Several weeks ago I posted the following comment about the difference between the CBO’s and RAND’s analysis of EMR/EHR value:
“The PROBABILITY is that when many organizations take on the very complicated, expensive and difficult task of implementing an EHR, some of them will fail. And those failures will bring down the average expected benefit. Since more failures occur early in a technology’s life cycle, the average benefit for all previously implemented EHRs may be very small, or even negative.
The demonstrated POSSIBILITY is that an individual hospital or system can realize substantial benefits from an EHR. The growing body of evidence of EHR benefits exists mainly in individual studies of components of an EHR (such as CPOE, safety alerts, or electronic clinical documentation), or specific types of benefits (such as ADE prevention, or order communication timing, or nursing efficiency). Taken together, these studies contain compelling proof of EHR benefits.“
To further develop this point: It’s difficult to distinguish between EHR failure rates that apply to physician practices vs. hospitals, but here are some of the percentages that are quoted:
- Oh no! Half of all current EMR’s fail! Technology for Doctors, 2007
- 19% of EMRs are uninstalled; 30 percent are not used by at least some physicians, Medical Records Institute, 2007
- Avoiding EMR meltdown: How to get your money’s worth (“About a third of practices that buy EMR systems stop using them within a year”) AMNews, 2006
- The failure rates of EMR implementations are…close to 50%.” Proceedings of the 11th Annual Symposium on Health Information Management Research, 2006
- “Industry experts estimate that failure rates of EMR implementations range from 50-80%.” A Commonsense Approach to EMRs, 2006
- “50% of EMR system implementations result in failure.” International Journal of Technology Assessment in Health Care, 1997
To use the lowest figure cited, let’s say that approximately twenty percent of hospital EHR installations fail and the system is removed. In these cases it’s likely that the entire investment in hardware, software, implementation costs, lost productivity (and other costs more difficult to measure) is lost. If this average investment is ten percent of annual revenues (or $100 million for a $1 billion IDN) the other 80 percent of EHR implementations must generate a return of 2.5 percent of annual revenues each just to get back to even.
And if the failure rate is 40 percent then the other 60 percent must generate a return of almost 7% of annual revenues for the collective return to be zero. But wait, there’s more: only the clear system failures, those that generate little or no benefit, are likely to be ripped out. The partial failures, those systems that are only partly used, where CPOE and physician documentation are never used, or where some physicians never use the system, may limp along for years, generating a slightly negative or slightly positive return. And the percentage of these “low benefit” installations is likely to be at least as large as those hospitals where the EHR is uninstalled.
Under these conditions, the relatively small number of very successful installations is not enough to pull the average EHR return into positive territory, and the CBO is right in concluding that “By itself, the adoption of more health IT is generally not sufficient to produce significant… savings”.
But what if you could increase the chances of success? Make it much more likely that your EHR would succeed than fail? Then the benefits demonstrated by the successes are well worth the investment. That’s the question at hand for hospital executives: “Can we do better than the average?” I believe the answer is yes, especially since the bar is set so low, but it will require a different approach to EHR implementation, one focused on system value (benefits) instead of just technical success or process changes.
