Published ahead of print on May 11, 2007, doi:10.1164/rccm.200701-026OC
© 2007 American Thoracic Society doi: 10.1164/rccm.200701-026OC
Carbon Monoxide PoisoningRisk Factors for Cognitive Sequelae and the Role of Hyperbaric Oxygen1 Department of Internal Medicine, Pulmonary and Critical Care Division, Intermountain Healthcare, Salt Lake City, Utah; 2 Department of Internal Medicine, University of Utah School of Medicine, Salt Lake City, Utah; 3 Institute for Health Care Delivery Research, Intermountain Healthcare, Salt Lake City, Utah; and 4 Department of Psychology and Neuroscience Center, Brigham Young University, Provo, Utah Correspondence and requests for reprints should be addressed to Lindell K. Weaver, M.D., Critical Care Medicine, LDS Hospital, 8th Avenue & C Street, Salt Lake City, UT 84143. E-mail: lindell.weaver{at}intermountainmail.org
Rationale: Carbon monoxide poisoning is common and causes cognitive sequelae. Hyperbaric oxygen (HBO2) reduces cognitive sequelae incidence, but which patients may benefit from HBO2 is unclear. Objectives: Risk factor determination for 6-wk cognitive sequelae from CO poisoning and risk modification with HBO2. Methods: Patients were from a randomized controlled trial, enrolling acutely CO-poisoned patients more than 15 years of age. Patients eligible but not enrolled in the randomized trial, and not receiving HBO2, were followed during the study interval. In patients not receiving HBO2, we performed univariate analyses including risk factors identified by randomized trial subgroup analyses. A multivariable analysis was performed using univariate results with and without HBO2. Measurements and Main Results: In 163 patients not receiving HBO2, 68 (42%) manifested sequelae. Risk factors for sequelae from subgroup analyses were loss of consciousness, age of 36 years or more, and carboxyhemoglobin levels greater than or equal to 25%. By univariate analyses, risks for sequelae were age of 36 years or more (odds ratio [OR], 2.6; 95% confidence interval [CI], 1.3–4.9; P = 0.005), and exposure intervals greater than or equal to 24 hours (OR, 2.4; 95% CI, 1.2–4.8; P = 0.019). Including 75 patients receiving HBO2, cognitive sequelae was reduced in patients age of 36 years or more (OR, 0.3; 95% CI, 0.2–0.6; P < 0.001). Exposure intervals greater than or equal to 24 hours are an independent risk factor for sequelae (OR, 2.0; 95% CI, 1.0–3.8; P = 0.046). Conclusions: HBO2 oxygen is indicated for patients with acute CO poisoning who are 36 years or older or have exposure intervals greater than or equal to 24 hours. In addition, subgroup analyses support that patients with loss of consciousness or higher carboxyhemoglobin levels warrant HBO2.
Key Words: carbon monoxide poisoning • hyperbaric oxygen therapy • cognitive outcome • neuropsychological outcome
Carbon monoxide poisoning is common, and causes cognitive sequelae (1). A recent double-blind clinical trial demonstrated a 46% reduction in cognitive sequelae 6 weeks after poisoning in patients treated with hyperbaric oxygen (HBO2) therapy compared with normobaric oxygen (1). An editorialist stated, "Hyperbaric oxygen should be the standard of care for acute CO poisoning" (2). Selection criteria for HBO2 in patients with acute CO poisoning are empiric. The accompanying editorial to our randomized trial (1) recommended that patients with a carboxyhemoglobin (COHb) level greater than 25%, or a base excess lower than –2 mmol/L, should receive HBO2 (3). Others recommend HBO2 if the COHb level exceeds 25% (4, 5) or 40% (6), regardless of signs or symptoms. In addition, loss of consciousness due to CO poisoning is an independent criterion for HBO2 (5–7). Clearly, there are differing opinions regarding which CO-poisoned patients should receive HBO2 (5). We reasoned that if we could identify available clinical criteria that were linked with higher risk for poisoning-related cognitive sequelae, and if HBO2 reduced that risk, then we could use that information to make recommendations regarding who should receive HBO2. In the present study, we assessed risk for cognitive sequelae in patients we have followed prospectively who were not treated with HBO2 (both randomized-trial [1] and nonrandomized-trial patients). We then assessed the risk-reduction in those patients treated with HBO2 from our randomized trial (1). To emphasize the importance of reducing the cognitive sequelae incidence with HBO2, we also assessed the magnitude of cognitive dysfunction in those patients with cognitive sequelae. Some of the results of this study have been previously reported in the form of an abstract (8).
