Drug Information Journal - March 2009 - (Page 143) Local Laboratory Reference Intervals MEDICAL INFORMATION 143 Local Laboratory Reference Interval False Negative Low 0%; N = 0 (A) URL False Negative Low 2.5%; N = 91 (D) LRL Agreement 97.5% N = 3,536 (G) False Positive High 3.1%; N = 336 (B) Agreement 97.2% N = 1,782 (C) FIGURE 2 Characterization of differences between local laboratory and central laboratory reference limits. Agreement 91.0% N = 9,906 (E ) False Negative High 2.8%; N =52 (F ) False Positive Low 5.9%; N = 640 (H ) LRL URL False Negative High 0%; N =0 (I ) Central Laboratory Reference Interval • Percentage false positive (low): This is the percentage of those results within the reference interval of the central laboratory that are below the LRL of the local laboratory [100*H/(B + E + H) using the notation defined in Figure 2]. Over the whole leucocyte count data set, 10,882 samples lay within the central laboratory reference interval and, of those, 640 (5.9%) were below the LRL of the local laboratory. The percentage false positive (low) is therefore 5.9%. • Percentage false positive (high): This is the percentage of those results within the central laboratory reference interval that are above the URL of the local laboratory [100*B/(B + E + H) using the notation defined in Figure 2]. Over the whole leucocyte count data set, 10,882 results lay within the central laboratory reference interval and, of those, 336 (3.1%) were above the URL of the local laboratory. The percentage false positive (high) is therefore 3.1%. • Percentage false negative (low): This is the percentage of those results below the central laboratory LRL that are not below the LRL of the local laboratory [100*(A + D)/(A + D + G) using the notation defined in Figure 2]. Over the whole leucocyte count data set, 3,627 samples lay below the central laboratory LRL and, of those, 91 (2.5%) were not below the LRL of the local laboratory. The percentage false negative (low) is therefore 2.5%. • Percentage false negative (high): This is the percentage of those results above the central laboratory URL that are not above the URL of the local laboratory [100*(F + I)/(C + F + I) using the notation defined in Figure 2]. Over the whole leucocyte count data set, 1,841 samples lay above the central laboratory URL and, of those, 52 (2.8%) were not above the URL of the local laboratory. The percentage false negative (high) is therefore 2.8%. It can be seen from Table 2 that there are many discrepancies between flagging based on the local laboratory reference intervals and flagging based on the central laboratory reference interval. Further, there is wide variation between laboratories. The question then arises: are local reference intervals set rationally (ie, based on the characteristics of the local population) or arbitrarily (in which case a centrally defined reference interval should be used)? A method for answering that question is proposed below. CONTEXT The description here is entirely in terms of the LRL and the fraction of results below that limit. Of course, the theory applies straightforwardly to the URL and the fraction of results with values above that limit. Drug Information Journal
For optimal viewing of this digital publication, please enable JavaScript and then refresh the page. If you would like to try to load the digital publication without using Flash Player detection, please click here.