Hypertonic saline was first used to treat hyponatremia in 1938 (1), and it is still generally accepted as the treatment of choice for hyponatremic emergencies (2,3). However, in many medical centers, hypertonic saline is rarely used; fear of complications—most notably the osmotic demyelination syndrome (ODS), a neurologic disorder thought to result from rapid correction of hyponatremia (4)—may be responsible for its limited use.
Clinicians have few data to guide them in the optimal use of hypertonic saline. Although guidelines for rates of infusion and monitoring procedures can be found in the literature (2,5),remarkably few studies have reported on dosage–response relationships. Formulas based on the apparent distribution of sodium in total body water (2,6–8) have been widely used clinically to predict the rise in serum sodium in response to hypertonic saline. However, it is not clear how accurately these calculations predict correction rates in clinical use of hypertonic saline: only one of these formulas was recently evaluated prospectively in a small number of patients (9).
We report our experience with hypertonic saline use for the treatment of hyponatremia in 62 patients during a 5-yr period in a 528-bed, acute care, teaching hospital. Previous observational studies reported from our institution have shown that ODS can usually be avoided in severely hyponatremic patients by limiting correction rates to no more than 12 mEq/L in 24 h and 18 mEq/L in 48 h (4,10,11). For many years, as a matter of policy, hypertonic saline use has been overseen by a single group of nephrologists to ensure adherence to these guidelines. This experience provided us with a unique opportunity to assess the pattern of hypertonic saline use and the clinical impact of these guidelines on outcomes and safety. In addition, we assess the accuracy of a commonly used formula in predicting the rise in serum sodium.
After institutional review board approval, paper charts and electronic health records of all cases of 3% saline use at Rochester General Hospital between December 1999 and December 2004 were reviewed. We excluded patients who were younger than 18 yr or those to whom hypertonic saline was administered for indications other than hyponatremia or was administered by any route other than intravenous.
Data collected included demographics, medical history, clinical presentation, physical and laboratory findings, urine output, suspected complications, and volume status as assessed by the treating physician. The volume and duration of hypertonic saline infusion and the clinical setting in which it was started were also noted. Physicians’ notes were reviewed for their sense of syndrome of inappropriate antidiuretic hormone secretion (SIADH) as being the cause of hyponatremia. Serum sodium concentrations within 4 h of the 24- and 48-h marks from the time of initiation of infusion were used to calculate the rates of change in serum sodium concentration.
Patients whose serum sodium concentrations, after hypertonic saline infusion, rose >12 mEq/L in 24 h or 18 mEq/L in 48 h were considered as having failed to have met the goals of correction of hyponatremia. The Adrogue[Combining Acute Accent]-Madias formula, shown as follows, is a widely used tool to predict the change in serum sodium levels in response to intravenous fluids (2,9):
Change is serum sodium concentration
with 1L of infusate = Infusate sodium concentration - Serum concentration/ Total body water + 1
To test the predictive accuracy of this formula, we looked at the patients whose charts contained sufficient data to apply this formula—namely the gender and weight of the patient, the sodium concentrations at the time intervals described previously, and volume of hypertonic saline infused. The formula was then applied, as recommended (2), to predict the change in sodium concentration that would be expected to be caused by the hypertonic saline infusion. This change in sodium level was termed the “expected rise in sodium.”
Any change in mental status was considered to be due to hyponatremia when the symptoms were present before and resolved after the correction of hyponatremia. Complications were considered to be due to hypertonic saline therapy when there was any documentation of seizures, pulmonary edema, or neurologic dysfunction during or after the infusion and when no alternative explanations for those complications were found. Charts and available imaging results were reviewed for any clinical or radiologic findings consistent with ODS.
The data were analyzed using Minitab 14 (Minitab Inc., State College, PA). Normally distributed variables are summarized with mean ± SEM, and for non-normal data, medians (interquartile range [IQR]) are used. Statistical significance was assessed for discrete variable by χ2 analysis or Fisher exact test and for continuous variables by t test or Kruskal-Wallis test, as appropriate. Linear regression was used to look for correlation between continuous variables. This study had no external funding source.
Ninety-two instances of 3% saline use were identified. Thirty patients met the exclusion criteria. The remaining 62 patients were included in the analysis. In 45 patients, the documentation contained sufficient information to apply the Adrogue[Combining Acute Accent]-Madias formula to calculate the effect of hypertonic saline infusion on serum sodium concentration.
