[Skip to Content]
[Skip to Content Landing]
July 13, 2005

Computers and Clinical Work—Reply

JAMA. 2005;294(2):181-182. doi:10.1001/jama.294.2.182

In Reply: Drs Harrison and Young criticize the source of the estimate for IT project failures. This is understandable because most of what we hear about, and most publicly reported information including the research literature, tends to be biased in favor of successes.1 In addition, when does a system “fail”—when it is rejected by users, when it does not perform as was promised, or when its implementation costs 3 times the budgeted figure? Good estimates are hard to find, and we opted for a conservative one. Other sources are comparable. Surveys of chief information officers conducted yearly since 1994 have suggested that only 15% to 30% of IT projects are completed successfully, on time, and within budget. About 30% are abandoned uncompleted, and the remainder have serious cost or time overruns (by a factor of 2-3) or are seriously deficient in their ultimate functionality.2,3 A Computer-based Patient Record Institute study in 1998 showed a dismal 95% failure rate in the case of electronic patient records.4 Finally, even if the proportion of successful systems is higher than we thought, that is no reason for comfort, because the system studied by Koppel et al5 would have been considered a “success” by most accounts. We agree, though, that more research is required into the area of success and failure. At the same time, we also argue that there is already much known about these failures, but that understanding is at risk of being ignored in the current climate of enthusiasm.