Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Key Words differences. Results: Among the patients with acute first-ev-
Hyponatremia Mortality Ischemic stroke First-ever er ischemic stroke, 107 (11.6%) were hyponatremic. Among
ischemic stroke stroke risk factors, the prevalence of diabetes mellitus was
significantly higher among hyponatremic patients (p !
0.001). Prevalence of chronic renal insufficiency was also
Abstract higher in the hyponatremic group (p = 0.002). Clinical pre-
Background: Hyponatremia is the most common electro- sentations, such as the length of acute ward stay, initial im-
lyte disorder in hospitalized patients, and is frequently a paired consciousness, and clinical course in acute stroke
marker of a significant underlying disease. The prognostic were similar among normo- and hyponatremic patients.
value of hyponatremia in patients with acute first-ever isch- Among the complications, pneumonia and urinary tract in-
emic stroke is not known. We aimed to analyze whether hy- fection were significantly higher in hyponatremic than in
ponatremia in the acute stroke stage contributed to the risk normonatremic patients. After multivariate logistic regres-
of mortality or recurrent stroke in these patients. Methods: sion analysis, diabetes mellitus and chronic renal insufficien-
We studied 925 patients presenting with acute first-ever cy were associated with hyponatremia in these patients.
ischemic stroke between 2002 and 2004. Sodium levels were Kaplan-Meier analysis indicated that the survival rate was
obtained on arrival at the emergency room within 3 days of significantly lower in hyponatremic patients than in nor-
acute stroke onset. Hyponatremia was defined as a serum monatremic patients (log rank test; p value !0.001). After
sodium concentration of 134 mmol/l or less. Clinical presen- multivariate Cox proportional hazards model analysis, hypo-
tation, stroke risk factors, associated medical disease, and natremia was a significant predictor of 3-year mortality in
outcome were recorded. All patients were followed for 3 these patients after adjustment for related variables (p val-
years for survival analysis. A multivariate Cox proportional ue = 0.003, hazard ratio = 2.23, 95% confidence interval:
hazards model was used to identify risk factors for 3-year 1.303.82). Conclusion: Hyponatremia in the acute stroke
mortality in these patients. We also constructed Kaplan-Mei- stage is a predictor of 3-year mortality in patients with acute
er survival curves, and compared groups with hyponatremia first-ever ischemic stroke that is independent of other clini-
and normonatremia by means of log rank tests for significant cal predictors of adverse outcome.
Copyright 2012 S. Karger AG, Basel
Hung /Chen
Katz formula, which increases sodium by 1.6 mmol/l for every risk ratio and 95% confidence intervals (CI) were measured. To
5.55 mmol/l (100 mg/dl) increase in glucose concentration above assess the relationship between hyponatremia and mortality,
5.55 mmol/l (100 mg/dl), and patients with pseudohyponatremia Kaplan-Meier curves were compared using the log rank test. The
were excluded [13]. Patients with extreme hypernatremia (serum Cox proportional hazards model was used to evaluate all variables
sodium concentration 1150 mmol/l) were also excluded from the and determine the significance of variables for predicting the all-
study. Hyponatremia was defined as a serum sodium concentra- cause 3-year mortality. To determine the risk of death, hazard
tion of 134 mmol/l or less and normonatremia was a sodium con- ratio (HR) and 95% CI were obtained using the Cox proportional
centration of greater than 134 mmol/l, on the basis of previously hazards model. A univariate Cox model that assessed all previ-
published reports [5, 6, 14]. ously identified important variables was used to calculate the HR
for mortality. A backward stepwise multivariate Cox regression
Definition and Clinical Subtypes of Ischemic Stroke model was also used to identify the risk factors for 3-year mortal-
Clinical subtypes of ischemic stroke were rated according to ity in these patients. All statistical calculations were performed
the classification of the Oxfordshire Community Stroke Project with SPSS for Windows (SAS Institute, Cary, N.C., USA).
