Postoperative Hyponatremic Encephalopathy in Menstruant Women J. Carlos Ayus, MD; James M. Wheeler, MD, MPH; and Allen I. Arieff, MD
• Objectives: To determine factors associated with the development of encephalopathy and with its clinical course in patients with postoperative hyponatremia. • Setting: Consultation and referral services of two university medical centers and community hospitals. • Design: Case-control study (risk factors for encephalopathy) and cohort study (clinical course among patients with encephalopathy). • Patients: Case patients included 65 adults with postoperative hyponatremic encephalopathy; controls included 674 adult patients who had postoperative hyponatremia without encephalopathy and who were selected from 76 678 consecutive adult surgical inpatients. • Measurements: Age, gender, menstrual status, neurologic symptoms, time to development and degree of hyponatremia, arterial blood gas determinations, serum chemistries, morbidity and mortality. • Results: Case patients included 40 women (62%) and 25 men (38%) (P > 0.05); controls included 367 women (54%) and 307 men (46%) (P > 0.1). Of the 34 case patients who developed permanent brain damage or died, 33 (97%) were women (P < 0.001). Among the women with brain damage, 25 (76%) were menstruant (P < 0.001). The relative risk for death or permanent brain damage from hyponatremic encephalopathy in women compared with men was 28 (95% CI, 5 to 141) and in menstruant women compared with postmenopausal women, 26 (CI, 11 to 62). Arterial P0 2 at diagnosis was significantly lower in female than in male case patients (34 ± 5 compared with 91 ± 3 mm Hg; P< 0.001). Further, of the 38 case patients who had respiratory arrest before the diagnosis of hyponatremic encephalopathy, 36 (95%) were women. Extent of or time to development of hyponatremia did not correlate with subsequent brain damage (P > 0.1). • Conclusions: Women and men are equally likely to develop hyponatremia and hyponatremic encephalopathy after surgery. However, when hyponatremic encephalopathy develops, menstruant women are about 25 times more likely to die or have permanent brain damage compared with either men or postmenopausal women.
Annals of Internal Medicine. 1992;117:891-897. From Baylor College of Medicine, Houston, Texas; and University of California, San Francisco, School of Medicine; and Veterans Affairs Medical Center, San Francisco, California. For current author addresses, see end of text.
x ostoperative hyponatremia is a frequent complication among adults in the United States and the United Kingdom (1, 2). Since 1935 (3), more than 260 cases of postoperative hyponatremia with associated neurologic involvement (hyponatremic encephalopathy) have been reported (4). At least 55 of these patients (21%) either died or developed permanent brain damage as a consequence of hyponatremia (3, 5-14). Brain damage associated with hyponatremia has commonly been related to both the magnitude of the hyponatremia and the rapidity of the fall in the plasma sodium level (15-17). However, in four recent studies, 36 of 38 patients with postoperative hyponatremic encephalopathy were women (7, 9, 13, 14). Among these 36 women, hyponatremia was present for intervals of up to 9 days, and the plasma sodium was as high as 128 mmol/L. All but one woman either died or experienced permanent brain damage, and both men recovered without experiencing neurologic sequelae. The reasons for the increased morbidity and mortality from postoperative hyponatremic encephalopathy among female patients and the relative absence of morbidity among men with the disorder are unclear. In our study, we evaluated the relation of morbidity and mortality from postoperative hyponatremic encephalopathy to several potential risk factors: degree and time to development of hyponatremia; associated medical conditions; type of surgery; arterial oxygen content; age; gender; and menstrual status.
Methods Patients Case patients included 65 adults who developed postoperative hyponatremic encephalopathy (plasma sodium, 128 mmol/L or less). All these patients were neurologically normal before surgery, had a postoperative plasma sodium level of 128 mmol/L or less, and developed postoperative neurologic symptoms of hyponatremic encephalopathy (all patients had nausea, emesis, headache, weakness or lethargy, and 38 of 65 had seizures and respiratory arrest) (6, 7). Control patients included 674 adults with postoperative hyponatremia. Controls were selected from among 76 678 consecutive adult surgical inpatients who had had operations during the same 3-year period in which the case patients had had surgery. A computer search of hospital records for the 76 678 surgical inpatients was done using the SAS database (18) to determine which surgical patients had had a postoperative plasma sodium level of 128 mmol/L or less. Demographic data were available for the case patients and controls. Case records were further reviewed for data on age, gender, menstrual status, preoperative medical problems, type of surgery and anesthesia, net postoperative fluid balance, type of hospital, type of physician coverage, therapy with hypertonic NaCl, arterial oxygen content, postoperative course, and clinical outcome. Menstruant women were defined as those who were having regular menstrual periods before surgery. Postmenopausal women were defined as those who had had no menstrual period for at least 1 year before surgery. In addition,
1 December 1992 • Annals of Internal Medicine • Volume 117 • Number 11
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19764/ by a University of California San Diego User on 02/21/2017
891
Figure 1. Effects of gender and menstrual status on brain damage from hyponatremic encephalopathy. Top panel. The proportions of men and women who did (cases) and did not (controls) develop permanent brain damage are compared. The relative risk for dying or developing permanent brain damage is 28 times as great in women as in men (95% CI, 5 to 141). Bottom panel. The proportion of women with hyponatremia who were and were not menstruant who did (cases) and did not (controls) have permanent brain damage. The relative risk for dying or developing permanent brain damage from postoperative hyponatremia is 26 times as great for menstruant women as for postmenopausal women (95% CI, 11 to 62). all menstruant women were also less than 50 years of age, whereas all post menopausal women were more than 50 years of age. None of the case or control patients have ever been described in previous reports from our laboratories (6, 7, 13, 14, 19, 20). Forty-nine percent of the case patients with hyponatremic encephalopathy were from hospitals affiliated with either Baylor University or the University of California, and the remainder were from other institutions. Statistical Analysis Data are presented as mean ± SE unless stated otherwise. Significance was determined by the appropriate test, either chi-square test, analysis of variance, or the unpaired Mest. The odds ratio as an estimate of the relative risk for encephalopathy among patients with postoperative hyponatremia was determined for various potential risk factors: gender; menstrual status; time to development of hyponatremia; and magnitude of hyponatremia. The 95% CIs for the relative risks were calculated using the logarithmic transformation (21). Results Case Patients Hyponatremia in the 65 case patients occurred secondary to the infusion of hypotonic fluids in patients known to have elevated plasma vasopressin levels. In 892
the 25 male patients, hypotonic fluids were administered either intravenously (usually 280 mmol/L glucose [5% dextrose/water]) or transurethrally (usually 200 mmol/L [1.5%] glycine). In the 40 female case patients, intravenous hypotonic fluids consisted primarily of either 280 mmol/L glucose or 280 mmol/L glucose plus 77 mmol/L NaCl. In the women, mean fluid intake was 6.45 ± 0.6L and mean fluid output was 1.6 ± 0.4L. The 65 cases included 25 men and 40 women (P > 0.05). At the time hyponatremia was diagnosed, the 25 male patients had a mean plasma sodium level of 111 ± 2 mmol/L, a mean osmolality of 227 ± 4 mmol/kg, a mean glucose level of 6.5 ± 0.6 mmol/L (116 ± 10 mg/dL), and a mean arterial P a 0 2 of 91 ± 3 mm Hg. Of the 25 male patients, 20 complained of headache and nausea, 14 were weak and lethargic, and 7 had seizure activity. Two male patients had respiratory arrest, which occurred before surgery was completed in one patient. Twenty-four of the 25 male patients recovered and were discharged from the hospital neurologically intact. In one patient, the plasma sodium level fell from 143 to 101 mmol/L in 35 minutes, resulting in seizure activity, respiratory arrest, and death. All 40 of the female case patients experienced nausea and vomiting, 34 had headache, and some had other variable symptoms, including weakness, slurred speech, lethargy, confusion, disorientation, bizarre behavior, urinary incontinence, dyspnea, and decorticate posturing. Thirty-six women sustained an abrupt respiratory arrest after surgery (range, 1 hour to 9 days). Seizure activity was seen before the arrest in 28 of these 36 women, but the other 8 women had not been closely observed. The diagnosis of hyponatremia was not made in these 36 women until a mean of 48 ± 8 hours after surgery. At the time of respiratory arrest, the plasma sodium level was 113 ± 1 mmol/L (range, 91 to 128 mmol/L), the potassium level was 3.1 ± 0.1 mmol/L, the glucose level was 5 ± 0.2 mmol/L (90 ± 3 mg/dL), and the osmolality was 228 ± 3 mmol/kg. As a consequence of respiratory arrest, P a 0 2 was 34 ± 5 mm Hg. Although the plasma osmolality and sodium level (Tables 1 and 2) were not different from those of the male case patients (P > 0.1), P2L02 was significantly lower (P < 0.001). After respiratory arrest, 25 female case patients were considered by their managing physicians to be brain dead, primarily because of the presence of apnea and fixed, dilated pupils. No attempt was made to correct their hyponatremia, and all died without regaining consciousness. Of these 25 women, 13 also developed central diabetes mellitus and insipidus (13).
