Hyponatremia is the most common electrolyte disorder among inpatients, and is classically associated with community-acquired pneumonia (CAP) (1–3). Moderate-to-severe hyponatremia per se may cause permanent neurological damage, and hyponatremia is also correlated with higher levels of inflammatory markers, mortality rates, and other indices of disease severity in a variety of conditions (4–10). Furthermore, the treatment of moderate-to-severe hyponatremia often requires close inpatient monitoring, as neurological sequelae may arise from inappropriate rates of correction of blood sodium levels, causing significant stress on scarce healthcare resources (11, 12).

An association between COVID-19 and hyponatremia is well-recognized, with studies reporting incidences of hyponatremia of up to 25% (13–20). Hyponatremia was associated with severity of pulmonary infiltrates, need for intubation and mechanical ventilation, and death. In addition, at least two different studies have correlated hyponatremia in COVID-19 with elevated IL-6 levels, which could theoretically stimulate anti-diuretic hormone (ADH) release from the pituitary (19, 21–23). Treatment with the IL-6-specific monoclonal antibody tocilizumab normalized sodium levels in patients with COVID-19, in conjunction with overall disease improvement (21).

Interestingly, hyponatremia often manifests as part of a specific clinical syndrome in CAP. Some bacterial infections are more associated with hyponatraemia, such as pneumonia caused by Legionella pneumophila (legionellosis), suggesting that some infections stimulate ADH release or other neurohormonal pathways more potently than others (24, 25). Similarly, the common clinical manifestations of COVID-19 have evolved during the pandemic depending on the dominant circulating variant at the time (26). Since all the reported studies so far were performed in the pre-Omicron era, we postulated that the epidemiology and clinical correlates of COVID-19-related hyponatremia might be different during the global Omicron outbreak. Here, we report on the territory-wide prevalence and clinical outcomes of COVID-19 patients with hyponatremia in Hong Kong, focusing on a major outbreak of the SARS-CoV-2 Omicron BA.2 subvariant in early 2022 (the “fifth wave” in Hong Kong). We also compared these findings with earlier outbreaks of COVID-19 due to other variants, as well as with the outbreak of Severe Acute Respiratory Syndrome (SARS) due to SARS-CoV-1 in 2003.

COVID-19 is a systemic disease with various clinical manifestations in multiple organ systems. In this territory-wide retrospective analysis, around a quarter of adult patients with COVID-19 had hyponatremia of any degree. The importance of hyponatremia is underscored by its correlation with mortality and ICU admission, suggesting that hyponatremia may be a cardinal feature and prognostic indicator of moderate-to-severe COVID-19.

The rate of hyponatremia appeared to be particularly high during the “fifth wave” in Hong Kong involving the Omicron BA.2 subvariant. Up to 31.7% of patients in the “fifth wave” were found to be hyponatremic, compared to only 7.3% of patients during the “second wave” predominated by the original SARS-CoV-2 virus (P < 0.001). While it is likely that different variants of SARS-CoV-2 are associated with distinct disease presentations, patient characteristics may have played a role as well. During earlier outbreaks in Hong Kong, the territory followed a strict policy of “zero COVID,” and all patients and contacts were rigorously traced, diagnosed by RT-PCR, and hospitalized (30). In contrast, admitting all infected patients was logistically impossible during the large-scale Omicron outbreak. Many infected patients did not present to health care services, and fewer still had laboratory investigations performed. The patient populations of earlier waves may have captured a disproportionately large percentage of patients with asymptomatic or mild disease, resulting in lower rates of hyponatremia. During the “fifth wave,” hyponatremic patients in our analysis were significantly older and had more medical comorbidities than previous waves, suggesting the possibility of selection bias (Supplementary Table 6).

Our analysis shows that both disease-specific factors and patient factors modulate the risk of hyponatremia in COVID-19. Importantly, our study is the first to demonstrate a correlation between viral load of SARS-CoV-2, as demonstrated by the Ct value from respiratory specimens, and hyponatremia. Adding to that, CRP, an end-product of the IL-6 inflammatory pathway, was tied with an increased risk of hyponatremia. This is consistent with previous studies correlating IL-6 levels with hyponatremia in COVID-19 and other inflammatory conditions, possibly through stimulation of pituitary ADH secretion (21, 22). These findings suggest that SARS-CoV-2 viral antigens may directly contribute to the pathogenesis of hyponatremia via activation of inflammatory pathways.

Patients who presented with hyponatremia tended to be older, male, and with more medical comorbidities. Although these same factors are also recognized risk factors for developing moderate-to-severe disease in COVID-19, multivariate analysis showed that certain comorbidities such as DM, active malignancy, CHF, dementia and CKD may particularly predispose patients to develop hyponatremia, independent of other risk factors and indices of systemic inflammation (31, 32). The mechanisms by which these conditions contribute to hyponatremia in COVID-19 remain speculative, with a paucity of reliable biochemical and clinical data available from large registry studies to explain the underlying pathophysiology.