Data from patients enrolled in our randomized clinical trial of acute CO poisoning (1) and nonrandomized-trial patients were included in this study. Nonrandomized-trial patients were derived from those eligible for our randomized trial, but who declined enrollment (1), or from those who were ineligible for the randomized trial because they presented more than 24 hours after poisoning (Table 1). The prime end-point of our randomized trial was cognitive sequelae 6 weeks after poisoning. (1) None of the nonrandomized-trial patients received HBO2.
Patients were eligible for study enrollment if they had a documented CO exposure (elevated COHb levels or measured elevations in ambient CO concentrations), and any of the following: loss of consciousness, confusion, headache, malaise, fatigue, forgetfulness, dizziness, visual disturbances, nausea, vomiting, cardiac ischemia, or metabolic acidosis (calculated base excess < –2.0 mmol/L, or a blood lactate level > 2.5 mmol/L). Nineteen percent (17/91) of nonrandomized-trial patients were not treated with supplemental oxygen because their COHb levels were normal, due to delay in presentation for medical care. Poisoning was confirmed by evidence of a CO poisoning source, the presence of symptoms, and no other plausible explanation for their symptoms. Patients were recruited from Utah, Idaho, and Wyoming from November 1992 through February 1999. The institutional review board at LDS Hospital approved this prospective outcome study. Written informed consent was obtained from patients or their surrogates before enrollment.
Data Collection A battery of neuropsychological tests identical to that used in our randomized trial (1) consisting of general orientation, digit span (9), Trail Making (Parts A and B) (10), digit-symbol (9), block design (9), and story recall (11) subtests was administered initially and at 6 weeks, 6 months, and 12 months. All tests were reliable and valid; were administered using standardized formats (12) in private, quiet examination rooms; and were converted to demographically corrected standardized T-scores (mean = 50; standard deviation = 10) (13, 14). A priori, cognitive sequelae were considered to be present if any 6-week neuropsychological subtest T-score was greater than two standard deviations below the mean, or if two or more subtest T-scores were each greater than one standard deviation below the mean. If the patient complained of memory, attention, or concentration difficulties, the required neuropsychological test decrement was decreased to greater than one standard deviation below the mean of demographically corrected standardized T-scores on any one subtest (13, 14). The magnitude of cognitive impairment in patients with cognitive sequelae at 6 weeks, 6 months, and 12 months after CO poisoning was assessed by summing the number of standard deviations greater than one standard deviation below the mean for all subtest T-scores (15). The oxygen therapy of randomized trial patients was described previously (1, 16). Briefly, patients breathed nonrebreather reservoir facemask oxygen followed by HBO2 or continued facemask oxygen. Intubated patients breathed 100% oxygen before and during HBO2 or the normobaric oxygen sessions. The mean duration of total high concentration oxygen therapy through chamber session one was 6.9 hours for the normobaric oxygen group and 7.0 hours for the HBO2 group (1, 16). The patients treated with HBO2 were initially compressed to 3.0 atmospheres absolute (atm abs) (304 kPa) for 50 minutes, followed by 60 minutes at 2.0 atm abs (203 kPa). In 6- to 12-hour intervals, they received two additional HBO2 sessions at 2.0 atm abs for 90 minutes. Five-minute air breathing periods were provided every 25 minutes while at 3.0 atm abs and every 30 minutes while at 2.0 atm abs (1). Eighty-one percent (74/91) of nonrandomized-trial patients were treated with supplemental oxygen. Our study prime endpoint was the incidence of cognitive sequelae at 6 weeks. Also, we reported CO-related cognitive sequelae at 6 months and 12 months after poisoning. A priori we assumed that cognitive sequelae developing after 6 weeks would not be caused by CO poisoning (1, 17–20).