Our study population was predominantly white (92%) and female (74.2%). Median age was 76.5 yr (IQR 62.8 to 84.0 yr). The women, as in the previously published series from our institution (10), tended to be older (median age 77.5 yr; IQR 67.5 to 85.0 yr) than men (median age 69 yr; IQR 56.2 to 80.7 yr; P = 0.075 by Kruskal-Wallis test). The most commonly encountered (see Table 1) associated conditions were hypertension (71%), psychiatric disorders (32.3%), congestive heart failure (24.2%), diabetes (21%), and malignancy (19.3%). Eighteen (29%) patients were taking a thiazide diuretic, 10 (16.1%) a loop diuretic, and 15 (24.2%) selective serotonin reuptake inhibitors (SSRI).
With the use of previously reported definitions (10,11), hyponatremia was chronic in 47 (76%) and acute (hospital acquired or associated with psychogenic polydipsia) in 15 (24%) patients. Six (10%) patients had seizures at presentation, none of which was ongoing at the time of hypertonic saline administration. Of these, five patients had preexisting seizure disorder. Nine (15%) patients were unresponsive, somnolent, or obtunded. The remaining patients had a variety of nonspecific complaints, including confusion, disorientation, lethargy, nausea, vomiting, dizziness, weakness, gait disturbances, and falls (two with fractures). Three patients had neurosurgical conditions (subdural hematoma and pituitary surgery) that, in addition to their symptoms, prompted hypertonic saline. The mean sodium concentration at presentation was 116.9 ± 0.9 mEq/L. Mean urine osmolality was 431 ± 22 mOsm/kg. Forty-six (74.2%) patients were euvolemic, and of these, 39 (62.9%) were considered to have SIADH by the treating physicians. One (1.6%) patient each was hypervolemic and hypovolemic; the remaining 14 (22.6%) patients had an indeterminate volume status and multiple coexisting risk factors for hyponatremia.
Hypertonic saline infusion was started in the emergency department (ED) in 19 (30.7%) patients, in the intensive care unit (ICU) in 9 (14.5%) patients, and on medical floors in 34 (54.8%) patients. In 84% of all patients and in 100% of patients with serum sodium <120 mEq/L, a nephrologist supervised the administration of hypertonic saline.
Because of either the absence of serious neurologic symptoms or unsuccessful attempts to increase the serum sodium concentration without hypertonic saline, there were substantial delays between the diagnosis of hyponatremia and the initiation of hypertonic saline therapy. In the ED, the median delay in the initiation of hypertonic saline therapy from the time of presentation was 5.6 h (IQR 4.3 to 8.0 h). This delay was significantly short when compared with the median delay of 16.5 h (IQR 12.0 to 27.6 h) on the medical floors (n = 43; P < 0.001) and with the ICU (n = 26; P = 0.005), where the median delay was 13.2 h (IQR 8.0 to 27.0 h). The delay in initiating therapy was significantly short (n = 52; P = 0.04 by Kruskal-Wallis test) when the sodium level at presentation was <110 mEq/L (7.3 h; IQR 4.0 to 10.0 h) than when it was ≥110 mEq/L (14.0 h; IQR 6.3 to 21.0 h). The physicians responded by ordering hypertonic saline approximately twice as quickly when the patients were either obtunded or having seizures (median response time 4 h; IQR 2 to 10 h) than when they were not (median 8.8 h; IQR 4.0 to 16.5 h), but this did not reach statistical significance (n = 58; P = 0.16 by Kruskal-Wallis test).
Response to Treatment
The median rate of administration of hypertonic saline was 23.5 ml/h (IQR 17.0 to 32.2 ml/h). In the 45 patients whose body weight was recorded, the median rates of infusion were 0.38 ml/kg per h (IQR 0.25 to 0.50 ml/kg per h) or 0.19 mmol/kg per h (IQR 0.13 to 0.25 mmol/kg per h). The median amount of hypertonic saline infused was 386.5 ml (IQR 241.0 to 707.0 ml). The mean serum sodium concentration before the administration of hypertonic saline was 117.5 ± 0.8 mEq/L (Figure 1).
