Heleen van der Sijs
Computerised physician order entry systems (CPOEs) are being widely implemented because they have been shown to result in enormous reductions in medical errors and positive effects on patient outcome.(1–6) This is due to drug safety alerts, among other things.
More safety alerts are often called for because alerts can prevent unnecessary duplicate orders and overdosages, unintended effects of interacting drugs and allergic reactions.(1,3,5) However, too many irrelevant warnings can result in “alert fatigue” and overriding of important alerts along with clinically unimportant ones.(1,2,6–9)
The main goal of CPOEs and decision support – improving patient safety – may be compromised in this way. Examples of irrelevant warnings are duplicate alerts for inpatient orders that match orders on the ambulatory medication list or alerts that warn for pregnancy in men as a contraindication. These alerts are “false positives” and should be overridden.
Overriding of drug safety alerts is very common, accounting for 49–96% of cases, except for high-level overdosage alerts, which are overridden in only 27% of cases.(10)
The most important reason for overriding is “alert fatigue” caused by poor signal-to-noise ratio – because the alert is not serious, is irrelevant or is shown repeatedly.
Other reasons include the importance of the treatment not allowing a drug change, physicians’ faith in their own knowledge or in other sources of information, incorrect information, patients’ resistance to drug change or lack of time.
Another factor that is also mentioned is alerts being overlong and difficult to interpret and clinical consequences being unclear.(10) Twenty-two percent of general practitioners admitted to overriding drug interactions without checking.(11)
These published data indicate that in order to achieve optimal patient safety, the alerting system needs to be improved considerably.
A safe alerting system has high specificity and sensitivity, presents clear information, does not needlessly disrupt workflow and facilitates safe and efficient handling.(10) A physician’s “wish list” regarding drug safety alerts is summarised in Table 1.
Attempting to improve drug safety alerting systems may lead to some problems:
- Improving sensitivity could result in unacceptable incidence of (nonspecific) alerts.
- Required entry of reasons for overriding – aimed at preventing unintended overriding – may result in an unacceptable time burden for physicians, although it provides information that is useful for system improvement.
- Noninquisitive alerts guarantee efficient handling but do not always result in higher sensitivity.
- Alternatives are preferred, but these should include the right alternative drug, dose and frequency for the patient’s particular situation. However, this is difficult to achieve.
- The ability to turn off alerts based on knowledge and specific to specialty is often asked for – but it is often forgetfulness and oversight instead of a knowledge deficit that causes an alert to be generated.
- Seriousness should be clear, but standardisation of severity levels is largely absent.
Furthermore, many aspects of the drug safety alerting process are still unclear and await research.(10,12) Examples include:
- How can drug safety alerts best be presented?
- Can several drug safety alerts be directed to other members of the healthcare team, such as nurses or pharmacists?
- How can existing (commercial) knowledge databases be made clinically valuable to users?
- Which pharmacy checks should be performed to guarantee optimal patient safety?
Although computerised drug safety alerts in CPOEs can have an enormous impact on patient safety, the ultimate goal of designing a safe alerting system that is accepted by users is still a long way off.
1. Kaushal R, Shojania KG, Bates DW. Effects of computerized physician order entry and clinical decision support systems on medication safety. Arch Intern Med 2003;163:1409-16.
2. Glassman PA, Simon B, et al. Improving recognition of drug interactions. Med Care 2002;40:1161-71.
3. Walton R, Dovey S, Harvey E, Freemantle N.
Computer support for determining drug dose.
Br Med J 1999;318:984-90.
4. Raschke RA, Gollihare B, Wunderlich TA, et al.
A computer alert system to prevent injury from adverse drug events. JAMA 1998;280:1317-20.
5. Hunt DL, Haynes RB, Hanna SE, Smith K. Effects of computer-based clinical decision support systems on physician performance and patient outcomes. JAMA 1998;280:1339-46.
6. Rind DM, Safran C, Phillips RS, et al. Effect of computer-based alerts on the treatment and outcomes of hospitalized patients. Arch Intern Med 1994;154:1511-7.
7. Goldstein MK, Hoffman BB, Coleman RW, et al. Patient safety in guideline-based decision support for hypertension management: Athena DSS. J Am Med Inform Assoc 2002;9:S11-6.
8. Shane R. Computerized physician order entry: challenges and opportunities. Am J Health-Syst Pharm 2002;59:286-8.
9. Peterson JF, Bates DW. Preventable medication errors: identifying and eliminating serious drug interactions. J Am Pharm Assoc 2001;41(2):159-60.
10. Van der Sijs H, Aarts J, Vulto A, Berg M. Overriding of drug safety alerts in computerized physician order entry. J Am Med Inform Assoc 2006;13:138-47.
11. Magnus D, Rodgers S, Avery AJ. GPs’ views on computerized drug interaction alerts. J Clin Pharm Ther 2002;27:377-82.
12. Kuperman GJ, Bobb A, Payne TH, et al. Medication-related clinical decision support in computerized provider order entry systems. J Am Med Inform Assoc 2007;14:29-40.