RESULTS

Section:

ChooseTop of pageAbstractMETHODSRESULTS <<DISCUSSIONReferencesCITING ARTICLES

Of the 17,733 patient discharges that met our inclusion criteria, we excluded 323 with only a secondary code indicative of PE (1.8%), 767 patient transfers from another hospital (4.3%), 265 subsequent transfers to another hospital (1.5%), 777 discharges without a match to key clinical findings (4.4%), 70 without a linkage to the National Death Index (0.4%), and 1,803 (10.2%) with an undocumented serum sodium at the time of presentation. The study cohort comprised 13,728 patient discharges with a diagnosis of PE from 185 Pennsylvania hospitals; 2,907 (21.1%) had hyponatremia (serum level ≤ 135 mmol/L). Of these, 2,598 (18.9%) had a sodium level 130 to 135 mmol/L and 309 (2.2%) had a sodium level less than 130 mmol/L. Compared with the 13,728 enrolled patients, the 1,803 excluded because of an undocumented serum sodium were significantly younger (median age, 64 versus 67 yr; P < 0.001), and were less likely to have a history of heart failure (10.7 versus 16.5%; P < 0.001) and chronic lung disease (14.7 versus 18.9%; P < 0.001). Compared with eligible patients with a documented serum sodium, these 1,803 excluded patients were also less likely to have a pulse of 110/minute or greater (11.1 versus 18.5%; P < 0.001), a systolic blood pressure less than 100 mm Hg (7.8 versus 10.8%; P < 0.001), a respiratory rate of 30/min or greater (8.3 versus 15.4%; P < 0.001), a body temperature of less than 36°C (14.4 versus 16.9%; P = 0.007), an altered mental status (5.9 versus 7.5%; P = 0.02), and an arterial oxygen saturation of less than 90% (3.9 versus 8.5%; P < 0.001) at presentation.

Association of Serum Sodium Level and 30-Day Mortality

Overall, 1,308 of 13,728 patients (9.5%) died at 30 days. As shown in Figure 1, patients with a serum sodium of greater than 135, 130–135, and less than 130 mmol/L had a cumulative 30-day mortality of 8.0, 13.6, and 28.5% (P < 0.001), respectively. Mortality in these three groups diverged 1 day after admission, and the difference continued to increase over the 30-day follow up. Mortality was also significantly increased among patients with lower serum sodium, stratified by each of the five PESI risk classes at presentation (Table 2).

Figure 1. Cumulative mortality by level of serum sodium. Kaplan-Meier estimates of 30-day mortality for patients with pulmonary embolism by baseline level of serum sodium. The cumulative mortality was 8.0, 13.6, and 28.5% for patients with a sodium level >135, 130–135, and <130 mmol/L (P < 0.001), respectively.

Download Figure | Download Powerpoint

[More] [Minimize]

TABLE 2. ASSOCIATION OF MORTALITY AND LEVEL OF SERUM SODIUM STRATIFIED BY SEVERITY OF ILLNESS

	Serum Sodium Level* (mmol/L)
PESI risk class	<130	130–135	>135	P-Value
I	0/12 (0.0)	10/334 (3.0)	22/2,185 (1.0)	0.02
II	7/45 (15.6)	28/463 (6.0)	62/2,386 (2.6)	<0.001
III	16/65 (24.6)	52/578 (9.0)	134/2,375 (5.6)	<0.001
IV	15/70 (21.4)	73/512 (14.3)	156/1,671 (9.3)	<0.001
V	50/117 (42.7)	190/711 (26.7)	493/2,204 (22.4)	<0.001

Definition of abbreviation: PESI = pulmonary embolism severity index.

^*The 30-day mortality, n/total n (%)

After adjustment for patient race, insurance, severity of illness by PESI risk class, administration of thrombolytic therapy, hospital region, and hospital size and teaching status (Table 3), the odds of 30-day mortality remained significantly increased for patients with a sodium level of 130–135 mmol/L (odds ratio [OR], 1.53; 95% confidence interval [CI], 1.33–1.76) and a sodium level less than 130 mmol/L (OR, 3.26; 95% CI, 2.48–4.29). When the PESI risk class was replaced in the model with the 11 variables comprising the PESI and laboratory parameters (serum glucose > 250 mg/dl, serum creatinine > 1.5 mg/dl, and serum troponin ≥ 0.1 ng/ml), or when the 5,951 patients with a history of cancer, chronic lung disease, or heart failure were excluded from analysis, the results remained similar.

