Article Text


Changes in serum sodium levels during treatment of hyperglycaemia
  1. A M Oudesluys-Murphy,
  2. P van Echtelt
  1. Department of Paediatrics, Medisch Centrum Rijnmond-Zuid, Rotterdam, The Netherlands
  1. Correspondence to:
    Dr Oudesluys-Murphy
  1. M L Halperin
  1. Chief, Division of Nephrology, University of Toronto, St Michael’s Hospital, 38 Shuter Street, Toronto, Ontario, Canada MSB 1A6; mitchell.halperin{at}

Statistics from

Carlotti et al1 state that fluid and electrolyte management might contribute to the development of cerebral oedema in hyperglycaemia. There is a simple rule of thumb, formulated by Katz, which may help calculate water and electrolyte deficits and predict the changes in sodium levels which accompany changes in glucose levels,2 namely that a decrease of 0.29 mmol/l in serum sodium may be expected for every 1.0 mmol/l increment in serum glucose.

This may be explained as follows: hyperglycaemia causes an osmotic movement of water out of the cells, which leads to hyponatraemia by dilution. Thus, at presentation, the patient is usually severely dehydrated intracellularly. However, the serum sodium is lower than would be expected because of this dilution of the extracellular fluid. When the patient is treated with insulin, glucose enters the cells, taking water with it, leading to a relative concentration of the extracellular fluid, and thereby a rise in serum sodium. This rise may be predicted and calculated using Katz’s formula.2

Carlotti et al also comment on the report of Glaser et al that the chance of cerebral oedema during treatment is increased in children who present with high initial serum urea levels and when there is a lack of an increase in serum sodium levels during treatment.3 This increased risk may be explained by the fact that the urea level rises in proportion to the degree of dehydration. Urea contributes to serum osmolality and if the fall in urea is not taken into account the serum osmolality may be allowed to drop too rapidly, thereby increasing the risk of cerebral oedema. Carlotti et al do not take this into account in their formula for calculation of osmolality. The calculation of serum osmolality as twice the sum of sodium and potassium plus the urea and glucose levels (all in mmol/l) corresponds better with the formally measured osmolality.4

By treating hyperglycaemia using hypotonic solutions or glucose alone, the serum osmolality will fall rapidly and thereby increase the risk of cerebral oedema.

Serum osmolality must be monitored frequently, either by direct measurement or calculation from the sodium, potassium, glucose, and urea levels. In this way, the effects of falling urea and glucose levels on the serum osmolality will be compensated to a large extent by the accompanying rise in sodium. Thus the osmolality falls slowly and in a controlled fashion at a rate of 1–2 mosmol/kg/hour thereby, reducing the risk of cerebral oedema.


Author’s reply

We thank Dr Oudesluys-Murphy for her letter in response to our article. In essence, two points were raised:

  1. Can one estimate the deficits of Na + and water if one applies the formula proposed by Katz? 1

    This calculation makes the presumption that one can predict the change in plasma sodium concentration (PNa) when water is drawn out of cells by hyperglycaemia. This assumption is not correct for a number of reasons.2

    • Glucose must be added as a pure solute. Glucose will be retained in the ECF compartment (normal 10 L in a 50kg person with 30 L of total body water). With the net retention of 600 mmol of glucose without water in the ECF compartment, the PGlu will rise by close to 57 mM if we assume that glucose distribution is only in the ECF compartment because water will shift from cells to the ECF. In more detail, the total number of osmoles in the body was 8550 milliosmoles (285 × 30 L) before the addition of glucose and 9150 milliosmoles after the addition of glucose (8550 + 600). Therefore the new Posm will be 305 mOsm/kg H2O (9150/30 L). The new ECF volume is equal to the total ECF osmoles (2850 + 600) divided by the new osmolality of 305 mOsm/L, or 11.3 L. Therefore 1.3 L of water will be drawn out of cells due to the high PGlu. Bottom line: The new PGlu is 57.5 mM, the new PNa is 124 mM, and the new ECFV is 11.3 L.

    • Addition of isosmotic glucose (285 mM) to raise the PGlu by close to 50 mM with all the same assumptions: No water is drawn out of or enters cells because an iso-osmotic solution of glucose was added to the ECF compartment and all added glucose remains in the ECF compartment. When 2.3 L of this glucose solution is in the ECF compartment, the new PGlu is 57 mM, the new PNa is 114 mM because water was retained in the ECF compartment without Na+, and the new ECF volume is 12.3 L. Bottom line: The new PGlu is 57 mM, the new PNa is 114 mM, and the new ECFV is 12.3 L. Overall, because the ECF volume was expanded by different amounts in calculations A and B above yet the rise in the PGlu was virtually identical, there is no constant relationship between the PGlu and the ECF volume. Moreover, there was no change in the content of Na in the ECF compartment in these two examples. In contrast, patients presenting with DKA have a contracted ECF volume and a deficit of Na+ when their PGlu is 57 mM. Conclusion: If you do not know what the ECF volume is in quantitative terms, you cannot deduce the ECF Na+ content from the PGlu. Accordingly, much as we would like to agree with the suggestion of Dr Oudesluys-Murphy, the facts do not support that view.

    • Potassium: Part of the deficit of K+ reflects the shift of K+ out of cells in a 1:1 relationship with a cation (Na+ and H+) of unpredictable amounts. The other major part of K+ loss from the ICF reflects the catabolic state (primarily a loss of K+ with organic phosphate (e.g. from RNA)). Since both of these components are not known with certainty, one cannot use the relationship described by Katz1 to help in this context.

    • Error in the assumption of Katz1: The volume of distribution of glucose is larger than the ECF volume even if there is a lack of insulin actions. Our reasoning is that, in cells that do not require insulin for glucose transport such as liver cells, cells of the proximal convoluted tubule, and red blood cells, the concentration of glucose is likely to be equal in their ICF and ECF compartments.3

  2. Urea should be included in calculations of effective osmolality.

    Urea is not an effective osmole across cell membranes when the change in the plasma urea concentration (Purea) is not abrupt.4 Given the time course for the fall in Purea in patients with DKA, we did not include urea in our calculation of effective osmolality. Therefore we believe that it is more prudent to keep the PGlu + 2 (PNa + PK) relatively constant in the first 12–18 hours of therapy. A gradual decline in the effective Posm should occur with time thereafter.


View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.