Control Patients Patients with postoperative hyponatremia who did not have encephalopathy served as controls. Because they were selected by computer search from among 76 678 surgical inpatients, only demographic and coded medical record data were available. A total of 680 patients had postoperative hyponatremia, representing 0.9% of the surgical inpatients. However, six of these patients had neurologic manifestations of hyponatremia and
1 December 1992 • Annals of Internal Medicine • Volume 117 • Number 11
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19764/ by a University of California San Diego User on 02/21/2017
were excluded from the control group. Among the controls, 367 were female (54%) and 307 were male (46%) (P > 0.1). Among the 367 female controls, 39 (11%) were menstruant and 328 (89%) were postmenopausal. All postmenopausal female controls were also more than 50 years of age. Among the 307 male controls, 46 (15%) were less than 50 years of age and 262 (85%) were more than 50 years of age. None of the controls developed neurologic manifestations or permanent brain damage. We assessed whether controls were representative of all such patients by seeing if the age and gender distributions among controls were similar to those of the 76 678 patients from which the controls had been selected. Proportions of men and women more than 50 years of age were 59% and 46%, respectively, and proportions of men and women less than 50 years of age were 41% and 54% respectively; thus, all proportions represented at least 41% of the total group. The gender distribution showed 49% men and 51% women. Thus, little evidence exists of bias or asymmetry related to either age or gender in the 76 678 patients from which the control population was selected. We tested whether the demographic characteristics of the 76 678 patients from whom the controls were selected were similar to those of other postoperative patient populations. Generalizability was tested by analyzing demographic data from all patients in the United States undergoing surgery during the same period in which controls were selected. Among all adult patients (age > 16 years) in the United States undergoing surgery (excluding uncomplicated obstetrical procedures but including cesarean sections), 60% of operations occurred in women and 40% in men (22, 23). For male
patients undergoing surgery in the United States, 32% of the men were less than 50 years of age and 68% were more than 50 years of age; for female patients, 44% were less than 50 years of age and 56% were more than 50 years of age (22, 23). The gender and age distributions among the 76 678 surgical inpatients from which our controls were selected are therefore similar to those of all surgical patients in the United States (22, 23).
Risk Factors for Encephalopathy Demographic and clinical characteristics of case patients are summarized in Tables 1 and 2 and in Figure 1; retrospectively derived demographic data on the controls have been described above. Of the 65 case patients, 34 had major morbidity: 28 died and 6 had permanent brain damage. Thirty-three of these patients (97%) with major morbidity were female. Of the 674 controls, 367 (54%) were female. The estimated relative risk for experiencing major neurologic morbidity in women compared with men was 28 (95% CI, 5 to 141; P < 0.001) (Figure 1). Of the 33 female case patients with major morbidity, 25 (76%) were menstruant, whereas only 39 of the female controls (11%) were menstruant (328 [89%] were postmenopausal). The estimated relative risk for experiencing major neurologic morbidity in menstruant compared with postmenopausal women was 26 (95% CI, 11 to 62; P < 0.001) (Figure 1). Assessment of confounding variables focused on clinical conditions that could have produced the marked differences in gender and menstrual status distribution between the control patients and the case patients with major morbidity. Overall, no differences were found
Figure 2. Relation between rapidity of development and magnitude of hyponatremia and death according to age and gender. Mortality from postoperative symptomatic hyponatremia in 65 cases (25 men and 40 women) is shown. Mortality is significantly greater in women than in men in all subgroups (P < 0.001). The differences in mortality between women with a plasma sodium level greater or less than 116 mmol/L or with a time to development of hyponatremia either greater or less than 24 hours were all not significant (P > 0.1). 1 December 1992 • Annals of Internal Medicine • Volume 117 • Number 11
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19764/ by a University of California San Diego User on 02/21/2017
893
between male and female case patients regarding the presence of important concomitant illnesses (see Table 2). No differences were found between female case patients who did survive and those who did not survive regarding associated medical conditions (see Table 2). However, we compared male case patients who survived with female case patients who did not survive. The male survivors had a significantly higher incidence of coronary artery disease, obstructive uropathy, occlusive vascular disease, and chronic obstructive pulmonary disease. These findings contradict common clinical sense, because they suggest that major comorbidity is not likely to be a factor in our case-control study. The types of surgeries were similar among the 65 case patients, the 674 controls, and the 76 678 inpatients from whom the controls were selected. Gynecologic, orthopedic, urologic, gastrointestinal, and cardiovasclar surgeries were among the 10 most common types in all three patient groups. Factors Related to the Clinical Course The magnitude of hyponatremia, duration of hyponatremia, and age distribution among all 65 cases are shown in Table 1 and Figure 2. The overall morbidity among female case patients with hyponatremia was greater than 75% for all subgroups except postmenopausal women. Mortality was lowest among postmenopausal female case patients with hyponatremic encephalopathy (see Table 1 and Figure 2). The overall mortality in postmenopausal women was 33%, which is lower than the mortality in menstruant women (82%) (P < 0.02) but still higher than the mortality in men (P < 0.01). However, no significant difference was observed either in mortality between female case patients with a plasma sodium level greater than 116 mmol/L and those with a plasma sodium level less than 116 mmol/L; or between female case patients with time to development of hyponatremia greater than 24 hours and those with a
time to development less than 24 hours (P > 0.1) (see Figure 2). To determine whether any possible clinical conditions contributed to the differences in outcome among male case patients who survived, female case patients who survived, and female case patients who died or had permanent brain damage, we examined other variables possibly related to outcome, including type of surgery and associated medical conditions (see Table 2). The surviving case patients had a higher Pa02 (Figure 3), were older, were treated more often with hypertonic NaCl, and had a lower incidence of respiratory arrest before initiation of therapy (P < 0.01). Male case patients had a significantly higher rate of genitourinary surgery and female cases and a higher rate of gynecologic surgery. The most common operation among male case patients was prostate resection and among female cases, hysterectomy. Among 23 patients who had an autopsy, no pathologic cause for death was found other than hyponatremic encephalopathy. The mean brain weight was 1381 ± 14 g, about 8% above the mean control value (± SD) of 1275 ± 73 g in adult women. Thus, even when time to development of hyponatremia was up to 9 days, the brain was unable to adapt. There were two distinct patterns of pathologic findings. In the 20 patients who died within 10 days of surgery, the major finding was diffuse cerebral edema, but 14 patients also had herniation of the cerebellum or uncus (or both). Pathologic lesions in addition to cerebral edema and herniation included diffuse cerebral demyelination (n = 9); pituitary infarction (n = 5); cortical infarction in region of middle cerebral artery (n = 2); brain stem infarction (n = 2); lateral pontine demyelination (n = 2); hemorrhage or necrosis (or both) of the occulomotor (third cranial) nerve (n = 5). The five patients with pituitary infarction all developed central diabetes mellitus and insipidus (13).
Table 1. Referred Patients with Symptomatic Postoperative Hyponatremia* Group
Male case patients Overall Time to development of hyponatremia of less than 24 hours Time to development of hyponatremia of more than 24 hours Sodium level less than 116 mmol/L Sodium level 116 to 128 mmol/L Age of more than 50 years Female case patients Overall Time to development of hyponatremia of less than 24 hours Time to development of hyponatremia of more than 24 hours Sodium level less than 116 mmol/L Sodium level 116 to 128 mmol/L Postmenopausal Menstruant
Initial Serum Sodium
Final Serum Sodium
Mortality
Brain Damage
Patients
Age
n
y
25 14
66 ± 2 64 ± 6
138 ± 1 139 ± 2
110 ± 2 104 ± 5
4 7
0 0
11
67 ± 3
136 ± 2
120 ± 2
0
0
15 10 24
64 ± 2 69 ± 4 67 ± 2
138 ± 1 137 ± 2 138 ± 1
105 ± 2 122 ± 2 111 ± 3
7 0 4
0 0 0
40 13
42 ± 2 32 ± 2
138 ± 1 138 ± 2
113 ± 1 117 ± 2
68 85
18 8
27
46 ± 3
139 ± 1
112 ± 2
59
19
19 21 12 28
46 37 63 33
138 137 137 139
107 120 108 116
63 71 33 82
26 5 33 7
±4 ± 3 ± 2 ±2
%
mmol/L
± 1 ± 2 ±3 ± 1
±2 ± 1 ± 3 ± 1
* Data presented as mean ± SE. P < 0.01 for all comparisons of corresponding male and female groups.
