Hyponatremia in heart failure (HF) is an established predictor of adverse outcomes in hospitalized patients with reduced ejection fraction (EF). However, there is a paucity of data in ambulatory patients with HF with preserved ejection fraction (HFpEF). We examined the prevalence, risk factors, and long-term outcomes of hyponatremia (serum sodium ≤135 mEq/L) in ambulatory HFpEF and HF with reduced EF (HFrEF) in a national cohort of 8,862 veterans treated in Veterans Affairs clinics. Multivariable logistic regression models were used to identify factors associated with hyponatremia, and multivariable Cox proportional hazard models were used for analysis of outcomes. The cohort consisted of 6,185 patients with HFrEF and 2,704 patients with HFpEF with a 2-year follow-up. Hyponatremia was present in 13.8% and 12.9% patients in HFrEF and HFpEF, respectively. Hyponatremia was independently associated with younger age, diabetes, lower systolic blood pressure, anemia, body mass index <30 kg/m 2 , and spironolactone use, whereas African-American race and statins were inversely associated. In multivariate analysis, hyponatremia remained a significant predictor of all-cause mortality in both HFrEF (hazards ratio [HR] 1.26, 95% confidence interval [CI] 1.11 to 1.44, p <0.001) and HFpEF (HR 1.40, 95% CI 1.12 to 1.75, p = 0.004) and a significant predictor of all-cause hospitalization in patients with HFrEF (HR 1.18, 95% CI 1.07 to 1.31, p = 0.001) but not in HFpEF (HR 1.08, 95% CI 0.92 to 1.27, p = 0.33). In conclusion, hyponatremia is prevalent at a similar frequency of over 10% in ambulatory patients with HFpEF and HFrEF. Hyponatremia is an independent prognostic marker of mortality across the spectrum of patients with HFpEF and HFrEF. In contrast, it is an independent predictor for hospitalization in patients with HFrEF but not in patients with HFpEF.
Hyponatremia (conventionally defined as a serum sodium concentration ≤135 mEq/L) has been established as a predictor of adverse outcomes and has been incorporated in prognostic models of heart failure with reduced ejection fraction (HFrEF). However, there is a paucity of data about the prognostic significance of hyponatremia in ambulatory patients with HF with preserved ejection fraction (HFpEF). Recently, a meta-analysis reported hyponatremia to be a predictor of mortality in patients with HFrEF and HFpEF; however, the analysis included a mixture of ambulatory and hospitalized patients with HF. Therefore, in a large, national cohort of ambulatory patients with HF, we examined the prevalence and the impact of hyponatremia on morbidity and mortality in patients with HFpEF and HFrEF. We hypothesized that hyponatremia is an independent predictor of mortality and hospitalization in ambulatory patients with HFrEF and HFpEF.
Methods
We performed a retrospective study of a national cohort of veterans with HF treated in ambulatory clinics of US Department of Veterans Affairs (VA) medical centers from October 1, 2000, to September 30, 2002. We used the VA quality performance External Peer Review Program (EPRP) data. As described previously, the sampling pool of outpatients for EPRP included ambulatory patients with chronic diseases including HF, diabetes, previous myocardial infarction, and chronic obstructive pulmonary disease (COPD), identified by International Classification of Disease-Ninth Revision codes. Abstractors reviewed electronic medical records for validation of inclusion criteria, including documentation by clinicians of the diagnosis of HF and other chronic diseases listed previously. Patient-level data from the EPRP HF cohort were linked with 5 existing national VA databases to obtain demographic, co-morbidity, laboratory, pharmacy, and outcome data. The left ventricular EF and date of its ascertainment were obtained from the EPRP database. Patients were classified as having HFpEF when the EF was ≥50% and as having HFrEF when the EF was <50%. Hyponatremia was defined as serum sodium concentration ≤135 mEq/L. As described previously, blood pressure, weight, height, and co-morbidities of previous myocardial infarction, diabetes mellitus, hypertension, and COPD were obtained from the EPRP database. Other co-morbidities were ascertained using International Classification of Disease-Ninth Revision codes from VA outpatient clinic files (containing demographics, diagnoses, and outpatient services) and patient treatment files (containing abstracts for patients discharged from VA hospitals) over a period of 2 years before and at the index clinic visit. The most recent laboratory data, including serum sodium levels, within 1 year before to 2 weeks after the index visit were used. Renal insufficiency was defined as an estimated glomerular filtration rate <60 ml/min/1.73 m 2 , calculated using the 4-variable Modification of Diet in Renal Disease formula. Of patients with known EF (n = 17,456), 9,451 patients who had an EF determination within 1 year before to 3 months after the clinic visit were included. Of these patients, 580 patients with missing sodium values, 4 patients with missing systolic blood pressure, and 5 patients with missing outcomes data were excluded. Thus, our sample size consisted of 8,862 patients. Variables with >20% missing values were excluded. Missing values for hemoglobin (12.6%), creatinine (6.9%), and serum blood urea nitrogen (BUN; 2.4%) were imputed. Missing values were imputed using linear regression with baseline variables as predictors and constraints applied on the basis of observed minimal and maximal values. Analyses were repeated by excluding observations with imputed values, and the results were found to be concordant. Thus, models using imputed data are shown. Follow-up data were available for mortality and the first hospitalization. The outcomes of interest were time to all-cause mortality and time to all-cause hospitalization.
