Table 2

Key failure modes for executing the do, study and act steps

PDSA stages	Key failure modes	Potential consequence
Do Implement the plan (including both the QI intervention and the data collection plan)	Failure to implement the QI intervention as intended27 36	Impossible to learn whether the planned QI intervention works as expected; wasted effort; disillusionment among staff involved with intervention design
	Failure to collect the data as intended27 36	Undercuts the Study phase; may be difficult or impossible to tell whether the intervention worked as expected; difficult or impossible to learn about the effectiveness of the original data collection plan
	Failure to capture unanticipated learning17 22 27	Missed learning opportunities (especially for qualitative learning about how and why the intervention did/did not work); project failure; unnecessary PDSA cycles
	Failure to abandon the Do phase despite manifest failure or severe negative side effects24	Wasted effort; excessive disruption; adverse outcomes from side effects
Study Analyse data and compare results to the definition of success; distil and communicate what has been learned from the formal data analysis and unanticipated learning	Failure to conduct a study5 or inappropriate failure to follow the study plan	No/limited opportunity to learn whether the intervention works as intended; potential for biased and misleading results
	Failure to communicate what has been learned27 46	Loss of stakeholder engagement; reinventing the same broken wheel in the service of other QI projects; loss of institutional knowledge if there is turnover among project leaders
Act Based on what has been learned, either: Revisit the investigation and problem framing phase Begin a new PDSA cycle at the Plan phase Fully implement and sustain the intervention or End the project without investing further effort	Failure to engage in ‘double loop learning’17 that questions the goals of the project in light of what has been learned	Wasted effort continuing to work on the wrong problem, or one that cannot realistically be solved; Excessive PDSA cycles spent trying to achieve a goal that is set too high, when a more realistic goal would deliver real improvement
	Moving too quickly from small-scale tests of change to full-scale implementation and sustainment5	Failure to uncover barriers to broader use prior to implementation; project failure; disruption associated with deimplementation; wasted resources/goodwill

PDSA, Plan-Do-Study-Act; QI, quality improvement.