Coma after a successful major surgical procedure is an. Postoperative Stupor and Coma

Transcription

1 ORIGINAL POSTOPERATIVE STUPOR ARTICLE AND COMA Postoperative Stupor and Coma ELSKE C. GOOTJES; EELCO F. M. WIJDICKS, MD; AND ROBYN L. MCCLELLAND, PHD OBJECTIVES: To identify predictive factors for postoperative coma or stupor and to examine the value of neuroimaging techniques in elucidating structural brain damage. PATIENTS AND METHODS: We performed a case-control study of surgical patients admitted to a Mayo Clinic affiliated hospital. We studied preoperative comorbidity, intraoperative hypotension, and postoperative data in patients with postoperative stupor or coma and compared the characteristics with control patients (surgical intensive care unit patients with neurologic consultations for other reasons). RESULTS: A total of 35 patients with stupor or coma after surgery and 31 control patients participated in this study. Comatose patients were older (P=.004) and had significantly more presurgical comorbidity (P<.001), cardiovascular surgical procedures (P<.001), and intraoperative hypotension (P=.03). Adjusted for age and comorbidity, intraoperative hypotension remained statistically significant but not after adjusting for cardiovascular surgery. Of the 34 computed tomograms obtained, 41% showed abnormal results; of the 12 magnetic resonance images obtained, 58% showed abnormal results. Both showed primarily infarctions. In 4 patients with normal computed tomographic results, magnetic resonance imaging showed multiple territorial infarctions. CONCLUSION: Prior comorbidity, older age, intraoperative hypotension, and cardiovascular surgery may predispose patients to postoperative coma. Widespread structural ischemic brain damage was often documented by neuroimaging. Metabolic causes for coma were uncommon. Mayo Clin Proc. 2005;80(3): CT = computed tomography; EEG = electroencephalography; ICU = intensive care unit; MRI = magnetic resonance imaging Coma after a successful major surgical procedure is an uncommon but devastating complication. 1 What initially appears as a delay in awakening after surgery may unfold as persistent coma. Often the prolonged effect of anesthetic agents or a metabolic derangement is held responsible for the decreased level of consciousness. When performed, neuroimaging may show abnormalities suggestive of structural brain damage. To our knowledge, no systematic study of coma or stuporous state after surgical procedures has been performed, and the causes and circumstances of postoperative coma are not well understood. We From the Department of Neurology (E.C.G., E.F.M.W.) and Division of Biostatistics (R.L.M.), Mayo Clinic College of Medicine, Rochester, Minn. Address reprint requests and correspondence to Eelco F. M. Wijdicks, MD, Department of Neurology, Mayo Clinic College of Medicine, 200 First St SW, Rochester, MN ( wijde@mayo.edu) Mayo Foundation for Medical Education and Research reviewed neurologic consultations in patients admitted to surgical wards or intensive care units (ICUs) to identify predictive factors for postoperative coma or stupor and to examine the value of neuroimaging techniques in elucidating structural brain damage. PATIENTS AND METHODS This retrospective case-control study was performed at the Mayo Clinic in Rochester, Minn, and was approved by the Mayo Foundation Institutional Review Board. We studied patients who had surgery under general anesthesia at Saint Marys Hospital, 1 of the 2 Mayo Clinic affiliated hospitals in Rochester. It contains 1157 licensed beds and 53 operating rooms. At these 2 hospitals, there are approximately 128,000 surgical cases a year. According to the 2001 and 2002 hospital statistics, there are approximately 48,000 surgical cases with general anesthesia annually. To select patients with altered mental status after general surgery with general anesthesia, we reviewed medical records of patients admitted to Saint Marys Hospital. The main database consisted of electronically entered neurologic consultations requested by surgical services from 2000 to Moreover, we identified potential cases from medical records that matched the criteria of coma and surgery and ICU using International Classification of Diseases, Ninth Revision codes through the Mayo linkage system. The records of patients were reviewed when the indication of the neurologic consultation included altered mental status or coma less than 3 weeks after surgery. We excluded seizures as a main reason for consultation, but records were reviewed for persistent changes in consciousness. Patients with postoperative cardiac arrest and resuscitation were excluded. Two patients with postoperative stupor and pituitary apoplexy have been described before 2 and were excluded. Coma or stupor was defined as a Glasgow Coma Scale score of 10 or less. If no Glasgow Coma Scale score was available, we defined coma or stupor according to the definitions proposed by Ropper. 3 Patients identified as having stupor or coma after surgery between 1996 and 2003 were compared with a control group chosen from a database of all neurologic consultations, excluding those for altered mental status in the surgical ICU between 2000 and We included every fifth consultation from this electronic listing of consultations. Control patients had a neurologic consultation for periph- 350