as partial anterior circulation syndrome, total anterior circulation
syndrome (TACS), posterior circulation syndrome, and lacunar
syndrome (LACS) [15]. TIA was defined by the new tissue-based
definition endorsed by the American Heart Association/Ameri- Results
can Stroke Association [16]. TIA mimics such as epileptic sei-
zures, complicated migraine, psychogenic hyperventilation, or Patient Characteristics
transient global amnesia were excluded from the study [17]. The Among 949 patients, 22 were excluded due to a final
clinical course in the acute stage of stroke, mean length of acute
ward stay, in-acute ward mortality, and frequency of medical diagnosis of TIA but not acute first-ever ischemic stroke,
complications were monitored. Clinical functional outcome upon and 2 patients were excluded due to extreme hypernatre-
discharge was assessed according to the modified Rankin Scale mia with unknown cause (sodium levels, 152 and 159
(mRS) [18]. Functionally dependent was defined as having an mmol/l, respectively). A total of 925 patients (486 men,
mRS score of 3, 4 or 5. 439 women) were analyzed (table1). The clinical proper-
Laboratory Measurements ties, including age, gender and comorbidity data, are list-
All laboratory values, including blood cell counts, and bio- ed in table1. Mean patient age was 69.48 8 11.62 years,
chemical data were measured by automated and standardized and 107 patients (11.6%) were hyponatremic (Na+ ^134
methods. All blood samples from patients were obtained on ad- mmol/l). There was no difference in age or gender be-
mission, centrifuged, and stored at 70 C until use in assays. Se-
tween the hypo- and normonatremic groups. Stroke risk
rum albumin, Cr, cholesterol, triglyceride, white blood cell count,
and hemoglobin levels were assayed and recorded. All other factors and their distributions in the 2 groups are also
markers were analyzed by standard automated laboratory meth- listed in table1. The prevalence of DM and CRI was sig-
ods. nificantly higher in hyponatremic patients than in nor-
monatremic patients. Differences in the prevalence of
Follow-Up other stroke risk factors and clinical presentations were
Patients were followed up for 3 years after initial assessment.
Follow-up consisted of clinical examinations at 1 and 3 months not statistically significant between normo- and hypona-
after first stroke and then every 3 months. End points of this study tremic patients.
included recurrent ischemic stroke or death. Recurrent stroke was
defined as any new focal neurological deficit of sudden onset last- Determinants of Hyponatremia in Patients with
ing at least 24 h for which no cause could be found other than Acute, First-Ever Ischemic Stroke
ischemic stroke. A diagnosis of recurrence was not made where
symptoms could be attributed to edema, mass effect, brain shift Univariate logistic regression analysis revealed that
syndrome, or hemorrhagic transformation, and could not be di- DM, CRI, and CAD were positively associated with hy-
agnosed within 24 h of the index stroke. Each death occurring ponatremia in patients with acute, first-ever ischemic
during the follow-up period was reviewed. stroke. After adjusting for these potential variables (p !
0.1) in the forward stepwise multivariate logistic regres-
Statistical Analysis
Unless otherwise stated, continuous variables are expressed as sion analysis, the presence of DM and CRI were positive-
mean 8 standard deviation (SD), and categorical variables are ly associated with hyponatremia (table2).
expressed as number or percentage of each parameter. Compari-
sons between the 2 patient groups were analyzed by 2 or Students Clinical Course in Patients with Acute Stage of
t test. The relative risks of independent associations between hy- First-Ever Ischemic Stroke
ponatremia and variables were analyzed by logistic regression. A
variable with p ! 0.1 in univariate logistic regression was consid- The length of acute ward stay, whether the patient
ered to be associated with hyponatremia, and was entered into experienced initial impaired consciousness, and clinical
backward stepwise multivariate logistic regression analysis. The course in the acute stroke stage were not significantly
* p < 0.05, univariate logistic regression; p < 0.05, multivariate logistic regression.
different between the normo- and hyponatremic groups. of discharge, in-hospital mortality rate, and rate of
Among complications, pneumonia and urinary tract in- stroke recurrence were not significantly different be-
fection were higher in hyponatremic patients than in tween the normo- and hyponatremic groups. However,
normonatremic patients (p value = 0.004 and 0.034, re- the death rate within 3 years of stoke onset was signifi-
spectively). In addition, the functional status at the time cantly higher in hyponatremic patients (p value !0.001).
Hung /Chen
Table 3. Clinical course in the acute stage of stroke, mortality, and stroke recurrence within 3 years of first-ever ischemic stroke onset
grouped according to sodium level
Mean length of acute-ward stay, days 14.7813.2 14.4810.2 1.32 (2.89 to 2.29) 0.818
Initial impaired consciousness 107 (13.1%) 20 (18.7%) 1.53 (0.902.59) 0.079
Course in acute stage of stroke
In evolution 201 (24.6%) 31 (29.0%) 1.25 (0.801.96) 0.191
Stationary 370 (45.2%) 52 (48.6%) 1.15 (0.771.71) 0.289
Improving 247 (30.2%) 24 (22.4%) 0.67 (0.411.08) 0.059
Complication
Pneumonia 79 (9.7%) 20 (18.7%) 2.15 (1.263.69) 0.004*
GI bleeding 66 (8.1%) 9 (8.4%) 1.05 (0.512.17) 0.510
UTI 96 (11.7%) 20 (18.7%) 1.73 (1.022.94) 0.034*
mRS score 3 at discharge 523 (63.9%) 73 (68.2%) 1.21 (0.791.86) 0.224
Range of follow-up duration, days 31,095 51,095
Median of follow-up time, days 1,095 1,095
In-hospital mortality 39 (4.8%) 6 (5.6%) 1.32 (0.543.19) 0.341
Stroke recurrence 129 (15.8%) 22 (20.6%) 1.38 (0.832.29) 0.132
Death 62 (7.6%) 19 (17.8%) 2.63 (1.514.61) <0.001*
Data are presented as mean 8 standard deviation or number (%). GI = Gastrointestinal; UTI = urinary tract infection. * p < 0.05,
2 or Students t test.
