21 May Data management: Capturing the data you really need
Madison, Wis. – Data classification modeling can be an integral part of any data storage strategy, but it’s an exercise that may involve as much emotion as it involves hardheaded business trade-offs.
How organizations classify different sets of data, distinguishable by value and age and legal (think e-discovery) factors, is part of an unavoidable business discussion when setting the policies and processes that guide data-archiving decisions.
Given the stress on computing systems, the relative affordability of storage, and the eagerness of physicians to defend professional reputations, there will be tough decisions in the healthcare realm.
In the view of healthcare CIOs like Aurora Health Care’s Phil Loftus, the sheer volume of data entering healthcare computer networks from picture archiving systems and electronic medical records will require periodic prioritization, or classification, exercises.
That’s because best practices never really stop developing. “What we are beginning to do is rethink what you really want in clinical record retention in the electronic age,” Loftus said. “The old maxim that all data is good and more data is better is not a good maxim.
“You have to think about what data you really want to capture and how long you should retain what you capture.”
Degrees of separation
John Barto carries the title “healthcare evangelist” with Sun Microsystems, but his advice on storage can apply to any business organization with established business processes.
“I think data must be classified based on how it’s used in your workflow,” he said. “It involves balancing the cost of keeping instant access versus its value in terms of workflow. Is the image of a broken arm from two years ago really that important in an emergency department?”
Vendors have competing versions of an approach that separates the main data system, which holds immediate critical data, from the storage system. Barto characterized it as more of a “policy-based model,” and the advantages of this model are that it requires thought leadership in the development of business processes, and it provides the flexibility to expand storage separately from the main data system.
In the healthcare realm, the policy discussion might pertain to a situation where an X-ray is taken by a 64-slice imaging machine, and the physician only uses three of those slices to make a diagnosis. That begs the following question: Are the 61 unused slices automatically archived?
“The policy-based model is integrated with business processes,” Barto said. “The policy is defined by the institution and it determines the value of data by type. You have to classify data in terms of where it should reside in various stages of its life cycle.
“The policy piece is definitely a business discussion that you have to have.”
Rethinking vital signs
During the recent Digital Healthcare Conference, Loftus said the volume of data cannot simply be allowed to grow forever without increasing the risk of system failure. Some data has to be archived to a backup system, he said, and data classification decisions should be evaluated with cost, risk, and clinical productivity in mind.
Citing the records of deceased patient and old vital sign data as examples of information that can be archived on a separate storage medium or even purged, he said not everything captured on electronic medical records must be permanently retained in the mission-critical systems that meet day-to-day computing needs.
“The ICU captures this [vital sign] information every 15 minutes,” Loftus said. “Once you’re discharged, who is going to want to know what your vital signs were at 15 minute intervals on any particular day?”
Even medical images lose their value over time, Loftus noted, and that should prompt healthcare organizations to think about when those space-consuming images should be archived.
When it comes to archiving, Loftus acknowledged there probably are legal barriers, but there are much larger emotional ones. It’s important for the IT department to be aware and sensitive to issues physicians might have regarding malpractice and the like, but understand that there are trade-offs to be made for the benefit of computer system performance.
Loftus’ mindset is that if the policy is to retain images for five years before they are placed in back-up storage, he wants to reduce that to four years.
“We see it as a multi-year journey, but I want to educate them [caregivers] as to the risks and options of the trade-offs,” he said.
Storage and archiving decisions also are impacted by the occasional necessity to retrieve electronic information as part of an e-discovery request. The revised Federal Rules of Civil Procedure have brought a new, electronic dimension to court proceedings and government investigations, requiring organizations to revisit document retention and incident-response policies.
This requires more in-depth thinking of data management and storage practices, with the possibility of a legal summons guiding those thoughts.
“They will go looking and they will try,” said Michael Best & Friedrich attorney Erik Phelps, on the likelihood that plaintiffs will want to comb through electronic documents. “You will be second-guessed, but the question, “Will we have discovery problems?” is a fact you will want to consider in your retention policy. Absolutely.”
Tale of the tape
When it comes to archiving, the use of “spinning disk” storage technology often is preferred to tape storage. It’s faster to retrieve records from the spinning disks and the magnetic quality of tape has a tendency to fade over time, requiring occasional migration to newer tapes that have greater density and better performance.
As part of quantifying savings from a storage-archiving strategy, Barto noted that tape has an interesting dynamic for companies looking to introduce more ecology-minded businesses processes – it reduces energy consumption. “You need power to spin those disks,” Barto said. “In terms of being a good ecological citizen, tape really gives you a lot of economy.”
• DHC 2008: Electronic data management a top CIO concern
• Gartner’s Top 10: Wisconsin companies progress to hot strategic technologies
• Tom Still: Medicaid data management a healthcare IT success