In this study we found that during the first two weeks of hospitalization 65% of 223 patients with severe COVID-19 developed sodium levels above 144 mmol/L and thus hypernatremia. We also found that patients with hypernatremia needed twice as long stay in ICU and had a higher risk of dying than patients without hypernatremia.
It cannot be excluded that the very high rate of hypernatremia we found could be iatrogenic and/or caused by activation of several endogenous regulating systems. The biochemical electrolyte shifts due to slight dehydration following treatment according to the acute respiratory distress syndrome (ARDS)-treatment protocol together with the virus activation of the RAAS would be expected to induce an increased reabsorption of sodium and an increased release of AVP. The latter could over time diminish the amount of available AVP and as a consequence decrease the reabsorption of water in the kidney, diabetes insipidus possibly in partial and transient form. In the present cohort, all patients with sodium levels above 149 mmol/L received active sodium-lowering treatment. Therefore, the very high sodium levels were unintended and likely caused by the disease.
The potential over-activation of RAAS with elevated angiotensin II levels could induce hyperaldosteronism. Aldosterone acts on the Na/K channels in the kidneys distal tubule mediating increased influx of sodium and outflow of potassium together with hydrogen ions. This results in hypertension, hypernatremia, hypokalemia and metabolic alkalosis.
Over time many patients in the study developed metabolic alkalosis with elevated pH and BE. Augmentation of BE is often a compensatory mechanism in respiratory acidosis, the body’s way of increasing low pH due to respiratory failure. In this cohort, though, the pH was alkalotic at admission. This respiratory alkalosis is suspected to be a result of the silent hypoxia, without hypercapnia, caused by the viral pneumonia.
Within a few days’ pH decreased to normal range in the hypernatremia patients but much slower in the non-hypernatremia group. The rise of BE did start at admission, in all groups, peaking days 7-11. This metabolic alkalosis was typed as chloride resistant, even hyperchloremic, also indicating possible aldosterone and dehydration/AVP involvement.
Of note, more patients in the subgroup who did not develop hypernatremia were receiving treatment for hypertension before the onset of COVID-19 and it can be speculated if the development of hypernatremia was inhibited by the anti-hypertensive drugs. In addition, the length of hospital stay was longer in the hypernatremia group and the mortality was higher compared to the non-hypernatremia group. Altogether, these clinical observations suggest that hypernatremia is an indicator of severe illness.
There was no difference in level of hyponatremia at admission between intubated and non-intubated. In the group of not intubated patients only one patient had sodium levels above 144 mmol/L. If this finding was due to less severe disease of the non-intubated patients or that the sedation, intubation and/or ventilation itself increases the risk of hypernatremia is an unanswered issue in this study.
A well-known cause of hyponatremia is SIADH where increased release of AVP elicits excess water reabsorption from aquaporins in the kidney’s tubule. Our study shows an incidence of hyponatremia of 63 % at admission. The main reason of admittance was atypical viral pneumonia with respiratory failure and, in this patient group, SIADH is known to be one of the most common underlying causes of hyponatremia.
In theory, patients presenting with hyponatremia at admission caused by SIADH could be of higher risk of developing severe hypernatremia later due to depletion of available AVP. This is difficult to substantiate due to patients seeking medical care at different time points and possible individual variation in AVP storage pool.
Synthetic AVP (desmopressin) is used as procoagulant treatment, e.g. in von Willebrand’s disease (vWD), as it increases platelet release of coagulation Factor VIII and von Willebrand Factor after administration. At admission the patients presented with low levels of sodium and suspected SIADH. The high release of AVP could be a part of the procoagulant state of COVID-19 further impaired by hyperosmolality which also increases the risk of venous thromboembolism(25). In our study very high levels of serum osmolality was seen a few days after admission, when sodium increased and the level of AVP released might have become insufficient.
There are some limitations to this study. Missing data is a hazard for retrospective studies and patients are not sampled according to a specific protocol leading to some variations in timing of blood tests between the patients. This study was based on routine blood tests and unfortunately no hormone biomarkers like aldosterone, angiotensin II or renin were analyzed. It would have been very interesting to see hormone results and also levels of sodium, potassium and osmolality in urine. This study, however, represents a large clinical consecutively recruited cohort of ICU patients, adding to the generalizability of our results.
Potassium infusions made it difficult to assess the untreated levels of hypokalemia present.
Another issue is that hemolysis is not registered by the machines used for blood gas analysis to the same extent and way as routine biochemistry lab does which potentially lead to a risk of falsely elevated potassium levels. The presence of CRRT is also a potential source of error due to its effect on electrolyte levels.
The presence or extent of hypertension during in-patient care was inaccessible owing to the majority of the patients receiving sedative drugs due to being intubated and mechanically ventilated.
In this retrospective study we studied a heterogeneous group of critically ill patients who received a multitude of intensive care and medications. Several factors and regulatory systems were involved to establish homeostasis and improve their condition. However, based on our findings with a high number of patients with hyponatremia at admission followed by hypernatremia the following weeks it can be speculated if this pattern of sodium could be a result of an over-activation of RAAS leading to a transient insufficiency of AVP. This would also be in line with the effect of SARS-cov2 on the ACE2 receptor. Furthermore, patients with hypernatremia displayed a more severe course of COVID-19 and the mortality rate was higher in this group compared to the patients not developing hypernatremia. To further understand the variations in sodium possible treatments, prospective studies with structured measurements of serum levels of angiotensin II, aldosterone, AVP, copeptin, urinary sodium, potassium and osmolality would be of great value.