Article Text
PDF
Abstract
Background and purpose By contrast with neurologic injury, myocardial injury associated with carbon monoxide (CO) poisoning has not been well investigated. Therefore, this study assessed features and predictors of myocardial injury in CO poisoned patients.
Subjects and methods 250 CO poisoning cases that were diagnosed and treated by the emergency department of Wonju Christian Hospital from January 2006 to February 2012 were retrospectively reviewed.
Results Fifty (20%) out of 250 patients with CO poisoning developed myocardial injury. Among those with elevated troponin I (Tn I), peak levels occurred at 11.0 (IQR, 4.5–18.5) h normalising by 65.0 (IQR 44.0–96.0) h. CO exposure time, and total and ICU admission length was longer (7.5 (IQR 3.7–10.0) h vs 3.0 (IQR 1.0–7.5) h, p<0.001; 3.5 (IQR 0.0–7.0) days and 0.0 (IQR 0.0–1.25) days vs 0.0 (IQR 0.0–2.0) days and 0.0 (IQR 0.0–0.0) days, p<0.001, respectively) in the myocardial vs non-myocardial injury group. The predictors of myocardial injury were male gender, Glasgow Coma Scale (GCS) ≤14, and CO exposure time ≥2 h (OR (95% CI) of 3.341 (1.171 to 9.531), 9.920 (3.763 to 26.150), and 7.743 (1.610 to 37.238), respectively).
Conclusions Myocardial injury developed in 20% of CO poisoned patients. Time to normalisation and of peak Tn I level in elevated Tn I group was 65.0 (IQR 44.0–96.0) h and 11.0 (IQR 4.5–18.5) h. Presence of myocardial injury was associated with poorer prognosis. Predictors of myocardial injury included male gender, GCS of 14 or less, or CO exposure times greater than 2 h.
- environmental medicine
Statistics from Altmetric.com
Introduction
In the USA, each year 40 000 patients are hospitalised with carbon monoxide (CO) poisoning with reported death rates from 0% to 31%.1–5 CO poisoning is the most common cause of poisoning-related deaths and complications; 500 individuals die every year from unintentional CO poisoning, and the number of those dying from intentional CO poisoning has increased by more than 5–10 times.1 ,6–8
Owing to the recent economic recession in South Korea, use of coal is on the rise, and the rate of CO poisoning as a method of suicide has increased, including some celebrity cases. Despite the increased demand, availability of hyperbaric oxygen (HBO) therapy has been reduced.
Toxicities of CO manifest through tissue hypoxia and direct cell damage. CO competes with oxygen for haemoglobin binding; affinity of haemoglobin for CO is 250 times higher than that for oxygen leading to reduction in oxygen delivery to tissues and ensuing cellular hypoxia.1 CO poisoning, therefore, is more damaging to organs that require more oxygen such as the brain, heart, muscle and kidney.
Most who survive CO poisoning show no neurologic symptoms, however, as documented in many studies, neurologic sequelae are possible, including severe neuropsychological deterioration, delayed encephalopathy, post-CO poisoning syndrome and Parkinson's disease.2 ,9 ,10 Myocardial injury, another well known major complication of CO poisoning, can be mediated by CO-Hb-induced hypoxia, direct damage to myocardial cell respiration and to coronary arteries.11–15 However, there are few studies on myocardial injury in the context of CO poisoning, and they have documented ECG changes, myocardial dysfunction and a case of myocardial infarction.16–22 In particular, the incidence of myocardial injury in patients with CO poisoning in South Korea is unknown.
Therefore, we evaluated the prevalence and characteristics of the myocardial injury group in a set of individuals who had been diagnosed with CO poisoning, including times to normalisation and of peak troponin I (Tn I) level in elevated Tn I subgroup.
Patients and methods
Patients
The study population included 281 patients who were diagnosed with CO poisoning from January 2006 to February 2012 in the emergency department (ED) of Wonju Christian Hospital, Wonju College of Medicine, Yonsei University. From 2006 to 2011, the number of patients recruited in our hospital ED was between 28 000 and 37 000.
