Predictive factors for high mortality in hypernatremic patients

Predictive factors for high mortality in hypernatremic patients

Predictive Factors for High Mortality in Hypernatremic Patients ANIL K. MANDAL, MD, MOHAMMAD G. SAKLAYEN, MD, NOSRAT M. HILLMAN, MD, RONALD J. MARKERT...

382KB Sizes 0 Downloads 34 Views

Predictive Factors for High Mortality in Hypernatremic Patients ANIL K. MANDAL, MD, MOHAMMAD G. SAKLAYEN, MD, NOSRAT M. HILLMAN, MD, RONALD J. MARKERT, PHD Hypernatremia (serum sodium level of >145 mEq/L) is associated with high mortality. This study reports an analysis of mortality in 116 patients with hypernatremia from two large university-affiliated teaching hospi. tals. The purpose was to identify factors predictive of high mortality in hypernatremic patients. Medical records were reviewed to obtain the following data: serum sodium (Na +) levels; systolic (S) and diastolic (D) blood pressure (BP) at the time of admission and throughout the hospital course; status of cognitive function; and type of fluid administered. The patients were divided into two groups: expired and survived. Seventyseven of 116 patients (66%)expired, while 39 patients (34%) survived and were discharged from the hospital. The mean age and gender for patients who died (70.9 +_15.4years, 90% men) were not different from those who survived (66.4 -+ 17.3years, 67% men). For the serum Na+ levels recorded at three different times (early, peak, and late), mean late serum Na ÷ level during hospital course was significantly higher in patients who died than in those who survived (151.2 + 9.2 v 143,1 +_8.0 mEq/L, respectively; P < .001). Mean admission serum Na+ level (154.9 -+ 5.5 v 155.1 + 7.7 mEq/L, respectively) and mean peak serum Na ÷ level (157.6 + 6.5 v 156.8-+ 9.4 mEq/L, respectively) were not different between the two groups. Both SBP and DBP at the time of admission (P < .05) and throughout the hospital course (P < .001)were significantly lower in the patients who died than in those who survived. The cognitive abnormalities consisting of confusion, obtundation, and speech abnormality were significantly (P < .05) higher in the expired patients than in those who survived. Normal (isotonic) saline was used significantly more frequently (P < .00001) in patients who expired than in those who survived. Thus, this study suggests that a persistently elevated serum Na+ level (possibly caused by prolonged infusion of normal saline) in association with protracted hypotension portends a dismal prognosis in hospitalized hypernatremic patients. (Am J Emerg ied 1997;15:130-132. Copyright © 1997 by W.B. Saunders Company) Hypernatremia (serum sodium level of > 145 mEq/L) is a common problem among hospitalized patients who are 65 years of age or older. 1,2 Most studies report mortality rates in excess of 50% in elderly patients with hypernatremia. None of these studies, however, found any relationship between the severity of hypernatremia and mortality.~-3 Some investigators have noted that mortality increases with increased

From the Division of Nephrology, Departments of Medicine and Pathology, Veterans Affairs Medical Center, and Wright State University Dayton, OH. Manuscript received November 12, 1995, returned December 14, 1995; revision received January 24, 1996, accepted February 22, 1996. Presented at the National Meeting of the American Federation for Clinical Research, April 30-May 2, 1994, Baltimore, MD. Address reprint requests to Dr Mandal, Department of Medicine (111W), Wright State University, VA Medical Center, 4100 W. Third Street, Dayton, OH 45428. Key Words: Elderly subjects, hypernatremia, mortality, predictive factors, cognitive function, blood pressure, saline infusion. Copyright © 1997 by W.B. Saunders Company 0735-6757/97/1502-000555.00/0 130

rates of fluid replacement, 1 while other investigators have stressed the need for greater awareness of the importance of active maintenance of hydration in susceptible patients.; The exact cause(s) of high mortality in hypernatremic patients have not yet been fully elucidated. Slight or no information is available regarding the relationship between the severity of the clinical picture (symptoms and signs) and high mortality in hypernatremic patients. We present an analysis of the relationships of serum sodium, blood pressure (BP), cognitive function, and therapeutic interventions to high mortality in hyperuatremic patients. The purpose of this study was to determine whether any clinical or laboratory findings and/or therapeutic manipulations could predict mortality in hypernatremic patients.

