Is independent double-checking superior to single-checking in medication administration error detection? A randomised controlled simulation trial.

Abstract

Background: Double-checking of medication administration is a safety practice used in hospitals around the world. Independence is recommended as the key to effectiveness. Independent double-checking (IDC) requires each nurse to separately check the five rights (eg, right drug, dose). There is no empirical evidence that IDC is more effective in error detection than a single-nurse check.

Objectives: To compare the effectiveness of IDC versus single-checking in detecting medication errors during administration, assess the time required and explore factors influencing performance, including nurse experience and social dynamics.

Methods: A multicentre randomised controlled simulation trial with 82 nurses from two paediatric hospitals. Forty-one 2-hour simulations each involved two nurses exposed to single-checking and IDC conditions administering medication sets containing embedded errors. Condition order and medication sets were randomised. Primary outcomes were error detection and time taken. Multivariable models compared primary outcomes by condition. Secondary analyses examined the influence of nurse experience and social loafing, defined as reduced individual performance when working in pairs. Social loafing was assessed by comparing error detection performance when nurses single-checked with error detection in the role of first checker in IDC.

Results: Across 1160 dose administrations with 640 error detection opportunities, nurses detected 72.2% (95% CI 68.6 to 75.5) of errors. Overall, IDC yielded an 11% higher error detection rate than single-checking (adjusted incident rate ratio (aIRR): 1.11, 95% CI 1.01 to 1.21; p=0.02; 77.7% (95% CI 72.9 to 81.9) for IDC vs 66.3% (95% CI 60.9 to 71.4) for single checks). However, this benefit was limited to experienced nurses. Among early career nurses (<5 years clinical experience), IDC produced no significant improvement in error detection (aIRR: 0.98, 95% CI 0.86 to 1.11; p=0.7; IDC 67.0% (95% CI 57.3 to 75.4) vs single checking 73.0% (95% CI 63.6 to 80.7) error detection). In contrast, experienced nurses showed a 26% improvement in error detection with IDC (aIRR 1.26, 95% CI 1.11 to 1.44; p<0.001; IDC 82.5% (95% CI 77.0 to 86.8) vs single-checking 63.2% (95% CI 56.5 to 69.4)). Evidence of social loafing was observed, particularly among early career nurses whose error detection performance was 26.0% higher when single-checking compared with their performance in the role of first checker in IDC. IDC required significantly more time per medication set (mean difference 4.96 min (95% CI 2.10 to 7.83; p<0.001)).

Conclusions: IDC improved error detection for experienced nurses, but not for early career nurses. Given its inconsistent benefits, resource demands and susceptibility to social loafing, IDC may be unsuitable as a universal safety strategy. Strengthening single-checking competence and supporting clinical judgement may offer a more effective, scalable approach to improving medication safety.