Background: The standard method for the determination of reference values in a population by testing a number of healthy volunteers is difficult within the paediatric age group; this study explores an alternative approach.
Methods: Biochemical blood test results collected by the Woman's and Children's Hospital over a 1 year period were used and the data selected to ensure, as far as possible, that the results were from patients who were 'healthy'. This was achieved using various selection criteria, such as the exclusion of patients with more than one test episode. Some of the data were skewed, making standard statistical approaches difficult. In such cases, transformations were used to ensure that the resulting information had a distribution that was approximately Gaussian.
Results: Data from different biochemical tests covering both genders and ages in the range 0-18 years were collected; this included more than 250 000 laboratory test results. After elimination of non-representative data and tests with insufficient results, there were approximately 23 500 results covering almost 3000 individuals and 16 different biochemical tests. There were no results from very young children (< 1 year) after the data selection process; hence the derivation of reference values from that age group was not possible.
Conclusions: This approach permitted better delineation of reference intervals for common biochemical tests performed on paediatric patients than is currently readily available. In addition, some important benchmarks for transformation of medical data have been found.
