Translating the implementation of computerized physician order entry (CPOE) into clinical transformation has long been a complex, tricky matter, and published studies have come to different conclusions regarding how much of a contribution CPOE can really make to clinical performance improvement in hospitals.
Significantly, clinicians at the 303-bed Lucile Packard Children’s Hospital (LPCH) at Stanford University in Palo Alto, Calif., have definitively researched and documented the steady lowering of mortality rates among inpatients at their facility. In the past, Healthcare Informatics has interviewed Christopher Longhurst, M.D., LPCH’s CMIO, regarding his leadership with fellow clinicians in this important area.
Longhurst recently made a formal presentation on the progress that LPCH clinicians have made to date, as part of the program at the AMDIS Physicians’ Symposium, convened annually by the Association of Medical Directors of Information Systems, the nation’s sole formal CMIO association. On July 13, Longhurst presented the latest findings from his and his colleagues’ research and performance improvement work, including citing statistics from recently published clinical journal articles he had authored, for his fellow CMIOs gathered in Ojai, Calif. for the symposium.
As Longhurst reported to his CMIO colleagues in Ojai, he and his fellow clinicians and analysts at LPCH performed a formal study that analyzed mortality rates at the hospital prior to and following CPOE implementation. The marquee finding: the hospital’s mean monthly adjusted mortality rate decreased by 20 percent following its 2007 CPOE implementation. That statistic translates to an extrapolated 36 fewer deaths in the 18-month post-implementation timeframe.
But as Longhurst noted in his July 14 presentation at the AMDIS symposium, the greatest potential in this area lies in a combination of continuous analysis, data reporting and feedback, and efforts to translate findings into changes in care delivery that will improve care delivery at the bedside. And in an interview he gave exclusively to HCI Editor-in-Chief Mark Hagland during the symposium, Longhurst discussed with Hagland the broader context of the work he and his colleagues are doing to leverage clinical data and reporting intelligently, which encompasses the concept of clinical resource management and clinical transformation. Below are excerpts from that interview.
One of the things you said in your presentation and have said to this publication is that CIOs, CMIOs and other healthcare IT leaders need to look beyond the tools to the strategic objectives involved. Can you elaborate on that thought?
Certainly. A lot of CIOs get all wrapped up in the sophistication of data warehouse products. But we didn’t have any super data warehouse; we built the data from the transaction system. But we built service line-level profiles, by APR DRG, and by looking at the variability within a given diagnosis, and looking at benchmarks from national children’s hospitals.
What percentage of people actually take the time after they implement CPOE to look at outcomes? And how can I alter the order sets, the rules and the alerts, the clinical decision support [CDS], to help drive outcomes change? Those are among the important questions here. And frankly, a lot of people want to do that on a CDS committee, but I’ve never seen it driven by a CMIO; it needs to be driven by a top-level intensivist or other clinician leader. Whereas the clinical resource management committee will say things like, how many resources are we using for this diagnosis? And, we need to drive towards this outcome.
And often on a CDS committee, CMIOs will say, we should put in this alert, and we should measure pre and post…the CRM committee, in contrast, drives towards interventions towards outcomes. So you put in a bundle, as from IHI, and it’s not about the science, it’s about the outcome. So this is similar.
Had anyone else used the term ‘clinical resource management’ in this kind of context, before you and your colleagues did so at LPCH?
Yes, the term first emerged in a 1997 article in the Journal of Quality and Safety, by Alan Rosenstein. It’s actually a fairly common term, but is often used to slap the doctors on the wrists. But if you read that article, you’ll find that Rosenstein says there are only four ways to affect costs—whether or not you admit patients, whether or not you do procedures and what types, ancillary resource utilization, and length of stay. And ancillary resource utilization is driven entirely physician orders.
So the full range of influences on resource use has not been thoroughly looked at?
What is clear to me is that CPOE can have affects beyond medication safety, which everyone focuses on. In fact, our hypothesis is that there’s a whole other world driven by physician utilization that can be driven by rules and alerts. So transfusion and lab utilization are examples. We also looked at respiratory therapy utilization, some nursing orders, the cost of pharmacy products, things like that.
Based on everything you’ve been saying, it seems that CIOs need to begin to think more broadly, more strategically, and in a more outcomes-focused way, correct?
Yes, that’s part of the message. I do think the secret sauce on the CRM committee is focusing on outcomes and driving towards that. But I also think there are several corollary messages there. One is that even without a fancy, sophisticated data warehouse, there’s a tremendous amount of information that can be gleaned from data reports. Another is that even if you have the data warehouse and you don’t have the clinician leadership driving this, not your CMIO, it doesn’t matter if you have [the best-known vendor in the data warehouse space]! And third, people have been talking about CPOE and medication safety for years. But let’s think beyond medication safety. And every order entered triggers resource utilization.
In a sense, some of that seems obvious, and yet as you’ve pointed out, few have pursued such questions.
There was a New England Journal article in the early 1990s that showed that if you put dollar signs on lab orders, you change ordering patterns. On antibiotic selection, we put one to four dollar signs on the microbiology susceptibilities involved, so that when a culture comes back and shows how resistant or sensitive that culture is to antibiotics, it will show the relative costs, and if there are two equivalent antibiotics, it will encourage you to choose the one that’s less expensive.
This is all more complex in children’s hospitals, isn't it?
Yes, because that's where pediatric drug dosing is tougher. But in everything else, the children’s hospital is a good model for other hospitals, because the kids who get admitted most often tend to have chronic illnesses, just as in the adult world.