0.6
tremic group and 19 patients (17.8%) from the hypona-
tremic group. Of the 19 hyponatremic patients with first-
ever ischemic stroke who died during the 3-year observa-
0.4 Na+ >134 mmol/l tion period, 7 (36.8%) died from infection, 6 (31.6%) from
Na+ 134 mmol/l
cardiovascular disease, 2 (10.5%) from recurrent stroke,
0.2 1 (5.3%) within the acute stage of stroke, 2 (10.5%) from
cancer and 1 (5.3%) from uremia. Kaplan-Meier survival
0
analysis indicated that the hyponatremic group had a
0 200 400 600 800 1,000 1,200 higher mortality rate than the normonatremic group (log
Time (days) rank test: p ! 0.001; fig.1).
Cases of short-term mortality with a follow-up time of
1 or 3 months were sub-analyzed, and no significant dif-
Fig. 1. Kaplan-Meier analysis of patient survival (all-cause mor-
tality) during the 3-year study. Log rank test: p ! 0.001.
ferences were seen between normonatremic and hypona-
tremic groups. At the 1-month follow-up, the mortality
rates were 5 patients (4.7%) in the hyponatremic group
and 35 patients (4.3%) in the normonatremic group (log
During the 3-year follow-up time, 4 (3.7%) patients were rank test: p = 0.680). At the 3-month follow-up, the mor-
lost to follow-up in the hyponatremic group, and 23 tality rates were 8 patients (7.5%) in the hyponatremic
(2.8%) patients were lost to follow-up in the normona- group and 60 patients (7.3%) in the normonatremic group
tremic group. The above findings are summarized in (log rank test: p = 0.735). Among the patients who sur-
table3. vived at the 1- or 3-month follow-ups and returned for
* p < 0.05, univariate Cox regression. p < 0.05, multivariate Cox regression.
the 3-year follow-up, the 3-year mortality rate was sig- risk factor for all-cause 3-year mortality in patients who
nificantly higher in hyponatremic patients (log rank test: survived at 1 and 3 months and returned for the 3-year
p ! 0.001 and p = 0.001, respectively). follow-up (HR, 2.86, 95% CI: 1.515.40, p = 0.001, and
HR, 2.80, 95% CI: 1.405.58, p = 0.003, respectively).
Cox Regression Multivariate Analysis for 3-Year
Mortality in Patients with Acute, First-Ever Stroke
Univariate Cox regression analysis indicated that age, Discussion
hyponatremia, history of CAD, history of atrial fibrilla-
tion, history of CHF, TACS, hyperlipidemia and CRI The current study demonstrated a novel association
were variables (p ! 0.05) likely to be associated with mor- between hyponatremia and increased 3-year mortality
tality in patients with acute, first-ever ischemic stroke. in patients with acute, first-ever ischemic stroke, even
These variables were entered into the multivariate Cox after adjustment for established clinical predictors of ad-
proportional hazards model (table4). A backward step- verse outcome, including age and CAD, and greater
wise multivariate Cox proportional hazards model dem- stroke severity. Furthermore, hyponatremia was not as-
onstrated that hyponatremia was a significant risk factor sociated with short-term mortality (either in-hospital,
for all-cause 3-year mortality in these patients after ad- 1-month or 3-month mortality), but was correlated with
justment for related variables (HR: 2.23; 95% CI: 1.30 long-term mortality in patients with acute, first-ever
3.82; p = 0.003; table4). Univariate Cox regression analy- ischemic stroke. Our results also indicated that hypona-
sis was also performed to sub-analyze short-term (1- and tremia in the acute phase of first-ever ischemic stroke
3-month) mortality; no significant differences were seen was more common in the patients with DM or CRI. In
between normonatremic and hyponatremic groups (HR, addition, pneumonia was more frequently a complica-
1.22, 95% CI: 0.483.11, p = 0.681, and HR, 1.14, 95% CI: tion in hyponatremic patients than in normonatremic
0.542.37, p = 0.736, respectively). Among patients who patients.
