The New Health Dialogue

Image via nffcnnr on Flickr, under a Creative Commons license.

Implementing new health IT may decrease hospital mortality due to medication errors and uncoordinated care, but is the effect great enough to justify multimillion-dollar investments in new technology?  Maybe not, according to a new study from Health Affairs.

Spencer Jones and colleagues from the RAND Corporation looked at hospital outcomes data for Medicare patients from 2007 to see whether use of CPOE technology (Computerized Physician Order Entry) was associated with lower mortality for heart attack, heart failure, and pneumonia.  They looked at 4,644 general acute care hospitals, categorizing them by the percentage of orders written using CPOE.  There was a small reduction in mortality for heart attack and heart disease, but not pneumonia, associated with CPOE use – really small.

Based on the results, the authors predicted “complete uptake by hospitals of electronic medication ordering at levels comparable to the requirements of stage 1 meaningful use (at least 30 percent of orders written using CPOEs) could result in 1.2 percent fewer deaths of hospitalized Medicare beneficiaries from heart failure and heart attack,” but this result was not statistically significant.

The authors’ model did predict a statistically significant 2.1 percent drop in mortality for hospitals that meet the future Stage 2 requirement of 60 percent CPOE usage.  These results give the authors, “measured optimism that computerized provider order entry, when used frequently, may have the potential to reduce hospital mortality rates.”

These disappointing results are at odds with the hype around electronic medical records in general, and CPOE systems in particular. A two percent reduction in heart attack mortality in hospitals is not insubstantial – with around 1.5 million heart attacks in the US each year leading to 500,000 deaths, CPOE could save 10,000 lives each year. That assumes, though, that all hospitals implement the technology and physicians use it for at least half of their orders.

Achieving a reduction of 10,000 heart attack deaths would be a success, sure, but the size of that success, relative to the investment and the hype that CPOE receives, throws cold water on the idea that electronic medical records are the sole solution to the problem of soaring medical costs.

There remains some ambiguity in the health effects of CPOE systems that could mean they’re more effective than they appear. It’s possible this study didn’t account for other factors that can affect mortality. Hospitals with high CPOE utilization may be fundamentally different from low utilizers in ways that explain their failure to see a larger drop in mortality. On the other hand, it could also be that what little reduction they did see was due to something other than CPOE, such as participation in other quality improvement efforts. 

Advocates for health IT have argued that the technologies could save many more lives and save money. Now, maybe CPOE and other health IT will have a bigger effect when it’s more completely implemented, but so far it’s hard to see a 2 percent decrease justifying the millions spent on these systems. And that makes it tough to know whether or not they save money. 

This is illustrative of a major problem in the US health care system: new technology is often perceived to be an unambiguous good, and hospitals routinely invest in innovations that produce, at best, marginal improvements in patient outcomes – and at worst, no improvement or even worse outcomes. Think DaVinci surgical robot for prostate surgery, which is more expensive than open surgery and thus far has not produced better results. This is not meant to reject the drive to innovate, but rather, to urge caution before sinking billions of dollars in new technology that might or might not improve outcomes.