Diagnostic Value of Urine Sodium Concentration in Hyponatremia Due to Syndrome of Inappropriate Antidiuretic Hormone Secretion Versus Hypovolemia

Abstract

Background

We are often left with the differential diagnosis of syndrome of inappropriate antidiuretic hormone secretion (SIADH) versus hypovolemic hyponatremia. It is difficult to tell who will respond to isotonic saline infusion and who will not, if the urine sodium value is not completely suppressed (>10 mEq/L).

Aim

To examine the diagnostic accuracy of the urine sodium value.

Design

A retrospective observation.

Methods

The diagnostic accuracy of the urine sodium value was compared to that of a complete work-up and hospital course, including a response to saline infusion in patients with a final diagnosis of SIADH or hypovolemic hyponatremia. We also examined the diagnostic value of urine sodium-to-BUN ratio which should improve separation between SIADH and hypovolemia since the urine sodium and BUN move in opposite directions in these two conditions.

Results

The urine sodium value of 50 mEq/L was the most accurate in separating SIADH from hypovolemic hyponatremia: sensitivity 0.89, specificity 0.69, and accuracy 0.82. The diagnostic utility for SIADH versus hypovolemia, as quantified by the areas under the ROC curves, was not statistically different between urine sodium alone (0.89, 95% CI 0.77–0.96) and urine sodium-to-BUN ratio (0.93, 95% CI 0.83–0.98); p-value 0.33.

Conclusions

When the underlying cause is inconclusive between SIADH and hypovolemia, and when only basic laboratory results are available at the time of initial evaluation, the urine sodium alone will be adequate to guide initial fluid management. In contrast to traditional teaching, elevated urine sodium levels up to 50 mEq/L demonstrated clinically meaningful responses to isotonic saline infusion.

Introduction

A small amount of literature suggests that the use of complex indices would be helpful to distinguish SIADH patients from others (e.g., combined use of fractional excretion of sodium and fractional excretion of urea).^1–4 However, in the real world, most physicians only order the urine sodium and then start intravenous fluids to correct hyponatremia once patients are noted to be hyponatremic. However, it is often difficult to tell who will respond to isotonic saline infusion and who will not.^5–7 While it is relatively easy to exclude hypervolemic hyponatremia due to renal disease, heart failure, or liver disease just by history and physical examination, generally we are left with the differential diagnosis of SIADH versus hypovolemic hyponatremia.^8,9 Urine sodium alone is likely enough to guide our management in patients who have extreme values (urine sodium concentration below 10 mEq/L, suggesting a good response to isotonic saline). But how about a urine sodium level of 40 mEq/L in the face of diuretic use?

The intention of this study was to examine whether the urine sodium can still be used to guide initial diagnosis and subsequent fluid management if the value is not completely suppressed (i.e. >10 mEq/L). Further, we investigated the usefulness of urine sodium-to-BUN ratio as a discriminative index between SIADH and hypovolemic hyponatremia. Since the urine sodium and BUN move in opposite directions (higher urine sodium and lower BUN levels tend to occur in SIADH as opposed to lower urine sodium and higher BUN in hypovolemia), the ratio should magnify the differences between these two conditions and improve separation of SIADH from hypovolemia.^10,11

Methods

The diagnostic accuracy of urine sodium value was compared to that of a complete work-up and hospital course. The final discharge diagnosis, which was based on all available clinical, laboratory, and imaging findings, and the overall hospital course, was made by consultant nephrologists. The study was a retrospective chart review conducted at a university-affiliated hospital. The study was approved by the local investigational review board. Cases were limited to ones in which nephrologists were involved in consultation to maintain uniformity and accuracy of the diagnosis of SIADH versus hypovolemia.

