Despite significant rates of overcorrection, very low rates of osmotic demyelination syndrome (ODS) were observed in a large cohort of hyponatremic patients, reported a study in NEJM Evidence (1).
Using the General Medicine Inpatient Initiative (GEMINI) database, which links electronic patient data with administrative hospital data for all patients admitted under general internal medicine, MacMillan et al. (1) conducted a multicenter cohort study of patients with hyponatremia, defined as an initial plasma sodium (PNa) <130 mmol/L, who were admitted to five academic hospitals in Toronto, Canada, over a period of approximately 10.5 years. Subsequent admissions for the same patient during the study period were included if the admissions met inclusion criteria. The researchers excluded patients who developed hyponatremia during hospitalization, patients with a plasma glucose ≥25 mmol/L (450 mg/dL), and patients with a history of diabetes insipidus because they may have been taking desmopressin.
The researchers identified 22,858 admissions (17,254 unique patients) meeting inclusion and exclusion criteria. The mean PNa in the entire cohort was 125 mmol/L, with 86.9% of patients with a PNa ≥120 mmol/L, a population already known to be at very low risk for ODS. Only 265 patients in the entire cohort had a PNa <110 mmol/L. Patients with acute, self-induced water intoxication, another group known to be at very low risk for ODS, were not excluded.
The primary outcome was the proportion of patients with hyponatremia who developed ODS on the index admission. ODS was identified by electronically searching for key words in radiology reports of magnetic resonance imaging (MRI) of the brain or computed tomography scans of the head. For any given positive screening report, the imaging report was reviewed. The investigators then identified a subset for manual chart review. In addition, using International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) codes for the diagnosis of ODS, a manual review was performed of all admissions, as well as readmissions, within 7 days of index admission with hyponatremia. ODS was reported in 12 patients in the entire cohort (0.05%). The rate of ODS in patients with a PNa <120 mmol/L was 0.3%. Patients with ODS had a lower initial PNa compared with patients without ODS (111 ± 10 mmol/L vs. 125 ± 4.6 mmol/L, respectively); 7 of the 12 patients with ODS had a PNa <110 mmol/L. Hypokalemia and alcohol use disorder occurred at a higher frequency in patients with ODS.
Secondary outcomes included overcorrection of hyponatremia, defined as an increase in PNa of >8 mmol/L in any 24-hour period. Since PNa measurements are not commonly obtained exactly at 24-hour time points, the investigators used the closest PNa within 6 hours before or after that time point. PNa values to estimate correction rates were available in 20,572 (90%) admissions in the entire cohort, and of these, 17.7% experienced overcorrection of hyponatremia. Overcorrection occurred in 184 (69.4%) patients with a PNa <110 mmol/L, with 81 patients correcting a PNa by ≥12 mmol/L. Of the 12 patients who developed ODS, 7 patients did not experience overcorrection of hyponatremia per the authors' definition. However, some of these patients became severely hypernatremic after correction of hyponatremia and had at least two other risk factors for ODS. It is now well known that rapid correction of hyponatremia is not the only osmotic challenge that can result in ODS; acute hypernatremia (2) and severe hyperglycemia (3) can also cause the disorder.
Although most of the identified cases of ODS had experienced either correction by >8 mmol/L in 24 hours or overcorrection resulting in hypernatremia, the investigators concluded that in most cases, overcorrection of hyponatremia is not causally related to ODS and infer that other factors that are as-yet unidentified must be implicated in the development of ODS in this setting. We believe this conclusion to be unwarranted.
The authors also concluded that ODS is an extremely rare complication of rapid correction of hyponatremia, citing an incidence of 0.05% in their cohort. We believe this conclusion to be extremely misleading. Most of the studied patients had a PNa >120 mmol/L, and some may have had acute hyponatremia from self-induced water intoxication. The recommended limit of 8 mmol/L in 24 hours does not apply to such patients because it is known that their risk of ODS is vanishingly low. For patients at higher risk, the 8-mmol/L limit was proposed since correction by 9 mmol/L can result in ODS (not because it commonly does), and because of the likelihood of “overshooting the mark” (4).
A more valid incidence of ODS can be found in patients with a PNa <110 mmol/L, the only group of participants in the study by MacMillan et al. (1) who were at a reasonably high risk of ODS. Seven of 265 patients with PNa <110 mmol/L (2.6%) developed ODS. Only 81 of the patients with a PNa <110 mmol/L were corrected by ≥12 mmol/L. Because of confidentiality safeguards, we do not know exactly how many of the patients with a PNa <110 mmol/L who developed ODS were corrected by ≥12 mmol/L. If there were five patients, the incidence of ODS was 6% (5 of 81). Inclusion of patients with a blood glucose as high as 450 mg/dL makes even these estimates suspect, as 1) the number of patients with a glucose-corrected PNa <110 mmol/L may have been <265, and 2) treatment of hyperglycemia could have raised PNa by as much as 6 mmol/L, resulting in an overestimate of the number of patients who were overcorrected.