2 eral nerve disease, evaluation of medication, suspicion of a movement disorder, and a variety of other queries. Controls were not matched with cases for type of surgery, age, or other risk factors because we wanted to evaluate these variables for their association with coma risk. We reviewed the neurologic examination results and recorded the time from surgery. This included the pupils and their reactivity to light, corneal reflexes, gaze preference, eye movements, nystagmus or absent doll s eyes, presence of hemiparesis, or myoclonus. The type of surgery was categorized as cardiothoracic, vascular, transplantation, orthopedic, abdominal, neck, or other. Emergency surgery was defined as surgery within 24 hours of an established diagnosis of a possibly life-threatening disorder. The data of both the case and the control group, including neuroimaging studies, were collected without knowledge of the intraoperative records and outcome. We reviewed comorbidity that could influence oxygenation or perfusion, such as chronic obstructive pulmonary disease, carotid disease, previous strokes (or transient ischemic attacks), congestive heart failure, and hypertension. Intraoperative data included duration of surgery, duration of anesthesia, whether cardiac arrhythmia occurred, and whether resuscitation was needed. We defined intraoperative hypotension as a systolic blood pressure decrease of 80 mm Hg or lower. We differentiated between an incidental decrease and persistent hypotension that lasted for more than 10 minutes. Systolic blood pressure measurements were not available during use of the cardiopulmonary bypass machine. We tabulated sedative drugs that could potentially affect the postoperative level of consciousness. We reviewed and graded abnormalities on computed tomography (CT) and magnetic resonance imaging (MRI). We categorized the findings as presence of infarction either in one of the major cerebrovascular territories or in multiple territories. We noted presence of parenchymal, subdural, subarachnoid, or multiple cerebral hemorrhages and cerebral edema. Electroencephalographic (EEG) abnormalities were classified according to Young et al. 4 Laboratory values at the time of neurologic examination (or within 1 day) were tabulated. These values included aspartate aminotransferase, alanine aminotransferase, alkaline phosphatase, serum urea nitrogen, serum creatinine, serum bilirubin, magnesium, calcium, sodium, glucose, arterial PCO 2, and arterial ph. In-hospital mortality was recorded for both study groups. Data were insufficient to grade neurologic disability. We performed unadjusted statistical comparisons by use of the χ 2 test (or Fisher exact test when sample sizes were limited), t test, and Wilcoxon rank sum test. The number of cases, percentages, and P values were calculated for the nominal variables. For important variables with sufficient sample size, we calculated odds ratios and 95% confidence intervals. Means, medians, and SDs were calculated for continuous variables; when appropriate, we calculated odds ratios and 95% confidence intervals. P<.05 was considered statistically significant. The statistical analyses were performed using JMP, version (SAS Institute Inc, Cary, NC). S-PLUS, 2000 release 3 (MathSoft Engineering and Education Inc, Cambridge, Mass) was used to perform multiple logistic regression to identify possible independent predictive factors. Age, sex, and univariately significant variables were included in this model. RESULTS This study consisted of 35 patients with stupor or coma after surgery occurring between 1996 and Of these 35 patients, 28 were treated between 2000 and These 35 patients were compared with 31 control patients. The preoperative and postoperative data for the comatose patients and controls are given in Table 1. The patients with postoperative coma were older (P=.004), with a mean age of 69 years (range, years) vs a mean age of 56 years (range, years) in the control group, and had more cardiovascular surgical procedures (P<.001). Intraoperative hypotension (systolic blood pressure of 80 mm Hg) was common in both groups but reported more often in comatose patients, 80% vs 61% (P=.03), with a higher frequency of hypotensive episodes (P=.05). In addition, persistent intraoperative hypotension, defined as a systolic blood pressure of less than 80 mm Hg for more than 10 minutes, was more common in the comatose patients, but the difference was not statistically significant (P=.23). The mean number of hypotensive episodes was 4.03 (SD, 3.41) for the comatose patients and 2.61 (SD, 3.16) for the control patients (P=.05, Wilcoxon rank sum test). In this subset of patients with prior hypertension, comatose patients had significantly more intraoperative hypotension (88% vs 40%; P=.01). This difference was not statistically significant in patients without prior hypertension (82% vs 71%; P=.43). The results of multiple regression analysis are given in Table 2. Intraoperative hypotension remained statistically significant after adjusting for age and any type of comorbidity. NEUROLOGIC EXAMINATION IN COMATOSE PATIENTS Neurologic consultation was obtained within a mean of 4.19 days (median, 4 days; range, 0-34 days); however, coma was reported in the medical records within a mean of 2.83 days (median, 1 day; range, 0-17 days). In total, 20 patients were comatose, and 15 were in a stuporous state. 351

3 TABLE 1. Preoperative and Postoperative Data in Patients With and Without Postoperative Coma* Postsurgical coma or stupor Control Odds ratio (n=35) (n=31) P value (95% CI) Sex Male 22 (63) 16 (52) ( ) Female 13 (37) 15 (48) Mean (SD) age (y) 69 (14) 56 (19) ( ) Mean (SD) comorbidity 1.37 (1.06) 0.55 (0.68) < ( ) Hypertension 18 (51) 10 (32).12 Diabetes mellitus 8 (23) 2 (6).06 Congestive heart failure 8 (23) 1 (3).02 Chronic obstructive pulmonary disease 6 (17) 3 (10).38 Stroke and/or transient ischemic attack 4 (11) 1 (3).21 Carotid disease 4 (11) 0 (0).05 Type of surgery Cardiothoracic 15 (43) 0 (0) Abdominal 7 (20) 14 (45) Orthopedic 7 (20) 11 (35) Vascular 4 (11) 2 (6) Transplantation 1 (3) 2 (6) Neck 1 (3) 1 (3) Other 0 (0) 1 (3) Cardiovascular 19 (54) 2 (6) < ( ) Intraoperative hypotension 28 (80) 19 (61) ( ) Mean (SD) duration of surgery (min) (150.22) (116.32).15 Mean (SD) duration of anesthesia (min) (135.01) (250.66).48 Emergency surgery 9 (26) 3 (10) ( ) Postoperative day of neurologic examination Mean (SD) 6.12 (6.94) 4.19 (4.95).21 Median (range) 2 (1-22) 4 (0-34) *Data are number (percentage) of patients unless indicated otherwise. CI = confidence interval. Odds ratios are from unadjusted logistic regression models. Cardiothoracic or vascular surgery. Wilcoxon rank sum test. Six (18%) of 33 patients had enlarged pupils. Eleven (32%) of 34 patients had pupils that were not reactive to light. Seven (33%) of 21 patients had absent corneal reflexes. Eleven (32%) of 34 patients had a gaze preference. Abnormal eye movements were reported in 12 (39%) of 31 patients. Eight of them presented with roving eye movements and 4 with absent doll s eyes. Five (15%) of 34 patients had hemiparesis, and 7 (21%) of 34 patients displayed myoclonus. Neurologic examination was potentially confounded in 13 (39%) of 33 patients because of sedation. LABORATORY TESTS Certain laboratory test results were not available at the time of neurologic examination. Twenty of 34 patients had in- TABLE 2. Multiple Logistic Regression Model* Odds ratio (95% CI) P value Age 1.05 ( ).01 Sex 0.93 ( ).90 Any comorbidity 4.18 ( ).03 Intraoperative hypotension 5.83 ( ).01 *CI = confidence interval. creased serum urea nitrogen levels (>30 mg/dl), 15 of 34 patients had increased serum creatinine levels (>1.5 mg/ dl), and 6 of 14 patients had increased serum bilirubin levels (>1.5 mg/dl). Eight of 15 patients had increased aspartate aminotransferase levels (>50 U/L), and 4 of 15 had increased alanine aminotransferase levels (>50 U/L). Alkaline phosphatase levels were increased (>150 U/L) in 6 of 14 patients. Five of 34 patients presented with hyponatremia (sodium, <134 meq/l), whereas 8 of them had hypernatremia (sodium, >145 meq/l). Hypocalcemia (calcium, <4.5 mg/dl) was found in 4 of 25 patients, and 5 of these 25 patients had hypercalcemia (calcium, >5.5 mg/ dl). No acidosis was reported, but 1 patient had an increased ph (>7.55). There was no hyperglycemia, and no abnormal values were found for serum magnesium and arterial PCO 2. EEG AND NEUROIMAGING In 12 patients with postoperative stupor or coma, EEG was performed, and all results were abnormal. Four had generalized delta or theta activity, 5 had triphasic waves, 2 had a burst suppression pattern, and 1 patient presented with a spindle coma. No epileptiform activity was found. 352