Univariate differences in baseline characteristics between patient groups with hyponatremia and nonhyponatremia with HFpEF or HFrEF were evaluated using chi-square tests for categorical variables and 2-sample t tests for continuous variables. All covariates were tested for normality by Shapiro-Wilk test. Log transformation was performed to normalize serum BUN levels. To assess the predictive role of hyponatremia, adjusted multivariable Cox proportional hazard models using fixed-effect models were generated. The covariates for multivariate models were selected on the basis of their clinical relevance and significance per previous studies. Associations were examined sequentially in models with incremental multivariable adjustments: model 1: adjusted for age, sex, and race; model 2: adjusted for model 1 + history of diabetes, hypertension, chronic renal insufficiency, peripheral vascular disease, cerebrovascular accident, atrial fibrillation, malignancy, dementia, COPD, liver disease, history of myocardial infarction, previous HF hospitalization, body mass index (BMI), hemoglobin, and serum BUN (log transformed); model 3: adjusted for model 2 + use of β blockers, angiotensin-converting enzyme inhibitors or angiotensin receptor blockers, spironolactone, and statins. Multivariate Cox proportional hazards models were run separately for the HFpEF and HFrEF groups to calculate the hazard ratios for hyponatremia for the end points of mortality and morbidity (all-cause hospitalization). Survival curves for all-cause mortality were generated using Kaplan-Meier survival estimates. All analyses were performed using SPSS version 19 (SPSS Inc., Chicago, Illinois). Data are presented as mean ± SD unless otherwise specified, and p values <0.05 were considered significant.
Results
The study cohort consisted of 8,862 veterans with HF, of whom 2,704 (30.5%) had HFpEF and 6,185 (69.5%) had HFrEF. Patients had a mean age of 70 years, and consistent with VA demographics, 95% were men. Mean serum sodium was 139 ± 3.5 mEq/L (median 139 mEq/L, range 114 to 158 mEq/L), and 1,195 patients (13.5%) were hyponatremic in the overall cohort. Hyponatremia was present in 847 (13.8%) and 348 (12.9%) patients in HFrEF and HFpEF groups, respectively (p = 0.26). Of note, in the overall cohort, only 178 patients (2%) had serum sodium concentrations >145 mEq/L. Of patients with HFrEF, 25% had mildly reduced EF (40% ≤EF <50%), 31% had moderately reduced EFs (30% ≤EF <40%), and 44% had severely reduced EFs (EF <30%). Baseline characteristics of patients with and without hyponatremia, within HFpEF and HFrEF groups, are listed in Table 1 . In patients with HFrEF, compared with patients with nonhyponatremia, those with hyponatremia were slightly younger and had higher prevalence of co-morbidities ( Table 1 ). In HFpEF, patients with hyponatremia did not differ significantly in age or several co-morbidities. However, compared with patients with nonhyponatremia, patients with hyponatremia with both HFpEF and HFrEF had lower systolic blood pressure, BMI, and serum hemoglobin; higher serum BUN and history of HF hospitalization within the previous 2 years, suggesting a more advanced HF status. On multivariable logistic regression that evaluated all the variables in Table 1 , in patients with HFpEF as well as HFrEF, hyponatremia was independently associated with younger age, diabetes, lower systolic blood pressure, anemia, BMI <30 kg/m 2 , and spironolactone use, whereas African-American race and statin use were inversely associated with hyponatremia. In addition, among patients with HFrEF, COPD, previous HF hospitalizations, and higher BUN were also associated with hyponatremia, whereas in HFpEF anemia was also associated with hyponatremia ( Table 2 ).
