Context: Hypothalamus-pituitary-adrenal function is associated with important physiological/pathological events in neonates. Plasma/serum cortisol levels have been used to assess hypothalamus-pituitary-adrenal function. Several noninvasive surrogate markers have been used without sufficient validation.
Objective: The objectives of the study were to investigate whether plasma cortisol levels are correlated with those in saliva and urine and whether these correlations are affected by procedural pain at blood sampling.
Design, setting, and patients: Fifty neonates were recruited from a tertiary neonatal intensive care unit. Saliva and urine samples were collected shortly before routine clinical blood sampling.
Main outcome measures: Cortisol levels were compared between plasma and noninvasive samples using a linear regression analysis for the entire study population and groups, whose blood was obtained via indwelling arterial catheters (group A) or by venipuncture (group V). Predictive values of salivary/urinary cortisol for low plasma cortisol levels less than 2.0 μg/dL were evaluated by receiver-operating characteristic analysis.
Results: Plasma cortisol showed linear correlations with salivary and urinary cortisol for the entire study population and patients in group A (all P < .0002) but not in group V. Areas under the curves of salivary and urinary cortisol to predict low plasma cortisol levels were 0.87 (95% confidence interval 0.78-0.97) and 0.84 (95% confidence interval 0.74-0.95), respectively.
Conclusions: Cortisol levels from saliva or urine samples were shown to be useful surrogate markers for plasma cortisol levels in neonates. Caution is required in interpreting the findings of plasma cortisol levels in young patients when blood samples are obtained by venipuncture because procedural pain may induce alteration of cortisol levels.