MATERIALS AND METHODS The names and identification numbers of the patients noted to have hypernatremia during a period of 12 months were initially obtained from the Laboratory Service of two large universityaffiliated teaching hospitals in Dayton, Ohio. Hypernatremia was detected in the blood samples of 120 patients. Medical records of these 120 patients were reviewed. Four patients with incomplete data were excluded, leaving data from 116 patients available for analysis. Data collection included age and gender; serum sodium levels; whether hypernatremia was first noted in the emergency department, during a clinic visit, or in the hospital; whether hypernatremia led to hospital admission; primary diagnosis; status of cognitive function (confusion, obtundation, abnormal speech); gastrointestinal symptoms; use of diuretics or nonsteroidal antiinflammatory drugs; BP levels; and type of fluid therapy. Statistical analysis. Univariate analysis was conducted using three statistical tests: t test, chi-square test, and Fisher's exact test. The t test was used to compare the discharged and expired groups on continuous variables (age and sodium values). The chi-square test with Yates' correction for continuity where necessary was used for 2 × 2 contingency tables comparing the two groups on categorical variables or for tables larger than 2 × 2 where the expected frequency for each cell was greater than five (eg, gender, primary diagnosis, signs and symptoms, BP). Fisher's exact test was used for tables larger than 2 × 2 with at least one cell expected frequency less than five (systolic and diastolic BP during hospitalization and type of fluid therapy). Inferences were made at the .05 level of significance. Stepwise logistic regression was used for variables significant with univariate analysis to determine the best prediction model for the outcome of death among hypernatremic patients.

RESULTS The 116 patients were divided into two groups: (1) those who died; (2) those who survived. Seventy-seven patients (66%) died, while 39 patients (34%) survived and were discharged from the hospital. One hundred three patients

MANDAL ET AL • MORTALITY IN HYPERNATREMIA

131

were men and 13 were women. The results of univariate analysis on age and serum Na + levels are presented in Table 1. Mean age and gender (70.9 +__ 15.4 years, 90% men) of those who died were not different from those who survived (66.4 + 17.3 years, 87% men). Mean admission and peak serum Na + levels were not different between the two groups, whereas mean late (during hospital course) serum Na + level was significantly higher in those who died than in those who survived (151.2 mEq/L v 143.1 mEq/L, P < .001). In addition, hypernatremia was noted significantly more commonly in the hospital in the expired group than the survived group (95% v 59%; P < .00001). Blood pressures (BPs) on admission, as well as during the hospital course, were significantly different between the two groups. Thus, admission systolic BP was <100 mm Hg in 25% of the expired group, compared to none in the survival group (P = .003). Systolic BP of <100 mm Hg during the hospital course was found in 53% of those who died, compared to only 5% among those who survived (P < .0001). Admission diastolic BP of <60 mm Hg was found in 35% of the expired group, compared to 19% of the discharged group (P = .040). However, diastolic BP of <60 mm Hg during the hospital course was found in 67% of the expired group, compared to only 19% of survived group. This difference was very significant (P < .001). To determine relationships between primary diagnosis and hypernatremic death, primary diagnoses were grouped into five major categories: malignancy, pulmonary disease, infectious process, gastrointestinal disorder, and others. Although there was no statistically significant relationship between primary diagnosis and hypernatremic death, those who died tended to have more malignancies (22% v 13%) and more pulmonary disease (21% v 8%) than those who survived. Gender, detection of hypernatremia that caused hospital admission, nausea, vomiting, diuretic or nonsteroidal anti-inflammatory drug therapy, and pulse rate showed no significant relationship to hypernatremic death. However, the expired group showed significantly more cognitive defects than the discharged group. These defects were confusion (57% v 36%; P = .031), obtundation (61% v 39%; P = .021) and abnormal speech (58% v 39%; P = .042). The types of fluid that were used to treat 116 patients were normal saline (isotonic saline), 5 % dextrose in water, or total parenteral nutrition (TPN). Sixty-two of 77 patients (81%) who died and 20 of 39 patients (51%) who survived received TABLE1. Univariate Analysis of Discharged Patients Versus Expired Patients on Variables Measured on a Continuous Scale (tTest) Discharged

