2006 OPEN FORUM Abstracts
K.I.S.S. CORRELATION SPREADSHEET FOR BLOOD GASES
Steimke, RCP, Coastal Communities Hospital, Santa Ana, CA & Michael H.
Terry, RCP, RRT, Loma Linda University Medical Center, Loma Linda, CA.
Purpose: Respiratory Care Departments frequently are responsible for the analysis of blood gasses but Respiratory Care Directors often struggle to find resources for information about these requirements. Some labs are currently only running the traditional QC programs, proficiency studies (With CTS or other organizations) and whole blood correlation studies. Recently regulatory agencies have asked some blood gas labs to present their analytical comparisons of their reportable ranges and of their correlation studies. Good Laboratory Practice and CLIA '88 require comparative studies to define inter-instrument and inter-method performance in the analysis of any analyte. CLIA '88 states this must be evaluated at least every six months The comparative method however, is not defined. Both JCAHO and CAP laboratory inspectors have asked to see linear regression used for these comparisons. This abstract presents a simple spreadsheet to collect, analyze and report this data.
Method: Duplicate analysis is performed on the two instruments for comparison throughout the six month period and entered on the data spreadsheet, (a minimum of twenty samples per reporting interval is required). The statistical analysis is performed automatically and dynamically. Thresholds for agreement in slope, intercept, correlation and R2 should be set in consultation with the Laboratory Medical Director. The chart worksheet presents the information graphically and this is printed and presented for review by the Medical Director every six months. In the event an institution has more than two instruments for comparison, all instruments should be compared to the primary instrument as designated on their proficiency testing system.
A copy of this spreadsheet is available on request.
Conclusion: Blood gas laboratory managers should become familiar with linear regression for comparing instruments. This simple spreadsheet can be used as a template to satisfy this requirement.