Leveraging the Cloud in Radiology
Health-care providers today have a heightened interest in streamlining operations and reducing capital outlays to maintain or improve margins. In an increasing number of cases, cloud-based solutions and software as a service are being embraced by the radiology sector to meet these challenges—but not before a careful analysis that goes beyond cost considerations has been performed. For some radiology players, a move to leverage the cloud stems from a desire to manage an increasing volume of imaging data more effectively, while simultaneously grappling with tight capital and operating budgets. Orlando Health in Florida—a 1,780-bed, not-for-profit private health network with 14,000 employees, multiple inpatient and outpatient facilities located across Central Florida, and 2,000 affiliated physicians—ranks among them. Approximately two years ago, Orlando Health’s IT department acknowledged that shared cloud storage might, based on perceived total cost of ownership (TCO), constitute a more viable means of addressing multiple data-storage challenges than a traditional product would. Brian Comp, MBA, PMP, is Orlando Health’s chief technology officer, information services. In calculating TCO, Comp and his team took into consideration not only the anticipated initial capital outlay, but maintenance costs and the very real possibility of the need to upgrade Orlando Health’s data-storage technology down the road. The health network currently maintains two self-hosted data centers in Orlando that handle more than 200 terabytes of data. About 25% of total data storage is needed for PACS and for a separate cardiology PACS. Comp says, “Besides these demands, our data rate had been growing by 50% annually and continues to do so. Storage growth allocations in the IT capital budget range from approximately $1 million to $1.5 million per year—a tight squeeze—and changes in reimbursement mean that capital is harder to find.” Comp and his team also looked at potential power savings in weighing whether Orlando Health would be better served by a cloud-storage solution or by an installed hardware/software configuration. He deems power and cooling for the data center significant issues in Florida, noting that these expenditures are likely to be three or four times higher than those incurred by providers in many other locations. “We figured that by reclaiming electricity that would otherwise be needed to power the systems and cool the equipment, we could do 25% better than now,” Comp reports. “That was a strong justification.” The department, however, wasn’t willing to base its decision entirely on economics; other concerns came into play. Orlando Health wanted reasonable assurance that only authorized parties would have access to data stored in the cloud, that cloud-storage performance levels would rival those provided by its own data centers, and that the vendor would, as Comp put it, be around for the long haul. “We are still in the midst of figuring out the best cloud-storage model for us and discussing these issues with potential vendor partners, but the analysis definitely pushed us in that direction,” Comp states. “In our long-range plan for the next five to 10 years, our first step is to move our long-term storage to the data center of the cloud provider. We reduce the requirement for capital investment, we should be able to reclaim some electric power, our data-center floor space will be freed, we are hoping that our technical-support needs will be at least somewhat reduced, and we will be able to maintain our failover capability. If we have a loss at a primary data center, we want to be able to continue to provide service to our clinicians.” He notes that Orlando Health’s planned transition to the cloud probably would not be possible if less-expensive, higher-bandwidth cloud-based solutions had not been introduced in recent years. “The original higher-bandwidth options were cost prohibitive—and thus, out of our reach,” he explains. “No analysis would have supported adopting them.” Access Insurance Alta Vista Radiology (Phoenix, Arizona) also saw the cloud as a panacea for its image-management and -storage ills. The provider offers teleradiology services to 80 hospitals in the Southwestern, Southeastern, and Midwestern regions of the United States, as well as along the East Coast. Alta Vista uses a PACS vendor’s software as a service to access a hosted version of the vendor’s PACS. The service is available on a pay-per-click basis. Robert Ortega, MD, is Alta Vista Radiology’s CEO; he and his team looked at a number of factors in conducting cost justification for a hosted PACS (rather than a traditional PACS). A TCO analysis took into consideration the initial financial outlay for equipping a data center, as well as the costs of ongoing management, technical support, and anticipated upgrades. Because Alta Vista Radiology has two administrative employees whose responsibilities touch IT, but has no dedicated IT staff, the potential cost of hiring someone to manage the data center and the PACS, along with the costs of finding and training a replacement for that individual (should he or she leave the organization), were factored into the equation. Ortega says, however, that the degree to which a hosted PACS would suit the provider’s business model and structure was a far more important criterion when contemplating a jump onto the cloud bandwagon. He notes, “Our model didn’t—and never will—include having a staff of network engineers, but our service promise necessitates near-perfect uptime. Our geographically diverse structure necessitated a data-center–based approach so that we are ready to read, no matter where hospitals are located, but we didn’t want to take the risk of having our own. Given the tight fit, we would have gone with the hosted PACS and the cloud even if that option cost us more than the do-it-yourself approach.” While there was no concern about potentially extricating Alta Vista Radiology from its relationship with the vendor, based on the latter’s longevity in the market, other considerations were addressed in advance of signing on the dotted line. Ortega questioned how flexible the vendor would be, in terms of setting and enforcing minimum-usage volume requirements. A guarantee from Alta Vista Radiology of a future uptick in usage volume enabled it to obtain the desired wiggle room at the beginning of the contract. Concerns about uninterrupted access to the hosted PACS were addressed by writing a guarantee of 99.99% uptime into the contract. The presence of a secure VPN connection between facilities and the data center appeased any worries about data compromise. Like Comp, Ortega believes that enhancements in Internet capabilities are pushing the cloud envelope, both within and beyond the radiology realm. He says, “Fiber-optic connectivity has done great things