Do plasma osmolality methods meet clinical requirements?
PJ Twomey1, AC Don-Wauchope1 & DA McCullough2
Plasma osmolality is used to investigate fluid and electrolyte abnormalities. The objective of this study was to investigate the degree of agreement between 2 routinely used osmometers and to compare this to clinical requirements.
49 plasma specimens were analysed once using the Advanced Micro-osmometer model 3MO and the Advanced Micro-osmometer plus model 3MO. These freezing-point depression based osmometers were calibrated using the same calibrator and were performing satisfactorily in the UKNEQAS scheme. Ordinary linear regression (OLR) analysis, absolute and percentage bias plots were performed.
The Pearson co-efficient was excellent with r = 0.99. The equation relating the two methods was y = 1.004 x - 0.387. The 95% confidence interval between the two methods was -5.2 to +6.7 mOsm/kg for the absolute plot and was -1.9% to +2.5% for the percentage bias plot.
In routine chemical pathology laboratories the difference between upper and lower reference ranges varies between 10 - 20 mOsm/kg. Accordingly a 95% confidence spread of 11.9 mOsm/kg between 2 routine osmometers is too large for clinical use. The current state of the art osmometers are based on the colligative property of freezing point depression and are on advance over previous methods based on the colligative property of vapour pressure. However, UKNEQAS data confirms that our findings are not unusual despite the majority of users employing freezing point depression osmometers.
There are also several algorithms that utilise at least 3 biochemical indices to calculate plasma osmolality but there are substantial biases between these algorithms in addition to an increased coefficient of variation that is associated with 3 or more variables. The potential therefore exists for significant misclassification of patients when undergoing water deprivation testing using both calculated and measured osmolality. We therefore conclude that plasma osmolality should be interpreted with caution especially around clinical cut-offs.