TABLE 3. ASSOCIATION BETWEEN LEVEL OF SERUM SODIUM AND OUTCOMES

Outcomes	Adjusted Odds Ratio*	95% Confidence Interval	P Value
30-d all-cause mortality			<0.001
Serum sodium > 135 mmol/L	1.00	—
Serum sodium 130–135 mmol/L	1.53	1.33–1.76
Serum sodium < 130 mmol/L	3.26	2.48–4.29
30-d hospital readmission rate†			<0.001
Serum sodium > 135 mmol/L	1.00	—
Serum sodium 130–135 mmol/L	1.28	1.12–1.46
Serum sodium < 130 mmol/L	1.44	1.02–2.02

^*The models were adjusted for patient race, insurance type, severity of illness use of the pulmonary embolism severity index (including age, sex, history of cancer, chronic lung disease or heart failure, systolic arterial blood pressure <100 mm Hg, pulse ≥110/min., respiratory rate ≥30 breaths/min, body temperature <36°C, arterial oxygen saturation <90%, altered mental status), administration of thrombolysis, hospital region within Pennsylvania, and hospital size and teaching status. Adjusted odds ratios were estimated with random-intercept logistic regression, with the two levels defined by patient and hospital site.

^†The 30-day readmission was estimated after the additional exclusion of 840 patients who died in the hospital, 91 who were still hospitalized 30 days after admission, and 35 with unknown readmission status, leaving a sample of 12,762 patients.

Patient reclassifications for 30-day mortality are summarized in Table 4. For 38 of 1,308 patients (2.9%) who died within 30 days of admission, classification improved with serum sodium level in addition to the PESI. Conversely, 154 of 12,420 patients (1.2%) who were alive within 30 days of admission were reclassified up. The net reclassification improvement was estimated at 1.7% (P < 0.001), corresponding to a modest, but statistically significant, improvement. The integrated discrimination improvement was also statistically significant (0.009; P < 0.001).

TABLE 4. RECLASSIFICATION OF PATIENTS WHO DIED AND WHO WERE ALIVE BY 30 DAYS BASED ON THE PULMONARY EMBOLISM SEVERITY INDEX WITH AND WITHOUT SERUM SODIUM LEVEL

PESI without Serum Sodium Level Frequency	PESI with Serum Sodium Level*
		<2.3%	2.3–5.0%	5.1–8.8%	8.8–17.5%	>17.5%	Total
Patients who died by 30 d, %
<2.3	32	0	0	0	0	32
2.3–5.0	0	90	0	7	0	97
5.1–8.8	0	0	186	16	0	202
8.9–17.5	0	0	0	229	15	244
>17.5	0	0	0	0	733	733
Total	32	90	186	252	748	1,308
Patients who were alive by 30 d, %
<2.3	2,487	12	0	0	0	2,499
2.3–5.0	0	2,759	0	38	0	2,797
5.1–8.8	0	0	2,767	49	0	2,816
8.9–17.5	0	0	0	1,954	55	2,009
>17.5	0	0	0	0	2,299	2,299
Total	2,487	2,771	2,767	2,041	2,354	1,2420

Definition of abbreviations: PESI = pulmonary embolism severity index.

^*<130, 130–135, >135 mmol/L.

Association of Serum Sodium Level and 30-Day Readmission

The 30-day readmission rate was estimated in 12,762 patients, after the exclusion of 840 patients who died in the hospital, 91 who were still hospitalized at 30 days after admission, and 35 with unknown readmission status. Of these, 1,596 (12.5%) were readmitted within 30 days. As shown in Figure 2, patients with a sodium level greater than 135, 130–135, and less than 130 mmol/L had a cumulative 30-day readmission rate of 11.8, 15.6, and 19.3% (P < 0.001), respectively. After adjustment (Table 3), patients with a sodium level 130–135 mmol/L (OR, 1.28; 95% CI, 1.12–1.46) and a sodium level less than 130 mmol/L (OR, 1.44; 95% CI, 1.02–2.02) had significantly increased odds of 30-day readmission. When the PESI risk class was replaced in the model with the 11 variables comprising the PESI and laboratory parameters (serum glucose > 250 mg/dl, serum creatinine > 1.5 mg/dl, and serum troponin ≥ 0.1 ng/ml), or when the 5,951 patients with a history of cancer, chronic lung disease, or heart failure were excluded from analysis, the results were unchanged.

Figure 2. Cumulative readmission rate by level of serum sodium. Kaplan-Meier estimates of 30-day readmission rate for patients with pulmonary embolism by baseline level of serum sodium. The cumulative readmission rate was 11.8, 15.6, and 19.3% for patients with a sodium level >135, 130–135, and <130 mmol/L (P < 0.001), respectively.

Download Figure | Download Powerpoint

[More] [Minimize]

DISCUSSION

Section:

ChooseTop of pageAbstractMETHODSRESULTSDISCUSSION <<ReferencesCITING ARTICLES

Our results demonstrate that a substantial proportion of patients with PE (21.1%) are hyponatremic at the time of presentation. After adjusting for potential patient- and hospital-related confounders, and the administration of thrombolytic therapy, we found that patients with hyponatremia had a significantly higher 30-day mortality than patients without hyponatremia, and that the mortality rate increased with the severity of hyponatremia. The higher mortality among patients with hyponatremia was observed across all PESI risk classes. Similarly, we also observed a significant association between hyponatremia and 30-day hospital readmission. To our knowledge, our study represents the first to demonstrate the prognostic importance of hyponatremia in patients with PE.