During the infusion of hypertonic saline, the increase in serum sodium level averaged 7.5 ± 0.7 mEq/L and the average rate of rise in sodium concentration was 0.47 ± 0.05 mEq/L per h. The average change in serum sodium concentration during the first 24 h after hypertonic saline therapy was 7.1 ± 0.6 mEq/L and during the first 48 h was 11.3 ± 0.7 mEq/L. As might be expected, the increase in sodium concentration was significantly related (Figure 2) to pretreatment sodium concentration (R2 = 0.43; P < 0.001). There were no significant differences in the rates of rise in sodium concentration among the clinical settings.
In seven (11.3%) of 62 patients, the increase in serum sodium concentration during the first 24 h exceeded 12 mEq/L and, in six (9.7%) patients, the increase in the first 48 h exceeded 18 mEq/L. The 10 (16.1%) patients whose correction rates exceeded either of these two limits (in three patients, the correction rate exceeded both limits) were designated “overcorrectors.” The mean pretreatment sodium concentration in overcorrectors was 111.9 ± 1.5 mEq/L, significantly lower than the pretreatment sodium concentration in patients whose correction rates remained within guidelines (mean 118.5 ± 0.8; P = 0.002 by t test; n = 62). Fifteen (28.9%) of the 52 nonovercorrectors missed overcorrection marks by 1 or 2 mEq/L; these were designated as “near misses.” Furthermore, all of the overcorrectors had pretreatment serum sodium concentrations <120 mEq/L. Therefore, we conducted a subgroup analysis in the 37 patients whose serum sodium concentration was <120 mEq/L to define the factors that were associated with excessive rise in sodium concentration.
Among patients with a pretreatment serum sodium concentration <120 mEq/L, the increase in serum sodium level during the infusion of hypertonic saline averaged 8.7 ± 0.8 mEq/L, and the average rate of correction of hyponatremia was 0.5 ± 0.1 mEq/L per h. The average change in serum sodium concentration during the first 24 h after hypertonic saline therapy was 8.5 ± 0.7 mEq/L and during the first 48 h was 14 ± 0.6 mEq/L. Overcorrectors received significantly less hypertonic saline (median 334 ml; IQR 94 to 520 ml) than patients whose correction rates remained within guidelines (median 640 ml; IQR 305 to 932 ml; P = 0.04; n = 31); however, the rate of hypertonic saline infusion was significantly higher in the overcorrected group (median 0.49 [IQR 0.44 to 0.59] versus 0.29 [IQR 0.25 to 0.42] ml/kg per h; P = 0.038; n = 31). Furthermore, the overcorrectors had significantly more serum sodium values obtained in the first 24 h (median 6.0; IQR 5.0 to 6.2) than patients who were not overcorrected (median 4; IQR 4 to 5; P = 0.01; n = 37). Four of the 10 overcorrectors had water diuresis that emerged during the course of therapy, documented by low urine osmolality or high urine output (Table 2). In three of the overcorrectors, there was documented use of 5% dextrose water in an effort to blunt the rapid increase in serum sodium concentration. In one patient, 5% dextrose water and DDAVP were administered after an increase in serum sodium concentration of 16 mEq/L in 18 h; with this intervention, the serum sodium concentration was successfully lowered and the patient remained within the 24- and 48-h goal limits.
Eleven (29.7%) patients received concurrent potassium chloride infusion (mean 55.5 mmol). Concurrent administration of potassium chloride was more prevalent in the overcorrected group, but the difference did not reach statistical significance (P = 0.09 by Fisher exact test). The mean pretreatment serum potassium concentration was 3.7 mEq/L ± 0.1, and only one patient had serum potassium concentration <2.5 mEq/L.
Predictive Accuracy of Adrogue[Combining Acute Accent]-Madias Formula
The increase in serum sodium concentration that could be expected from the hypertonic saline infusion was calculated in 45 patients (31 of these had pretreatment sodium concentrations <120 mEq/L) using the formula proposed by Adrogue[Combining Acute Accent] and Madias (2,9) and is referred to as the “expected increase” in serum sodium concentration. The ratio of the actual increase in serum sodium concentration to the expected increase in serum sodium concentration during therapy is shown in Figure 3. A value of 1 in this ratio indicates that the entire increase in serum sodium concentration can be accounted for by the administered hypertonic saline; a value <1 indicates that the actual increase is less than would be predicted; and a value >1 indicates that the actual increase in serum sodium concentration exceeds the predicted increase. As shown in Figure 3, in patients with serum sodium concentrations <120 mEq/L, the actual increase in serum sodium concentration exceeded the expected increase in 74.2% of the patients, and the average ratio of actual to expected was 1.66 ± 0.2. In overcorrectors, the ratio was significantly higher than in patients who were not overcorrected (2.4 [IQR 1.6 to 2.9] versus 1.3 [IQR 0.7 to 1.9]; P = 0.006 by Kruskal-Wallis test; n = 31). Among patients with a pretreatment serum sodium concentration <120 mEq/L, there were no significant differences between the overcorrectors and the nonovercorrectors in use of medications or comorbidities, with the exception of psychiatric disorders, which were more prevalent among the overcorrectors (P = 0.049 by Fisher exact test).
