Electronic prescribing and patient safety: a rigour-stratified systematic review of medication error, adverse drug events, workflow, and communication

Abstract

Background Electronic prescribing is promoted to improve medication safety. Whether its benefit differs between process failure and patient harm is rarely asked. Objective To synthesise the evidence on electronic prescribing and medication error, adverse drug events, workflow and communication against the designs able to establish each. Methods We searched PubMed, Scopus, Web of Science, CINAHL, Embase and ProQuest (1994 to 2024, English), reporting to PRISMA 2020 and PRISMA-S. Risk of bias was assessed at domain level with design-matched tools. Non-poolable metrics precluded meta-analysis; findings were synthesised following SWiM. Results Forty studies were included, 31 quasi-experimental. None of the 31 was at low risk of bias; 27 moderate, four serious. Twenty-nine reported medication error, most a reduction of 20% to 98%. Four could be read against background change: the reduction held in three, and in one exceeded the crude difference, errors having risen beforehand. Three others had a concurrent comparison only. The remaining 22 measured no trend, so magnitude is uninterpretable. No design able to test an effect on harm showed one. Eight studies documented error classes introduced by the system, two quantifying them at 68% and 35% of residual errors. Workflow effort was redistributed rather than reduced, and communication disrupted short-term, nurses most affected. Conclusions Electronic prescribing reduces measured medication errors; the reduction holds in three of the four studies able to test it. An effect on patient harm has not been demonstrated. Certainty was very low for both quantitative outcomes, low for workflow, moderate for communication.