4 The results of neuroimaging studies are given in Table 3. Only 1 CT was performed in the control patients, which showed no abnormalities. All but 1 of the comatose patients had postoperative brain CT performed at a mean of 5.88 days after surgery. Eleven had evidence of infarctions, and 2 had hemorrhages. Infarctions in the territories of the middle and posterior cerebral artery and the vertebral basilar artery were seen in 2 patients each. Five patients had multiple cerebral infarctions in different territories. Hemorrhages were rare, but 1 patient receiving postoperative intravenous heparin presented with massive cerebellar and intraventricular hemorrhage, and 1 patient had caudate hemorrhage. One CT showed postanoxic cerebral edema. In 12 (34%) of 35 patients, an additional MRI was performed at a mean of 13.5 days after surgery. Abnormalities revealed on MRI were in both anterior and posterior cerebral vascular territories (Figure 1). None of these abnormalities were clearly delineated on CT, and in 4 patients with a normal CT result, the MRI showed abnormalities. One patient had possible central pontine myelinolysis. Of the patients with intraoperative hypotension, 11 (39%) of 28 patients had abnormal CT results, and 4 (57%) of 7 patients had abnormal MRI results, as opposed to 1 (20%) of 5 and 2 (67%) of 3 patients, respectively, without intraoperative hypotension. MORTALITY Sixteen (46%) of 35 comatose patients died compared with none in the control group (P<.001). Of the patients with an abnormal MRI result, 6 (86%) of 7 died vs 1 (20%) of 5 patients with a normal MRI result (P=.046). Patients who underwent MRI did not have a significantly higher mortality rate compared with those who had no MRI performed. Seven (58%) of 12 died compared with 9 (39%) of 23 without MRI (P=.28). Of the patients with an abnormal CT result, 9 (69%) of 13 died compared with 6 (29%) of 21 with a normal CT result (P=.02). Of 19 surviving patients, only 18 had awakened at the time of discharge. None of the patients awoke and later died of other medical causes. TABLE 3. Neuroimaging Findings in Comatose Patients* CT MRI (n=34) (n=12) Postoperative day Mean (SD) 5.88 (5.82) 13.5 (9.26) Median Abnormal result 14 (41) 7 (58) Infarction 11 (32) 7 (58) Cerebral artery Anterior 0 (0) 0 (0) Middle 2 (6) 0 (0) Posterior 2 (6) 0 (0) Vertebral basilar 2 (6) 0 (0) Multiple 5 (15) 7 (58) Hemorrhage 2 (6) 0 (0) Parenchymal 1 (3) 0 (0) Multiple 1 (3) 0 (0) Edema 1 (3) 0 (0) *Data are number (percentage) of patients unless indicated otherwise. CT = computed tomography; MRI = magnetic resonance imaging. DISCUSSION In this comprehensive study spanning 7 years, we found only 35 patients with postoperative coma or stupor of more than 275,000 surgical procedures performed with the patient under general anesthesia. This may be partially explained by a large proportion of patients who undergo elective surgery in which we would expect the number of neurologic complications to be low. Previous studies show that emergency surgery predisposes patients to postoperative complications. 5 Emergency surgery as a result of hemodynamic instability (eg, cardiac arrhythmia with subsequent insufficient cardiac output or inability to maintain blood pressure) or failure to oxygenate adequately may increase the risk of neurologic complications postoperatively. Although not statistically significant, we found a higher percentage of emergency surgery in comatose patients. This case-control study presents new important observations and possible predictive factors for postoperative stupor or coma. Review of intraoperative records showed that repetitive periodic decreases in blood pressure were tolerated poorly and were possibly more consequential than a persistent decrease in blood pressure. This finding FIGURE 1. Magnetic resonance images of a comatose patient after vascular repair. Multiple hyperintensities are evident on fluid-attenuated inversion recovery sequences. (The computed tomogram, not shown, showed hints of early hypodensity.) 353

5 could be explained by the ability of the cerebral vessels to autoregulate but not when repetitively challenged with marked hypotension. This failure to compensate could be particularly relevant in a patient with prior hypertension and a blood pressure cerebral blood flow curve shifted to the right. Why some, evidently most, patients tolerate this blood pressure challenge and awaken after surgery cannot be answered and requires a prospective study, preferably with serial MRI. Intraoperative hypotension was more commonly associated with an abnormal CT result but not with abnormal MRI results. This comparison of hypotension with neuroimaging should be considered tentative because of the small number of patients and does not necessarily negate the statistical robustness of the association between intraoperative hypotension and postoperative coma and stupor. Other mechanisms for infarction in multiple territories are emboli as a result of cardiopulmonary bypass or associated with clamping of the aorta. However, 2 other studies on adverse events after surgery found a low incidence of intraoperative hypotension and no relationship with outcome. 