The reliability of the serum dioxin measurement in veterans of operation Ranch Hand

A study was conducted on the reliability of the serum dioxin measurement of enlisted Ranch Hands veterans participating in the Air Force Health Study using paired serum dioxin measurements. The 46 veterans were not randomly selected, but their demographic characteristics, health, and dioxin levels were similar to those of 404 other enlisted Ranch Hand veterans who had a single dioxin measurement made in 1987. The average time between the measurements was 0.61 years, the first measurement made from blood drawn on 10 April 1987 and the second from blood collected at a subsequent physicaI examination. In original units, the coefficient of reliability was 0.87 (95% confidence interval: 0.76, 0.94) when the first measurement was at or below 50 parts per trillion. The measurement had no reliability in original units when the first measurement was greater than 50 parts per trillion. After a logarithmic transformation, the coefficient of reliability was 0.96 (95% confidence interval: 0.93 to 0.98). These results suggest that the serum dioxin measurement should not be used in original units for any purpose when she value exceeds 50 parts per trillion. The measurement is, however highly reliable after a logarithmic transformation over the entire range of concentrations. Other studies using the same analytical method to measure dioxin in serum could similarly benefit if the measurement used is on the natural logarithm scale.