Expired

Variable

N

Mean + SD

N

Mean ± SD

P Value

Age (yrs) Sodium--admission Sodium---peak Sodium--late*

39 39 39 26

66.4 155.1 156.8 143.1

77 77 77 48

70.9 154.9 157.5 151.2

.157 .904 .684 <.001

-+ 17.3 -+ 7.7 ± 9.4 _+ 8.0

___15.4 _+ 5.5 +-- 6.5 _+ 9.2

NOTE: Sodium values listed in mEq/L (mean -+ standard deviation [SD]), *Late = during hospital course.

normal saline infusion throughout the hospital course. Conversely, 1% who died versus 33% who survived received no fluids by parenteral route. These differences between the two groups were highly significant (P < .00001). Eight percent who died and 10% who survived received infusion of 5% dextrose in water. Ten percent who died and 5% who survived were treated with TPN via a central line, These differences were not significant. Similarly, volume of fluid administered showed no significant relationship to mortality. Stepwise logistic regression was conducted using variables that were significant univariate predictors: serum sodium level during hospital course, location where hypernatremia first noted, cognitive dysfunction (confusion, obtundation, and speech abnormalities), systolic BP and diastolic BP at admission and during hospital course, and type of fluid therapy. The resulting model retained location where hypernatremia was first noted and diastolic BP during hospital course as significant predictors. Specifically, patients whose hypernatremia was first noted in the hospital versus the clinic or emergency department and with low diastolic BP during the hospital course were more likely to expire. The two-variable logistic regression model correctly predicted death in those patients who expired 96.6% of the time (only 3.4% false positives) but did less well predicting survival (only 28.6 percent of survivors were predicted correctly) (ie, 71.4% false negatives). DISCUSSION

The results of this analysis are consistent with previous studies t-3 and further confirm that hypernatremia is associated with a high probability of mortality in patients older than 65 years. However, we have presented new information showing significant relationships between hypernatremic death and serum sodium level, BE and type of fluid administered. Thus, a persistently elevated serum sodium level and low BPs during the hospital course are associated with high mortality rate. Our data suggest that the persistently elevated Na + level was most likely caused by the continuous infusion of normal saline, because 62 of 77 patients (81%) who died received normal saline infusion throughout their hospital course. In contrast, 20 of 39 patients (51%) who survived received normal saline infusion throughout their hospital course. This difference was highly significant (P < .00001). This finding suggests that continuous infusion of normal saline solution, except in the initial stage, could adversely influence outcome in hypernatremic patients. 4 As an alternative, it could also mean that patients with hypernatremia and hypotension are more sick than patients with hypernatremia but without hypotension. Since the former patients are more likely to be infused with normal saline, this could explain the association of persistently elevated Na + level, hypotension, and normal saline infusion to high mortality. The exact mechanism by which late (ie, persistently) elevated serum sodium level is associated with high mortality could not be ascertained from this study. Nevertheless, some logical considerations can be proposed. Hypernatremic patients who died were more often confused, more frequently obtunded, and more often had abnormal speech