894
1 December 1992 • Annals of Internal Medicine • Volume 117 • Number 11
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19764/ by a University of California San Diego User on 02/21/2017
Table 2. Characteristics of Patients Who Survived and of Patients Who Experienced Morbidity or Mortality* Variables Potentially Related to Outcome
dale Suivivors
in = 24) Serum sodium, mmol/L Net fluid retention, mL/kgper day Time for hyponatremia to develop, h Pa 02 at the time of diagnosis, mm Hg Age,y Treatment with hypertonic NaCl, % Respiratory arrest before therapy, % Private hospital, % Private physician, % Availability of intensive care, % General anesthesia, % Spinal anesthesia, % Type of surgery, % Gynecology Genitourinary Orthopedic Plastic or cosmetic Gastrointestinal Neurosurgery Cardiovascular Associated medical conditions, % Coronary artery disease Benign uterine tumor Obstructive uropathy or urolithiasis Chronic obstructive pulmonary disease Cholelithiasis Occlusive vascular disease Fracture or arthropathy Chronic renal failure Diabetes mellitus Benign breast tumor Miscellaneous gynecologic disorders
112 ± 2 35 ± 2 35 ± 8 91 ± 3 66 ± 2 75 4 29 33 100 62 38 46 21 4 17 0 12 21 46 12 8 21 17 12 8 0
Female Survivors (n = 7) 115 ± 2 41 ± 5 37 ± 6 55 ± 2 58 ± 5 100 43 100 100 100 71 29
Women Who Died or Had Permanent Brain Damage (n = 33) 113 ± 2 39 ± 3 48 ± 8 31 ± 5 39 ± 2 21 100 100 100 100 88 12
Statistical Significance When Compared with Women Who Died or Had Permanent Brain Damage Male Survivors Female Survivors
P P P P P P
NS NS NS < 0.001 < 0.01 < 0.001 < 0.01 < 0.01 < 0.025 NS NS NS
P P P P
NS NS NS < 0.05 < 0.01 < 0.01 < 0.05 NS NS NS NS NS
14 14 29 14 29 0 0
45 9 12 15 12 6 0
NS P < 0.05 NS NS NS NS NS
NS NS NS NS NS NS NS
14 14 14 0 14 0 29 0 0 0 14
3 18 9 0 6 3 12 3 3 6 45
P < 0.05 NS P 0.05).
Discussion Our study shows that the likelihood of developing postoperative hyponatremia appears to be similar among men and women. Hyponatremic encephalopathy also has a similar distribution among men and women. However, our case-control study showed that menstruant women with hyponatremic encephalopathy are significantly more likely to either die or experience permanent brain damage than are either men or postmenopausal women. This increased risk does not appear to depend on either the rapidity of development or magnitude of the hyponatremia, because fatalities occurred in menstruant women with plasma sodium concentrations as high as 128 mmol/L and a time to development of hyponatremia of as great as 9 days. The data from our case-control study suggest that, with no obvious source of confounding, the patients with the greatest likelihood of permanent brain damage or death from postoperative hyponatremia are menstruant women, most of whom were undergoing elective gynecologic surgery (see Table 2). Although occult confounding variables or biases are possible in case-control research, the statistical magnitude of our findings suggests that both gender and menstrual status have a real association with morbidity from postoperative hyponatremic encephalopathy. Among the 40 women studied, the brain was exam-
ined by either computed tomography (n = 21) or magnetic resonance imaging (n = 8) when the plasma sodium concentration was 128 mmol/L or less, and all showed cerebral edema. In addition, postmortem examination was done in 23 women 1 to 9 days after surgery, and all had gross evidence of cerebral edema. Brain weight was increased by a mean of 8% above control values, which exceeds the capacity of the brain to expand within the rigid confines of the skull (about 5%) (24). Thus, the female case patients were not able to adapt to hyponatremia even, in some cases, after more than a week. All but two of the male case patients were clinically able to adapt to hyponatremia within several hours. Studies in our laboratory have shown that male rodents were better able than female rodents to adapt to hyponatremia by extruding sodium to decrease braincell osmolality (25, 26). The more effective adaptation resulted in significantly less brain swelling in hyponatremic male rodents (26). Thus, the reasons for the increased morbidity and mortality in women compared with men may include a diminished ability of the female brain to adapt to hyponatremia by limiting the amount of brain swelling. The much higher mortality from postoperative hyponatremia in menstruant women relative to that in either postmenopausal women or men may be due, at least in part, to physical factors that affect the ability of
1 December 1992 • Annals of Internal Medicine • Volume 117 • Number 11
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19764/ by a University of California San Diego User on 02/21/2017
895
Figure 3. Pa0l in surviving case patients compared with those case patients who died or had permanent brain damage. The Pa 02 was significantly greater in both the male and female survivors compared with female case patients who died or had permanent brain damage. *P < 0.01; **P < 0.05.
the brain to adapt to hyponatremia. As persons age, brain volume declines progressively, whereas skull size remains constant in adult life (27). Further, intracerebral loss of cerebrospinal fluid to decrease brain size is a major adaptive component of the volume-regulatory decrease in hyponatremia (28, 29). The percent of cerebrospinal fluid in the brain increases with age (30), and much more so in men (27). Thus, elderly persons of both genders have more room in the rigid skull for the brain to expand than do younger ones (27, 31), but the finding is more marked in men, which may explain in part why elderly men have much less morbidity from hyponatremic encephalopathy than do women. The clinical presentation of men was distinctly different from that of women. Even though the plasma sodium level in the 25 male case patients decreased from 137 to 111 mmol/L in 34 ± 8 hours, only 7 men had grand mal seizures and 2 had respiratory arrest. On the other hand, plasma sodium in the female case patients decreased essentially the same amount, but all developed symptoms, with 36 of 40 women (90%) experiencing respiratory arrest before hyponatremia was diagnosed. The differing clinical presentations appear to be a major reason why men experience so much less morbidity compared with females. In 23 of 25 male case patients, consultation was obtained and the hyponatremia diagnosed without the occurance of respiratory arrest. The more timely diagnosis usually enabled appropriate therapy to be undertaken, thus preventing the possible occurrence of respiratory arrest. The occurrence of respiratory insufficiency before therapy is a major factor militating against survival without permanent brain damage. Hyponatremia can result in both a decline of cerebral blood flow and respiratory depression (28, 31-33). In our study, P a 0 2 less than 40 mm Hg was seen in the female patients who had permanent brain damage. Such a degree of 896
1 December 1992 • Annals of Internal Medicine
hypoxemia has been associated with coma and permanent brain damage in healthy humans without hyponatremia (34-36). Preliminary data from our laboratory show that in laboratory animals, the combination of systemic hypoxemia and hyponatremia is far more deleterious than either factor alone because hypoxia produces several additional maladaptive effects in the hyponatremic brain (37). We did not directly study whether encephalopathy could be prevented. It is clear, however, that when menstruant women develop hyponatremic encephalopathy, their risk for death or permanent brain damage is significantly greater than that of either men or postmenopausal women. This increased risk does not depend on either the rapidity of development or the magnitude of the hyponatremia. Thus, every reasonable effort should be made to avoid hyponatremia in such patients. Elevated plasma levels of vasopressin are essentially a universal postoperative occurrence in the initial 2 to 4 postoperative days (1, 38, 39), so it may be important to avoid the use of hypotonic intravenous solutions in the immediate postoperative period. When a patient, especially a menstruant woman, is receiving hypotonic fluids and manifests either headache, emesis, nausea, lethargy, or evidence of increased intracranial pressure, hyponatremic encephalopathy should be suspected and the plasma sodium level measured. These symptoms are somewhat nonspecific. However, the diagnosis is easily established by evaluation of the plasma sodium level, with virtually no risk to the patient and at minimal cost. When hyponatremic encephalopathy is diagnosed, the focus should be on prevention of brain damage or death. This can best be accomplished by initiation of treatment before occurrence of a hypoxic or anoxic episode, such as respiratory arrest (9, 20, 32, 33). Acknowledgments: The authors thank Maxine A. Papadakis, MD, for assisting in the study design; Warren S. Browner, MD, MPH, and Susan M. Clarkson, PhD, for helping with the statistical interpretation of the data; and D. Subramanian, MD, for helping to evaluate the control groups. Grant Support: In part by grant RO1AG08575 from the National Institute on Aging, Department of Health and Human Services, Bethesda, Maryland, and by the Research Service of the Veterans Affairs Medical Center, San Francisco, California. Requests for Reprints: J. Carlos Ayus, MD, 4 Brompton Court, Houston, Texas 77024. Current Author Addresses: Dr. Ayus: 4 Brompton Court, Houston, Texas 77024. Dr. Wheeler: Department of Obstetrics and Gynecology, Baylor College of Medicine, 1 Baylor Plaza, Houston, Texas 77030. Dr. Ariefif: Department of Medicine (111 G), Veterans Affairs Medical Center, 4150 Clement Street, San Francisco, CA 94121.
