Presenting and defending a quality improvement plan — OSCE

Station brief (8–10 minutes)

The neonatal unit has a 40% rate of infants receiving antibiotics beyond 48 hours without a proven infection. A registrar has proposed simply "educating everyone and reauditing in a year." The head of department has asked you, the senior registrar, to present a proper quality improvement plan to the unit quality committee and to defend your method. You will be questioned on the aim, the measures, how you would test a change, how you would read the data, and how you would ensure the gains sustain and do not widen an equity gap. Do not invent local mandated targets. [1] [8]

Tasks for the candidate

1. Reframe the problem as a quality improvement project and state a specific, measurable, time-bound aim. [1]
2. Define a balanced set of outcome, process and balancing measures and explain why each matters. [8]
3. Describe how you would test a change idea with a PDSA cycle before any unit-wide rollout. [2]
4. Explain how you would read the data on a run chart, distinguishing special-cause from common-cause variation. [9]
5. Outline how you would sustain the improvement, protect against an equity blind spot, and report the results appraisably. [10] [13]

Expected performance

Must hit. States a clear aim with population, metric, target and deadline (e.g. reduce from 40% to 20% in 6 months); names the Model for Improvement and its three questions; defines outcome, process and balancing measures including a re-start rate to detect under-treatment; describes a small-scale PDSA test before any rollout; identifies a special-cause rule (shift, trend, run or astronomical point) and states that common-cause variation should not be chased; commits to disaggregating data by subgroup and to a sustainability plan with an owner and ongoing data. [1] [2] [9]

Merit. Builds a driver diagram linking aim to primary drivers to change ideas; explains why a before-after design is weak and why a time-series run chart is stronger; names SQUIRE as the reporting standard; cites a paediatric exemplar such as the Vermont Oxford Network or the Dukhovny antibiotic-stewardship collaborative; distinguishes QI from clinical audit and research when discussing oversight; engages a family partner in the team. [5] [8] [10] [13]

Fail. Accepts "educate and reaudit" as a plan; proposes a big-bang rollout with no small testing; measures only process without outcome; has no balancing measure; reacts to common-cause variation; reports only an aggregate average without checking subgroups; or invents local mandated targets. [8] [9]

Sample candidate structure

"This is a quality problem, not a research question, and 'educate everyone and reaudit in a year' is not a quality improvement project — it has no aim, no measure and no test of change. My aim is to reduce the proportion of infants on this unit receiving antibiotics beyond 48 hours without a proven infection from 40% to 20% within 6 months. I will use the Model for Improvement: three questions — aim, measure, change — then PDSA cycles. I will track a balanced set: the outcome measure is the over-48-hour exposure rate; the process measure is the proportion of courses with a documented 48-hour stop-or-review decision; the balancing measure is the re-start rate for infants with a subsequently proven infection, plus length of stay, so that I do not under-treat real sepsis in pursuit of a lower exposure number. I will establish a baseline run chart, then test a change idea — a 48-hour stop-review checklist — on one team for one shift: plan what data I collect and what I predict, do it, study the result, and act to adopt, adapt or abandon before scaling. When I read the data I will look for special cause — a shift of six or more points on one side of the median, a trend, a run, or an astronomical point — and I will not chase common-cause noise. To sustain this I will assign an owner and keep the data flowing, and I will disaggregate the results by subgroup so an improving average does not hide an equity gap for Indigenous or disadvantaged infants. I will report with SQUIRE so another unit can appraise and adapt what we learned." [1] [2] [9] [10]