Statistical Analysis
Fisher's exact tests were conducted on categorical factors to determine if the incidence of cognitive sequelae varied across factor levels. For continuous factors, two-sampled t tests were performed to determine if mean values with and without cognitive sequelae were different. Skewed factors were log-transformed before conducting two-sample t tests, and geometric means were reported. Factors statistically significant at a 0.1 two-sided significance level, and factors often used to make treatment decisions for acute CO poisoning (e.g., loss of consciousness, initial COHb levels, and initial base excess measurements), were analyzed further. The initial COHb level and base excess are continuous factors, which are used for HBO2 treatment decisions. They were categorized as COHb level less than 25% or higher than or equal to 25% and base excess lower than –2 mmol/L or greater than or equal to –2 mmol/L. (3) Analysis on each statistically significant continuous factor (P < 0.1) was performed to assess if a factor would be more appropriately represented as a categorical variable. For this analysis, the factor was categorized into intervals with similar number of cases. Logistic regression was performed with 6-week cognitive sequelae as the dependent variable and the newly categorized factor as the independent variable. The coefficients from the logistic regression model versus the midpoint of the categorized intervals were plotted. A factor remained as continuous if its associated plot was linear; otherwise, final categories were determined by reviewing the plot of sequelae rates across categories. In addition, receiver operating characteristic (ROC) curves confirmed were created to confirm the appropriateness of a categorized factor. Multivariable logistic regression using the backward stepwise likelihood ratio technique was performed on the remaining factors. From this multivariable analysis we retained potential risk factors with P < 0.05. The multivariable analysis was then repeated for these identified risk factors along with their two-way interactions. This latter analysis determined those risk factors and interactions associated with 6-week cognitive sequelae (P < 0.05). Of the univariate factors retained in the multivariable analyses, if two or more of these factors were highly associated, then only one factor was selected for further modeling to avoid collinearity. To determine if HBO2 reduced risk for cognitive sequelae, we performed separate multivariable logistic regression using the backward stepwise likelihood ratio technique on the entire sample (n = 238) with each identified risk factor, the specific therapy (HBO2 versus no HBO2), and the interaction between therapy and risk factor as independent variables. All risk factors and risk factor–therapy interaction from each multivariable analysis with P < 0.1 were retained. A final multivariable logistic regression analysis using the retained risk factors and risk factor–treatment interactions was performed. Risk factors, therapy, and interactions with P < 0.05 were considered significant and used to determine efficacy of HBO2. All P values are two-sided.
A total of 238 patients were enrolled in this study; 147 patients were enrolled in the randomized trial, 75 of whom received HBO2 (1), 72 normobaric oxygen (1), and 91 who were eligible for, but were not enrolled in, the randomized trial (Table 1). Of the 238 patients, cognitive sequelae were found in 37% (87/238) at 6 weeks, 26% (54/204) at 6 months, and 17% (35/211) at 12 months after acute CO poisoning. (The reason we had more patients follow-up at 12 months compared with 6 months was award of a grant that funded 6- and 12-month follow-up. Grant funding was received after some patients had passed their 6-month follow-up evaluation, but not their 12-month evaluation.) Baseline characteristics of the 75 patients treated with HBO2 (1), those not treated with HBO2 (n = 163), and the total group of patients (n = 238) are reported in Table 3. Of the 163 patients not treated with HBO2, 42% (69/163) had cognitive sequelae at 6 weeks, 30% (44/146) at 6 months, and 18% (27/149) at 12 months. Of the group treated with HBO2 (n = 75), 24% (18/75) had cognitive sequelae at 6 weeks, 17% (10/58) at 6 months, and 14% (9/62) at 12 months (1). Of these 163 patients not treated with HBO2, there was no difference in cognitive sequelae by treatment group (nonrandomized versus randomized normobaric oxygen). In the 146 patients treated with oxygen (but not HBO2), 60 (41%) had 6-week cognitive sequeale. In the 17 patients that were not treated with supplemental oxygen, 9 (53%) had 6-week sequelae.
In patients with cognitive sequelae, the magnitude of the cognitive impairment 25 to 75% interquartile range was one to four standard deviations below the mean (median 2.0) of demographically corrected T-scores (mean of 50, stand deviation of 10; Figure 1). The magnitude of cognitive impairment was the same regardless of patient group (hyperbaric or normobaric oxygen or nonrandomized trial). There was no difference in the magnitude of cognitive sequelae at 6 weeks, 6 months, or 12 months after poisoning.
Risk Factors for Cognitive Sequelae without HBO2 Univariate analyses performed on data from the 163 patients not treated with HBO2 identified the following potential risk factors for 6-week cognitive sequelae: age of 36 years or more, CO exposure interval 24 hours or longer, intermittent exposure, and initial memory complaints (Table 4). Eighty-nine percent (41/46) of patients with CO exposure interval 24 hours or longer had intermittent CO exposures.