Exclusion criteria were age less than 18 years (none of the patients screened), previous coronary disease history (eight patients), Tn I not checked (19 patients) and ECG not obtained (two patients), and arrest on arrival at ED (two patients). Therefore, 250 patients were included in the study.
Methods
Data were retrospectively collected from medical records and reviewed. The following parameters were assessed: age, gender, cause of poisoning, CO exposure time, past history, smoking status, initial symptoms, initial mental status (Glasgow Coma Scale (GCS)), initial vital signs and number of HBO therapy sessions. Arterial blood gas, lactate, complete blood count (CBC), and levels of CO-Hb, cardiac enzymes (creatine kinase total and myocardial band (MB): CK and CK-MB; and Tn I), B-type natriuretic peptide (BNP), aspartate aminotransferase (AST), and alanine aminotransferase (ALT) were evaluated. ECG obtained at the time of CO poisoning diagnosis was classified as normal sinus rhythm (NSR), sinus tachycardia, ischaemic ECG changes and non-specific ST-T segment changes.
Myocardial injury was defined as elevation of Tn I levels (ie, ≥0.7 ng/ml) and/or ischaemic ECG changes including new ST-segment elevation greater than 1 mm, new ST-segment depression greater than 0.5 mm, or T-wave inversion greater than 2 mm.23
Indications for HBO included altered mental state, seizures, focal neurologic deficits, ischaemic chest pain, ECG changes, newly developed arrhythmia or hypotension regardless of CO-Hb concentration, and CO-Hb concentration greater than 40% or CO-Hb concentration greater than 25% plus a history of cardiovascular or cerebrovascular disease, 60 years of age or older, 2 h or more of CO exposure, and Hb level less than 10 g/dl. However, we could not perform HBO for patients with unstable vital signs, or who were intubated because of inability to monitor patients inside the tank.
ICU and total admission length, and mortality were investigated to compare patient prognosis.
Statistical analysis
Statistical analyses were performed using SPSS software for Windows (V.18.0K, SPSS Inc, Chicago, Illinois, USA). Nominal data are presented as frequencies and percentages, and continuous variables as median and IQR. For nominal variables, χ2 test and Fisher's exact test were used for comparison of nominal variables, while the Mann–Whitney test was used for continuous variables. For multivariate analysis, logistic regression analysis was used. A p value less than 0.05 was considered statistically significant.
Results
General characteristics of patients with CO poisoning
A total of 250 patients were selected for this study. There were 129 male patients (51.6%). Age ranged from 18 to 87 years with a median of 51 (IQR 37–68) years; 42 patients (16.8%) had intentionally exposed themselves to CO. Median CO exposure time was 4.0 (IQR 1.0–8.0) h, and initial CO-Hb level was 13.55% (IQR 4.90–27.50).
A total of 66 patients (26.4%) had a smoking history, and there were 31 patients (12.4%) with diabetes, 69 patients (27.6%) with hypertension and one patient (0.4%) with hyperlipidaemia. Most common symptoms at initial presentation included headache (83 patients, 33.2%), dizziness (82 patients, 32.8%), altered mental status (80 patients, 32.0%), and dyspnoea (41 patients, 16.4%). Other symptoms included chest pain, nausea, vomiting, seizure, blurred vision, paralysis, aphasia and memory loss.
In ECG, NSR (136 patients, 54.4%) was the most common finding followed by sinus tachycardia (38 patients, 15.2%); ischaemic ECG changes were seen in nine patients (3.6%). Median GCS was 15 (IQR 13–15) (table 1).
General characteristics and laboratory findings of CO poisoning patients
Median CK-MB and Tn I level was 1.63 (IQR 0.68–5.35) ng/ml and 0.017 (IQR 0.006–0.101) ng/ml, respectively. In arterial blood gas analysis, arterial pH and base excess (BE) was 7.41 (IQR 7.38–7.43) and −2.50 (IQR −4.77–−0.60) mmol/l, respectively. Lactate level was 2.06 (IQR 1.44–3.94) mmol/l. In CBC analysis, WBC was 9200 (IQR 6200–12 880) and Hb level was 14.0 (IQR 12.7–15.2) g/dl. AST was 28.0 (IQR 22.0–45.0) U/l and ALT was 23.0 (IQR 15.0–31.0) U/l (table 1).