Complications of Hypertonic Saline Use
There were no documented cases of ODS. Ten (16.1%) patients had preexisting neurologic dysfunction that persisted after hypertonic saline infusion. Each of these had an alternative cause for persistent neurologic dysfunction (Table 3), and none of these patients was overcorrected. There were no respiratory arrests before or during hypertonic saline infusion despite slow rates of rise in sodium concentration and long delays in starting the infusions. One patient, patient 58, also developed fluid overload because of progressive renal failure. Two patients died during the same admission from causes unrelated to hypertonic saline therapy (cardiac arrest and progressive azotemia from bilateral hydronephrosis).
Some authorities have suggested that hypertonic saline should be used only in critical care units (12) or that it be limited to patients with repeated convulsions, agitated confusion, or coma (13). At our institution, hypertonic saline is permitted outside the critical care units, but oversight by nephrologists is strongly encouraged and is required for administration of >200 ml. Although hypertonic saline is accepted as a treatment for hyponatremic seizures, this was rarely the indication. Rather, hypertonic saline was used when it was believed that isotonic saline or water restriction alone was unlikely to increase the serum sodium concentration promptly. In most patients, the rate of infusion was modest, the median being 0.38 ml/kg per h—a rate that would be expected to increase the serum sodium concentration by <0.5 mEq/L per h (2). Hypertonic saline was administered in three different clinical settings: ED, ICU, and medical wards. Although there were significant differences in the time taken by the physicians to respond to low sodium levels—likely a result of the inherent differences between these settings with regard to the rapidity of phlebotomy, laboratory processing times, and medication dispensing as well as nurse staffing—the clinical setting did not seem to have an impact on either the rates of overcorrection or the neurologic outcomes.
We were successful in maintaining a rate of rise in sodium concentration ≤12 mEq/L per d and 18 mEq/L in 48 h in 84% of patients, and in all patients correction remained well below 25 mEq/L in 48 h, a rate that is associated with a high incidence of severe posttherapeutic neurologic complications (14). Fortunately, there were no complications in patients who were overcorrected. The favorable outcome depended on frequent interventions by medical housestaff guided by nephrology consultants to modify the rate of infusion of hypertonic saline and in some cases to administer 5% dextrose water and/or DDAVP (in the same dosages used to treat central diabetes insipidus) as “antidote.”
Patients who were overcorrected actually received less hypertonic saline than those who remained within therapeutic guidelines. This seeming paradox is explained by the downward adjustments in the rate of infusion that occurred in response to more rapid increases in the serum sodium concentration. Indeed, in some patients, not only was hypertonic saline stopped, but also therapeutic rescue with 5% dextrose water and/or DDAVP was required because of the unanticipated emergence of a water diuresis during therapy.
Predictive equations are based on the relationship among the serum sodium concentration, total body exchangeable sodium and potassium, and total body water as defined empirically by Edelman et al. (6). Although clinically popular formulas are reasonably accurate in predicting the increase in serum sodium concentration from hypertonic saline infusion, in our experience, the formula tended to underestimate the increase, and in overcorrectors, the actual increase in sodium concentration was up to five times the predicted rate. The predictive formula that we used omits both potassium and the intercept in Edelman's empirical relationship (6,15,16). Severe hypokalemia was infrequent and potassium replacement was modest in this series.
