Taxonomy of delays in the implementation of hospital computerized physician order entry and clinical decision support systems for prescribing: a longitudinal qualitative study

Hajar Mozaffar, Kathrin M. Cresswell, Lisa Lee, Robin Williams, Aziz Sheikh, On behalf of the NIHR ePrescribing Programme Team

Implementation delays are common in health information technology (HIT) projects. In this paper, we sought to explore the reasons for delays in implementing major hospital-based HIT, through studying computerized physician order entry (CPOE) and clinical decision support (CDS) systems for prescribing and to develop a provisional taxonomy of causes of implementation delays. We undertook a series of longitudinal, qualitative case studies to investigate the implementation and adoption of CPOE and CDS systems for prescribing in hospitals in the U.K. We used a combination of semi-structured interviews from six case study sites and two whole day expert roundtable discussions to collect data. Interviews were carried out with users, implementers and suppliers of CPOE/CDS systems. We used thematic analysis to examine the results, drawing on perspectives surrounding the biography of artefacts. We identified 15 major factors contributing to delays in implementation of CPOE and CDS systems. These were then categorized in a two-by-two delay classification matrix: one axis distinguishing tactical versus unintended causes of delay, and the second axis illustrating internal i.e., (the adopting hospital) versus external (i.e., suppliers, other hospitals, policymakers) related causes. Our taxonomy of delays in HIT implementation should enable system developers, implementers and policymakers to better plan and manage future implementations. More detailed planning at the outset, considering long-term strategies, sustained user engagement, and phased implementation approaches appeared to reduce the risks of delays. It should however be noted that whilst some delays are likely to be preventable, other delays cannot be easily avoided and taking steps to minimize these may negatively affect the longer-term use of the system.