Study objective: We sought to assess physicians' ability to accurately determine the presence or absence of sodium fluorescein (SF) in urine at a concentration corresponding to that present after ingestion of a toxic amount of commercial automotive antifreeze.
Methods: We studied 2 different urine specimen evaluation formats--one presenting isolated specimens, and the other presenting specimens grouped for comparison--to determine whether the visual clues afforded by grouped comparison aided the accuracy of the evaluation. On each study day, 3 urine specimens (1 control specimen obtained before SF administration and 2 specimens obtained after SF administration) were obtained from each of 9 or 10 volunteers. Each of these 27 or 30 urine specimens were presented sequentially and in random order to 2 emergency physicians during separate evaluation time periods. Each physician was asked to classify each specimen as fluorescent or nonfluorescent (sequential format). After a rest period, each physician, again separately, was asked to look at the same 27 or 30 urine specimens, this time all together in a test tube rack so that grouped comparisons were possible. The physicians again classified each sample as either fluorescent or nonfluorescent (grouped format). We assessed sensitivity, specificity, and accuracy of the evaluation by each presentation format (sequential or grouped).
Results: Mean examiner sensitivity, specificity, and accuracy for detecting the presence of SF in urine using the sequential presentation format were 35%, 75%, and 48%, respectively, whereas the same test performance indices were 42%, 66%, and 50%, respectively, when the grouped format was used.
Conclusion: Wood's lamp determination of urine fluorescence is of limited diagnostic utility in the detection of SF ingestion in an amount equivalent to toxic ingestion of some ethylene glycol--containing automotive antifreeze products.