Choosing an Improvement with Conflicting Constraints
A Green Belt improving a clinic check-in process has a baseline: average wait 34 minutes, 90th percentile 71 minutes, target average ≤20 minutes. Process mapping shows three steps: registration (avg 6 min, one clerk), triage form completion by patient (avg 9 min, self-service on paper), and nurse vitals (avg 8 min, two nurses). Arrivals average 22 per hour at peak. The clinic cannot add staff (budget freeze), cannot change the triage form content (clinical governance), and cannot extend opening hours. The clerk's registration step is done on a single terminal, and patients queue for it before starting the paper form. The Green Belt proposes: (a) give patients the paper form on arrival to complete while queuing for registration; (b) add a second terminal at the same desk; (c) digitise the form.
1. Evaluate the three proposals against the constraints and the baseline, and recommend a course of action.
Read the worked answer
Constraint check first:
- (a) Reorders work - no new staff, no form change, no hours change. Permitted.
- (b) A second terminal is a capital item; under a budget freeze this is likely blocked, and the bottleneck may not be the terminal anyway (one clerk can only register one patient at a time regardless of terminals). Likely blocked and possibly ineffective.
- (c) Digitising changes the form's delivery mechanism; 'content' is protected, but digitisation is a project beyond a Green Belt's scope and timeline. Out of scope.
Flow reasoning: patients currently queue for registration then complete the 9-minute form, so the form time is serial and adds to wait. Under (a), the 9-minute form overlaps with the registration queue - for a patient waiting >9 min, the form time is effectively free. Given the 34-min average wait, most patients wait far longer than 9 min, so (a) could remove a large share of the 9 min from the critical path.
Recommendation: implement (a) as a low-cost, reversible change; pilot for 4 weeks; measure average and 90th-percentile wait on the same definitions. Do not pursue (b) or (c) in this project - record them as recommendations for the sponsor with a business case, not as project actions.
Caveat: verify that the registration clerk does not need form data to register; if they do, (a) breaks the dependency and must be redesigned.
2. How would you structure the pilot and what would tell you to stop or scale?
Read the worked answer
Design:
- Run the change on selected sessions (e.g. two morning peaks per week) with the rest as comparison, or a before/after with a stable baseline period - randomised assignment is ideal but operationally hard in a clinic.
- Duration long enough to cover normal variation: at least 4 weeks, ideally 6, to avoid a single atypical week.
- Same operational definitions of 'wait' (arrival to start of vitals) and same data capture method as baseline.
Decision rules (set in advance):
- Scale if average wait falls toward ≤20 min and the 90th percentile does not worsen - a change that improves the average but worsens the tail is not a win for patients.
- Stop/redesign if the form-on-arrival creates errors (patients completing it in the queue with no surface), or if registration is delayed because patients hold the only clipboard.
- Investigate if average improves but variation increases - suggests the change interacts with arrival patterns.
Use an individuals/moving-range chart on session-level wait times to distinguish real shift from common-cause noise before declaring success. Pre-register the rules so the result is not rationalised after the fact.