Biomedical subjects
Publications and source records attributed to R K Gilbert.
Interlaboratory quality control programs provide an important contribution to achieving and maintaining accuracy in clinical laboratories. The various strategies for providing accurate target values in the two principal types of interlaboratory programs are reviewed.
Samples of lyophilized human serum were circulated to participants in the CAP Chemistry Survey program. The participants measured the concentrations in the samples of calcium, iron, lithium, magnesium, potassium, and sodium. Samples were also sent to the National Board of Standards (NBS), and the concentrations of these seven analytes were measured; definitive methods were used. The participant values and the NBS values were compared, providing a measure of the accuracy of the participating laboratories and of the individual methods used by the participants. There was, in general, close agreement between the NBS values and the participant values. A bias that varied with the concentration of the analyte was a common feature.
The CAP Interlaboratory Survey data from the non-core laboratories identified by Paule and Mandel were examined. Some laboratories were designated non-core because they submitted one or more values that were widely discrepant and were caused by unit problems, decimal shifts, or reversal of the specimens. A second group of laboratories submitted a number of discrepant results that appeared to be caused by problems within the laboratories. A third group submitted results that appeared acceptable, and the exclusion and the effectiveness of the criteria differed for each of the three analytes. The findings suggest that examination of data from excluded laboratories can provide a valuable source of information to supplement the central data.
Six paired specimens of lyophilized human serum were circulated to participants in the CAP Comprehensive Chemistry Program. The intervals between circulations ranged from 13 to 33 weeks. Each specimen was analyzed for 21 constituents, and the six pairs of specimens provided 126 paired sets of data for statistical comparison. The mean values for all constituents reported by the participants for each pair showed close agreement. The median difference between the two mean values for 126 pairs was 0.6% of the concentrations of the constituents. Twenty-six individual pairs showed differences that were sufficiently large to be significant (p = less than .0005). The majority of these stemmed from differences reported by participants using the SMA 12/60. Although significant, these differences were so small that they exerted no demonstrable effect on the dispersion of the results reported by the SMA 12/60 users. One vial pair showed a loss of glucose. All six pairs showed a loss of triglyceride. There was a suggestion of loss of cholesterol and uric acid in one pair. These differences were small. This study demonstrates that a comparison of mean values in a large Survey program can provide a very sensitive method of monitoring the stability of the specimens. The study also suggests the repeatability of mean values in a large Interlaboratory Survey program far exceeds clinical needs, and the agreement may be closer than can be obtained by any other system of measurement.
Explore the source record for details and available documents.
The accuracy of the methods commonly used for calcium analysis in clinical laboratories in the United States was studied with the aid of the National Bureau of Standards (NBS). Samples from six large serum pools that had been analyzed by 4,258 clinical laboratories were sent to NBS. At NBS the calcium contents of the samples were determined using isotope dilution and mass spectrometry (ID-MS). The findings show that the bias of the routine methods varies from levels that are insignificant to levels that would have an impact on patient care. The accuracy of the automated methods is comparable to the accuracy of the best manual methods. The accuracy of the proposed calcium reference method is comparable to the accuracy of a number of the methods in current routine use. There is evidence of a widespread lack of standardization at the extremes of the clinical range; this effect is not shown by the results obtained by automated methods. The study underlines the value of developing technology, such as ID-MS, that is capable of providing definitive values. The definitive values can then be used to assess the bias of routine methods on a nationwide scale, using existing programs.
Explore the source record for details and available documents.
The data that have been accumulated by the Survey program of the College of American Pathologists for analysis of 12 serum constituents for the five-year period 1969-1973 are reviewed. The data include approximately 2 million individual analyses provided by the majority of the clinical laboratories in the United States. The data show clear evidence of progressive improvement. The values reported by the participants for each analysis have agreed more closely in each successive year. The improvements have stemmed partly from improved performance of individual methods and partly from increased use of methods with good precision. Closer standardization between laboratories may have played a role. The findings are used for proposing a set of goals for the analysis of these constituents in serum. Comparison of the goals with the actual levels of current performance shows that for many analyses the current levels are adequate, while for others the current levels fall short of the desired goals. Achieving the goals from some analyses may be beyond the present limits of the technology of clinical laboratory analysis.