Complications after CO poisoning were rhabdomyolysis (37 patients, 14.8%), burn (seven patients, 2.8%), pneumonia (five patients, 2.0%), and acute kidney injury (AKI) (two patients, 0.8%). Median total admission length was 0 (IQR 0–3) days. one patient (0.4%) died (table 1).
Characteristics according to the presence or absence of myocardial injury
Tn I elevation (ie, level ≥0.7 ng/dl) was seen in 43 patients (17.2%) and ischaemic ECG changes were seen in nine patients (3.6%) from ER visit to discharge. Among ischaemic ECG changes, ST-segment elevation was seen in two patients (0.8%), ST-segment depression in four patients (1.6%), and T-wave inversion in three patients (1.2%). Among patients with myocardial injury, two showed concurrent ECG changes and Tn I elevation. Thus, a total 50 patients (20.0%) showed signs of myocardial injury.
In the subgroup of patients with elevated Tn I level, elevation was seen at presentation in 31 patients (72.1%), and after admission in 12 patients (27.9%). Time to normalisation and of peak Tn I level in this subgroup was 65.0 (IQR 44.0–96.0) h and 11.0 (IQR 4.5–18.5) h, respectively (table 2). Median Tn I level at admission and on days 1–4 was 1.216 (IQR 0.673–2.675) ng/dl, 2.977 (IQR 1.375–5.498) ng/dl, 1.648 (IQR 0.685–3.570) ng/dl, 0.973 (IQR 0.737–2.116) ng/dl, and 0.533 (IQR 0.310–1.192) ng/dl (figure 1).
Time of normalisation and peak level of troponin I in troponin I elevated group
Daily pattern of troponin I in patients with elevated troponin I levels.
There were differences between myocardial versus non-myocardial injury groups in terms of intentionality (12 patients (27.9%) vs 30 patients (15.5%) with intentional poisoning in the myocardial vs non-myocardial injury group, p=0.053) and length of exposure to CO (7.5 (IQR 3.7–10.0) h vs 3.0 (IQR 1.0–7.5) h in the myocardial vs non-myocardial injury group, p<0.001), although only the latter differences achieved statistical significance (table 3).
General characteristics and laboratory findings of CO poisoning patients according to myocardial injury
Among initial symptoms, altered mental status was apparent in 35 patients (70.0%) and 45 patients (22.5%) in the myocardial and non-myocardial injury groups (p<0.001), respectively. Initial GCS was 11 (IQR 7–15) and 15 (IQR 15–15) in the myocardial and non-myocardial injury groups, respectively (p<0.001). None versus 15 (7.5%) patients had chest pain in the myocardial versus non-myocardial injury group, respectively (p=0.047) (table 3).
Initial CK-MB, Tn I, BNP (assessed in 67 patients, 26.8%) and CK level was significantly higher in the myocardial versus non-myocardial groups (12.78 (IQR 6.29–25.33) ng/ml vs 1.14 (IQR 0.60–2.51) ng/ml, p<0.001; 0.94 (IQR 0.36–2.21) ng/ml vs 0.01 (IQR 0.01–0.03) ng/ml, p<0.001; 68.7 (IQR 30.4–134.0) pg/dl vs 16.6 (IQR 8.9–56.7) pg/dl, p=0.002; 717.0 (IQR 187.5–3331.5) U/l vs 124.0 (IQR 78.5–216.7) U/l, p<0.001 in the myocardial vs non-myocardial injury group, respectively). But, there were no difference between myocardial versus non-myocardial injury groups in terms of elevated BNP (>100 pg/ml) patients (four patients (26.7%) vs seven patients (13.5%) with elevated BNP (>100 pg/ml) patients in the myocardial vs non-myocardial injury group, p=0.248) (table 3).