References 1. Chung HM, Kluge R, Schrier RW, Anderson RJ. Postoperative hyponatremia. A prospective study. Arch Intern Med. 1986; 146: 333-6. 2. Guy AJ, Michaels JA, Flear CT. Changes in the plasma sodium concentration after minor, moderate and major surgery. Br J Surg. 1987;74:1027-30. 3. Helwig FC, Schultz CB, Curry DE. Water intoxication: report of a fatal human case, with clinical, pathologic and experimental studies. JAMA. 1935;104:1569-75. 4. Fraser CL, Arieff AI. Metabolic encephalopathy associated with water, electrolyte, and acid-base disorders. In: Maxwell MH, Kl-
• Volume 117 • Number 11
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19764/ by a University of California San Diego User on 02/21/2017
5. 6.
7.
8.
9. 10. 11.
12. 13.
14.
15.
16.
17.
18. 19.
20.
21. 22.
23.
eeman CR, Narins RG, eds. Clinical Disorders of Fluid and Electrolyte Metabolism. New York: McGraw-Hill; [In press]. Scott JC Jr., Welch JS, Herman IB. Water intoxication and sodium depletion in surgical patients. Obstet Gynecol. 1965;26:168-75. Arieff AI, Llach F, Massry SG. Neurological manifestations and morbidity of hyponatremia: correlation with brain water and electrolytes. Medicine (Baltimore). 1976;55:121-9. Arieff AI. Hyponatremia, convulsions, respiratory arrest, and permanent brain damage after elective surgery in healthy women. N Engl J Med. 1986;314:1529-35. Charlton AJ. Cardiac arrest during transurethral prostatectomy after absorption of 1.5% glycine. A case report and review of the literature. Anaesthesia. 1980;35:804-6. Worthley LI, Thomas PD. Treatment of hyponatraemic seizures with intravenous 29.2% saline. Br Med J. 1986;292:168-70. Zimmermann B, Wangensteen OH. Observations on water intoxication in surgical patients. Surgery. 1952;31:654-669. Harrison RH 3d, Boren JS, Robison JR. Dilutional hyponatremic shock: another concept of the transurethral resection reaction. J Urol. 1956;75:95-110. Ryder KW, Olson JF, Kahnoski RJ, Karn RC, Oei TO. Hyperammonia after transurethral resection of the prostate: report of 2 cases. J Urology. 1984;132:995-7. Fraser CL, Arieff AI. Fatal central diabetes mellitus and insipidus resulting from untreated hyponatremia: a new syndrome. Ann Intern Med. 1990;112:113-9. Tien R, Arieff AI, Kucharczyk W, Wasik A, Kucharczyk J. Hyponatremic encephalopathy: is central pontine myelinolysis a component? Am J Med. 1992;92:513-22. Andreoli TE. Disorders of fluid volume, electrolytes and acid-base balance. In: Wyngaarden JB, Smith LH, Bennett JC, eds. Cecil Textbook of Medicine. Philadelphia: W.B. Saunders; 1992:499-513. Narins RG, Krishma GG. Disorders of water balance. In: Stein JH, ed. Internal Medicine. Boston: Little, Brown and Company; 1990: 832-44. Levinsky NG. Sodium and water: hyponatremia; clinical features and diagnosis. In: Wilson JD, et al., eds. Harrison's Principles of Internal Medicine. New York: McGraw-Hill; 1991:278-89. SAS Institute. SAS Users Guide: Basics, Version 5. Cary, North Carolina: SAS Institute Inc.; 1985:1290. Ayus JC, Olivero JJ, Frommer JP. Rapid correction of severe hyponatremia with intravenous hypertonic saline solution. Am J Med. 1982;72:43-8. Ayus JC, Krothapalli RK, Arieff AI. Treatment of symptomatic hyponatremia and its relation to brain damage. A prospective study. N Engl J Med. 1987;317:1190-5. Rothman KJ. Case control data. In: Rothman KJ, ed. Modern Epidemiology. Boston: Little, Brown and Company; 1986:174-5. Kozak LJ. Hospital inpatient surgery: United States, 1983-87. Hospital Inpatient Surgery: United States, 1983-87. Hyattsville, Maryland: Department of Health and Human Services. 1989. DHHS publication No. PHS 89-1250. U.S. Department of Commerce. Health and nutrition. In: U.S. De-
24.
25.
26. 27.
28. 29.
30.
31.
32. 33.
34.
35.
36.
37.
38.
39.
partment of Commerce. Statistical Abstracts of the United States, 1990. 110th edition. Washington, DC: Bureau of the Census; 1990: 110-111. Garcia JH, Conger KA, Morawetz R, Halsey JH Jr. Postischemic brain edema: quantitation and evolution. In: Cervos-Navarro J, Ferszt R, eds. Brain Edema: Pathology, Diagnosis, and Therapy. New York: Raven Press; 1980:147-69. Fraser CL, Sarnacki P. N a + - K + ATPase pump function in male rat brain synaptosomes is different from that of females. Am J Physiol. 1989;257:E284-9. Ayus J C , Arieff AI. Brain adaptation in hyponatremic rodents: effects of gender and age [Abstract]. Clin Res. 1992;40:223A. Gur RC, Mozley PD, Resnick SM, Gottlieb GL, Kohn M, Zimmerman R, et al. Gender differences in age effect on brain atrophy measured by magnetic resonance imaging. Proc Natl Acad Sci USA. 1991;88:2845-9. Rosomoff HL, Zugibe FT. Distribution of intracranial contents in experimental edema. Arch Neurol. 1963;9:36-44. Melton JE, Nattie EE. Brain and CSF water and ions during dilutional and isomotic hyponatremia in the rat. Am J Physiol. 1983; 244:R724-32. Nattie EE, Edwards WH. Brain and CSF water in newborn puppies during acute hypo- and hypernatremia. J Appl Physiol. 1981 ;51: 1086-91. Arieff AI, Ayus J C , Fraser CL. Hyponatraemia and death or permanent brain damage in healthy children. Br Med J. 1992;304:121822. Keating J P , Schears GJ, Dodge PR. Oral water intoxication in infants. An American epidemic. Am J Dis Child. 1991;145:985-90. Sarnaik AP, Meert K, Hackbarth R, Fleischmann L. Management of hyponatremic seizures in children with hypertonic saline: a safe and effective strategy. Crit Care Med. 1991;19:758-62. Hornbein TF, Townes BD, Schoene RB, Sutton JR, Houston CS. The cost to the central nervous system of climbing to extremely high altitude. N Engl J Med. 1989;321:1714-9. Oelz O, Regard M, Wichmann W, Walavanis A, Witztum A, Brugger P, et al. Cognitive impairment, neurologic performance and MRI after repeat exposure to extreme altitude. In: Sutton JR, Coates J, Remmers J, eds. Hypoxia: The Adaptations. Philadelphia: B.C. Becker, Inc.; 1990:206-9. Shimojyo S, Scheinberg P, Kogure K, Reinmuth OM. The effects of graded hypoxia upon transient cerebral blood flow and oxygen consumption. Neurology. 1968;18:127-33. Vexler ZS, Ayus J C , Kucharczyk J, Guntupalli K, Arieff AI. Effects of hypoxia on brain adaptation in experimental hyponatremia [Abstract]. Clin Res. 1992;40:306A. Deutsch S, Goldberg M, Dripps RD. Postoperative hyponatremia with the inappropriate release of antidiuretic hormone. Anesthesiology. 1966;27:250-6. Moran WH Jr., Miltenberger FW, Shuayb WA, Zimmermann B. The relationship of antidiuretic hormone secretion to surgical stress. Surgery. 1964;56:99-108.
