A SYSTEM TO MONITOR A PORTION OF THE TOTAL TESTING PROCESS IN MEDICALCLINICS AND LABORATORIES - FEASIBILITY OF A SPLIT-SPECIMEN DESIGN

Objective.-The purpose of this study was to assess the feasibility of using a prototype split-specimen design to assess integrity of a porti on of the total testing process in medical clinics and laboratories. D esign.-Two or three tubes of venous blood were collected from 177 pati ents for analysis of one of three analytes (serum potassium, serum tot al cholesterol, and whole-blood hemoglobin). Patients were seen at one of the nine clinics participating in this study. In all cases, one tu be of blood from each patient was sent to a commercial referral labora tory, and the other tube(s) forwarded to the laboratory that routinely tested specimens for the clinic (participating laboratory) for analys is. Each participating laboratory removed a preanalysis and sometimes a post-analysis aliquot from each specimen and forwarded these to the referral laboratory for analysis. Setting.-The study was conducted in six physician office laboratories (three serving 1 to 4 [mean, 2.7] in ternists and three serving 3 to 24 [mean, 12] family physicians) and t hree hospital laboratories (serving hospitals with 100 to more than 70 0 beds). Patients.-Study patients were voluntary participants and prov ided informed consent. Patient age ranged from 18 to 80 years, and for all the laboratory test was specifically ordered for clinical reasons . Patients who were unable or unwilling to provide informed consent, t hose for whom testing would require that they provide more than by fin gerstick, and those with results that were part of a laboratory test p rofile were excluded. Main Outcome Measures.-Two main outcome measures were assessed: (1) percent differences between split-specimen results exceeding the maximum allowable imprecision level, which was based on published biological variation data (defined as one-half of the intra individual percent coefficient of variation), for each analyte (result discrepancies); and (2) all ''problems'' (defined as departures from standard operating procedures) that could be documented by retrospecti ve review of all relevant medical and laboratory records. Results.-The rate of result discrepancies was 1 in 20 (5%) for patients in whom he moglobin was analyzed, 12 in 57 (21%) for patients in whom potassium w as analyzed, and 1 in 60 (2%) for patients in whom total cholesterol w as analyzed. Results of samples obtained during the aliquoting and sto rage phases of the total testing process were subject to study-induced problems and were generally not useful in tracing problems to specifi c stages of the testing process. A total of 28 problems (involving 26 patients) were documented, but only 6 problems were due to routine tes ting professes. Conclusions.-The feasibility and limitations of a spli t-specimen design to detect result discrepancies were demonstrated. Mo st documented problems (22 of 28, or 79%) were study induced. To asses s integrity of the total testing process, such problems need to be avo ided.