The authors identified a total of 78 cases of hyponatremia that were evaluated by independent nephrologists in consultation over a 5-year period (2005–2009). Forty five cases met the predefined inclusion and exclusion criteria. Inclusion criteria were as follows: 1) Sodium concentration below 130 mEq/L at the time of nephrology consultation. 2) Serum osmolality less than 280 mOsm/kg H₂O. 3) A patient with a final diagnosis of either SIADH or extracellular volume depletion documented by consultant nephrologists based on the entire hospital course. 4) A patient who received an isotonic saline infusion. Hyponatremia due to causes other than SIADH and hypovolemia were excluded from the study (adrenal, thyroid, or pituitary insufficiency, chronic kidney disease, nephrotic syndrome, cirrhosis, congestive heart failure, polydipsia, or low solute “beer potomania,” etc). Since the primary interest of this study lay in cases where SIADH or hypovolemia could not easily be separated, obvious cases were excluded: urine sodium <10 mEq/L with consistent clinical findings. Those who did not have the urine sodium measured in a timely manner were excluded. In the absence of a single gold standard diagnostic test, the response to an infusion of isotonic saline and the subsequent hospital course yield invaluable information, which often unmasks the cause of the hyponatremia. The threshold for a “good response” to an isotonic saline infusion was not pre-specified because this was a retrospective study. Consultant nephrologists were aware of the results of the urine sodium and BUN values. However, they were not aware that there would be a retrospective study in using urine sodium and BUN values at that time. The rate of initial isotonic saline infusion varied anywhere from 75 ml/h to 1 L of bolus. Some of the urine and serum laboratory values were obtained during saline infusion rather than pre-infusion, but none of the values were obtained after correction of hyponatremia.

Statistical Analysis

Continuous variables are presented as means (±SD), and categorical variables as numbers and percentages. Continuous variables were compared with the use of the Welch two sample t-test, and categorical variables with the use of Pearson's Chi-squared test and Fisher's exact test. Logistic regression analysis was used to adjust confounders and to examine correlations among urine sodium, BUN, renal function, diuretic use, and other pertinent parameters. Receiver-operating-characteristic (ROC) curves were constructed to assess the sensitivity and specificity of urine sodium value and urine sodium-to-BUN ratio, and ability to diagnose SIADH and hypovolemic hyponatremia at the time of hospitalization. The comparison of areas under the ROC curves (AUC) was performed as a nonparametric approach. Statistical analyses were performed using R software 2.9.0.

Results

A total of 45 cases met the predefined criteria for inclusion. Twenty nine cases were assigned the final diagnosis of SIADH, and 16 cases were hypovolemic hyponatremia. Baseline characteristics are shown in Table 1. At the start of hospitalization, 24 out of 45 patients (53%) were thought to have hypovolemic hyponatremia. However, based on the hospital course, including an inadequate response to saline infusion, and additional workup, the final diagnosis of hypovolemic hyponatremia at the time of discharge was reduced to 16 cases out of 45 (36%).

Table 1.

Characteristics of Patients with SIADH and Hypovolemia

	SIADH	Hypovolemia	P-value
	N = 29	N = 16
Age	81.0 ± 9.2	79.3 ± 8.1	0.514
Gender - male sex - no. (%)	12 (41)	6 (37.5)	1.0
Chief complaints
Altered mental status - no. (%)	8 (27.6)	2 (12.5)	0.292
Fall - no. (%)	8 (27.6)	9 (56.3)	0.115
Bone fracture - no. (%)	5 (17.2)	2 (12.5)	1.0
Acute pain - no. (%)	14 (48.3)	5 (31.3)	0.429
Intracranial event - no. (%)	2 (6.9)	0 (0)	0.531
Vomiting - no. (%)	5 (17.2)	3 (18.8)	1.0
Underlying condition
Pulmonary disease (non malignant) - no. (%)	10 (34.5)	1 (6.2)	0.067
Neoplasm - no. (%)	6 (20.7)	2 (12.5)	0.692
Use of antipsychotics antidepressants - no. (%)	8 (27.6)	2 (12.5)	0.292
Diuretic use - no. (%)	5 (17.2)	10 (62.5)	0.006
Hydrochlorothiazide	4	6
Furosemide	2	4
Spironolactone	0	3
Biochemical data
Serum creatinine - mg/dL	0.78 ± 0.23	1.06 ± 0.28	0.001
Estimated GFR (MDRDb) - ml/min	94.1 ± 31.4	66.8 ± 25.0	0.003
BUN - mg/dL	13.2 ± 6.3	24.1 ± 12.3	0.003
			0.061a
Serum Na (0h) - mEq/L	120.8 ± 5.3	123.3 ± 5.7	0.153
Serum Na (24h) - mEq/L	123.8 ± 5.7	131.2 ± 3.6	<0.001
Serum Na (48h) - mEq/L	126.9 ± 4.5	135.0 ± 2.1	<0.001
Initial Urine Na - mEq/L	85.1 ± 31.4	40.3 ± 23.0	<0.001 0.002a
Urine Na/BUN ratio	8.6 ± 6.7	2.0 ± 1.3	<0.001 0.009a
Urine osmolarity - mOsm/kg	424.8 ± 158.4	307.1 ± 192.6	0.053