1 December 1992 • Annals of Internal Medicine • Volume 117 • Number 11
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19764/ by a University of California San Diego User on 02/21/2017
897
• Objectives: To determine factors associated with the development of encephalopathy and with its clinical course in patients with postoperative hyponatremia. • Setting: Consultation and referral services of two university medical centers and community hospitals. • Design: Case-control study (risk factors for encephalopathy) and cohort study (clinical course among patients with encephalopathy). • Patients: Case patients included 65 adults with postoperative hyponatremic encephalopathy; controls included 674 adult patients who had postoperative hyponatremia without encephalopathy and who were selected from 76 678 consecutive adult surgical inpatients. • Measurements: Age, gender, menstrual status, neurologic symptoms, time to development and degree of hyponatremia, arterial blood gas determinations, serum chemistries, morbidity and mortality. • Results: Case patients included 40 women (62%) and 25 men (38%) (P > 0.05); controls included 367 women (54%) and 307 men (46%) (P > 0.1). Of the 34 case patients who developed permanent brain damage or died, 33 (97%) were women (P < 0.001). Among the women with brain damage, 25 (76%) were menstruant (P < 0.001). The relative risk for death or permanent brain damage from hyponatremic encephalopathy in women compared with men was 28 (95% CI, 5 to 141) and in menstruant women compared with postmenopausal women, 26 (CI, 11 to 62). Arterial P0 2 at diagnosis was significantly lower in female than in male case patients (34 ± 5 compared with 91 ± 3 mm Hg; P< 0.001). Further, of the 38 case patients who had respiratory arrest before the diagnosis of hyponatremic encephalopathy, 36 (95%) were women. Extent of or time to development of hyponatremia did not correlate with subsequent brain damage (P > 0.1). • Conclusions: Women and men are equally likely to develop hyponatremia and hyponatremic encephalopathy after surgery. However, when hyponatremic encephalopathy develops, menstruant women are about 25 times more likely to die or have permanent brain damage compared with either men or postmenopausal women.
Annals of Internal Medicine. 1992;117:891-897. From Baylor College of Medicine, Houston, Texas; and University of California, San Francisco, School of Medicine; and Veterans Affairs Medical Center, San Francisco, California. For current author addresses, see end of text.
x ostoperative hyponatremia is a frequent complication among adults in the United States and the United Kingdom (1, 2). Since 1935 (3), more than 260 cases of postoperative hyponatremia with associated neurologic involvement (hyponatremic encephalopathy) have been reported (4). At least 55 of these patients (21%) either died or developed permanent brain damage as a consequence of hyponatremia (3, 5-14). Brain damage associated with hyponatremia has commonly been related to both the magnitude of the hyponatremia and the rapidity of the fall in the plasma sodium level (15-17). However, in four recent studies, 36 of 38 patients with postoperative hyponatremic encephalopathy were women (7, 9, 13, 14). Among these 36 women, hyponatremia was present for intervals of up to 9 days, and the plasma sodium was as high as 128 mmol/L. All but one woman either died or experienced permanent brain damage, and both men recovered without experiencing neurologic sequelae. The reasons for the increased morbidity and mortality from postoperative hyponatremic encephalopathy among female patients and the relative absence of morbidity among men with the disorder are unclear. In our study, we evaluated the relation of morbidity and mortality from postoperative hyponatremic encephalopathy to several potential risk factors: degree and time to development of hyponatremia; associated medical conditions; type of surgery; arterial oxygen content; age; gender; and menstrual status.
Methods Patients Case patients included 65 adults who developed postoperative hyponatremic encephalopathy (plasma sodium, 128 mmol/L or less). All these patients were neurologically normal before surgery, had a postoperative plasma sodium level of 128 mmol/L or less, and developed postoperative neurologic symptoms of hyponatremic encephalopathy (all patients had nausea, emesis, headache, weakness or lethargy, and 38 of 65 had seizures and respiratory arrest) (6, 7). Control patients included 674 adults with postoperative hyponatremia. Controls were selected from among 76 678 consecutive adult surgical inpatients who had had operations during the same 3-year period in which the case patients had had surgery. A computer search of hospital records for the 76 678 surgical inpatients was done using the SAS database (18) to determine which surgical patients had had a postoperative plasma sodium level of 128 mmol/L or less. Demographic data were available for the case patients and controls. Case records were further reviewed for data on age, gender, menstrual status, preoperative medical problems, type of surgery and anesthesia, net postoperative fluid balance, type of hospital, type of physician coverage, therapy with hypertonic NaCl, arterial oxygen content, postoperative course, and clinical outcome. Menstruant women were defined as those who were having regular menstrual periods before surgery. Postmenopausal women were defined as those who had had no menstrual period for at least 1 year before surgery. In addition,
1 December 1992 • Annals of Internal Medicine • Volume 117 • Number 11
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19764/ by a University of California San Diego User on 02/21/2017
891
Figure 1. Effects of gender and menstrual status on brain damage from hyponatremic encephalopathy. Top panel. The proportions of men and women who did (cases) and did not (controls) develop permanent brain damage are compared. The relative risk for dying or developing permanent brain damage is 28 times as great in women as in men (95% CI, 5 to 141). Bottom panel. The proportion of women with hyponatremia who were and were not menstruant who did (cases) and did not (controls) have permanent brain damage. The relative risk for dying or developing permanent brain damage from postoperative hyponatremia is 26 times as great for menstruant women as for postmenopausal women (95% CI, 11 to 62). all menstruant women were also less than 50 years of age, whereas all post menopausal women were more than 50 years of age. None of the case or control patients have ever been described in previous reports from our laboratories (6, 7, 13, 14, 19, 20). Forty-nine percent of the case patients with hyponatremic encephalopathy were from hospitals affiliated with either Baylor University or the University of California, and the remainder were from other institutions. Statistical Analysis Data are presented as mean ± SE unless stated otherwise. Significance was determined by the appropriate test, either chi-square test, analysis of variance, or the unpaired Mest. The odds ratio as an estimate of the relative risk for encephalopathy among patients with postoperative hyponatremia was determined for various potential risk factors: gender; menstrual status; time to development of hyponatremia; and magnitude of hyponatremia. The 95% CIs for the relative risks were calculated using the logarithmic transformation (21). Results Case Patients Hyponatremia in the 65 case patients occurred secondary to the infusion of hypotonic fluids in patients known to have elevated plasma vasopressin levels. In 892
the 25 male patients, hypotonic fluids were administered either intravenously (usually 280 mmol/L glucose [5% dextrose/water]) or transurethrally (usually 200 mmol/L [1.5%] glycine). In the 40 female case patients, intravenous hypotonic fluids consisted primarily of either 280 mmol/L glucose or 280 mmol/L glucose plus 77 mmol/L NaCl. In the women, mean fluid intake was 6.45 ± 0.6L and mean fluid output was 1.6 ± 0.4L. The 65 cases included 25 men and 40 women (P > 0.05). At the time hyponatremia was diagnosed, the 25 male patients had a mean plasma sodium level of 111 ± 2 mmol/L, a mean osmolality of 227 ± 4 mmol/kg, a mean glucose level of 6.5 ± 0.6 mmol/L (116 ± 10 mg/dL), and a mean arterial P a 0 2 of 91 ± 3 mm Hg. Of the 25 male patients, 20 complained of headache and nausea, 14 were weak and lethargic, and 7 had seizure activity. Two male patients had respiratory arrest, which occurred before surgery was completed in one patient. Twenty-four of the 25 male patients recovered and were discharged from the hospital neurologically intact. In one patient, the plasma sodium level fell from 143 to 101 mmol/L in 35 minutes, resulting in seizure activity, respiratory arrest, and death. All 40 of the female case patients experienced nausea and vomiting, 34 had headache, and some had other variable symptoms, including weakness, slurred speech, lethargy, confusion, disorientation, bizarre behavior, urinary incontinence, dyspnea, and decorticate posturing. Thirty-six women sustained an abrupt respiratory arrest after surgery (range, 1 hour to 9 days). Seizure activity was seen before the arrest in 28 of these 36 women, but the other 8 women had not been closely observed. The diagnosis of hyponatremia was not made in these 36 women until a mean of 48 ± 8 hours after surgery. At the time of respiratory arrest, the plasma sodium level was 113 ± 1 mmol/L (range, 91 to 128 mmol/L), the potassium level was 3.1 ± 0.1 mmol/L, the glucose level was 5 ± 0.2 mmol/L (90 ± 3 mg/dL), and the osmolality was 228 ± 3 mmol/kg. As a consequence of respiratory arrest, P a 0 2 was 34 ± 5 mm Hg. Although the plasma osmolality and sodium level (Tables 1 and 2) were not different from those of the male case patients (P > 0.1), P2L02 was significantly lower (P < 0.001). After respiratory arrest, 25 female case patients were considered by their managing physicians to be brain dead, primarily because of the presence of apnea and fixed, dilated pupils. No attempt was made to correct their hyponatremia, and all died without regaining consciousness. Of these 25 women, 13 also developed central diabetes mellitus and insipidus (13).