The overall mortality of COVID-19 patients with hyponatremia was high, and was significantly greater than those without hyponatremia. Thirty-day mortality was up to 9.7% for hyponatremic patients compared to 5.7% for normonatremic patients. Regression analysis showed that the excess risk for death was primarily driven by patients with moderate and severe hyponatremia. Furthermore, deaths associated with hyponatremia tended to occur early, and there was little difference between the mortality rate at 30 and 90 days. Hyponatremia was also a predictor of ICU admission, especially with more severe degrees of hyponatremia. Paradoxically, patients with severe hyponatremia actually had shorter ICU stays than normonatremic patients, likely due to the extremely high rates of early mortality among patients with severe hyponatremia. In general, hyponatremic patients had longer hospital stays than normonatremic ones, which could be due to greater severity of disease as well as the need for close monitoring to avoid over-rapid correction of hyponatremia. All these factors conspire to burden heavily overstretched healthcare systems during large-scale COVID-19 outbreaks.

In our study, we further compared the prevalence and outcomes of hyponatremia among patients with COVID-19 with their counterparts with SARS during the 2003 outbreak of SARS-CoV-1. This comparison is salient, as the SARS-CoV-1 pandemic was the most widespread and significant coronavirus outbreak before the current COVID-19 pandemic, and the two causative viruses share significant homology (33). In our analysis, hyponatremia was not a significant predictor for either mortality or ICU admission during the SARS-CoV-1 outbreak after adjusting for other demographic variables and biochemical parameters, contrasting with the strong dose-response relationship for mortality noted in COVID-19. These findings suggest that respiratory tract infections due to different coronaviruses, including different variants of the same species, may manifest different clinical and biochemical phenotypes. These findings are analogous to the differing rates of hyponatremia with different causes of bacterial pneumonia, with higher rates observed in atypical infections such as legionellosis (24, 25).

This study has certain limitations. First, paired plasma and urine biochemistries were unavailable for many patients with hyponatremia. Volume status could not be accurately ascertained and the etiology of hyponatremia in many cases could not be determined with confidence. To overcome this, we attempted to evaluate the volume status by the percentage change of serum creatinine from hyponatremia to eunatremia. Second, this study was observational in nature. Without specific interventions directed at correcting hyponatremia, a direct causal effect of hyponatremia on clinical outcomes could not be established. Third, morbidity from severe hyponatremia often arises from inappropriate correction of hyponatremia. However, the method of correction of hyponatremia, including solute therapy, fluid restriction, saline infusions, or other pharmacological interventions, was not available from our registry analysis. Fourth, we also used the measured plasma sodium before adjusting for glucose level in the analysis. In addition, the measurement of plasma sodium by the indirect ion selective electrode (ISE) methods may further be affected when the plasma glucose reaches an exceedingly high level (34, 35). Fifth, the time lag between onset of symptoms and presentation to health care services could not be accounted for. Delayed presentation may have contributed to a higher prevalence of hyponatremia and severe disease, although these discrepancies also reflect real-world variations in healthcare access and evolutions in health policy. Sixth, due to the limitation of the registry, the disease coding is based in ICD-9, but not the most updated ICD-10. Nevertheless, this database has a high diagnostic accuracy, and has been widely used in several publications (36, 37). Finally, we could not exclude an impact on the prevalence and outcomes of hyponatremia related to the availability of COVID-19 vaccines and use of immunomodulatory or antiviral treatments in later waves. In Hong Kong, in earlier waves of COVID-19, treatment consisted of combinations of interferon-1β, lopinavir-ritonavir, ribavirin, and corticosteroids; treatment evolved to encompass antivirals including remdesivir, molnupiravir and nirmatrelvir-ritonavir, and novel immunomodulatory agents including baricitinib and tocilizumab in later waves. Crucially, hyponatremia was a convincing risk factor for mortality and other adverse outcomes regardless of the wave of COVID-19 and the availability of COVID-19 treatments.

A key strength of our study is the large number of patients included in this territory-wide analysis, all of who were diagnosed by RT-PCR. Data on viral loads were available for all patients, allowing for correlation between viral burden and important clinical parameters, including hyponatremia. All patients with COVID-19 were followed up for at least 90 days or until death, and received similar standards of medical care, providing insight into real-world outcomes. Furthermore, our analysis included comparisons across different waves of COVID-19 locally, each attributable to a different circulating variant, as well as a comparative analysis with a historical cohort of patients with SARS-CoV-1 infection. This allowed for the characterization of a syndrome of hyponatremia specific for COVID-19, particularly for the dominant Omicron BA.2 subvariant during the largest local outbreak thus far.

Hyponatremia is common among COVID-19 patients, and is associated with adverse outcomes including mortality, prolonged hospitalization, and ICU admission. Further research is required to clarify the causes of hyponatremia in COVID-19, and identify appropriate strategies for the treatment of hyponatremia and optimization of clinical outcomes in COVID-19.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Institutional Review Board of the University of Hong Kong/Hospital Authority Hong Kong West Cluster, Hong Kong SAR, China. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

GC, CW, BS, KC, and DY conceptualized the work. GC, CW, and BS drafted the original manuscript. JN, GL, IH, HT, ST, TC, KC, and DY reviewed and edited the manuscript. All authors contributed to the article and approved the submitted version.