A distribution method for analysing the baseline of pulsatile endocrine signals as exemplified by 24-hour growth hormone profiles.
Matthews DR., Hindmarsh PC., Pringle PJ., Brook CG.
OBJECTIVE: To develop a method for quantifying the distribution of concentrations present in hormone profiles, which would allow an observer-unbiased estimate of the time concentration attribute and to make an assessment of the baseline. DESIGN: The log-transformed concentrations (regardless of their temporal attribute) are sorted and allocated to class intervals. The number of observations in each interval are then determined and expressed as a percentage of the total number of samples drawn in the study period. The data may be displayed as a frequency distribution or as a cumulative distribution. Cumulative distributions may be plotted as sigmoidal ogives or can be transformed into discrete probabilities (linear probits), which are then linear, and amenable to regression analysis. Probability analysis gives estimates of the mean (the value below which 50% of the observed concentrations lie, which we term 'OC50'). 'Baseline' can be defined in terms of percentage occupancy--the 'Observed Concentration for 5%' (which we term 'OC5') which is the threshold at or below which the hormone concentrations are measured 5% of the time. PATIENTS: We report the use of applying this method to 24-hour growth hormone (GH) profiles from 63 children, 26 adults and one giant. RESULTS: We demonstrate that GH effects (growth or gigantism) in these groups are more related to the baseline OC5 concentration than peak concentration (OC5 +/- 95% confidence limits: adults 0.05 +/- 0.04, peak-height-velocity pubertal 0.39 +/- 0.22, giant 8.9 mU/l). CONCLUSIONS: Pulsatile hormone profiles can be analysed using this method in order to assess baseline and other concentration domains.