Thoughts on Clinical Systems: introduction

You may know HIStalk mostly for news and opinion, but I'm going to write a few related articles on clinical systems and clinical decision support. That's a topic I know fairly well, and one that I'm stimulated to report on by another excellent report from The Advisory Board Company's True North division. This one is called Hardwiring the Evidence: Customizing Systems to Drive Breakthrough Clinical Performance. I hope they don't mind that I'm using their book as a springboard. I encourage you to contact them for full information beyond the tiny bit that I'll be citing.

I have hands-on experience with a few clinical systems: CPOE, order entry, physician portals, medication management, documentation, etc. I won't reveal a lot of details about any of them and I certainly won't divulge the vendors. Needless to say, my opinions are as Mr. HIStalk and an experienced clinical IT person, not necessarily those of my employer or anyone else. If you work for a vendor and wonder if I'm going to hammer endlessly on specific systems, the answer is no. Once I give you some background in this article, I'll be zooming out to a more generalized level. There's plenty of work to be done, no matter which vendor you work for or buy sustems from.

Once upon a time, I worked for a health system with a CPOE application whose name you would recognize instantly. We were lucky enough to have quite a few physicians entering their orders into it directly. It was pushed from top administration down, an investment to help us deliver better, safer care. Many of our strategic initiatives (not just IT initiatives) involved very public implementation of clinical systems and related biomedical technology products. To their credit, adminstration understood the potential of IT to support clinical practice and allocated resources accordingly. From that lofty altitude, it sounded easy: pick a good product, implement it, and check those quality problems off as fixed. To an outsider, we were near the fnish line before many hospitals had even started the race.

Down in the trenches, it got ugly fast. The system was primitive, difficult to support, and so highly configurable that no one on the vendor side seemed to grasp the big picture of how to set it up properly. No one seemed to know why we chose this particular one, how it stacked up against the competition, and (most importantly) why we thought we could make it better than it obviously was. It was like an annoying puppy that barks all night and soils the carpet to the point that whatever family member pleaded to bring it home pretends they don't remember doing so. Outside the corner offices, it was not brought up in polite conversation for fear of instigating vocal clinical staff members to loudly and emotionally proclaim how much less safe it had made us. The IT people fled ashen-faced to their cubicles when cynical nurses accused them of causing patient harm with their poor technology decisions.

I can't say whether it was a weak product or the way we implemented it that was the problem. Probably some of each. But I can tell you that instead of the highly intelligent physician partner that vendors always claim they're selling, it was pretty much an IBM Selectric. Other than flagging patient allergies, it had no clinical intelligence. It could not detect duplicate orders, did no dose range checking, couldn't see lab results or use them for warnings or guidance, had no ability to link to outside content, contained no clinical rule-building capability, maintained no history of past orders even from the current visit (much less prior ones,) had no clinical reminder capability, didn't support standard protocols like Core Measures that are generated from diagnosis and history, and had to be interfaced to everything (often in the form of printing paper.) Like most systems, it was generally unable to support key areas with any degree of sophistication: oncology, ambulatory, surgery, pediatrics, behavioral, and ED.

All you RHIO fans take note: this is the situation in many hospitals, so don't get your hopes up that opening up a shared pipe will suddenly loose a torrent of valuable data. I think most hospitals want on board simply because they figure that their competitors surely have better clinical information than they do.

Getting back to our system. On the plus side, it was relatively easy to use. It gave physicians an easy way to find the orderable items they need, with most choices predetermined to avoid the "I have no idea, so I'll make something up" syndrome. Because everything was electronic, it eliminated legibility issues, although as I always say, so would a copy of Microsoft Word. It handled order sets pretty well, which is a topic I'll be writing on in the future. The vendor was responsive and had a good plan for gradual improvement. It was an OK foundation, although no more than that.

Still, only the Kool Aid drinkers believed it had reduced errors. In fact, we had created whole new categories of clinical gaffes. The best patient safety weapons we had -- nurses and pharmacists -- said it took time away from their patient care and error prevention activities. In their minds, it was a mistake that made the situation worse. This isn't my opinion or a criticism of the vendor, the product, or those who chose or implemented it. It's what I heard from the people on the front lines, right or wrong.

CPOE isn't a panacea. The mistakes CPOE prevents are the ones you were already catching. As AHRQ studies have indicated, the real value of CPOE is having a clinical decision support platform, getting information in front of the doc at the golden moment of order entry when far-reaching decisions are made and large amounts of money are spent. Otherwise, the act of having an MD tickle his or her own ivory just isn't worth much, which is why they resist it so vigorously. That's where I'm going with these articles ... the fact that as hard as CPOE is to implement, it's really just merging on the Interstate without having gone anywhere yet.

Back to True North. They mention some sobering facts: well-documented, no-brainer interventions for pneumonia, AMI, and other life-threatening conditions are missed much of the time, even with electronic systems. Despite CPOE posturing by Leapfrog and others, the jury's still out on whether it really improves patient outcomes. Implementation failures are legend. Several years after the IOM report that supposedly opened our eyes, clinical systems really haven't delivered on the expectations. They haven't made us much better.

Here's a True North stat (culled from a recent JAMA article) that says it all. 87% of decision support systems don't improve practitioner performance. Bet you won't find that number on the side of a HIMSS bus. Furthermore, almost all of those few systems that worked were homegrown or custom built, not the stuff that those nice guys in suits knock on your door about. Everyone is buying and implementing and improving, but the patient doesn't seem to get much benefit from all this clinical system churning. We're still plagued with poor integration, sloppily designed bolt-ons to old products, and outdated architecture. We're also not good at changing business processes on the provider side, so let's take a few lumps along with our vendors. We're equally guilty.

That's the background for now. Topics I'll address include the lack of diagnosis information, the use and mis-use of clinical alerts, the need to link to outside content, why order set development and maintenance is so important, why medical practice varies between academic and community hospitals, why vital clinical information is spread over multiple systems and is therefore unusable, the importance of clinical evidence,and why vendors probably won't be able to deliver functional systems on the old platforms they're selling.

I don't claim to be an expert, so I welcome your input. If you want to write an article, I'll run it. I created HIStalk as a vehicle to say things that need to be said, topics that won't otherwise see the light of day in advertiser-supported publications and organizations (including HIMSS.) I don't know that I'm the best person to lead the charge, but I'm one of few people waving a flag for like-minded others to gather around.