Several explanations are possible for the association between hyponatremia and increased 30-day mortality in patients with PE. Hyponatremia is a well known marker of neurohormonal activation in patients with left ventricular heart failure, and reflects nonosmotic release of vasopressin consequent to activation of the sympathetic and renin–angiotensin–aldosterone system axis (6, 15). Neurohormonal activation has also been demonstrated in patients with pulmonary arterial hypertension, and occurs in proportion to the degree of right ventricular dysfunction (16, 17). Therefore, hyponatremia may indicate neurohormonal activation in the context of PE-related pulmonary arterial hypertension and subsequent right heart dysfunction, a known factor of adverse prognosis in patients with PE (10). Indeed, a prior prospective study of patients with chronic pulmonary arterial hypertension demonstrated a strong association between hyponatremia and the presence of right ventricular dysfunction and short-term survival (7). Future prospective studies should examine whether hyponatremia reflects neurohormonal activation and right ventricular dysfunction in patients with acute PE.

Another explanation for the observed association between hyponatremia and adverse outcomes may be the presence of unmeasured prognostic factors that are unrelated to right ventricular dysfunction. In a recent study of middle-aged and elderly community subjects without a history of heart failure, hyponatremia was an independent predictor of death, regardless of concomitant diuretic use, renal and hepatic disease, and hyperglycemia (18). In large, retrospective and prospective cohort studies, hyponatremia also significantly increased short-term mortality in unselected inpatients (19–22). Indeed, the relationship between hyponatremia and adverse outcomes is incompletely understood because of its association with a multitude of underlying disease states, and its multiple causes with different pathophysiologic mechanisms (23).

Our study also shows that, after adjustment for patient and hospital characteristics and the administration of thrombolytic therapy, hyponatremia in the acute phase of PE is an independent predictor for future hospital admissions, reinforcing the prognostic importance of a low serum sodium level. Our findings are consistent with the results from prior studies that found significantly higher hospital readmission rates in hyponatremic patients with acute myocardial infarction and heart failure (2, 12, 24).

Our findings have both clinical and research implications. Clinically, patients with PE who are hyponatremic at presentation carry a higher risk of short-term mortality and hospital readmission, and may, therefore, potentially benefit from more intensive surveillance in the hospital and after discharge. Further research is warranted to determine whether correction of hyponatremia (e.g., with vasopressin receptor antagonists or other treatments) is associated with improved outcomes for patients with PE. Evidence suggests that patients with left ventricular heart failure who have persistent hyponatremia have a significantly higher risk of short-term mortality or hospitalization for heart failure than patients in whom hyponatremia was corrected (24). Similarly, resolution of admission hyponatremia during hospitalization appears to attenuate the increased mortality risk conferred by hyponatremia among unselected inpatients (21).

Our work has several potential limitations. First, patients in the study sample were identified by use of International Classification of Diseases, 9th Clinical Modification codes for PE rather than standardized radiographic criteria, and patient eligibility may, therefore, be subject to study selection biases due to hospital coding procedures. In prior studies, up to 96% of patients with specific codes for PE had objectively documented disease on the basis of chart review criteria (25–27). Second, our sample excluded 10.2% of younger, healthier, and less severely ill patients in whom serum sodium was not measured at the time of admission. However, because hyponatremia adversely affects outcomes across the full spectrum of severity of illness (PESI risk classes I–V), the exclusion of these lower-risk patients is unlikely to change our study results. Third, because measures of right ventricular function and neurohormonal activation were not available in our database, we could not examine whether these factors are linked to the pathophysiology of hyponatremia. Moreover, we had no information about diuretic use and hypervolemic hyponatremic states, such as cirrhosis or nephrotic syndrome. Thus, we could not explore whether these conditions are associated with the higher observed mortality among patients with hyponatremia. Fourth, because we had no information on sodium levels after hospital discharge, the prognostic implications of transient versus persistent hyponatremia could not be analyzed. Finally, we can only detect associations, not causality, from these data. Thus, we cannot determine whether hyponatremia has a specific effect on the pathophysiology of PE, or if it is simply a marker of adverse outcome.

In conclusion, in this large sample of patients hospitalized with acute PE, hyponatremia at presentation was common, and was associated with a higher risk of 30-day mortality and readmission. Hyponatremia may serve as an easy-to-use marker to identify patients with PE who are at high risk of adverse medical outcomes. Future studies should examine whether hyponatremia reflects neurohormonal activation and right ventricular dysfunction in patients with acute PE, and whether correction of this electrolyte abnormality is associated with improved patient outcomes.