Aims The term ‘medication error’ has numerous definitions, impeding comparison between studies and is susceptible to subjectivity.¹ The Delphi Process is widely used in health research to achieve consensus and has been previously used to define and specify medication error scenarios in both paediatric and adult settings.^2,3 Novel technology-generated errors are emerging with increasing use of health information technology (HIT).⁴ Application of earlier Delphi studies to novel errors and those common in the prescribing of infusions in paediatric intensive care is limited. This study aims to achieve consensus on medication error scenarios identified in a paediatric intensive care unit (PICU) that have not been previously defined.

Methods Stage 1 identified the scenarios requiring consensus. These were grouped into 3 error categories: electronic prescribing, smart-pump and prescribing of PICU infusions. Stage 2 selected a multidisciplinary expert panel using both purposive and convenience sampling. Stage 3 involved iterative rounds of consensus using paper-based and newer e-Delphi techniques. Participants independently scored on a 9-point scale their extent of agreement on the inclusion of each scenario as an error. Median and inter-quartile ranges were used to assess group consensus and to provide controlled feedback after each round.

Results 19 scenarios requiring consensus were identified. A panel of 37 participants was selected, comprising of 15 doctors, 13 nurses and 9 pharmacists. 35 participants were from the study site, 1 pharmacist from a local PICU and 1 from a local NICU. Round 1 achieved consensus on 11 scenarios, increasing to 14 in Round 2. Round 3 consisted of 2 scenarios, both electronic prescribing related. Individual opinion on these was diverse, with 1 remaining equivocal after round 3. Some differences between healthcare professionals were found, but were only significant (p<0.05) for two and three scenarios in rounds 2 and 3 respectively.

Conclusion The Delphi Process can successfully be employed to reach consensus on HIT-generated novel errors. The complexity of electronic prescribing systems is evident in the included errors and the difficulty in obtaining consensus. In contrast, the broad consensus reached on all smart-pump scenarios reflects the known risks associated with infusion pumps. The included scenarios highlight the limitation of smart-pump technology as a single intervention. Further similar studies are likely to be required as more novel errors emerge with increased HIT implementation across the entire medication use process. This extended tool should add to the quality of future paediatric medication error studies across a broad range of settings.

References

1. Lisby M, Nielsen LP, Brock B, et al. How are medication errors defined? A systematic literature review of definitions and characteristics. Int J Qual Health Care2010;22(6):507–18.

2. Dean B, Barber N, Schachter M. What is a prescribing error?Qual Health Care2000;9(4):232–7.

3. Ghaleb MA, Barber N, Dean Franklin B, et al. What constitutes a prescribing error in paediatrics?Qual Saf Health Care2005;14(5):352–7.

4. Walsh KE, Landrigan CP, Adams WG, et al. Effect of computer order entry on prevention of serious medication errors in hospitalised children. Paediatrics2008;121(3):e421–7.