Application of TEDPED to Analyze Serum Enzyme Distributions in Rabbits

摘要

The most common assumption regarding the natural probability distribution of data is that the data are normally distributed. However, many natural phenomena are not normally distributed. The most frequently assumed one tail distribution for continuous data is the lognormal distribution. A computer code, TEDPED, was written for testing the hypothesis that a data set is normally or lognormally distributed. The application of this code to analyze the distribution of specific serum enzymes as determined by commercial assays is illustrated. Three clinically important enzymes: alkaline phosphatase, AP; serum alanine amino transaminase, SAAT; and lactate dehydrogenase, LDH, were selected for analysis with standard commercial assays. Experimental data were determined from healthy mature New Zealand white rabbits. The AP assay was from General Diagnostic. The LDH and SAAT assays were from Fisher Scientific. The standard employed to calibrate the assays was Validate by General Diagnostic. The alkaline phosphatase data fitted a normal distribution with an r exp 2 = 0.979. However, these same data also fitted a lognormal distribution with logarithmically transformed distribution with r exp 2 = 0.957. The results for the other two enzymes were similar. The LDH data fit a normal distribution with r exp 2 = 0.974 and logarithmically transformed distribution with r exp 2 = 0.978. The SAAT data fit a normal distribution with r exp 2 = 0.935 and a log transformed distribution with r exp 2 = 0.961. The interpretation of this analysis indicates that although the enzyme concentrations can only be positive and appear with reasonable r exp 2 values to be lognormally distributed, they can also be assumed to be normally distributed for the purpose of applying t-tests to compare either independent or paired samples. (ERA citation 06:002974)