Childhood systemic lupus erythematosus with negative antinuclear antibody test

mined. These dual determinations are to assure accurate analysis of the sample, not accuracy of the diagnosis of CF. Applications of the point and interval estimation equations presented here to values in the "border l ine" region yield practically identical point estimates for the equations derived from CF patients and controls. Interval estimates were less precise when based on the equation for CF patients; the intervals were about twice as wide. We recommend using the narrower limits based on the equation of the controls for value in the "border l ine" region because of the higher relative prevalence of nonCF patients in this region (approximately five to one in our experience). The slope o f both equations was not greatly different from one, although we observed slightly higher values of chloride than sodium in CF patients and a reversal of this ratio in the controls.