We compared five methods for the determination of total and direct bilirubins in serum samples from normal controls, subjects with
Gilbert's syndrome, and serum pools containing about 50 and 150 mg of total
bilirubin per liter. The Kodak Ektachem method and a diazotized sulfanilic
acid method with 0.15 mmol/L
sodium nitrite concentrations are the only methods that gave accurate direct
bilirubin values, as judged by liquid-chromatographic results. The aca method that involved
p-nitrobenzene diazonium tetrafluoroborate and another diazotized sulfanilic
acid method with a higher concentration of
sodium nitrite (0.8 mmol/L) yielded falsely high values for direct
bilirubin, which could lead to clinical
confusion. The more recently introduced diazotized sulfanilic
acid method of the aca gave substantially better results than the
p-nitrobenzene diazonium tetrafluoroborate method but was still inaccurate. Systematic investigation of these procedures revealed that the overestimation of direct
bilirubin by the diazotized sulfanilic
acid method was related to the amount of unconjugated
bilirubin present and its ability to react as direct
bilirubin in the presence of higher concentrations of
sodium nitrite. Inherent properties of
p-nitrobenzene diazonium tetrafluoroborate appeared to be responsible for inaccuracies in that method, which could not be corrected by varying
reagent concentration or the reaction conditions.