Understanding physicians' behavior toward alerts about nephrotoxic medications in outpatients: a cross-sectional analysis

This is a review of the 2014 paper by Cho, et al.

Abstract

Background

Although most outpatients are relatively healthy, many have chronic renal insufficiency, and high override rates for suggestions on renal dosing have been observed. To better understand the override of renal dosing alerts in an outpatient setting, we conducted a study to evaluate which patients were more frequently prescribed contraindicated medications, to assess providers' responses to suggestions, and to examine the drugs involved and the reasons for overrides.

Methods

We obtained data on renal alert overrides and the coded reasons for overrides cited by providers at the time of prescription from outpatient clinics and ambulatory hospital-based practices at a large academic health care center over a period of 3 years, from January 2009 to December 2011. For detailed chart review, a group of 6 trained clinicians developed the appropriateness criteria with excellent inter-rater reliability (κ=0.93). We stratified providers by override frequency and then drew samples from the high- and low-frequency groups. We measured the rate of total overrides, rate of appropriate overrides, medications overridden, and the reason(s) for override.

Results

A total of 4120 renal alerts were triggered by 584 prescribers in the study period, among which 78.2% (3,221) were overridden. Almost half of the alerts were triggered by 40 providers and one-third was triggered by high-frequency overriders. The appropriateness rates were fairly similar, at 28.4% and 31.6% for high- and low-frequency overriders, respectively. Metformin, glyburide, hydrochlorothiazide, and nitrofurantoin were the most common drugs overridden. Physicians' appropriateness rates were higher than the rates for nurse practitioners (32.9% vs. 22.1%). Physicians with low frequency override rates had higher levels of appropriateness for metformin than the high frequency overriders (P=0.005).

Conclusion

A small number of providers accounted for a large fraction of overrides, as was the case with a small number of drugs. These data suggest that a focused intervention targeting primarily these providers and medications has the potential to improve medication safety.

Background

Medication errors are common in patients with renal insufficiency, and assisting with nephrotoxic medications represents a high-yield area for Clinical Decision Support (CDS) pharmacy intervention. The authors seek to study the patterns of prescriber override in nephrotoxic medications to help determine which alerts should be displayed, and which can be deleted.

Methods

Records were reviewed, and random warning overrides sampled for 200 warnings from high-override prescribers, and 100 warnings from low-override prescribers. The appropriateness of the override was evaluated by a panel of 6 expert clinicians.

Results

A small percentage of prescribers account for the vast majority of overrides. This group initiated an override in 90% of alerts. The appropriateness of overrides was similar in the high- and low- override frequency groups. (approximately 30%)

Conclusions

There is a high rate of alert override in the ambulatory setting, with a nephrotoxic agent CDS system. A small group of providers generated the bulk of overrides, and these were almost always inappropriate. CDS may not be adequate to address this issue, these prescribers may require more direct intervention to change patterns.

Comments

An interesting paper that reminds us that even the most carefully thought out CDS can have limited effect, if its recommendations are consistently ignored by providers. This seems to get at a deeper problem than alert fatigue, rather, the "high-override" prescribers have likely generated a near-automatic reflex to disregard CDS alerts. It remains a significant clinical problem as the risk of ADE with nephrotoxic agents is a persistent problem. This may underline the need for an effective informatics program to have some muscle, in regards to enforcing compliance through the medical staff leadership.