Computer-assisted quality control: considerations for program design.

A well-designed laboratory computer system has the potential of enhancing all phases of quality control. Of most importance is the choice of an underlying approach that is statistically valid and robust. Specification of quality goals, participation in regional programs, and use of a flexible multi-rule approach are recommended. Careful attention to general and specific design is necessary to achieve full system potential. The design process must include detailed consideration of file and record structure, data integrity, flexibility of input and output, and display and report formats. For statistical run evaluation, theoretic validity, operational process, and ease of understanding by the user must be considered.

Stability of mean assay values of magnesium and iron in lyophilized quality control serum: a study based on data from the quality assurance service (QAS) of the College of American Pathologists.

Data from magnesium and iron analyses of pools of lyophilized quality control serum were used to evaluate stability of analyte mean values in the pre-reconstitution period. Information was received from laboratories in Regional Quality Control Programs between 1977 and 1981, using CAP Quality Assurance Service data processing. For magnesium, 28 of 41 (68%) pool-method combinations were stable, 11 (27%) showed decreases, and 2 (5%) showed increases. Decreases averaged 0.008 mg/dL/month (all methods). A pronounced effect of method (automated methylthymol blue) and year correlated with decreases in measurable magnesium. For iron, 45 of 56 (80%) pool-method combinations were stable, 8 (15%) showed decreases, and 3 (5%) showed increases. Decreases averaged 0.349 g/dL (all methods). In most cases, changes in measured concentrations are attributed to methodologic factors, rather than intrinsic changes in analyte concentrations.

Glucose stability in lyophilized chemistry quality control serum. A study of data from the quality assurance service (QAS) program of the College of American Pathologists.

Data from 2.5 million glucose analyses on pools of lyophilized human quality control serum were used to evaluate analyte stability in the prereconstitution phase. Input information was from laboratories in Regional Quality Control Programs that use CAP Quality Assurance Service (QAS) data processing. Of 31 pools in use between 1977 and 1981, decreasing glucose concentration was detected by, at least, one method in 26 pools, and by two or more methods in 21 pools. Method-associated average decrease in concentration varied from 0.13 mg/dL/month (glucose oxidase-electrode) to 0.19 mg/dL/month (automated glucose oxidase-colorimetric). Bidirectional instability as a function of analytic method, i.e., increase with "mild" methods, decrease with "rigorous" methods that was noticed previously with pools analyzed between 1973 and 1977, was no longer seen. Dominant directional changes in the later pools were downward by all methods, when statistically significant trends were demonstrated.

Analytic clinical laboratory precision. State of the art for selected enzymes.

Results of precision estimates are reported for selected enzyme analytes. Laboratories that participated in regional quality control programs supplied data to a centralized data processing service (College of American Pathologists Computer Center). These data, representing several regional groups which used different commercial sources of control serum, were statistically analyzed by regression of coefficient of variation (CV) upon concentration of analyte using a modified polynomial regression technic. Tolerance limits for long-term within-laboratory precision were developed from this data. Estimated CVs were extracted at selected analyte concentrations, and tolerance limits for CVs at these concentrations were developed. In addition, selected individual methods within the larger groups were individually examined to evaluate effect of control material matrix upon clinical estimates of enzyme precision. Differences in selected method groups are discussed.

Stability of sodium and potassium in lyophilized quality control serum. A study of data from the Quality Assurance Service Program of the College of American Pathologists.

The long-term prereconstitution stability of sodium and potassium in large pools of lyophilized quality control serum used in Regional Quality Control Programs, conducted between 1977 and 1980 is reviewed. In approximately one third of 27 pools studied, minimal but significant increases in sodium concentration were detected, these increases averaged less than 0.50 per year and confirmed our previously reported results from pools analyzed between 1973 and 1976. Results relating to potassium stability have been inconsistent. The previously reported tendency for potassium to increase in some pools was again suggested by data obtained from laboratories using automated flame-emission photometry procedures. This was not confirmed, however, by the data reflecting manual/semi-automated flame-emission photometry procedures or automated electrode methods. It is postulated, that the source of sodium causing the increases in concentration is the glass containers in which control materials are stored between the time of manufacture and reconstitution. Regression analysis against time of monthly interlaboratory means, from participants in Regional Quality Control Programs, is a useful tool for evaluating postmanufacture stability of analytes in pools used for daily internal quality control procedures.

Additional day-to-day precision estimates based on regional chemistry quality control data.

State-of-the-art precision values are presented for the following serum constituents: aldolase (EC 4.1.2.13), alpha-hydroxybutyrate dehydrogenase (EC 1.1.1.30), cholinesterase (EC 3.1.1.8), cortisol, gamma glutamyl transferase (EC 2.3.2.2), haptoglobin, immunoglobulins, lactic acid, leucine aminopeptidase (EC 3.4.1.1), total lipids, osmolality, protein fractions, T3 uptake, thyroxine and vitamin B12. Precision estimates are based on values reported for four lyophilized serum pools analyzed by participants in the Pennsylvania Association of Clinical Pathologists regional quality control program for clinical chemistry, during 1976, 1977 and 1978. Use of the upper limit of the "most common range" of precision (that range including the 75 percent most precise laboratories) as a warning level for trouble-shooting is advocated.