In initial arterial blood gas analysis, PaO2 was 93.7 (IQR 73.8–133.0) mm Hg and 103.9 (86.4–138.1) mm Hg in the myocardial and non-myocardial injury group, respectively (p=0.065). BE was −4.8 (IQR −7.5–−1.5) mmol/l and −2.0 (IQR −3.9–−0.5) mmol/l in the myocardial and non-myocardial injury group, respectively (p<0.001). Lactate level was 3.84 (IQR 1.88–6.36) mmol/l and 1.90 (IQR 1.35–3.23) mmol/l in the myocardial and non-myocardial injury group (p<0.001) (table 3).
In initial CBC, WBC was 12 340 (IQR 7225–17 285) cells/ml and 8800 (IQR 6212–11 712) cells/ml in the myocardial and non-myocardial injury group, respectively (p=0.001). AST was 46 (IQR 32–85) U/l and 26 (IQR 21–40) U/l in myocardial and non-myocardial injury group, respectively (p<0.001) (table 3).
There were differences in complication rates between both groups. Rhabdomyolysis, pneumonia and AKI were more frequent in the myocardial versus non-myocardial injury group (24 patients (54.5%) vs 13 patients (8.4%), p<0.001; four patients (9.1%) and one patient (0.6%), p=0.009; and two patients (4.5%) and 0 patients (0%), p=0.048 in myocardial and non-myocardial injury groups, respectively) (table 3).
Predictors of myocardial injury
Among parameters studied, predictors of myocardial injury were male gender (OR 3.341, 95% CI 1.171 to 9.531), GCS≤14 (OR 9.920, CI 3.763 to 26.150), and CO exposure time greater than 2 h (OR 7.743, CI 1.610 to 37.238, all p<0.05) (table 4).
Predictors of myocardial injury
The relationship between the presence of myocardial injury and prognosis
Total admission length was 3.5 (IQR 0.0–7.0) days and 0.0 (IQR 0.0–2.0) days in the myocardial and non-myocardial injury group, respectively (p<0.001). ICU admission length was 0.0 (IQR 0.0–1.25) days and 0.0 (IQR 0.0–0.0) days in the myocardial and non-myocardial injury group (<0.001). One patient (2.0%) died in the myocardial injury group while there were no deaths in the non-myocardial injury group (p=0.203) (table 3).
Discussion
In general, myocardial injury secondary to CO poisoning is apparent as ECG changes or cardiac enzyme level elevation. As in the study by Satran et al, in this study, we defined myocardial injury as ECG changes or cardiac enzyme elevation; however, in contrast with Satran et al,23 CK-MB level was excluded due to its low specificity.
Satran et al23 showed that myocardial injury was apparent in 37% of patients with moderate to severe CO poisoning who underwent HBO therapy. By contrast, in this study, myocardial injury was found in patients with mild poisoning, namely not indicated for HBO therapy. The lower prevalence of myocardial injury of 20% in this study as compared with that in Satran et al's study might reflect the fact that this study included patients with lesser severity of poisoning, and that CK-MB was excluded from the definition of myocardial injury due to its low specificity. The latter differences may also account for the finding of 17.2% Tn I level elevation in our study versus 35% CK-MB or Tn I elevation in Satran et al.
ECG analysis showed NSR in 136 patients (54.4%), sinus tachycardia in 38 patients (15.2%), and ischaemic ECG changes in 18 patients (3.6%). Proportion of patients with ischaemic ECG changes was relatively lower compared with other studies. For instance, the study by Sahin et al reported NSR, sinus tachycardia and ischaemic ECG changes in 48.2%, 26.5% and 14.4% of patients, respectively.8
In this study, both Tn I elevation and ischaemic ECG changes were only seen in two patients (0.8%). Therefore, to determine myocardial injury in patients with CO poisoning, we have to evaluate both ECG and Tn I.