The following were not risk factors for 6-week cognitive sequelae: sex, education level, poisoning etiology, if work-related, interval between the end of CO exposure and clinical oxygen therapy, duration of clinical oxygen, loss of consciousness, initial COHb level, estimated COHb level when CO poisoning ended, hypertension, history of psychological disorder, and initial symptoms of headache, nausea or vomiting, weakness, malaise, or lethargy, and dizziness. Results did not change when subsequent analyses omitted: (1) 33 patients with COHb levels less than 10%, (2) 26 with suicide attempt, or (3) 46 with CO exposure duration longer than 24 hours.
Multivariable Logistic Regression without HBO2 Only age greater than or equal to 36 years (odds ratio [OR], 2.6; 95% confidence interval [CI], 1.3–4.9; P = 0.005), and CO exposure interval 24 hours or longer (OR, 2.4; 95% CI, 1.2–4.8; P = 0.02) remained as risk factors for 6-week cognitive sequelae.
Risk Factor Reduction with HBO2
We demonstrated previously that HBO2 reduces cognitive sequelae from acute CO poisoning, and that CO-related sequelae are common after poisoning (1). In that report, we did not identify risk factors for CO-related cognitive sequelae. In the present study, only two risk factors for cognitive sequelae were identified by univariate and multivariable analyses: older age and longer CO poisoning interval. The presence of identified factors increase the patient's risk for sequelae, but patients without these risk factors may still develop cognitive sequelae. Results indicate that symptomatic CO-poisoned patients 36 years or older should receive HBO2, regardless of loss of consciousness, initial COHb, or base excess levels. Older poisoned patients have a higher risk for cognitive sequelae, possibly analogous to risk for worse outcomes in older patients with traumatic brain injury (23). Possible age-related mechanisms that influence recovery after brain injury include apoptosis (24), and older individuals with closed head injury possessing the apolipoprotein epsilon 4 genotype (25). Longer CO exposure intervals were associated with increased risk for cognitive sequelae, in agreement with some investigations (26), but not others (27). We could not estimate the absolute CO exposure interval in some patients, yet, even if intermittent, the longer CO exposure intervals were associated with increased risk for sequelae. None of five patients with poisoning exposures 24 hours or longer treated with HBO2 manifested cognitive sequelae, but our study was underpowered to demonstrate that HBO2 reduced cognitive sequelae in patients with longer CO exposures. However, patients with longer CO exposures should be treated with HBO2, since their risk for cognitive sequelae is high. The magnitude of neuropsychological impairments in patients with cognitive sequelae is substantial and not different between patients treated with or without HBO2. The severity of the cognitive impairments underscores the importance for HBO2 to reduce the probability of cognitive sequelae. Possible beneficial mechanisms for HBO2 in CO poisoning include prevention of inflammation (28), reduced lipid peroxidation (29), and preservation of adenosine triphosphate function (30). It is conceivable that alternatives to HBO2 could be developed or tried that might prevent some of these pathophysiologic processes, such as brain inflammation. In patients 36 years or older, regardless of loss of consciousness, or initial COHb, four patients would need to be treated with HBO2 to prevent one case of 6-week cognitive sequelae. There appear to be 40,000 to 50,000 Emergency Department visits annually in the United States due to CO poisoning (31, 32). Of this number of patients, approximately half would be 36 years or older. Therefore, without HBO2, we estimate there would be at least 10,000 new cases of cognitive sequelae per year in the United States due to CO poisoning and HBO2 should reduce the number who develop cognitive sequelae by 50% (1). The subgroup analyses from our randomized trial suggest HBO2 reduces cognitive sequelae in poisoned patients with loss of consciousness, in agreement with others (5–7). This observation is supported by animal studies indicating that loss of consciousness is required for brain lipid peroxidation-induced injury (33), which is reduced by HBO2. However, multivariable logistic regression found loss of consciousness and loss of consciousness duration were not risk factors for cognitive sequelae. The anesthetic properties of CO (34) may explain why loss of consciousness may not be a risk factor for cognitive sequelae. Therefore, brain injury may be due to hypoxia in some poisoned patients, with or without loss of consciousness, and not all loss of consciousness may be attributable solely to brain hypoxia. The subgroup analyses from our randomized trial suggest that HBO2 reduces cognitive sequelae in patients with higher COHb levels. In our larger study, the initial and estimated COHb levels when CO exposures ended were not risk factors for cognitive sequelae, in agreement with other investigators (35). The COHb level may not be associated with mechanisms of injury such as lipid