| Variable | HFrEF (n = 6,158) | HFpEF (n = 2,704) | ||||
|---|---|---|---|---|---|---|
| Hyponatremia (n = 847; %) | Nonhyponatremia (n = 5,311; %) | p Value | Hyponatremia (n = 348; %) | Nonhyponatremia (n = 2,356; %) | p Value | |
| Age (yrs) | 68.6 ± 10.7 | 69.6 ± 10.3 | 0.02 | 69.9 ± 10.1 | 70.9 ± 10.0 | 0.11 |
| Men | 96.9 | 96.4 | 0.44 | 91.1 | 90.9 | 0.92 |
| European American | 76.8 | 78.4 | 0.02 | 77.0 | 74.2 | 0.53 |
| African-American | 11.9 | 13.5 | 12.6 | 14.0 | ||
| Others/unknown | 11.2 | 8.1 | 10.3 | 11.8 | ||
| Hypertension | 60.1 | 63.1 | 0.09 | 68.7 | 71.2 | 0.33 |
| Diabetes mellitus | 48.4 | 39.5 | <0.001 | 48.3 | 44.9 | 0.23 |
| Peripheral arterial disease | 32.1 | 27.9 | 0.01 | 27.6 | 27.6 | 0.99 |
| Cerebrovascular accident | 24.3 | 21.3 | 0.05 | 21.6 | 21.4 | 0.95 |
| Atrial fibrillation | 38.8 | 35.2 | 0.04 | 37.6 | 34.7 | 0.29 |
| History of MI | 40.7 | 40.5 | 0.90 | 28.2 | 27.1 | 0.67 |
| Renal insufficiency | 53.8 | 51.8 | 0.27 | 47.1 | 50.3 | 0.26 |
| COPD | 31.1 | 26.2 | 0.003 | 37.1 | 33.6 | 0.21 |
| Moderate-to-severe liver disease | 0.8 | 0.6 | 0.36 | 1.1 | 0.6 | 0.28 |
| Malignancy without metastasis | 19.0 | 18.7 | 0.84 | 23.3 | 21.5 | 0.45 |
| BMI (kg/m 2 ) | 27.9 ± 5.8 | 28.7 ± 6.0 | <0.001 | 30.1 ± 7.4 | 31.5 ± 7.2 | <0.001 |
| Systolic blood pressure (mm Hg) | 118.7 ± 21.5 | 125.3 ± 20.7 | <0.001 | 128.3 ± 21.7 | 132.9 ± 20.9 | <0.001 |
| Serum hemoglobin (g/dl) | 13.2 ± 1.9 | 13.5 ± 1.8 | <0.001 | 12.8 ± 2.3 | 13.3 ± 1.9 | 0.001 |
| Serum BUN (mg/dl) | 29.9 ± 15.0 | 25.9 ± 13.9 | <0.001 | 27.2 ± 17.2 | 24.8 ± 13.3 | 0.01 |
| Serum creatinine (mg/dl) | 1.51 ± 0.81 | 1.45 ± 0.71 | 0.06 | 1.38 ± 0.69 | 1.40 ± 0.69 | 0.57 |
| HF hospitalization within previous 2 years | 33.9 | 22.3 | <0.001 | 21.3 | 17.0 | 0.05 |
| β blockers | 63.8 | 65.5 | 0.33 | 54.0 | 56.1 | 0.46 |
| ACEI/ARBs | 85.7 | 85.8 | 0.96 | 75.0 | 73.2 | 0.48 |
| Spirinolactone | 36.6 | 28.0 | <0.001 | 24.7 | 15.8 | <0.001 |
| Statin | 46.9 | 52.2 | 0.004 | 37.1 | 46.4 | 0.001 |
| Variables | Odds Ratios (95% CI) ∗ | p Value |
|---|---|---|
| HFrEF | ||
| Age, per 10-year increase | 0.84 (0.77–0.91) | <0.001 |
| African-American | 0.54 (0.42–0.70) | <0.001 |
| Diabetes mellitus | 1.47 (1.25–1.73) | <0.001 |
| COPD | 1.21 (1.03–1.43) | 0.02 |
| Previous HF hospitalization | 1.50 (1.27–1.77) | <0.001 |
| Serum blood urea nitrogen (per 1 mg/dl) | 1.32 (1.10–1.53) | 0.003 |
| Spironolactone use | 1.30 (1.11–1.53) | 0.001 |
| Systolic blood pressure (per 10 mm Hg increase) | 0.87 (0.84–0.91) | <0.001 |
| BMI <30 kg/m 2 | 1.35 (1.14–1.60) | 0.001 |
| Statin use | 0.79 (0.67–0.92) | 0.002 |
| HFpEF | ||
| Age, per 10-year increase | 0.85 (0.74–0.96) | 0.009 |
| African-American | 0.60 (0.39–0.94) | 0.02 |
| Diabetes mellitus | 1.29 (1.01–1.67) | 0.04 |
| Spironolactone use | 1.76 (1.33–2.33) | <0.001 |
| Anemia | 1.33 (1.04–1.71) | 0.02 |
| Systolic blood pressure (per 10 mm Hg increase) | 0.92 (0.87–0.97) | 0.004 |
| BMI <30 kg/m 2 | 1.62 (1.26–2.08) | <0.001 |
| Statin use | 0.67 (0.52–0.86) | 0.002 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