Kalay et al24 reported that myocardial dysfunction assessed by echocardiography mostly normalised after 24 h in patients with decreased myocardial function. However, there are no reports on normalisation time and peak level time for elevated Tn I in CO poisoning. In this study, normalisation times of elevated Tn I was longer than those assessed by echocardiography in the study by Kalay et al.24
In this study, although the difference did not achieve statistical significance, there was a numerically higher proportion of patients after intentional poisoning in the myocardial injury group. CO exposure times were also longer in the myocardial injury group as had been reported by Kalay et al.24 In their study, CO exposure time and the initial CO-Hb concentration was related to myocardial damage and dysfunction.24 However, in this study, CO-Hb was not significantly different between groups. To compare between initial CO-Hb concentration and myocardial injury, the half-life of CO should be taken into account. The normal half-life of CO in the atmosphere is about 6 h, and oxygen supply shortens the half-life of CO. Because this was not considered in this study, it was difficult to assess the relationship between initial CO-Hb level and myocardial injury in both groups.
Altered mental status and lower GCS were more frequent in the myocardial injury group, which is consistent with previous observations of the brain and heart being more readily prone to damage by CO poisoning. However, in this study, chest pain was more frequent in the non-myocardial injury group rather than myocardial injury group. Therefore, it would appear that chest pain is not a good indicator of myocardial injury in CO poisoning patients.
The higher lactate levels in the myocardial injury group documented in this study reflect tissue hypoperfusion making the heart more vulnerable to injury.
Kalay et al reported that high BNP levels were related to low left ventricular ejection fraction, or to direct effects of CO poisoning.13 Although we found higher BNP levels in the myocardial injury group, this difference in BNP is clinically not important because both median values were within the normal range. And there were no differences between myocardial vs non-myocardial injury groups in terms of elevated BNP (>100 pg/ml) patients. We did not directly evaluate myocardial function by echocardiography.
There were also differences in complication rates between groups. Rhabdomyolysis, pneumonia and AKI developed more often in the myocardial injury group. There may be difficulties of interpreting significance of elevated cardiac enzymes in the context of rhabdomyolyis. Thus, we have excluded CK-MB and other enzymes, which are important in interpreting the definition of myocardial injury, due to its low specificity unlike other studies. In the myocardial injury group, aspiration pneumonia due to altered mental status and nosocominal pneumonia may be more common than in the non-myocardial injury group.
In this study, predictors of myocardial injury were male gender, GCS≤14, and CO exposure time greater than 2 h. The reason, which was the selection of male gender as predictor of myocardial injury, is that maybe there was a difference between men and women in terms of intentionality (29 patients (24.0%) vs 13 patients (11.2%) with intentional poisoning in the male vs female group, p=0.011). In this study, CO exposure time was divided into 2-h intervals because the receiver operating characteristic curve according to CO exposure time and presence of myocardial injury showed that the 2–4 h interval has higher sensitivity and specificity. In Satran et al,23 male gender, GCS≤14 and hypertension were positive predictive factors of myocardial injury. Our study included a broader spectrum of CO exposure intensity.
To compare prognosis between groups, we analysed total and ICU admission length and mortality. Henry et al25 had reported that myocardial injury is an important predictor of mortality in moderate to severe CO poisoning and recommended assessing for myocardial injury in CO poisoned patients. However in this study, death rate did not differ between groups even though hospital and ICU stay was longer in the myocardial injury group.
The current study is limited by its retrospective design and by inclusion of only one hospital. Not all relevant assessment parameters could be included, and large multicentre studies, including more comprehensive laboratory, imaging and other assessments are warranted.
Conclusions
Myocardial injury developed in 20% of CO poisoned patients. Time to normalisation and of peak Tn I level in elevated Tn I group was 65.0 (IQR 44.0–96.0) h and 11.0 (IQR 4.5–18.5) h. Presence of myocardial injury was associated with poorer prognosis including more frequent occurrence of other complications (rhabdomyolysis, pneumonia, burn and AKI) and longer hospital stays. Predictors of myocardial injury included male gender, GCS of 14 or less, or CO exposure times greater than 2 h.
References
Footnotes
-
Contributors The conception and design of the study: HK Acquisition of data: YSC, KCC, OHK Drafting the article: YSC, HK Revising draft critically for important intellectual contents: KHL, SOH Final approval of the version: Hyun Kim
-
Competing interests None.
-
Ethics approval Wonju College of Medicine, Yonsei University, Wonju, Republic of Korea.
-
Provenance and peer review Not commissioned; externally peer reviewed.