Correlation Between Partial Pressure of Arterial Carbon Dioxide and End Tidal Carbon Dioxide in Patients with Severe Alcohol Withdrawal

INTRODUCTION

The management of patients with alcohol withdrawal is guided by the Clinical Institute Withdrawal Assessment of Alcohol Scale, Revised (CIWA-Ar) that is scaled from 0-67.¹ The goal of therapy is to prevent adverse outcomes of alcohol withdrawal, in particular delirium tremens (DT), and to support patients on their path to long-term recovery. The CIWA-Ar score plays an important role in determining the dose and frequency of medication and assessing whether the patient needs to be admitted to the intensive care unit (ICU). Patients with a score >20 are admitted to the ICU to effectively manage their withdrawal symptoms. High doses of benzodiazepines are routinely administered to manage symptoms of psychomotor agitation and to avert advancement of withdrawal symptoms. Benzodiazepines bind to and excite the gamma aminobutyric acid receptor, leading to a decrease in neuronal activity and inducing sedation. A common adverse reaction to benzodiazepines is respiratory depression, defined as insufficient ventilation to achieve the necessary gas exchange for normal metabolic function. During such states of hypoventilation, patients retain carbon dioxide (CO₂), become hypercapnic, and experience respiratory acidosis. To monitor for this unwanted outcome, the arterial blood gas (ABG) test is the standard of care to track the partial pressure of arterial CO₂ (PaCO₂).

The ABG test provides valuable information, although its application has several limitations. ABG tests are painful, expensive, time consuming, and not obtained without risk. Capnography is rapidly replacing the ABG test as a noninvasive measure of end tidal CO₂ (ETCO₂), the maximum CO₂ concentration or partial pressure at the end of each tidal breath. To assess disease severity and response to treatment, monitoring the changes in ETCO₂ with respirations is common practice. Capnography offers practitioners immediate information about metabolism, perfusion, and ventilation or how effectively CO₂ is produced, transported back to the lungs, and eliminated by the alveoli. Capnography has proved to be a valuable tool for patients receiving general anesthesia and care in the emergency department and in the ICU. In intubated patients, capnography can help verify proper placement of the endotracheal tube, predict the success of resuscitation during cardiac arrest, and help transition patients off mechanical ventilation. However, capnography is not limited to intubated patients. In fact, in spontaneously breathing patients, capnography can help determine response to treatment during acute respiratory stress, offer prognostic signs in patients with sepsis, and determine the adequacy of ventilation in patients who are sedated. While it is known that abnormal changes in ETCO₂ levels can be found in patients with respiratory depression caused by excessive sedation,² research comparing ABG PaCO₂ to capnography for monitoring adverse respiratory events in this patient population is limited.

ETCO₂ can be monitored via mainstream or sidestream monitors. Mainstream monitors detect the concentration of air flowing directly through the endotracheal tube, whereas sidestream monitors analyze a sample of the expired air. Mainstream monitors are more accurate than sidestream monitors but are less versatile and more difficult to supply. Sidestream monitors are slightly less accurate than mainstream monitors—because the air sample is smaller—but are more versatile and easier to supply. Sidestream monitors can be used on prone and awake patients and do not require repeated sterilization.

This retrospective study was conducted to assess the value of ETCO₂ levels compared to PaCO₂ levels from ABG tests in monitoring respiratory depression secondary to benzodiazepine treatment in patients with alcohol withdrawal. Our aim was to investigate for a surrogate marker that is cost effective, rapid, and less labor intensive than the ABG test.

RESULTS

The Table displays demographics and clinical parameters for the study population (n=36). The sample was predominantly male (n=31, 86.1%), with a mean age of 46.0 years (SD=10.2). The mean CIWA-Ar score was 14.8 (SD=9.9), the mean temperature was 36.8°C (SD=2.8), the mean heart rate was 87.0 beats per minute (SD=18.0), and the mean respiratory rate was 20.0 breaths per minute (SD=7.0).

Mean differences between PaCO₂ and ETCO₂ are displayed in Figure 1 for the first 3 time points when these scores were measured. PaCO₂ was 9.4-10.3 mmHg higher on average than ETCO₂: for time 1, the mean difference was 10.3 mmHg (95% confidence interval [CI]=7.9-12.6); for time 2, the mean difference was 9.7 mmHg (95% CI=6.7-12.6); for time 3, the mean difference was 9.4 mmHg (95% CI=6.3-12.6). A statistically significant correlation between PaCO₂ and ETCO₂ was seen for time 1 (r=0.74, P<0.01) and time 3 (r=0.52, P=0.02) but not for time 2 (r=0.22, P=0.31) (Figure 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Mean differences between partial pressure of arterial carbon dioxide (PaCO₂) and end tidal carbon dioxide (ETCO₂). Mean difference scores with 95% confidence interval between PaCO₂ and ETCO₂ levels for the first 3 time points were measured.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Scatterplots of the correlation between partial pressure of arterial carbon dioxide (PACO₂) and end tidal carbon dioxide (ETCO₂) levels at time points 1, 2, and 3.

