able to patch servers and all that, but we were relying on vendors to do the patching. And they are years behind, compared to where the industry is. And these are life-saving devices. So that’s where the challenge becomes more difficult for us. And then comes Petya, right? The next hit. And that took [a number of vendor solutions and] devices down. And we realized, you can’t outsource the responsibility to vendors. It’s your baby. And I heard a few people saying, I’ve given the responsibility to the vendor. But OCR [the Office for Civil Rights in the Department of Health and Human Services] isn’t going to knock on your vendor’s door—it’s going to knock on yours.”
“And,” Eggleston noted, “if you’re counting on regulation to save you and give you a framework, please don’t. Regulations take years to go through their processes. And the bad guys are updating their strategies daily.”