6,7 We found that comorbidity was more frequent in comatose patients compared with controls, but we have no satisfactory biological explanation why this would increase the risk. A surprising finding was the high incidence of MRI abnormalities, which raises an important practical issue. Despite use of confounding sedation (in more than a third of patients) and metabolic derangements (in more than two thirds of patients), structural damage to the brain may be present. The role of postoperative EEG in evaluating postoperative coma was small, at least in this study. One prior study of postoperative MRI confirmed the sensitivity of MRI in revealing infarction but was limited to cardiac surgery. In that study, MRI showed cerebral infarction in multiple vascular territories in patients who presented with encephalopathy, some of whom had focal symptoms. 8 Outcome was poor in patients who underwent MRI and in those who did not, suggesting that withdrawal of support was not strongly influenced by these results. Despite a 3-year, fully operational electronic registration of all in-hospital consultations, our study is possibly hampered by insufficient case ascertainment. We identified more patients in the electronic database than in the Mayo linkage system. Moreover, neurologic consultation requires a formal request by the surgical service endorsed by the attending surgeon, and it is possible that our comparatively young index patients (mean age, 69 years) are a direct consequence of early withdrawal of intensive care support in the old comatose patients. All patients had neurologic examination performed by neurology staff, but nonetheless data entry was not always complete. We were unable to exactly note the onset of postoperative coma. We found a significant age mismatch between our comatose patients and controls. This finding may be relevant because postoperative complications, at least in 1 prospective study, were seen predominantly in patients older than 70 years. 9 In our study, patient mean age was 69 years vs 56 years for the controls, but age remained statistically significant after adjusting for possible confounding factors. Our study may have practical consequences for consulting neurologists. In evaluating patients with postoperative coma, structural abnormalities should be actively sought several days after surgery. In addition, MRI should be considered despite initial reassurance from normal CT results or from possible confounding factors, such as sedation, metabolic derangements, and a nonspecifically slowed activity on EEG. Stupor or coma several days after surgery with the patient under general anesthesia is a sign of poor prognosis. Only when further studies in other tertiary care centers provide similar results can these observations (persistent coma and abnormalities on neuroimaging) become incorporated in decisions about level of care. CONCLUSION Our study found that prior comorbidity, older age, intraoperative hypotension, and cardiovascular surgery may predispose patients to postoperative coma. Neuroimaging often revealed widespread structural ischemic brain damage. Metabolic causes for coma were uncommon. REFERENCES 1. Gamil M, Fanning A. The first 24 hours after surgery: a study of complications after 2153 consecutive operations. Anaesthesia. 1991;46: Tang-Wai DF, Wijdicks EFM. Pituitary apoplexy presenting as postoperative stupor. Neurology. 2002;58: Ropper AH. Lateral displacement of the brain and level of consciousness in patients with an acute hemispheral mass. N Engl J Med. 1986;314: Young GB, McLachlan RS, Kreeft JH, Demelo JD. An electroencephalographic classification for coma. Can J Neurol Sci. 1997;24: Wolters U, Wolf T, Stutzer H, Schroder T, Pichlmaier H. Risk factors, complications, and outcome in surgery: a multivariate analysis. Eur J Surg. 1997;163: Leung JM, Dzankic S. Relative importance of preoperative health status versus intraoperative factors in predicting postoperative adverse outcomes in geriatric surgical patients. J Am Geriatr Soc. 2001;49: Ouchterlony J, Arvidsson S, Sjostedt L, Svardsudd K. Preoperative and immediate postoperative adverse events in patients undergoing elective general and orthopaedic surgery: the Gothenburg study of perioperative risk (PROPER), part II. Acta Anaesthesiol Scand. 1995;39: Wityk RJ, Goldsborough MA, Hillis A, et al. Diffusion- and perfusionweighted brain magnetic resonance imaging in patients with neurologic complications after cardiac surgery. Arch Neurol. 2001;58: Polanczyk CA, Marcantonio E, Goldman L, et al. Impact of age on perioperative complications and length of stay in patients undergoing noncardiac surgery. Ann Intern Med. 2001;134: