Sustaining Lab Turnaround After the Gains Slip
A hospital laboratory reduced median sample turnaround time from 55 to 32 minutes over four months. Eight weeks later the median is 41 minutes and the 90th percentile has risen from 58 to 78 minutes. Three of 14 technicians are new, and the night shift has returned to batching samples before loading the centrifuge. An audit of five standard steps found variation in labeling, centrifuge loading and result release. The improvement team had produced standard work, but the document sits in a binder; the supervisor reviews a dashboard weekly but no one has defined what to do when a point is high. The control chart shows six consecutive points above the previous average, and the lab manager wants to recalculate the limits so the chart 'looks normal again.' A Yellow Belt who supported the original project is asked to help the lab supervisor build a control plan and respond to the signal.
1. Draft the core elements of a control plan for lab turnaround time, including the metric definition, owner, monitoring frequency and response.
Read the worked answer
Metric: TAT per sample, defined as receipt timestamp in the lab system to verified result release timestamp. Report median and 90th percentile by shift and by priority (stat vs routine). Target: median at most 32 minutes, 90th percentile at most 58 minutes; define the baseline and control limits from the stable improvement period, not from the eroded period.
Data source: laboratory information system timestamps; check that timestamps are captured automatically and not edited. Frequency: daily chart review by shift, weekly summary by the lab supervisor. Owner: lab supervisor for the process; quality lead for the chart; Yellow Belt supports data validation.
Control method: individuals/moving-range chart for daily median TAT, or a run chart by shift; annotate new staff, centrifuge changes and batching. Response plan: if a special-cause signal occurs, contain (check night shift and new staff), investigate the five standard steps, record the cause, and take action within one business day. Standard work: visual standard at each bench, training sign-off for new technicians, and a daily huddle. Audit labeling, centrifuge loading and result release monthly. Review the control plan quarterly.
2. Interpret the six consecutive points and advise the lab manager who wants to recalculate the limits. What would you do next?
Read the worked answer
Six consecutive points above the previous average is a special-cause signal under standard run-chart rules, not normal variation. Recalculating the limits to make the chart 'look normal' would hide a real change and is a form of tampering; it would also make future signals unreliable. Keep the original limits and investigate. The timing coincides with three new technicians and the night shift returning to batch loading, and the audit found variation in labeling, centrifuge loading and result release. These are plausible assignable causes, but they are not yet proven.
Next steps: stratify the chart by shift and by technician tenure. If the rise is concentrated on nights and among new staff, observe the night process and compare batching versus the standard. Check whether the new staff were trained on the standard work and whether the centrifuge loading rule is feasible at night. Verify data timestamps to rule out a measurement change. If a cause is confirmed, fix it and monitor; once the process is stable again at the improved level, recalculate limits from the new stable period. Distinguish common-cause variation from this special-cause shift.
3. What should the Yellow Belt do to sustain the gain, and what should they not do?
Read the worked answer
Sustain: help the lab supervisor turn the control plan into daily practice - a visible chart, a short huddle, a named owner, a documented response plan, and standard work at each bench. Support training sign-off for the three new technicians and make sure the night shift uses the same loading rule. Set up a monthly audit of the five steps with a simple checklist and feed results back to the team. Help the team annotate the chart when changes occur, so future signals can be interpreted. Transfer ownership to the lab supervisor and quality lead; the Yellow Belt supports data collection, coaching and visual management.
Do not: unilaterally change the control limits, declare the project finished because the median improved once, or blame the new staff. Do not rely on the binder standard alone; if it is not visible and trained, it is not standard work. Do not collect more data without a decision rule. Escalate to the Black Belt or sponsor if the night-shift batching is a staffing or policy constraint that the supervisor cannot resolve.