The authors' methodology for defining overcorrection may also have led them to erroneous conclusions. Since most PNa is not measured exactly at a 24- or 48-hour time point, the investigators used the closest PNa within 6 hours of the time points. However, there could be considerable change of PNa in 6 hours affecting the 24-hour change of the PNa estimate. To address this issue, other studies have used an estimated PNa at the 24-hour time point (PNa24h) using a formula developed by Geoghegan et al. (5): PNa24h = PNaA + [(PNaB − PNaA) × (24 − TA)/(TB − TA)], in which PNaA and TA are the nearest PNa and time values before the 24-hour mark, respectively, and PNaB and TB are the nearest PNa and time values after the 24-hour mark, respectively. Although the authors also considered the maximum rate of correction in any 24-hour period, they did not provide their methodology for determining this rate, leaving the reader with unanswered questions: Was the increase in PNa for all 24-hour intervals during the entire hospitalization determined, and if so, how? If the PNa was immediately re-lowered after an increase of >8 mmol/L, was this still defined as overcorrection?
Other studies, in which most of the patients had mild to moderate hyponatremia or acute hyponatremia, have also reported low rates of ODS (6, 7). Studies of patients with chronic severe hyponatremia report very different findings. In a study by Vu et al. (8), 15% of patients with a PNa ≤120 mmol/L were corrected by >12 mmol/L over the first 24 hours; 4 of the 37 overcorrected patients (11%) developed ODS. All of the patients with ODS had a PNa ≤105 mmol/L; although the denominator of overcorrected patients with a PNa ≤105 is not given, their incidence of ODS must have been considerably higher. In two other studies (9, 10), approximately half of the overcorrected patients with a PNa ≤105 mmol/L developed ODS.
ODS is a clinical diagnosis, and its severity varies considerably. Patients with evidence of central pontine and extrapontine myelinolysis on MRI represent the more severe end of the spectrum, and even these cases can be missed when the case finding is based on radiology reports obtained during hospitalization for hyponatremia. Images are typically negative at the onset of clinical symptoms and may not be positive until weeks later. Some patients with a delayed onset of transient, apparently self-limited neurologic symptoms that emerge after discharge may not require readmission and may never have a positive MRI. For these reasons, the number of cases with ODS could easily have been underestimated in the study by MacMillan et al. (1).
Based on the results of this study, do we need to rethink our current approach to overcorrection of hyponatremia? Do we need to relax our PNa correction limits? Is it safe to rapidly correct all patients with hyponatremia? We believe the answer to these questions is no. Detractors of the current approach (e.g., use of desmopressin) point to drawbacks of a slow PNa correction, such as more frequent blood draws for PNa monitoring and increasing the length of hospital stays, but we should ask ourselves what percentage risk of ODS would we be willing to accept for our patients or family members so that they can be discharged from the hospital 1 day or 2 days earlier. The answer is probably none. There is a need for a multicenter study of patients with a PNa ≤105 mmol/L (a population likely to have a relatively high incidence of ODS) with meticulous manual chart review, not data mining using diagnostic codes and radiology results in a population with mild hyponatremia, which can be very misleading.
ODS [osmotic demyelination syndrome] is a clinical diagnosis, and its severity varies considerably.
References
- 1. ↑
MacMillan TE, et al. Osmotic demyelination syndrome in patients hospitalized with hyponatremia. NEJM Evid 2023; 2:1–9. https://evidence.nejm.org/doi/full/10.1056/EVIDoa2200215
- 2. ↑
Shah MK, et al. Osmotic demyelination unrelated to hyponatremia. Am J Kidney Dis 2018; 71:436–440. doi: 10.1053/j.ajkd.2017.10.010
- 3. ↑
Burns JD, et al. Central pontine myelinolysis in a patient with hyperosmolar hyperglycemia and consistently normal serum sodium. Neurocrit Care 2009; 11:251–254. doi: 10.1007/s12028-009-9241-9
- 4. ↑
Adrogue HJ, Madias NE. Hyponatremia. N Engl J Med 2000; 342:1581–1589. doi: 10.1056/NEJM200005253422107
- 5. ↑
Geoghegan P, et al. Sodium correction practice and clinical outcomes in profound hyponatremia. Mayo Clin Proc 2015; 90:1348–1355. doi: 10.1016/j.mayocp.2015.07.014
- 6. ↑
Aegisdottir H, et al. Incidence of osmotic demyelination syndrome in Sweden: A nationwide study. Acta Neurol Scand 2019; 140:342–349. doi: 10.1111/ane.13150
- 7. ↑
Nzerue CM, et al. Predictors of outcome in hospitalized patients with severe hyponatremia. J Natl Med Assoc 2003; 95:335–343. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2594506/
- 8. ↑
Vu T, et al. Grossmann M. Patients presenting with severe hypotonic hyponatremia: Etiological factors, assessment, and outcomes. Hosp Pract (1995) 2009; 37:128–136. doi: 10.3810/hp.2009.12.266
- 9. ↑
Sterns RH. Severe symptomatic hyponatremia: Treatment and outcome. A study of 64 cases. Ann Intern Med 1987; 107:656–664. doi: 10.7326/0003-4819-107-5-656
- 10. ↑
Sterns RH, et al. Neurologic sequelae after treatment of severe hyponatremia: A multicenter perspective. J Am Soc Nephrol 1994; 4:1522–1530. doi: 10.1681/ASN.V48152