View source for Implementation of a clinical decision support system for computerized drug prescription entries in a large tertiary care hospital

This is an original article by Zenziper et al. (2014) in the Israel Medical Association Journal, entitled “Implementation of a Clinical Decision Support System for Computerized Drug Prescription Entries in a Large Tertiary Care Hospital”. <ref name="Zenziper"> Zenziper et al. (2014). Implementation of a Clinical Decision Support System for Computerized Drug Prescription Entries in a Large Tertiary Care Hospital. Israel Medical Association Journal, 16:289-94.  http://www.ncbi.nlm.nih.gov/pubmed/24979833.</ref>

== Introduction ==

On average, every hospitalized patient becomes a victim of medication errors once daily.  Medication errors, especially prescription errors, are most commonly associated with adverse drug events. This not only increases morbidity, mortality and healthcare costs, but increases the length of hospital stay. Implementation of a computerized physician order entry system alone, does not provide necessary alerts, such as drug-drug interactions, dose errors, or duplicate therapy. Therefore, [[computerized physician order entry]] (CPOE) coupled with a clinical decision support system ([[CDS]]) ensures [[alerts]], creating potential to decrease the risks associated with erroneous prescriptions. However, altered workflows and excessive alerts are common usability challenges, which desensitize the physicians for meaningful use of EHRs, so-called “alert fatigue,” a variant of the “cry wolf” effect. 

== Purpose ==
The paper discusses the customization of clinical decision support systems in a large tertiary care hospital in order to minimize alert fatigue, as well as increase acceptance of full EHR adoption, aside from incentivization by the government. Customizing CDSS to individual departments needs and reducing the volume of alerts are vital for improving the CDSS’s signal-to-noise ratio, thereby decreasing barriers for compliance. 

== Methods ==

The methods are divided into five sections: 

*Description of the drug prescription CDSS
*Alert types
**Dose alerts
**Renal and dose adjustments alerts
**Drug-drug interaction alerts
**Duplicate-drug therapy alerts
*System review and adjustability
*Implementation and CDSS customization
**Moderate interactions
**Severe interactions
**Specific recommendation for follow-up
**Changes of alert wording and typography
*CDSS alert volume

== Results ==

When alerts were prioritized by minimizing drug-drug interaction alerts, alert fatigue declined. For example, combinations that are familiar and easy to recall, such as aspirin and clopidogrel in the internal medicine department, were silenced.  These were categorized as ‘soft alerts’.  On the other hand, dosing alerts that signify potential threat to the patients were expanded. For instance, both poor renal and hepatic function, and low body mass index triggered dosage alerts, which were classified as ‘hard alerts’. In addition, alerts without meaningful recommendations were not significant for clinical decision making.  Rather, alerts accompanied with specific recommendations for follow-up enhance the effectiveness of patient care. 

== Conclusion ==

The large scale of drug prescriptions is common among hospitalized patients. A prescription error is the most common cause of medication errors and the resultant adverse consequences.  Despite the advancements in information technology, comprehensive benefits of electronic health records have not been fully achieved due to common challenges, such as alert fatigue and disrupted workflow. Tailoring of commercially available CDSS to suit needs is the best way to overcome the common challenges of EHR use. This mission is possible only if the interprofessional team works together with end-users. Future endeavors should be focused on forming guidelines for drug alert silencing and exploring other user-friendly features in clinical decision support systems. 

== Comments ==

The authors conducted a quasi-experimental study involving a mean of 339 ± 13 patients per month per department for 7 months. Some 11.2 prescriptions were issued per patient, 30% triggering ≥1 CDSS alerts, commonly drug-drug (43%) and dosing (38%) alerts. The intervention during the period consisted of silencing or modifying 3981 alerts, based upon detailed prior knowledge of important vs clinically insignificant interactions and very important dosage adjustments in the presence of renal or hepatic disease. The review committee silenced or modified 3981 alerts which enhanced clarity, and provided specific dosing modifications with recommendations for follow-up, in place of vague warnings. With the collaboration of end-users, the researchers accomplished their goal of reducing the number of alert notices to a ≤30% threshold necessary to avoid alert fatigue, lower significant medication errors, and increase acceptance of the CDSS alert system. 
This particular study was chosen because a common problem was identified and corrected using universally available and inexpensive tools in a realistic manner.



== Related articles ==

*[[CPOE]]
*[[CDS]]
*[[Alerts]]
*[[Electronic health record–based decision support to improve asthma care: a cluster-randomized trial]]

==References==
<references/>

[[Category: Reviews]]
[[Category: EHR]]
[[Category: CPOE]]
[[Category: CDS]]