Liamis et al. (9) recently prospectively evaluated the Adrogue[Combining Acute Accent]-Madias formula in patients with dysnatremias, including hypertonic saline with intravenous furosemide, in 10 patients with SIADH. The degree of hyponatremia, however, was much less severe (average pretreatment sodium was 122.6 ± 5 mEq/L, compared with 117.4 ± 0.8 mEq/L in our series and 111.9 ± 1.5 mEq/L in the overcorrectors). In addition, the investigators’ concurrent use of furosemide may have helped prevent spontaneous water diuresis that frequently accompanied overcorrection in our series. In patients with SIADH, the difference between the actual and predicted serum sodium after 12 h of 3% saline did not reach statistical significance; nevertheless, the actual increase was 1.5 mmol/L higher than the 4.8 mmol/L increase predicted—a 31% discrepancy. The discrepancy was of similar magnitude in patients with thiazide-induced hyponatremia and was much greater in patients with polydipsia. These discrepancies, because of the small number of patients, did not reach statistical significance, but these data, although limited, highlight the limitation of the Adrogue[Combining Acute Accent]-Madias formula in predicting change in sodium concentration with hypertonic saline therapy.
Although many of the overcorrectors in our series were perceived to be euvolemic and as having definite SIADH by the attending nephrologists, their response to hypertonic saline infusion is suggestive of multifactorial causes for the hyponatremia, possibly including undiagnosed hypovolemia and other reversible impairments of water excretion. Indeed, as is seen in practice, an accurate clinical assessment of volume status is often extremely difficult (17), especially in elderly patients with altered mental status. In such cases, the rapid volume expansion from hypertonic saline infusion can appropriately suppress ADH secretion, effect a water diuresis, and result in a rapid rise in serum sodium concentration. Although establishing volume status remains an important part of the traditional diagnostic approach to hyponatremia, when approaching a patient from the therapeutic standpoint, the clinician needs to be cognizant that establishing volume status accurately may not be possible. One such case is presented next to illustrate this point.
An 87-yr-old woman who was treated long term with hydrochlorothiazide for hypertension was admitted with a serum sodium of 106 mEq/L and a serum potassium of 2.6 mEq/L 2 wk after starting treatment with an SSRI. The thiazide and the antidepressant were stopped, and she was given isotonic saline and potassium replacement; however, her serum sodium increased only by 3 mEq/L during her first 14 h in the hospital, and her urine output was 60 ml/h. Therefore 3% saline was prescribed at 20 ml/h. A few hours later, diuresis developed with a recorded urine output of 1950 ml in 7 h and urine osmolality of 90 mOsm/kg. Hypertonic saline was discontinued when the serum sodium was 118 mEq/L, after 120 ml had been administered (enough to increase the serum sodium by a calculated 3 mEq/L). However, primarily because of the water diuresis, the serum sodium increased from 109 to 124 mEq/L in 18 h. Therefore, 5% dextrose in water was prescribed, successfully preventing any further increase in sodium concentration. This case illustrates several of the problems that we have encountered at our hospital while treating patients for hyponatremia with hypertonic saline:
- There is often ambiguity as to the cause of hyponatremia; was this patient's hyponatremia the result of thiazide diuretic and volume depletion, or was it due to SIADH caused by an SSRI?
- The patient's condition often changes over time; initially this patient failed to respond to isotonic saline, suggesting that her hyponatremia was due to SIADH, but then a water diuresis emerged, reflecting discontinuation of the diuretic or the SSRI or a response to volume repletion.
- Attempts to predict the increase in serum sodium to be expected from 3% saline are often inaccurate; in this case, the calculated increase was one fifth the increase that actually occurred.
Some authors have drawn a distinction between “symptomatic” and “asymptomatic” hyponatremia, arguing that patients with symptoms should be treated with hypertonic saline by at least 1 to 2 ml/kg per h (12). Our experience suggests that for most patients with symptomatic hyponatremia, this aggressive approach may not be necessary, and we are unaware of any data showing that it produces better results than the more conservative approach that we have used. As others have reported (18,19), most patients with hyponatremia have at least subtle symptoms, and in our patients, these symptoms did not worsen or lead to respiratory arrests during treatment despite rates of correction that are considerably slower than what others have reported. We do not dispute the recommendation that patients with active seizures or impending herniation be given 100-ml bolus infusions of hypertonic saline (12); however, we did not encounter any patients with such a presentation in this series. Similarly, case series (14,20) reporting a favorable outcome with more aggressive therapy did not include such patients.
In our series, although the overcorrectors received less hypertonic saline, they also received it at a more rapid rate than patients whose correction rates remained within guidelines. Moreover, the infusion rate in both of these groups was well below the 1- to 2-ml/kg per h rate as has been recommended. On the basis of this experience, especially when uncertainty regarding the volume status exists, it may be desirable to initiate hypertonic saline therapy with even slower rates than those formerly recommended and those predicted by the popular formulas, particularly in chronically hyponatremic patients with modest symptoms.