Note: Plus-minus values are means ± SD

^aLogistic regression model, adjusted for glomerular filtration rate and diuretic use

^bGlomerular Filtration Rate, Four-variable Modification of Diet in Renal Disease (MDRD)

Conversion factors for units: BUN in mg/dL to mmol/L, x0.357; serum creatinine in mg/dL to mol/L, x88.4

Baseline characteristics were well matched. The majority of the patients were elderly Asians. Falls and loss of balance with subsequent trauma were frequent causes for seeking medical attention. Unspecified weakness was also a common chief complaint. There was no statistical difference in the presence of neoplasm, pulmonary disease, and use of antipsychotics and antidepressants, but they trended toward a diagnosis of SIADH. The most common neoplasm was acute myeloid leukemia and lymphoma (3 cases), followed by pancreatic carcinoma (2 cases), and non-small cell lung cancer (1 case). The breakdown of pulmonary disease consisted of pneumonia (7 cases), exacerbation of chronic obstructive pulmonary disease or interstitial lung disease (4 cases), and adult respiratory distress syndrome (2 cases). Often the underlying etiology of vomiting was not determined. There was a statistical difference in the greater use of diuretics with hypovolemia as compared with SIADH. Furosemide was used in 6 cases, hydrochlorothiazide in 10 cases, and spironolactone in 3 cases. Some patients were on multiple diuretics. Diuretic use had little effect on urine sodium concentration (Table 2). There was a difference in initial serum creatinine values and estimated glomerular filtration rates with MDRD (Modification of Diet in Renal Disease). However, serum creatinine levels improved with volume repletion in the hypovolemic group, such that, there was no statistical difference in renal function between the two groups on hospital day 2. Serum sodium levels improved faster in hypovolemic hyponatremia (Figure 1).

Table 2.

Urine Sodium Values in Patients with SIADH and Hypovolemia in Relation to Diuretic Use

Urine Na value	Diuretic use = Yes	Diuretic use = No	p-value
SIADH - mEq/L	78.6 (n=5)	86.5 (n=24)	0.503 a
Hypovolemia - mEq/L	44.2 (n=10)	33.8 (n=6)	0.367 a

Figure 1.

Effects of Isotonic Saline Infusion on Serum Na

The urine sodium value at presentation was significantly higher in patients with SIADH. The BUN value alone was not robust enough after adjusting for the use of diuretics and glomerular filtration rates (p-value 0.061). Urine sodium-to-BUN ratio was significantly higher in patients in whom SIADH was the final diagnosis than in patients in whom the final diagnosis was hypovolemic hyponatremia (Table 1 and Figure 2). However, the diagnostic utility for SIADH versus hypovolemia, as quantified by the AUC, was not statistically different between urine sodium alone (AUC 0.89, 95% CI 0.77–0.96) and urine sodium-to-BUN ratio (AUC 0.93, 95% CI 0.83–0.98) (Figure 3). The urine sodium value of 50 mEq/L conferred the best accuracy in separating SIADH from hypovolemic hyponatremia: sensitivity 0.89, specificity 0.69, and accuracy 0.82.

Figure 2.

The Initial Values of Urine Na, BUN, and Urine Na/BUN Ratio in Patients with Final Diagnosis of Hypovolemia and SIADH

Figure 3.

The Diagnostic Utility of Urine Na and Urine Na/BUN to Differentiate Between Hypovolemia and SIADH

Discussion

The primary purpose of our study was to compare the diagnostic accuracy of urine sodium value to that of a complete work-up and hospital course including a response to isotonic saline infusion. We also examined the diagnostic value of the urine sodium-to-BUN ratio which should magnify the differences between SIADH and hypovolemia.^{10, 11} The urine sodium value of 50 mEq/L conferred the best accuracy in separating SIADH from hypovolemic hyponatremia. The study demonstrated that the urine sodium-to-BUN ratio was not particularly useful to distinguish SIADH from hypovolemic hyponatremia when compared with the urine sodium alone as quantified by the AUC.