DISCUSSION

In normal healthy lungs, CO₂ is an end product of tissue metabolism that is transported to the lungs where it diffuses across the capillary membrane via perfusion. Alveolar ventilation then eliminates CO₂ from the body. This close relationship between ventilation and perfusion subsequently manifests in a constant and predictable gradient between PaCO₂ and ETCO₂ levels.^2,3 To calculate the ETCO₂–PaCO₂ gradient—normally 2-5 mmHg—an ABG test must be sampled at the same time an ETCO₂ measurement is acquired.⁴ In patients who do not have significant cardiopulmonary disorders or a ventilation/perfusion (V/Q) mismatch, ETCO₂ levels can be used as a surrogate marker for PaCO₂ levels.

In our review of the literature, we found no studies indicating that ETCO₂ levels could not be used as a surrogate marker for PaCO₂ levels in patients who are facing alcohol withdrawal symptoms and are being treated with benzodiazepines. Our hypothesis is that this cannot be disproved because although these patients are experiencing respiratory depression secondary to therapy, no V/Q mismatch is present, and therefore the ETCO₂–PaCO₂ gradient will be normal and consistent at approximately 2-5 mmHg. Other studies assessed patients with cardiopulmonary pathology in whom a V/Q mismatch was present.^3,5,6 In these patients, the ETCO₂–PaCO₂ gradient increases, and therefore correlating the two measurements is not a reliable measure of the true CO₂ level in the blood. In conditions with increased dead space ventilation, such as a pulmonary embolism, parts of the lungs are well ventilated but not perfused. On the other hand, in conditions such as atelectasis or increased alveolar fluid, areas of the lung are well perfused but not ventilated. Studies have shown that ETCO₂ levels are a poor predictor of PaCO₂ levels in these conditions of respiratory failure.^3,7

Based on a 2013 study, an emerging trend in the treatment of acute alcohol withdrawal syndrome is combined barbiturate and benzodiazepine therapy for prevention of severe DT.⁸ In that study, only 1 patient of a possible 102 patients was intubated. It is unclear if this intubation was because of respiratory depression or refractory agitation. Another study showed that escalating doses of phenobarbital decreased the amount of intubations required.⁹ To our knowledge, no studies have been performed on ETCO₂ monitoring of patients receiving barbiturates for alcohol withdrawal.

Patients experiencing respiratory depression from excessive sedation require careful monitoring of their oxygenation and ventilation. In respiratory depression, patients have decreased ventilation that leads to an abnormal retention of CO₂. To monitor the adverse effects of hypoventilation, ICU protocols require trends in CO₂ levels to be tracked using ABG tests and ETCO₂ measurements. The purpose of this study was to assess whether ETCO₂ levels could be used as a reliable surrogate for PaCO₂ levels once a baseline gradient between the two values was established. Tracking ETCO₂ levels may reduce the need for frequent ABG sampling, allowing for continuous, safe, comfortable, and less-invasive monitoring. Our results showed a positive correlation between the ETCO₂–PaCO₂ gradient in patients undergoing alcohol withdrawal and requiring benzodiazepine therapy. The results indicate a constant gradient of 9.5 mmHg (95% CI=6.7-12.6) between the two values, a higher gradient than has been reported in previous studies on other populations.^3,4,7 This difference was consistent throughout our population's hospital course, as 3 different data sets indicate a mean difference of approximately 9.5 mmHg. The gradient trend is significant because it can be used as an active clinical tool to assess the adequacy of treatment. A reduction in the gradient can indicate that treatment has been effective. An elevation in the gradient can indicate that the pulmonary condition is worsening, and further diagnostic and therapeutic measures should be considered. The ETCO₂ level may also be used reliably to indicate PaCO₂ level trends in patients with healthy lungs. For example, declining ETCO₂ levels may indicate a developing underlying process such as metabolic acidosis.

The limitations of this study were as follows: this study was performed using a patient population at 1 hospital and may not represent the population as a whole, the study did not compare ETCO₂ levels from a healthy population to the ETCO₂ levels of patients enrolled in the study, and this was a retrospective study with a relatively small sample size.

CONCLUSION

This study confirms a positive correlation between PaCO₂ and ETCO₂ levels in patients experiencing severe alcohol withdrawal. However, the gradient between the two measurements was larger in our study than in previous studies that have indicated a normal gradient of 2-5 mmHg. The ETCO₂ level in our study was, on average, 9.5 mmHg lower than the PaCO₂ level and was consistent throughout the patients' hospital course. Capnography can be a potential form of cardiopulmonary monitoring for patients who do not have a V/Q mismatch and offers more convenient, noninvasive monitoring than ABG testing.