peroxidation (36), oxidative stress from neutrophil activation (37), immune-mediated damage (38), or premature apoptosis (39). Nevertheless, the initial COHb level is associated with putamen volume loss, but not cognitive dysfunction (40), and for that reason patients with high COHb levels may benefit from HBO2. From our randomized controlled trial, initial cerebellar dysfunction was associated with 6-week cognitive sequelae (OR, 5.71) (1). Therefore, one could reason that poisoned patients with cerebellar dysfunction should receive HBO2. However, even in patients with initial normal cerebellar function, HBO2 reduced 6-week cognitive sequelae (P = 0.05) (1), so the presence of cerebellar dysfunction should not be used as the sole criteria for HBO2. We did not have cerebellar examination results from the majority of patients from nonrandomized trials, so we did not include this factor in our analyses. We cannot specify the interval from cessation of CO poisoning to HBO2 that would convey reduced cognitive sequelae. Over 60% of patients in our randomized clinical trial (1) were treated with HBO2 in less than 6 hours from poisoning, so our data is underpowered to determine if treatment with HBO2 beyond 6 hours will also reduce cognitive sequelae. The lack of association between initial symptoms and cognitive sequelae is an important finding. It appears that CO poisoning may initiate a cascade of pathophysiological process in the brain (38), which may lead to cognitive sequelae, yet these processes are not necessarily linked to the physical expression at the time of poisoning. Contrary to expectation (36), for patients not treated with HBO2, the time from CO removal to supplemental oxygen, the concentration of inhaled oxygen, or the duration of oxygen therapy were not associated with cognitive sequelae. Since oxygen is inexpensive, readily available, and probably does not worsen outcomes, we endorse the administration of 100% normobaric oxygen to patients with acute CO poisoning pending HBO2. However, clinicians need to be aware that HBO2, not necessarily 100% normobaric oxygen, reduces cognitive sequelae. It is important to reiterate that the results we found are derived from patients enrolled in a randomized trial (1) and from poisoned patients not treated with HBO2, but followed similarly. Inferences may be strengthened by including both patients from randomized trials and those from nonrandomized trials. In addition, the inclusion of patients from nonrandomized trials makes results more generalizeable. All patients enrolled in our prior study (1) and the present study had symptomatic CO poisoning. Almost half the patients had an interval of unconsciousness and the initial mean COHb level was greater than 20%, with an estimated COHb level when poisoning stopped of 35%. In a study conducted in France, one of three patients with apparent milder poisoning appear to have persistent neurologic symptoms 1 month after poisoning (35). Comparisons of this study to ours is difficult, since they did not conduct standardized cognitive assessments and HBO2 dosing was different (35). Medical decision-making regarding HBO2 for patients with acute CO poisoning needs to be balanced by side effects of HBO2 and transport risk. The incidence of side effects and complications of HBO2 is low (41). The most common side effect of hyperbaric pressurization is middle ear pressure equalization difficulties, which occur in approximately 2% of patients (41). The risk of seizures in large series of HBO2-treated patients is approximately 0.02%. In patients with CO poisoning, the seizure rate ranged from 0.3 to 2.5%, without residua (41). With dedicated, experienced transport teams, risk of transport is extremely low (42). There were no transport-related mishaps or events in our randomized trial of HBO2 for CO poisoning (43). Our findings support that HBO2 reduces cognitive sequelae in poisoned patients that are older or have longer exposures to CO. In addition, we agree with consensus opinion that mirrors our subgroup data: that HBO2 is indicated for patients with loss of consciousness or higher carboxyhemoglobin levels.
The authors thank Susan Churchill, N.P. for study-related assistance; Kayla Deru, B.A., for secretarial assistance; Diane Haberstock for database entry and study coordination; and C. Gregory Elliott, M.D., and James Walker, Ph.D., for manuscript review. The authors appreciate the neuropsychological evaluators: Andrea Chiba, Ph.D.; Jennifer Duncan-Davis, Ph.D.; Thane Freemow, Ph.D.; Mary Beth Hart, M.S.; Shawn D. Gale, Ph.D.; Debra Johnson, Ph.D.; Shelli Kesler, Ph.D.; Jeff Long, Ph.D.; and Laurie M. Rilling, Ph.D. This study was supported by grants (247 and 305) from the Deseret Foundation, LDS Hospital.
Supported by grants (247 and 305) from the Deseret Foundation, LDS Hospital. Originally Published in Press as DOI: 10.1164/rccm.200701-026OC on May 11, 2007 Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form January 4, 2007; accepted in final form May 11, 2007
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
AT A GLANCE COMMENTARY
60 min), initial COHb level 25% or higher, and initial base excess lower than –2 mmoles/L. 