Control Patients Patients with postoperative hyponatremia who did not have encephalopathy served as controls. Because they were selected by computer search from among 76 678 surgical inpatients, only demographic and coded medical record data were available. A total of 680 patients had postoperative hyponatremia, representing 0.9% of the surgical inpatients. However, six of these patients had neurologic manifestations of hyponatremia and
1 December 1992 • Annals of Internal Medicine • Volume 117 • Number 11
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19764/ by a University of California San Diego User on 02/21/2017
were excluded from the control group. Among the controls, 367 were female (54%) and 307 were male (46%) (P > 0.1). Among the 367 female controls, 39 (11%) were menstruant and 328 (89%) were postmenopausal. All postmenopausal female controls were also more than 50 years of age. Among the 307 male controls, 46 (15%) were less than 50 years of age and 262 (85%) were more than 50 years of age. None of the controls developed neurologic manifestations or permanent brain damage. We assessed whether controls were representative of all such patients by seeing if the age and gender distributions among controls were similar to those of the 76 678 patients from which the controls had been selected. Proportions of men and women more than 50 years of age were 59% and 46%, respectively, and proportions of men and women less than 50 years of age were 41% and 54% respectively; thus, all proportions represented at least 41% of the total group. The gender distribution showed 49% men and 51% women. Thus, little evidence exists of bias or asymmetry related to either age or gender in the 76 678 patients from which the control population was selected. We tested whether the demographic characteristics of the 76 678 patients from whom the controls were selected were similar to those of other postoperative patient populations. Generalizability was tested by analyzing demographic data from all patients in the United States undergoing surgery during the same period in which controls were selected. Among all adult patients (age > 16 years) in the United States undergoing surgery (excluding uncomplicated obstetrical procedures but including cesarean sections), 60% of operations occurred in women and 40% in men (22, 23). For male
patients undergoing surgery in the United States, 32% of the men were less than 50 years of age and 68% were more than 50 years of age; for female patients, 44% were less than 50 years of age and 56% were more than 50 years of age (22, 23). The gender and age distributions among the 76 678 surgical inpatients from which our controls were selected are therefore similar to those of all surgical patients in the United States (22, 23).
Risk Factors for Encephalopathy Demographic and clinical characteristics of case patients are summarized in Tables 1 and 2 and in Figure 1; retrospectively derived demographic data on the controls have been described above. Of the 65 case patients, 34 had major morbidity: 28 died and 6 had permanent brain damage. Thirty-three of these patients (97%) with major morbidity were female. Of the 674 controls, 367 (54%) were female. The estimated relative risk for experiencing major neurologic morbidity in women compared with men was 28 (95% CI, 5 to 141; P < 0.001) (Figure 1). Of the 33 female case patients with major morbidity, 25 (76%) were menstruant, whereas only 39 of the female controls (11%) were menstruant (328 [89%] were postmenopausal). The estimated relative risk for experiencing major neurologic morbidity in menstruant compared with postmenopausal women was 26 (95% CI, 11 to 62; P < 0.001) (Figure 1). Assessment of confounding variables focused on clinical conditions that could have produced the marked differences in gender and menstrual status distribution between the control patients and the case patients with major morbidity. Overall, no differences were found
Figure 2. Relation between rapidity of development and magnitude of hyponatremia and death according to age and gender. Mortality from postoperative symptomatic hyponatremia in 65 cases (25 men and 40 women) is shown. Mortality is significantly greater in women than in men in all subgroups (P < 0.001). The differences in mortality between women with a plasma sodium level greater or less than 116 mmol/L or with a time to development of hyponatremia either greater or less than 24 hours were all not significant (P > 0.1). 1 December 1992 • Annals of Internal Medicine • Volume 117 • Number 11
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19764/ by a University of California San Diego User on 02/21/2017
893
between male and female case patients regarding the presence of important concomitant illnesses (see Table 2). No differences were found between female case patients who did survive and those who did not survive regarding associated medical conditions (see Table 2). However, we compared male case patients who survived with female case patients who did not survive. The male survivors had a significantly higher incidence of coronary artery disease, obstructive uropathy, occlusive vascular disease, and chronic obstructive pulmonary disease. These findings contradict common clinical sense, because they suggest that major comorbidity is not likely to be a factor in our case-control study. The types of surgeries were similar among the 65 case patients, the 674 controls, and the 76 678 inpatients from whom the controls were selected. Gynecologic, orthopedic, urologic, gastrointestinal, and cardiovasclar surgeries were among the 10 most common types in all three patient groups. Factors Related to the Clinical Course The magnitude of hyponatremia, duration of hyponatremia, and age distribution among all 65 cases are shown in Table 1 and Figure 2. The overall morbidity among female case patients with hyponatremia was greater than 75% for all subgroups except postmenopausal women. Mortality was lowest among postmenopausal female case patients with hyponatremic encephalopathy (see Table 1 and Figure 2). The overall mortality in postmenopausal women was 33%, which is lower than the mortality in menstruant women (82%) (P < 0.02) but still higher than the mortality in men (P < 0.01). However, no significant difference was observed either in mortality between female case patients with a plasma sodium level greater than 116 mmol/L and those with a plasma sodium level less than 116 mmol/L; or between female case patients with time to development of hyponatremia greater than 24 hours and those with a
time to development less than 24 hours (P > 0.1) (see Figure 2). To determine whether any possible clinical conditions contributed to the differences in outcome among male case patients who survived, female case patients who survived, and female case patients who died or had permanent brain damage, we examined other variables possibly related to outcome, including type of surgery and associated medical conditions (see Table 2). The surviving case patients had a higher Pa02 (Figure 3), were older, were treated more often with hypertonic NaCl, and had a lower incidence of respiratory arrest before initiation of therapy (P < 0.01). Male case patients had a significantly higher rate of genitourinary surgery and female cases and a higher rate of gynecologic surgery. The most common operation among male case patients was prostate resection and among female cases, hysterectomy. Among 23 patients who had an autopsy, no pathologic cause for death was found other than hyponatremic encephalopathy. The mean brain weight was 1381 ± 14 g, about 8% above the mean control value (± SD) of 1275 ± 73 g in adult women. Thus, even when time to development of hyponatremia was up to 9 days, the brain was unable to adapt. There were two distinct patterns of pathologic findings. In the 20 patients who died within 10 days of surgery, the major finding was diffuse cerebral edema, but 14 patients also had herniation of the cerebellum or uncus (or both). Pathologic lesions in addition to cerebral edema and herniation included diffuse cerebral demyelination (n = 9); pituitary infarction (n = 5); cortical infarction in region of middle cerebral artery (n = 2); brain stem infarction (n = 2); lateral pontine demyelination (n = 2); hemorrhage or necrosis (or both) of the occulomotor (third cranial) nerve (n = 5). The five patients with pituitary infarction all developed central diabetes mellitus and insipidus (13).
Table 1. Referred Patients with Symptomatic Postoperative Hyponatremia* Group
Male case patients Overall Time to development of hyponatremia of less than 24 hours Time to development of hyponatremia of more than 24 hours Sodium level less than 116 mmol/L Sodium level 116 to 128 mmol/L Age of more than 50 years Female case patients Overall Time to development of hyponatremia of less than 24 hours Time to development of hyponatremia of more than 24 hours Sodium level less than 116 mmol/L Sodium level 116 to 128 mmol/L Postmenopausal Menstruant
Initial Serum Sodium
Final Serum Sodium
Mortality
Brain Damage
Patients
Age
n
y
25 14
66 ± 2 64 ± 6
138 ± 1 139 ± 2
110 ± 2 104 ± 5
4 7
0 0
11
67 ± 3
136 ± 2
120 ± 2
0
0
15 10 24
64 ± 2 69 ± 4 67 ± 2
138 ± 1 137 ± 2 138 ± 1
105 ± 2 122 ± 2 111 ± 3
7 0 4
0 0 0
40 13
42 ± 2 32 ± 2
138 ± 1 138 ± 2
113 ± 1 117 ± 2
68 85
18 8
27
46 ± 3
139 ± 1
112 ± 2
59
19
19 21 12 28
46 37 63 33
138 137 137 139
107 120 108 116
63 71 33 82
26 5 33 7
±4 ± 3 ± 2 ±2
%
mmol/L
± 1 ± 2 ±3 ± 1
±2 ± 1 ± 3 ± 1
* Data presented as mean ± SE. P < 0.01 for all comparisons of corresponding male and female groups.