for uptime.” That, in turn, has bolstered the appeal of the cloud. Share and Share Alike Other health-care providers see the cloud as the solution for image sharing. They include Baystate Health (Springfield, Massachusetts), which operates four hospitals in the state—the 659-bed Baystate Medical Center in Springfield, Baystate Children’s Hospital within Baystate Medical Center, Baystate Mary Lane Hospital in Ware, and the Baystate Franklin Medical Center in Greenfield. In addition, the system operates Baystate Radiology and Imaging and a number of other services and health centers. Baystate Medical Center is Western New England’s only level I trauma center and tertiary-care center. Its emergency department is the second-busiest such facility in Western Massachusetts, according to Michael Favreau, director of imaging and special projects. In January 2010, after inking a deal with a startup image-sharing company, Baystate Health took a first step by installing the company’s product on its private network; this includes disc viewers, in addition to in-boxes that allow authorized users to hold data files obtained from patients’ CDs, as well as the tools needed to view, store, and share these files. Along with medical images, the files can include physicians’ notes and reports. Favreau explains, “We were looking for a better way to manage inbound CDs. Although they may have been DICOM compatible, they couldn’t always be read if we put them into a computer, leading to delays in patient care (and often, the need to expose patients to repeat studies).” The solution helped immensely, making it possible to view 95% of the CDs that accompanied patients to Baystate Medical Center. It wasn’t enough, though: Baystate Health wanted to facilitate the sharing of images with community physicians and hospitals outside its network. Favreau and his colleagues quickly concluded that implementing any solution other than one centered on the cloud would not make sense, from a logistical standpoint. The health system had neither the staff resources nor the capital to build or buy (and manage) its own image-sharing application—especially one to which round-the-clock access was a must. Decision makers, however, determined that the hospital network’s IT budget and department could absorb the cost of a subscription-based image-sharing solution that was—by virtue of requiring minimal intervention by IT personnel and having minor equipment requirements—well within financial reach. Accordingly, over the past few months, Baystate Health has moved image sharing to the vendor’s cloud, enabling referring hospitals to send imaging exams, reports, and associated documents to specialized departments at Baystate Medical Center through a secure online conduit. The data arrive in online drop boxes accessible only to authorized users who enter a password. Drop boxes, to which two hospitals can currently upload data, have recently been established for trauma and vascular services. Ultimately, the platform will serve as a conduit among five hospitals and for six to eight service lines, Favreau reports. “It’s still evolving,” he says. Favreau emphasizes that a decision to handle image sharing via cloud should involve more than logistical and financial considerations and analyses. Open lines of communication must be established with any entity that will be sharing images with a given provider, and security concerns must be addressed with the solution vendor. When the idea of migrating to the cloud first appeared on Baystate Health’s agenda, Favreau and his colleagues approached the potentially connected hospitals to discuss the project and explain how online image exchange would expedite and enhance patient care. The IT department requested (and received) from the image-sharing vendor a detailed description of security standards and protocols. Tighter security around the Internet has made it a far more appealing platform for data sharing of any kind, Favreau concedes, emphasizing that the financial aspects are not all important. “Security, however, has to be scrutinized up front and is not worth sacrificing,” he adds. Expanding Reach Radiologic Associates of Fredericksburg (RAF) in Virginia has turned to the cloud in an effort to support an expansion of its services reach. The provider has a staff of 32 radiologists and two vascular surgeons; data pertaining to patients served at two hospitals, four outpatient imaging centers, and an interventional-radiology clinic reside in a conventional hospital-owned PACS—managed by a hospital-based IT department and backed up by an on-site database. At one time, RAF provided the bulk of its interpretation services to MediCorp Health System (Fredericksburg), now known as Mary Washington Healthcare, which owns the hospitals to which RAF provides service in Fredericksburg and surrounding cities. Not long ago, however, it contracted to provide radiology services to an entity outside the hospital network. Neil Green, MD, director of nuclear medicine, says, “Because we were dealing with a group od urgent-care facilities not affiliated with Mary Washington Healthcare, we could not incorporate patient images and data into our hospital’s PACS and RIS. We asked ourselves how we could interpret the images easily.” The practice looked at the cost of adding dedicated workstations and a full-blown RIS; in addition to the hardware expenditure, this would have required maintaining a separate database on-site, with expensive server hardware (and no dedicated IT staff to support it), Green notes. Also factored in was the cost of system upgrades, which would probably become necessary as other nonaffiliated entities sought RAF’s services. The TCO did not compare favorably with that of a hosted, turnkey, browser-based PACS, leading RAF to implement a Web-based image-management solution that allows on-site and off-site image availability. Along with the PACS capability, Green says, the solution offers the functionality of a RIS by enabling referring physicians to perform order entry. A number of preimplementation meetings were held to iron out what Green calls “critical issues and concerns, when you’re talking about the cloud.” For example, the vendor was asked to provide training on the system for RAF’s staff, along with assurance that the solution would be sufficiently scalable to accommodate data-volume growth. An exit strategy—in case RAF ever wants to migrate images out of a separate system—was also developed, as were set triage procedures (in the event of system difficulties). “The cloud is not an answer for everyone,” Green says. Nonetheless, he adds, if the price of training, maintenance, upgrades, troubleshooting, and related factors is compared with the cost of a turnkey solution, it becomes clear that the cloud is an excellent alternative. Julie Ritzer Ross is a contributing writer for Radiology Business Journal.