Strategy for 90% autoverification of clinical chemistry and immunoassay test results using six sigma process improvement

Six Sigma involves a structured process improvement strategy that places processes on a pathway to continued improvement. The data presented here summarizes a project that took three clinical laboratories from autoverification processes that allowed between about 40% to 60% of tests being auto-verified to more than 90% of tests and samples auto-verified. The project schedule, metrics and targets, a description of the previous system and detailed information on the changes made to achieve greater than 90% auto-verification is presented for this Six Sigma DMAIC (Design, Measure, Analyze, Improve, Control) process improvement project.


Data accessibility
Raw data is maintained with the corresponding author.

Value of the data
Provides outline for Six Sigma process improvement design for auto-verification processes. Provides benchmarks and metrics to monitor and assess auto-verification processes. Describes test specific auto-verification parameters and consistency checks to achieve 90% autoverification.
Provides brief notes to medical laboratory technologists and basic strategies to address delta check and extreme values held for manual review.

Data
The data presented is from three clinical chemistry laboratories in Newfoundland and Labrador where Six Sigma process improvement methodology was used to improve the efficiency of autoverification (AV) processes affecting clinical chemistry and immunoassay tests. Data includes baseline data from all three laboratories (HSC-Health Science Centre; WMH-Western Memorial Hospital; and SCH-St. Clare's Mercy Hospital), test specific parameters for the new AV system, and other tools to assist with operation of the new AV program which achieved greater than 90% sample AV at the three sites examined. The original AV system is described, specific changes made, and some effects on the changes.

Experimental design, materials and methods
A Six Sigma process improvement effort carried out to improve AV processes at the three sites [1]. All sites had similar AV routines starting out. An outline of the Six Sigma process improvement schedule based on DMAIC (Design, Measure, Analyze, Improve, Control) methodology is provided in Table 1. The project team consisted of thirteen-members representing managers, clinical biochemists, front line staff and others. The process metrics and benchmarks/targets were established during the "Design and Measurement" phases. Various process maps including Fig. 1 which outlines the patient      [3]. c New rules with no occurrence in the data set were assigned a predicted frequency o 0.0001. DB/TB ratio DB/TB ratio ¼ Direct Bilirubin/Total Bilirubin (4 1 will flag) 4 A/P ratio A/P ratio ¼ Albumin/Total Protein (beyond 0.25 or 1 will flag) 5 Transam ratio Transam ratio ¼ ALT/AST (beyond 0.25 or 4 will flag) 6 T4 high rule Both TSH and fT4 greater than upper reference limit 7 T4 low rule Both TSH and fT4 less than lower reference limit 8 HIL all positive All indices (H, I, L) of one plus or greater.  result verification workflow were also constructed to better understand the AV process. The reliability and reproducibility of all process metrics were validated and are listed in Table 2 along with baseline and benchmarks or targets for each metric. Baseline values for most metrics were mainly determined from download and analysis of test order specific information from Instrument manager (IM) middleware. An exception was test manual verification time which was determined by an observer who timed by stop watch the manual verification activities by medical laboratory technologists (MLTs) both during the Measurement Phase but also later during the Control phase. The new AV scheme (parameters detailed in Supplementary Table 3) was developed following review of process metrics and examination of the original system, and by several rounds of meetings with MLTs at the three sites in order to gain insight on manual verification activities. The key changes made and their predicted impact on test hold rates are summarized in Table 3. The predicted impact of various rules and consistency checks on proportions of tests held for manual review and verification were evaluated using downloaded patient test results from the laboratory information system. A description of consistency check rules and calculations are summarized in Table 4 and the notes back to MLTs for each are summarized in Table 5. Following implementation of the new AV system several new tools were implemented in order to allow continuous monitoring of the impact of the new system on error detection (Fig. 2) and in order to standardize evaluation of extreme values (Fig. 3A) and delta checks (Fig. 3B) to compliment the automated comments to MLTs concerning consistency checks and HIL failures. The impact of the new AV system compared to the original one relative to time spent by MLTs for review and release of held tests are summarized in Table 6.    Average time for release of samples by MLTs during manual verification. Manual result verification time studies were conducted at HSC site by an observer using a stop watch and timing technologists as they went about manual review activities. Verification time was determined from point of first appearance of result profile to release of results to the electronic record. Appearance of critical results were sporadic but these time periods were removed as they were very variable in length, proportionately more common during the post-improvement stage, and tended to skew average time per sample verified.