894
1 December 1992 • Annals of Internal Medicine • Volume 117 • Number 11
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19764/ by a University of California San Diego User on 02/21/2017
Table 2. Characteristics of Patients Who Survived and of Patients Who Experienced Morbidity or Mortality* Variables Potentially Related to Outcome
dale Suivivors
in = 24) Serum sodium, mmol/L Net fluid retention, mL/kgper day Time for hyponatremia to develop, h Pa 02 at the time of diagnosis, mm Hg Age,y Treatment with hypertonic NaCl, % Respiratory arrest before therapy, % Private hospital, % Private physician, % Availability of intensive care, % General anesthesia, % Spinal anesthesia, % Type of surgery, % Gynecology Genitourinary Orthopedic Plastic or cosmetic Gastrointestinal Neurosurgery Cardiovascular Associated medical conditions, % Coronary artery disease Benign uterine tumor Obstructive uropathy or urolithiasis Chronic obstructive pulmonary disease Cholelithiasis Occlusive vascular disease Fracture or arthropathy Chronic renal failure Diabetes mellitus Benign breast tumor Miscellaneous gynecologic disorders
112 ± 2 35 ± 2 35 ± 8 91 ± 3 66 ± 2 75 4 29 33 100 62 38 46 21 4 17 0 12 21 46 12 8 21 17 12 8 0
Female Survivors (n = 7) 115 ± 2 41 ± 5 37 ± 6 55 ± 2 58 ± 5 100 43 100 100 100 71 29
Women Who Died or Had Permanent Brain Damage (n = 33) 113 ± 2 39 ± 3 48 ± 8 31 ± 5 39 ± 2 21 100 100 100 100 88 12
Statistical Significance When Compared with Women Who Died or Had Permanent Brain Damage Male Survivors Female Survivors
P P P P P P
NS NS NS < 0.001 < 0.01 < 0.001 < 0.01 < 0.01 < 0.025 NS NS NS
P P P P
NS NS NS < 0.05 < 0.01 < 0.01 < 0.05 NS NS NS NS NS
14 14 29 14 29 0 0
45 9 12 15 12 6 0
NS P < 0.05 NS NS NS NS NS
NS NS NS NS NS NS NS
14 14 14 0 14 0 29 0 0 0 14
3 18 9 0 6 3 12 3 3 6 45
P < 0.05 NS P 0.05).
Discussion Our study shows that the likelihood of developing postoperative hyponatremia appears to be similar among men and women. Hyponatremic encephalopathy also has a similar distribution among men and women. However, our case-control study showed that menstruant women with hyponatremic encephalopathy are significantly more likely to either die or experience permanent brain damage than are either men or postmenopausal women. This increased risk does not appear to depend on either the rapidity of development or magnitude of the hyponatremia, because fatalities occurred in menstruant women with plasma sodium concentrations as high as 128 mmol/L and a time to development of hyponatremia of as great as 9 days. The data from our case-control study suggest that, with no obvious source of confounding, the patients with the greatest likelihood of permanent brain damage or death from postoperative hyponatremia are menstruant women, most of whom were undergoing elective gynecologic surgery (see Table 2). Although occult confounding variables or biases are possible in case-control research, the statistical magnitude of our findings suggests that both gender and menstrual status have a real association with morbidity from postoperative hyponatremic encephalopathy. Among the 40 women studied, the brain was exam-
ined by either computed tomography (n = 21) or magnetic resonance imaging (n = 8) when the plasma sodium concentration was 128 mmol/L or less, and all showed cerebral edema. In addition, postmortem examination was done in 23 women 1 to 9 days after surgery, and all had gross evidence of cerebral edema. Brain weight was increased by a mean of 8% above control values, which exceeds the capacity of the brain to expand within the rigid confines of the skull (about 5%) (24). Thus, the female case patients were not able to adapt to hyponatremia even, in some cases, after more than a week. All but two of the male case patients were clinically able to adapt to hyponatremia within several hours. Studies in our laboratory have shown that male rodents were better able than female rodents to adapt to hyponatremia by extruding sodium to decrease braincell osmolality (25, 26). The more effective adaptation resulted in significantly less brain swelling in hyponatremic male rodents (26). Thus, the reasons for the increased morbidity and mortality in women compared with men may include a diminished ability of the female brain to adapt to hyponatremia by limiting the amount of brain swelling. The much higher mortality from postoperative hyponatremia in menstruant women relative to that in either postmenopausal women or men may be due, at least in part, to physical factors that affect the ability of
1 December 1992 • Annals of Internal Medicine • Volume 117 • Number 11
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19764/ by a University of California San Diego User on 02/21/2017
895
Figure 3. Pa0l in surviving case patients compared with those case patients who died or had permanent brain damage. The Pa 02 was significantly greater in both the male and female survivors compared with female case patients who died or had permanent brain damage. *P < 0.01; **P < 0.05.
the brain to adapt to hyponatremia. As persons age, brain volume declines progressively, whereas skull size remains constant in adult life (27). Further, intracerebral loss of cerebrospinal fluid to decrease brain size is a major adaptive component of the volume-regulatory decrease in hyponatremia (28, 29). The percent of cerebrospinal fluid in the brain increases with age (30), and much more so in men (27). Thus, elderly persons of both genders have more room in the rigid skull for the brain to expand than do younger ones (27, 31), but the finding is more marked in men, which may explain in part why elderly men have much less morbidity from hyponatremic encephalopathy than do women. The clinical presentation of men was distinctly different from that of women. Even though the plasma sodium level in the 25 male case patients decreased from 137 to 111 mmol/L in 34 ± 8 hours, only 7 men had grand mal seizures and 2 had respiratory arrest. On the other hand, plasma sodium in the female case patients decreased essentially the same amount, but all developed symptoms, with 36 of 40 women (90%) experiencing respiratory arrest before hyponatremia was diagnosed. The differing clinical presentations appear to be a major reason why men experience so much less morbidity compared with females. In 23 of 25 male case patients, consultation was obtained and the hyponatremia diagnosed without the occurance of respiratory arrest. The more timely diagnosis usually enabled appropriate therapy to be undertaken, thus preventing the possible occurrence of respiratory arrest. The occurrence of respiratory insufficiency before therapy is a major factor militating against survival without permanent brain damage. Hyponatremia can result in both a decline of cerebral blood flow and respiratory depression (28, 31-33). In our study, P a 0 2 less than 40 mm Hg was seen in the female patients who had permanent brain damage. Such a degree of 896
1 December 1992 • Annals of Internal Medicine
hypoxemia has been associated with coma and permanent brain damage in healthy humans without hyponatremia (34-36). Preliminary data from our laboratory show that in laboratory animals, the combination of systemic hypoxemia and hyponatremia is far more deleterious than either factor alone because hypoxia produces several additional maladaptive effects in the hyponatremic brain (37). We did not directly study whether encephalopathy could be prevented. It is clear, however, that when menstruant women develop hyponatremic encephalopathy, their risk for death or permanent brain damage is significantly greater than that of either men or postmenopausal women. This increased risk does not depend on either the rapidity of development or the magnitude of the hyponatremia. Thus, every reasonable effort should be made to avoid hyponatremia in such patients. Elevated plasma levels of vasopressin are essentially a universal postoperative occurrence in the initial 2 to 4 postoperative days (1, 38, 39), so it may be important to avoid the use of hypotonic intravenous solutions in the immediate postoperative period. When a patient, especially a menstruant woman, is receiving hypotonic fluids and manifests either headache, emesis, nausea, lethargy, or evidence of increased intracranial pressure, hyponatremic encephalopathy should be suspected and the plasma sodium level measured. These symptoms are somewhat nonspecific. However, the diagnosis is easily established by evaluation of the plasma sodium level, with virtually no risk to the patient and at minimal cost. When hyponatremic encephalopathy is diagnosed, the focus should be on prevention of brain damage or death. This can best be accomplished by initiation of treatment before occurrence of a hypoxic or anoxic episode, such as respiratory arrest (9, 20, 32, 33). Acknowledgments: The authors thank Maxine A. Papadakis, MD, for assisting in the study design; Warren S. Browner, MD, MPH, and Susan M. Clarkson, PhD, for helping with the statistical interpretation of the data; and D. Subramanian, MD, for helping to evaluate the control groups. Grant Support: In part by grant RO1AG08575 from the National Institute on Aging, Department of Health and Human Services, Bethesda, Maryland, and by the Research Service of the Veterans Affairs Medical Center, San Francisco, California. Requests for Reprints: J. Carlos Ayus, MD, 4 Brompton Court, Houston, Texas 77024. Current Author Addresses: Dr. Ayus: 4 Brompton Court, Houston, Texas 77024. Dr. Wheeler: Department of Obstetrics and Gynecology, Baylor College of Medicine, 1 Baylor Plaza, Houston, Texas 77030. Dr. Ariefif: Department of Medicine (111 G), Veterans Affairs Medical Center, 4150 Clement Street, San Francisco, CA 94121.
