Can CPOE Actually Increase Mortality?
Pediatrics recently published a study by the Children’s Hospital of Pittsburgh (CHP) and the School of Medicine at the University of Pittsburgh, Pennsylvania which blamed a recently-introduced Computerized Physician Order Entry System (CPOE) for an unexpected increase in mortality among children transported for specialized care. These findings understandably caused quite a stir; after all previous studies found positive outcomes in terms of (sometimes dramatically) reducing medication prescribing errors (MPEs), adverse drug events (ADEs) and rule violations (RV) in pediatric inpatients (though ADEs were reduced less significantly). Even the very same hospital had good things to say in another article published less than two months ago.
Since it has been estimated that up to 2% of US inpatients are harmed by medication errors, including serious ones sometimes leading to death, it might seem logical that reducing them would lower mortality as well. Yet this study adds up to a hard-to-dismiss list of articles about Health information technology (HIT) projects gone wrong. After all, many IT projects in other industries fail to deliver some or all of their intended benefits despite towering costs, and reduced feature scopes as well as budget and planning overruns are frequently heard of.
Within the past couple of weeks though, several people involved in health care and/or healthcare IT disputed the study’s conclusion. Contributors from VA Palo Alto Health Care System, Stanford University School of Medicine, Pacific Business Group on Health, The Leapfrog Group and Harvard Medical School (no less) responded in this post-publication peer review by saying other system changes and implementation problems have introduced a bias in the study which thus cannot prove causation between CPOE and mortality. The issue is less with the body of the study itself (which did disclose plenty of caveats) than with its summary and lead conclusion. The CHP introduced a centralized pharmacy in parallel with the CPOE, and several technical and workflow problems with the hospital’s implementation of the CPOE got also in the way of delivering drugs in time to the children who needed them. Several methodological limitations might also have affected the study’s accuracy, such as an unexplained much shorter post-implementation period than the timeframe observed before the CPOE rollout (which itself seems to have been done unusually fast).
The HIStalk weblog also chimed on this issue, with interesting points about CPOE in general and the Cerner software used in this particular case. Though supportive of CPOE in principle and not overly critical of Cerner, the blog’s author and his readers agreed that vendors generally do not pay due attention to usability and ease-of-use which can make tasks longer and more error-prone. The CHP’s problems with its CPOE echo issues and disappointments met by many commercial companies implementing Customer Relationship Management (CRM) software. That does not mean CPOE in itself cannot lead to better health practice including lower mortality, if properly designed and implemented.
December 19, 2005 Related topics: Quality, Safety, Errors, Pediatrics, IT & software