The therapeutic response to isotonic saline is the best available tool to distinguish SIADH from hypovolemic hyponatremia.^{1–3 , 11–14} However, the distinction should not be based solely on the response to isotonic saline. The clinical scenario in a patient with hyponatremia is often complex, and there may be a component of both SIADH and hypovolemia.¹⁰ Without various factors taken into account, it is impossible to arrive at a proper diagnosis. It is not uncommon to see some improvement in the serum sodium concentration with isotonic saline infusion in the setting of SIADH.^1,15 Overall, the magnitude of the response to saline infusion was clearly higher in patients with hypovolemia as opposed to patients with diagnosis of SIADH. We did not incorporate patients with urine sodiums below 10 mEq/L. In the presence of extremely low urine sodiums and a low to normal volume status, there is no doubt in the usefulness of saline infusion. Since we wanted to know the value of saline infusion in cases in which the urine sodium was not completely suppressed, we omitted those who had extremely low urine sodium values. In actuality, only 3 cases were removed from analysis because of undetectable urine sodium levels. This may be a reflection of selection bias in that simple, obvious cases of hypovolemic hyponatremia supported by very low urine sodiums did not come to the attention of nephrologists.

In patients with urine sodiums below 50 mEq/L, the mean increase in serum sodium within a 48 hour period was 12.1 mEq/L. With urine sodiums above 50 mEq/L, the mean increase was 5.3 mEq/L. The results corresponded well to the magnitude of increase in serum sodium level based on the final diagnosis: 11.7 mEq/L of increase in patients with hypovolemia, 6.1 mEq/L of increase in patients with SIADH. For all practical purposes, the urine sodium can be used as a good indicator as to what to expect with an initial saline infusion. In other words, the diagnostic accuracy of 82% (urine sodium at 50 mEq/L) is clinically satisfactory from the standpoint of initial fluid management. Complicated diagnostic tools such as urine sodium-to-BUN, fractional excretion of sodium, uric acid, and urea will have little additional impact clinically.^1–4 Traditionally, a urine sodium of 20 or 30 mEq/L is used as a cutoff value to differentiate whether a patient would respond to isotonic saline infusion or not.^5,10,16 In our study, the urine sodium at 20 mEq/L provided a very low specificity and failed to distinguish SIADH and hypovolemia (Figure 3). This difference may be due to the average age of the study population, which at 80 years, may have slow adaptation to hypovolemia with reduction of urinary sodium.^14,17,18

Finally, there are several limitations in our study. It is a retrospective study, and we were unable to control several aspects. All of the patients received isotonic saline; however, the rate of IV fluid varied from 75 ml/L to 1 liter of bolus. Most received an isotonic saline infusion at a rate of 100 to 150 ml/hour initially. The timing of the nephrology consultation varied, and measurement of relevant laboratory values was not exactly every 24 hours. Another limitation of the study is that these patients were all selected cases seen by nephrology specialists. These cases were probably more complicated and may not be generalizable to ones seen by emergency department physicians or general internists. Consultant nephrologists were aware of the results of the urine sodium and BUN values, and one might question the validity of this study if they heavily relied on these values in making the diagnosis. The nephrologists were not aware that there would be a retrospective study in using the urine sodium and BUN values at that time. Also, the fact that many patients were categorized into hypovolemia despite a high urine sodium level (in a traditional perspective) speaks against this concern.

Conclusion

The ability to distinguish SIADH and hypovolemia using the urine sodium alone was reasonably high (the accuracy with a cutoff value of 50 mEq/L was 82%). In contrast to traditional teaching, elevated urine sodium levels up to 50 mEq/L demonstrated clinically meaningful responses to isotonic saline infusion. When the underlying cause is inconclusive between SIADH and hypovolemia, and when only basic laboratory results are available at the time of initial evaluation, the urine sodium alone will be adequate to guide initial fluid management.