Laboratory TAT Control Chart After a Lean Redesign
A hospital central laboratory tracks median turnaround time for STAT chemistry panels from specimen receipt to result verification. Before a Lean redesign, 30 daily medians had a mean of 42 minutes and a standard deviation of 6 minutes. After standard work, a 20-day pilot shows a mean of 35 minutes and a standard deviation of 5 minutes. An I-MR chart using limits calculated from the pre-improvement 30 days shows eight consecutive points below the centerline and one point above the upper control limit on day 14. The laboratory manager says the chart is out of control and wants to remove it because the process is clearly better. The quality lead wants to recalculate limits using only the post-improvement data and declare victory. A night-shift supervisor notes that day 14 had a computer system slowdown and two staff called in sick. The lab has three shifts; the pilot only covered day shift. The emergency department has a target of 30 minutes. The lab director asks the Black Belt whether the process is stable, improved, and ready to standardize.
1. Interpret the chart signals. Is the process stable, improved, and ready to standardize?
Read the worked answer
Chart interpretation: Eight consecutive points below the pre-improvement centerline is a signal of a sustained shift, not random variation. The process is not out of control in the sense of unpredictable; it has moved to a new level. However, one point above the upper limit on day 14 coincides with an IT slowdown and two sick calls, so it is a plausible special cause. Stability and improvement: The pilot mean dropped from 42 to 35 minutes, a 7-minute improvement, but the chart uses old limits; stability of the new process is not yet established. The process appears improved in central tendency but not yet stable enough to standardize. Reasoning: Control charts detect change; they do not by themselves prove a new stable process. The correct next step is to investigate the day-14 signal and collect a clean post-improvement period before recalculating limits.
2. Evaluate the manager's request to remove the chart and the quality lead's request to recalculate limits immediately.
Read the worked answer
Manager's request: Removing the chart because the process is clearly better is wrong. The chart is the tool that shows whether the improvement is sustained; removing it eliminates the feedback loop. Quality lead's request: Recalculating limits from the 20 post-improvement days immediately would be premature. Phase limits should be recalculated only after the special cause is understood, the process is stable, and enough subgroups are available, typically 20-25 or more, to estimate limits. The day-14 point should not simply be deleted; it should be explained and, if confirmed as a special cause, handled with a documented rule. Recommendation: Keep the chart with old limits for now, annotate the shift and the day-14 event, investigate the event, then establish new phase limits for the improved process. Reasoning: Recalculating limits to fit the data can hide instability and make any process look stable.
3. What control plan and further study are needed before the improvement is standardized across the laboratory?
Read the worked answer
Control plan: Monitor median STAT turnaround time daily with an I-MR chart, stratified by shift; if subgroup size is adequate, use Xbar-S for within-shift subgroups. Add a reaction plan for turnaround time above the new upper control limit: check analyzer status, IT systems, staffing, and specimen transport; escalate to the lab supervisor. Track the proportion of results within 30 minutes and the 90th percentile, not just the median. Further study: Extend the pilot to evening and night shifts before standardizing; the day-shift-only pilot may not represent the lab. Verify the measurement definition of turnaround time and the data feed. Standardization: After at least 20-25 stable subgroups and confirmation that the 30-minute target is met, update the control plan, training, and visual management. Reasoning: Control-phase monitoring must cover the full process and include a reaction plan; a day-shift pilot is not evidence for all shifts.