132

AMERICAN JOURNAL OF EMERGENCY MEDICINE • Volume 15, Number 2 • March 1997

than those who survived. These cognitive defects suggest gross derangement of the central nervous system. A recent study in rats has shown that sustained levels of hypernatremia could induce brain myelinolysis similar to the histologic changes observed in hyponatremia-related brain myelinolysis.5 Therefore, hypernatremic death can be attributed to brain myelinolysis, which may develop as a result of persistent elevation of serum sodium. Previous studies have stressed excessive free water loss accompanied by inadequate replacement as a common cause of hypernatremia. 1,3,6In our patients, febrile illness was not common but hypernatremia was almost always first noted in the hospital among those who died (95% of the cases). Because many of these patients were confused and obtunded, inadequate fluid therapy (both parenteral and oral) cannot be completely ruled out. This problem may be encountered in hospitals for patients who are physically handicapped or mentally ill, including psychogeriatric units, as found in a British study. 6 American elderly veterans, who constituted the vast majority of this study population, resemble patients in geriatric hospitals in the United Kingdom. Therefore, our findings are likely to be similar to those in British studies. A study from the Mayo Clinic reported that rate of correction of hypernatremia did not influence s u r v i v a l . 7 Another report without any data analysis suggests that nursing and medical staff should be made more aware of this problem and encouraged to initiate early treatment of dehydration. 8 In our study, persistently elevated serum sodium in association with protracted low BPs, in particular diastolic BE was found to be the most significant predictable factor for high mortality in the hypernatremic patients. Since hyperuatremia is frequently associated with excellular fluid volume depletion, and therefore, low BP and renal function impairment are common accompaniments, infusion of normal saline is a common therapeutic recommendation. Although normal saline repletes volume and stabilizes hemodynamic state, serum Na + concentration tends to increase. After hemodynamic state is stabilized with normal saline or in hemodynamically stable hyperuatremic patients, we recommend continuation of fluid therapy with infusion of 5% dextrose in water, which provides free water as dextrose is oxidized. With this infusion, glomerular filtration rate increases, causing increased excretion of Na + and nitrogenous waste products and resulting in decrease of serum Na +, urea nitrogen, and creatinine levels. The patient's mental state also improves. In this context, a previous study

has shown that prolonged high-volume intravenous water load in five hyperuatremic patients resulted in biochemical correction and was paralleled by progressive clinical improvement, notably in peripheral circulation, tissue turgor, and cerebral function. 9 The mortality rate in our series is not different from those reported previously. Thus, among adult patients whose serum sodium levels were increased to 160 mEq/L in less than 24 hours, the reported mortality rate was greater than 70%J ,w The high mortality associated with hypernatremia, which develops in the hospital and is accompanied by central nervous manifestations, is rather intriguing. However, diffuse demyelinating lesions of the brain in hyperuatremic animals and humans have been reported,4,n,~2 and may account for the high mortality rate in hypernatremic patients. The high incidence of impaired cognitive function in these patients is further supportive evidence of that pathological lesion.

REFERENCES 1. Snyder NA, Feign DW, Arieff Al: Hypernatremia in elderly patients. Ann Intern Med 1987;107:309-319 2. Long CA, Marvin P, Bayer AJ, et al: Hypernatremia in an adult inpatient population. Postgrad Med J 1991 ;67:643-645 3. Mahowald JM, Himmelstein DU: Hypernatremia in the elderly: Relation to infection and mortality. J Am Geriatr Soc 1981 ;29:177-180 4. Morrison G: Hypernatremia. In Szerlip HM, Goldfarb S: Workshops in Fluid and Electrolyte Disorders. New York, Churchill Livingston, 1993, pp 49-69 5. Soupart A, Peaninckx R, Stenuit A, et al: Brain myelinolysis following hypernatremia in rats. J Am Soc Nephro11993;4:829 (abstr) 6. MacDonald NJ, McConnell KN, Stephen MR, Dunnigan MG: Hypernatremic dehydration in patients in a large hospital for the mentally handicapped. BMJ 1989;299:1426-1429 7. Moder KG, Hurley DL: Fatal hypernatremia from exogenous salt intake: Report of a case and review of the literature. Mayo Clinic Proc 1990;65:1587-1594 8. Bhatnagar D, Weinkove C: Serious hypematremia in a hospital population. Postgrad Med J 1988;64:441-443 9. Taylor KM, Morgan HG, Thomson RM, et al: Hypernatremic dehydration following tricuspid valve replacement, d Cardiovasc Surg 1978;19:449-454 10. Himmelstein DU, Jones AA, Woolhandler S: Hypernatremic dehydration in nursing home patients: An indicator of neglect. J Am Geriatr Soc 1983;31:466-471 11. Clark WR: Diffuse demyelinating lesions of the brain after the rapid development of hypernatremia. Western J Med 1992;157:571573 12. McComb RD, Pfeiffer RF, Casey RH, et al: Lateral pontine and extrapontine myelinolysis associated with hypernatremia and hyperglycemia. Clin Neuropatho11989;8:284-288