[Patient data for evaluating reference values].

Objective: To explore the use of data in patients as a way to evaluate the values of reference in clinical assays. Serum calcium (Ca) and phosphorus (P) were selected for this purpose.

Material: A total of 836 consecutive admissions in a 10 week period were revised. Seventy percent (580 cases) had data of Ca and P in the first week following admission. A total of 424 were excluded by anyone of the following criteria: a) serum urea greater than 50 mg/dL, b) serum creatinine greater than 1.5 mg/dL; c) serum albumin less than 4 g/dL; d) lack of data of either urea, creatinine or albumin. These cutoff values were selected on the basis of a significant difference (t test) in the mean of Ca and/or P in patients grouped according to levels of the available data. The remainder of 156 cases is the selected population of study, and the 580 with Ca and P is the total population.

Methods: Ca and P were assayed in an analyzer (Coulter Chemistry) using the DAM and the Jaffé methods respectively. The precision during the period of study was adequate (CVs of 2.6% for Ca and 2.9% for P).

Results: Table 1 compares the mean, SD and CV of Ca and P in the selected versus the total population. There are significant reductions (F test) in the variability of Ca and P in the selected population (from a CV 9 to 5% in Ca and from 35 to 13% in P). Table 2 compares the age and sex distributions of the selected versus the total population. There are no significant differences (chi square test) although there is a lower proportion of people above 80 years of age in the selected population. The reference intervals of Ca and P in the selected population were obtained with the mean +/- 2 SD as both showed Gaussian distributions. These limits are compared in table 3 versus the values of the institution and of the manufacturer of the analyzer used. The P limits of this study fall intermediate to the other two, but in Ca they are the lowest of the three. Shortly after the completion of this study, the institutional values of Ca were changed to 8.5-10.5 mg/dL on the basis of the one year results in a WHO quality assessment scheme. The modification made the Ca limits of this study also intermediate to the other two series. The small differences in the limits have clinical and economic repercussions. Table 4 shows there is a 20% reduction in the number of abnormal tests (206 vs 259) and in the number of abnormal individuals (165 vs 203) using the limits of this study versus the institutional limits before modification.

Conclusions: The data of Ca and P in a selected but representative population of patients proved to be a reliable way of evaluating its reference intervals. The reference intervals of this study were intermediate to those of the manufacturers of the analyzer and the laboratory (after modification). This finding suggests the strategy may be as good as more conventional approaches to establish reference values.