References 1. Chung HM, Kluge R, Schrier RW, Anderson RJ. Postoperative hyponatremia. A prospective study. Arch Intern Med. 1986; 146: 333-6. 2. Guy AJ, Michaels JA, Flear CT. Changes in the plasma sodium concentration after minor, moderate and major surgery. Br J Surg. 1987;74:1027-30. 3. Helwig FC, Schultz CB, Curry DE. Water intoxication: report of a fatal human case, with clinical, pathologic and experimental studies. JAMA. 1935;104:1569-75. 4. Fraser CL, Arieff AI. Metabolic encephalopathy associated with water, electrolyte, and acid-base disorders. In: Maxwell MH, Kl-
• Volume 117 • Number 11
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19764/ by a University of California San Diego User on 02/21/2017
5. 6.
7.
8.
9. 10. 11.
12. 13.
14.
15.
16.
17.
18. 19.
20.
21. 22.
23.
eeman CR, Narins RG, eds. Clinical Disorders of Fluid and Electrolyte Metabolism. New York: McGraw-Hill; [In press]. Scott JC Jr., Welch JS, Herman IB. Water intoxication and sodium depletion in surgical patients. Obstet Gynecol. 1965;26:168-75. Arieff AI, Llach F, Massry SG. Neurological manifestations and morbidity of hyponatremia: correlation with brain water and electrolytes. Medicine (Baltimore). 1976;55:121-9. Arieff AI. Hyponatremia, convulsions, respiratory arrest, and permanent brain damage after elective surgery in healthy women. N Engl J Med. 1986;314:1529-35. Charlton AJ. Cardiac arrest during transurethral prostatectomy after absorption of 1.5% glycine. A case report and review of the literature. Anaesthesia. 1980;35:804-6. Worthley LI, Thomas PD. Treatment of hyponatraemic seizures with intravenous 29.2% saline. Br Med J. 1986;292:168-70. Zimmermann B, Wangensteen OH. Observations on water intoxication in surgical patients. Surgery. 1952;31:654-669. Harrison RH 3d, Boren JS, Robison JR. Dilutional hyponatremic shock: another concept of the transurethral resection reaction. J Urol. 1956;75:95-110. Ryder KW, Olson JF, Kahnoski RJ, Karn RC, Oei TO. Hyperammonia after transurethral resection of the prostate: report of 2 cases. J Urology. 1984;132:995-7. Fraser CL, Arieff AI. Fatal central diabetes mellitus and insipidus resulting from untreated hyponatremia: a new syndrome. Ann Intern Med. 1990;112:113-9. Tien R, Arieff AI, Kucharczyk W, Wasik A, Kucharczyk J. Hyponatremic encephalopathy: is central pontine myelinolysis a component? Am J Med. 1992;92:513-22. Andreoli TE. Disorders of fluid volume, electrolytes and acid-base balance. In: Wyngaarden JB, Smith LH, Bennett JC, eds. Cecil Textbook of Medicine. Philadelphia: W.B. Saunders; 1992:499-513. Narins RG, Krishma GG. Disorders of water balance. In: Stein JH, ed. Internal Medicine. Boston: Little, Brown and Company; 1990: 832-44. Levinsky NG. Sodium and water: hyponatremia; clinical features and diagnosis. In: Wilson JD, et al., eds. Harrison's Principles of Internal Medicine. New York: McGraw-Hill; 1991:278-89. SAS Institute. SAS Users Guide: Basics, Version 5. Cary, North Carolina: SAS Institute Inc.; 1985:1290. Ayus JC, Olivero JJ, Frommer JP. Rapid correction of severe hyponatremia with intravenous hypertonic saline solution. Am J Med. 1982;72:43-8. Ayus JC, Krothapalli RK, Arieff AI. Treatment of symptomatic hyponatremia and its relation to brain damage. A prospective study. N Engl J Med. 1987;317:1190-5. Rothman KJ. Case control data. In: Rothman KJ, ed. Modern Epidemiology. Boston: Little, Brown and Company; 1986:174-5. Kozak LJ. Hospital inpatient surgery: United States, 1983-87. Hospital Inpatient Surgery: United States, 1983-87. Hyattsville, Maryland: Department of Health and Human Services. 1989. DHHS publication No. PHS 89-1250. U.S. Department of Commerce. Health and nutrition. In: U.S. De-
24.
25.
26. 27.
28. 29.
30.
31.
32. 33.
34.
35.
36.
37.
38.
39.
partment of Commerce. Statistical Abstracts of the United States, 1990. 110th edition. Washington, DC: Bureau of the Census; 1990: 110-111. Garcia JH, Conger KA, Morawetz R, Halsey JH Jr. Postischemic brain edema: quantitation and evolution. In: Cervos-Navarro J, Ferszt R, eds. Brain Edema: Pathology, Diagnosis, and Therapy. New York: Raven Press; 1980:147-69. Fraser CL, Sarnacki P. N a + - K + ATPase pump function in male rat brain synaptosomes is different from that of females. Am J Physiol. 1989;257:E284-9. Ayus J C , Arieff AI. Brain adaptation in hyponatremic rodents: effects of gender and age [Abstract]. Clin Res. 1992;40:223A. Gur RC, Mozley PD, Resnick SM, Gottlieb GL, Kohn M, Zimmerman R, et al. Gender differences in age effect on brain atrophy measured by magnetic resonance imaging. Proc Natl Acad Sci USA. 1991;88:2845-9. Rosomoff HL, Zugibe FT. Distribution of intracranial contents in experimental edema. Arch Neurol. 1963;9:36-44. Melton JE, Nattie EE. Brain and CSF water and ions during dilutional and isomotic hyponatremia in the rat. Am J Physiol. 1983; 244:R724-32. Nattie EE, Edwards WH. Brain and CSF water in newborn puppies during acute hypo- and hypernatremia. J Appl Physiol. 1981 ;51: 1086-91. Arieff AI, Ayus J C , Fraser CL. Hyponatraemia and death or permanent brain damage in healthy children. Br Med J. 1992;304:121822. Keating J P , Schears GJ, Dodge PR. Oral water intoxication in infants. An American epidemic. Am J Dis Child. 1991;145:985-90. Sarnaik AP, Meert K, Hackbarth R, Fleischmann L. Management of hyponatremic seizures in children with hypertonic saline: a safe and effective strategy. Crit Care Med. 1991;19:758-62. Hornbein TF, Townes BD, Schoene RB, Sutton JR, Houston CS. The cost to the central nervous system of climbing to extremely high altitude. N Engl J Med. 1989;321:1714-9. Oelz O, Regard M, Wichmann W, Walavanis A, Witztum A, Brugger P, et al. Cognitive impairment, neurologic performance and MRI after repeat exposure to extreme altitude. In: Sutton JR, Coates J, Remmers J, eds. Hypoxia: The Adaptations. Philadelphia: B.C. Becker, Inc.; 1990:206-9. Shimojyo S, Scheinberg P, Kogure K, Reinmuth OM. The effects of graded hypoxia upon transient cerebral blood flow and oxygen consumption. Neurology. 1968;18:127-33. Vexler ZS, Ayus J C , Kucharczyk J, Guntupalli K, Arieff AI. Effects of hypoxia on brain adaptation in experimental hyponatremia [Abstract]. Clin Res. 1992;40:306A. Deutsch S, Goldberg M, Dripps RD. Postoperative hyponatremia with the inappropriate release of antidiuretic hormone. Anesthesiology. 1966;27:250-6. Moran WH Jr., Miltenberger FW, Shuayb WA, Zimmermann B. The relationship of antidiuretic hormone secretion to surgical stress. Surgery. 1964;56:99-108.
1 December 1992 • Annals of Internal Medicine • Volume 117 • Number 11
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19764/ by a University of California San Diego User on 02/21/2017
897
