VA's computerized medication system does not fix the problem of adverse drug reactions

In this month's Archives of Internal Medicine an article describes what so far has not been accomplished by the VA's highly touted computerized interventions in medication delivery.
The authors ( Nebeker JR et al " High Rates of Adverse Drug Events in a Highly computerized Hospital; Arch Intern Med/Vol 165, May 23, 2005 p 1111-1116) quote the literature that indicate that specific computerized interventions can reduce medication errors and hopefully reduce adverse drug events (ADEs). The VA system has earned praise for their efforts to reduce medication errors by computerized physician order entry (CPOE) and bar coding medication delivery and EMR, automated drug interaction checking and allergy tracking and alerting. These authors deserve praise for looking at how well the emperor is clothed. They studied the ADEs in a 20 week period at the VA Hospital in Salt Lake City.
The authors address the question "Do CPOE and related systems reduce ADEs?" Since they did not compare a control period-i.e. ADE rate prior to the introduction of these various systems-they cannot determine conclusively from their data if these systems bring about a reduction in ADE rate.
Interesting they report higher rates than do other studies (the author quote their incidence density of 70 ADEs per 1000 patient days as being 5 to 19 time higher than generally previously reported). The authors doubt their rate is really higher but that they merely found more cases because of their diligence in case finding and use of clinical pharmacists to find cases and legible and accessible medical records. Understandably, they reject the notion that their systems lead to an increase in ADEs.
On a positive note, they found that their system had apparently virtually eliminated ADEs due to transcription error and bar code administration. The three most common events were constipation, low potassium and hypotension.
The authors describes the ADE rate as high the ADEs as serious and recommend improvements in computerized interventions. For example, they suggest that CPOE could suggest a order for supplemental potassium and for monitoring potassium and creatinine when loop diuretics are ordered.( I cannot help but wonder why house officers or their attendings don't already do this, Isn't that what you do on rounds? I am more concerned that house officers don't already know some pretty basic material than I am that we have not yet perfected a computer system to relieve doctors of the burden to know what they are doing) Generally they recommend more computerized decision support with such computerized systems and warn that purchasers of generic or off the shelf CPOE and bar code systems against expecting dramatic reduction in ADE rates. The entire article is worth reading for its detail both in regard to the data found and the methods used to look at errors and adverse reactions and to serve as a precautionary tale to those who think adverse drug events will be banished by buying a CPOE and having a few training sessions.