Accuracy of the calculated serum osmolarity to screen for hyperosmolar dehydration in older hospitalised medical patients.

Abstract

Simple hyperosmolar dehydration, also termed water-loss dehydration (HD), is common in older hospitalised patients, thus increasing the risk of morbidity and mortality. Directly measured serum osmolality is the reference standard to determine HD; however, it is not a routine test due to its complexity and cost. Thus, a simple valid objective diagnostic tool to detect HD is needed. Consequently, we aimed to validate the agreement between measured s-osmolality (mOsm/kg) and calculated s-osmolarity (mOsm/L). Patients aged >65 were included from the emergency medical department at Herlev Hospital, Copenhagen, Denmark. Exclusion criteria were: eGFR< 30mmol/L, severe heart failure, decompensated cirrhosis, alcohol intake or initiated rehydration treatment. We obtained data for measured s-osmolality as well as calculated osmolarity, using the by ESPEN recommended equation [1.86x (Na++K+)+1.15 ∗glucose+urea+14]. To determine accuracy, we used cut-off values of >295 mOsm/L versus >300 mOsm/kg. A total of 90 patients (female 53%), age median 78yrs (72-86yrs) were included. According to the measured mOsm/kg, impending HD was evident in 32% (n=10), of these 11% (n=10) had current HD. There was a significant association between calculated mOsm/L and measured Osm/kg (r2=0.7513, p<0.0001). A sensitivity of 90% (95% CL: 56%-100%), a specificity of 68% (95% CL: 56%-78%), Positive predictive value (PPV) of 26% (95% CL: 12%-43%), and Negative predictive value (NPV) of 98% (95% CL: 90%-100%) were observed. Notably, only 20% (n=2) of the patients who were dehydrated according to the measured Osm/kg were correctly clinically diagnosed with dehydration. The equation recommended by ESPEN to calculate osmolarity was found to be an accurate objective diagnostic tool to assess HD in older hospitalised medical patients. The method is markedly superior to the current clinical practice.