Even with careful oversight by medical residents and nephrology consultants and despite rates of infusion that are considerably slower than what others have used, the number of “near misses” in our series was disturbingly high. Regardless of how much hypertonic saline is used or how fast it is infused, clinicians should recognize that the increase in serum sodium concentration cannot be reliably predicted by formulas and that frequent monitoring of the urine output and serum sodium concentrations is mandatory.
Some of these data were presented as an abstract at the annual meeting of the American Society of Nephrology; November 14 through 19, 2006; San Diego, CA.
Published online ahead of print. Publication date available at www.cjasn.org.
See related editorial, “The Adrogue-Madias Formula Revisited,” on pages 1098–1099.
Access to UptoDate on-line is available for additional clinical information at http://www.cjasn.org/
1. Helwig FC, Schutz CB, Kuhn HP: Water intoxication: Moribund patient cured by administration of hypertonic salt solution. JAMA110 :644– 645,1938
2. Adrogue HJ, Madias NE: Hyponatremia. N Engl J Med342 :1581– 1589,2000
3. Gross P, Reimann D, Neidel J, Doke C, Prospert F, Decaux G, Verbalis J, Schrier RW: The treatment of severe hyponatremia. Kidney Int Suppl64 :S6– 11,1998
4. Sterns RH, Riggs JE, Schochet SS Jr: Osmotic demyelination syndrome following correction of hyponatremia. N Engl J Med314 :1535– 1542,1986
5. Culpepper RM, Clements BD, Pence SR: Hypertonic saline: Patterns of and guidelines for use. South Med J87 :1203– 1207,1994
6. Edelman IS, Leibman J, O’Meara MP, Birkenfeld LW: Interrelations between serum sodium concentration, serum osmolarity and total exchangeable sodium, total exchangeable potassium and total body water. J Clin Invest37 :1236– 1256,1958
7. Spital A, Sterns RD: The paradox of sodium's volume of distribution. why an extracellular solute appears to distribute over total body water. Arch Intern Med149 :1255– 1257,1989
8. Adrogue HJ, Madias NE: Aiding fluid prescription for the dysnatremias. Intensive Care Med23 :309– 316,1997
9. Liamis G, Kalogirou M, Saugos V, Elisaf M: Therapeutic approach in patients with dysnatraemias. Nephrol Dial Transplant21 :1564– 1569,2006
10. Sterns RH: Severe symptomatic hyponatremia: Treatment and outcome. A study of 64 cases. Ann Intern Med107 :656– 664,1987
11. Sterns RH, Cappuccio JD, Silver SM, Cohen EP: Neurologic sequelae after treatment of severe hyponatremia: A multicenter perspective. J Am Soc Nephrol4 :1522– 1530,1994
12. Achinger SG, Moritz ML, Ayus JC: Dysnatremias: Why are patients still dying? South Med J99 :353– 362, quiz 363–364,2006
13. Laureno R, Karp BI: Myelinolysis after correction of hyponatremia. Ann Intern Med126 :57– 62,1997
14. Ayus JC, Krothapalli RK, Arieff AI: Treatment of symptomatic hyponatremia and its relation to brain damage. A prospective study. N Engl J Med317 :1190– 1195,1987
15. Nguyen MK, Kurtz I: New insights into the pathophysiology of the dysnatremias: A quantitative analysis. Am J Physiol Renal Physiol287 :F172– F180,2004
16. Nguyen MK, Kurtz I: Role of potassium in hypokalemia-induced hyponatremia: Lessons learned from the Edelman equation. Clin Exp Nephrol8 :98– 102,2004
17. McGee S, Abernethy WB 3rd, Simel DL: The rational clinical examination: Is this patient hypovolemic? JAMA281 :1022– 1029,1999
18. Chow KM, Kwan BC, Szeto CC: Clinical studies of thiazide-induced hyponatremia. J Natl Med Assoc96 :1305– 1308,2004
19. Renneboog B, Musch W, Vandemergel X, Manto MU, Decaux G: Mild chronic hyponatremia is associated with falls, unsteadiness, and attention deficits. Am J Med119 :71.e1– 71.e8,2006
20. Ayus JC, Arieff AI: Chronic hyponatremic encephalopathy in postmenopausal women: Association of therapies with morbidity and mortality. JAMA281 :2299– 2304,1999