Comparison of potential risk factors for medication errors with and without patient harm

Purpose: To compare determinants for medication errors leading to patient harm with determinants for medication errors without patient harm.
Methods: A two-way case-control design was used to identify determinants formedication errors without harm (substudy 1) and determinants for medication errors causing harm (substudy 2). Data of patients admitted to five internalmedicinewards of two Dutch hospitals during 5months were collected prospectively by chart review. Medication errors were detected and classified by two pharmacists. Consensus between five pharmacists was reached on the causal relationship betweenmedication errors and patient harm.Data analysis was performed bymultivariate logistic regression.
Results: We included 7286 medication orders, of which 3315 without errors (controls), and 5622 medication errors without harm (cases substudy 1) and 102 medication errors causing harm (cases substudy 2) were identified. Hospital, ward and the therapeutic class antiinfectives were associated with both medication errors without harm (hospital odds ratio (OR) 1.40; 95% confidence interval (CI) 1.21-1.63), TweeSteden hospital (TSh) geriatrics OR 2.03; 95% CI 1.73-2.38, TSh general internal medicine OR 1.44; 95% CI 1.23-1.69 and anti-infectives OR 1.28; 95% CI 1.06-1.56) and medication errors with harm (hospital OR 4.91; 95% CI 3.02-7.79, TSh geriatrics OR 5.76; 95% CI 2.52-13.15, TSh general internal medicine OR 6.51; 95% CI 2.82-15.02 and anti-infectives OR 4.20; 95% CI 2.24-7.90). - Conclusions: This study shows that organisational determinants (hospital, ward) are comparable for medication errors with and without harm. For conclusions on patient- andmedication-related determinants studies with larger sample sizes are needed.