Skip to main content

Table 1 Rates of medication administration errors

From: The impact of a novel medication scanner on administration errors in the hospital setting: a before and after feasibility study

Type of error Pre-intervention number of errors (% of OE) Post-intervention number of errors (% of OE) 95% CI* p value* 95% CI (adjusted for heterogeneity) for the coefficient in the fitted logit model corresponding to the `period' indicator. For the difference in error rate pre- and post-intervention† p value (adjusted for heterogeneity) †
Total errors 739 (69.1) 302 (69.9) − 4.6 to 5.9% 0.773 − 2.2 to 0.1 0.0753
Non-Timing errors 51 (4.77) 11 (2.55) − 4.2% to 0.1% 0.050 − 2.0 to 0.2 0.115
Omission errors 17 (1.6) 4 (0.9) − 1.9% to 0.9% 0.326 − 2.3 to 1.3 0.579
Omission errors due to lack of ward stock 9 (0.8) 3 (0.7) − 1.1% to 1.3% 1.0 − 3.4 to 3.3 0.976
(% of total number of omission errors in that period) − 52.9 − 75 − 30.6% to 55.6% 0.486   
Other administration errors 34 (3.2) 7 (1.6) − 3.2% to 0.4 0.096 − 2.6 to 0.8 0.295
Wrong Dose 1 (0.09) 0 (0.0) − 0.6% to 0.8% 1.0 Result could not be generated -
Documentation Error 28 (2.6) 7 (1.6) − 2.5% to 0.9% 0.263 − 3704.1 to 3728.2 0.9949***
Wrong Form 5 (0.5) 0 (0.0) − 1.1% to 0.5% 0.183 − 1714.3 to 1694.2 0.9908***
Timing errors 688 (64.4) 291 (67.4) − 2.6% to 8.3% 0.282 − 1.6 to 0.5 0.318¢
Early 1–2 h% of timing errors 25 (2.3) 34 (7.9) 3.1% to 8.6%  < 0.00001 3.1 to 15.7 0.00341
  − 3.6 − 11.7 4.4% to 12.6%  < 0.00001   
Early > 2 h% of timing errors 3 (0.28) 0 (0.0000) − 0.9% to 0.7% 0.286 − 247,877.8 to 247,831.9 0.9999
  (0.44) (0.0000) 1.3% to 1.0% 0.271   
Late 1–2 h% of timing errors 568 (53.1) 229 (53.0) − 5.8% to 5.5% 0.9711 − 1.3 to 2.7 0.493
  − 82.6 − 78.7 − 9.7% to 1.5% 0.159   
Late > 2 h% of timing errors 92 (8.6) 28 (6.5) − 4.9% to 1.0% 0.174 − 8.0 to − 0.7 0.0188
  − 13.4 − 9.6 − 7.9% to 0.9% 0.103   
  1. OE- opportunity for error
  2. *p value reported using the Z-pooled Exact Test (exact unconditional tests for 2 × 2 contingency tables)
  3. p value adjusted for heterogeneity including possible correlation effects within nurses, patients, and observers
  4. ¢ We did not identify a significant impact of MedEye on the overall rate of timing errors but did note a significant decrease with nurse time on duty. It is possible this was due to the busier morning drug rounds, which increased the likelihood of nurses making a mistake
  5. We also noted a strongly significant decreasing effect of nurse time on duty on late 1–2 h errors. This was possibly associated with calmer and quieter drug rounds that occurred later in the day
  6. ***Fitted models do not include Nurse time on duty as models do not fit otherwise
  7. °There were no reports of the following ‘other—error subtypes’: wrong patient, wrong administration equipment used, wrong medication error, administration without order, route error, failure to recognise drug-drug interaction, patient had a documented allergy to medication, directions/ monitoring error