survived the 1- and 3-month follow-ups, the mortality Determinants of hyponatremia in patients with acute,
rate at the 3-year follow-up was higher in hyponatremic first-ever stroke were also evaluated. From univariate and
patients (HR, 3.15, 95% CI: 1.675.92, p ! 0.001, and HR, multivariate logistic regression analyses, the presence of
3.10, 95% CI: 1.566.16, p = 0.001, respectively). After DM and CRI (Cr 6132.6 mol/l or 1.5 mg/dl) were sig-
multivariate Cox regression analysis and adjustment for nificant determinants of hyponatremia in first-ever isch-
related variables, hyponatremia remained a significant emic stroke patients. DM was reported to be associated
Hung /Chen
with hyponatremia in previous studies [2, 5, 6]. Renal in- This study has several limitations. First, we did not
sufficiency was also noted to be associated with hypona- survey the etiology of hyponatremia in all patients,
tremia in patients with pulmonary arterial hypertension although most of the hyponatremic patients were eu-
[7] and acute coronary syndrome [6]. volemic. The syndrome of inappropriate antidiuretic
The exact cause for the higher long-term mortality hormone secretion (SIADH) and central salt wasting
rate observed in patients with hyponatremia associated syndrome are common causes of hyponatremia in pa-
with acute first-ever ischemic stroke remains to be deter- tients with neurologic disease. However, in a previous
mined and requires further investigation. Only one pre- Japanese study, the incidence of SIADH in patients with
vious study has shown that all-cause and non-cardiovas- cerebral infarction was only 2.2% [22], and we did not
cular mortality were significantly increased at serum so- check urinary sodium concentration and urine osmolar-
dium levels ^138 mmol/l in stroke patients, but the study ity. Second, we did not collect information about serum
included all types of stroke (including hemorrhage) and sodium levels after discharge; therefore, we cannot clari-
was performed only in middle-aged male patients [10]. In fy whether the hyponatremia was transient or persistent
the present study, the reason for the association of hypo- in our patients. Third, we did not evaluate the stroke se-
natremia with long-term, but not short-term, mortality in verity by National Institutes of Health Stroke Scale score
acute first-ever ischemic stroke patient is unknown. Re- since a number of patients did not receive this scoring
cent hyponatremia treatment guidelines state the follow- while in admission. Instead, we used the classification of
ing: hyponatremia remains incompletely understood the OSCP to evaluate stroke severity; LACS was regarded
because of its association with a plethora of underlying as a less severe and TACS as a more severe stroke. We had
disease states, and its multiple etiologies with differing forced added TACS into the Cox regression analysis. Al-
pathophysiologic mechanisms [5, 19]. In cardiovascular though some limitations existed in our investigation, we
diseases, hyponatremia is frequently encountered in pa- clearly demonstrated that hyponatremia was an indepen-
tients with advanced heart failure and is an established dent factor associated with increased 3-year mortality in
indicator of heart failure progression and death; this re- acute, first-ever ischemic stroke patients.
lationship is probably due to the activation of the renin- In conclusion, this is the first study that demonstrated
angiotensin-aldosterone system [1, 2, 3, 6]. In patients that hyponatremia is an independent predictor of 3-year
with MI, the development of hyponatremia may reflect mortality in patients with acute, first-ever ischemic
neurohormonal activation, which affects left ventricular stroke. In addition, hyponatremia occurred more fre-
remodeling and leads to higher long-term risk for heart quently in patients with DM or CRI, and pneumonia was
failure and mortality [4, 7]. more frequently observed in hyponatremic patients than
Our study failed to show an association between hy- in normonatremic patients at the time of hospital admis-
ponatremia and a higher risk of recurrent stroke in pa- sion. Larger groups and further investigations are re-
tients with acute first-ever ischemic stroke. Only one pre- quired to confirm our clinical observations and to deter-
vious study has suggested that the risk of stroke rose mine whether correcting the blood sodium level during
significantly when the sodium concentration decreased the acute stroke stage could improve clinical outcomes in
below 144 mmol/l, but the study included patients with acute ischemic stroke patients. In addition, the mecha-
all types of stroke (including hemorrhage), and was per- nisms behind hyponatremia in acute ischemic stroke pa-
formed only in middle-aged male patients [10]. Further tients require further investigation.
study is warranted to clarify whether sodium level is as-
sociated with risk of recurrent stroke.
Hyponatremia frequently accompanies pulmonary Acknowledgements
diseases, both infectious and neoplastic [20]. Recent stud-
This work was supported by grants from the Chang-Gung Me-
ies have suggested that hyponatremia is highly prevalent
morial Hospital CMRPG 270331, CMRPG 270332 and CMRPG
in pneumonia patients and frequently accompanies lon- 270333.
ger hospitalization times [14, 21]. Therefore, our finding
that the incidence of pneumonia was higher in hypona-
tremic patients than in normonatremic patients was in
agreement with previous reports. However, the clinical
course in the acute stage of stroke or ward stay did not
differ between the 2 groups of patients.
Hung /Chen
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.