Prevalence of Patients Admitted to Intensive Care After Administration of Chlordiazepoxide in the Emergency Room
Funding: The authors received no specific funding for this work.
ABSTRACT
Chlordiazepoxide is effective in treating alcohol withdrawal syndrome, but it poses a risk of long-term sedation. The prevalence of this side effect and its risk factors remain uncertain. This retrospective cross-sectional study aimed to estimate both using data from Aarhus University Hospital's BI portal. We identified and manually reviewed patient records from 1 September 2019 to 31 August 2021, including the treating physicians' conclusions on ICU admissions to determine whether they were likely due to chlordiazepoxide toxicity. Chlordiazepoxide was administered to 1363 unique patients in the study period. We identified 32 ICU admissions preceded by chlordiazepoxide administration, 5 of which (16%) were likely related to chlordiazepoxide toxicity. Patients with chlordiazepoxide-induced admissions received higher cumulative doses compared to other admissions (425 mg vs. 150 mg, p = 0.01), had longer ICU stays (median 8 vs. 2 days, p = 0.01) and required higher doses of flumazenil (p = 0.04). Their median age was above 60 years, and not all had known liver disease. The overall incidence of long-term chlordiazepoxide toxicity was approximately 0.35%, with risk factors including higher doses and age above 60. Our findings suggest increased caution when treating not only patients with liver disease but also elderly patients with chlordiazepoxide for alcohol withdrawal symptoms.
Summary
- Chlordiazepoxide is used for alcohol withdrawal but can cause prolonged sedation.
- However, incidence and risk factors other than liver failure are poorly understood.
- We reviewed records of patients treated with chlordiazepoxide in the emergency department and later admitted to the ICU at our hospital.
- Over 2 years, 1363 patients were treated.
- We found 32 ICU admissions, five likely due to chlordiazepoxide.
- Those admitted due to chlordiazepoxide were typically > 60 years and had received higher doses.
- Thus, although prolonged sedation is rare, caution is needed with using high doses of chlordiazepoxide in elderly patients in addition to those with liver disease.
1 Introduction
Chlordiazepoxide, the first benzodiazepine to reach the market, was introduced in 1960 [1]. It is prescribed not only for the management of anxiety disorders but also used to treat alcohol withdrawal syndrome due to its anxiolytic and sedative properties and long half-life [2, 3]. In Denmark, chlordiazepoxide is the first-line medical treatment of alcohol withdrawal symptoms [4].
Though effective in treating alcohol withdrawal syndrome, one concerning side effect is the potential for prolonged sedation and respiratory insufficiency as documented by several case reports [1, 5-8]. In the most extreme cases, it can lead to unresponsive patients staying in intensive care units (ICU) for weeks to receive mechanical ventilation [1], exposing the patients to a high risk of morbidity and mortality associated with a long stay in ICU [9, 10].
The mechanisms and patient factors underlying prolonged sedation induced by chlordiazepoxide remain incompletely understood. Factors such as drug accumulation due to its long half-life and its active metabolites [11], and reduced liver function is believed to contribute [6], as patients with liver cirrhosis have a threefold lower clearance of chlordiazepoxide than in healthy individuals [12]. Furthermore, polymorphism of the CYP3A4 gene that metabolizes chlordiazepoxide [13, 14] could affect chlordiazepoxide clearance and the risk of toxicity [15].
However, to our knowledge, neither the prevalence nor the potentially predictable or modifiable patient factors associated with prolonged stay in the ICU after administration of chlordiazepoxide have been reported. Understanding these aspects is crucial for optimizing therapeutic strategies and minimizing the risk of prolonged sedation in patients undergoing treatment with chlordiazepoxide.
Thus, this study aimed to explore the occurrence and patient characteristics of patients needing intensive care after treatment with chlordiazepoxide at a Danish hospital.
2 Methods
The study was a retrospective cross sectional study based on patient records at Aarhus University Hospital. We chose Aarhus University Hospital as this is the largest hospital in the Central Denmark Region with more than 80 000 admissions per year, of which more than 50 000 are acute admissions [16].
2.1 Identification and Inclusion of Patients
Patients treated with chlordiazepoxide for alcohol withdrawal syndrome are admitted to Aarhus University Hospital through the emergency department. We identified relevant patients in the BI portal of Region Midtjylland Denmark. The BI portal is a data system that automatically collects and manages patient data for hospitalized patients including administrations of medications and dates of admittance at the hospital's different departments. Patients who both had administrations of chlordiazepoxide in the emergency department and a contact to the ICU from 1 September 2019 to 31 August 2021 were initially selected and listed with the dates of their administrations and dates of admittance to ICU. Subsequently, two authors independently reviewed the list manually to include patients admitted to ICU within 21 days after chlordiazepoxide administration. From the BI portal, we also retrieved the total number of unique patients receiving chlordiazepoxide during the study period in the emergency department.
2.2 Collection of Data
We manually reviewed patient records and entered data in REDCap electronic data capture tool [17]. The following data were retrieved: sex, height, weight, body mass index (BMI), known diseases, use of medication and medication during hospitalization. When transferred to the ICU the following data was retrieved: Glasgow Coma Scale (GCS) score, vital parameters (blood pressure, heart rate, saturation, respiratory rate and temperature), fluid balance (sodium, potassium, magnesium, phosphate, creatinine and urea), liver function tests (alanine aminotransferase and bilirubin), ammonium, infection markers (C-reactive protein [CRP] and leucocyte count), coagulation tests, results of relevant radiological and microbiological examinations, need for mechanical ventilation, need for dialysis, length of stay in ICU and mortality.
2.3 Evaluation of the Cause of Admittance to ICU
We read and evaluated the whole patient record related to the relevant admission and included the treating physicians' conclusions on the indication for admittance to ICU to evaluate whether the admittance was most likely due to chlordiazepoxide (CDPX group), most likely not due to chlordiazepoxide or not determinable. If the attending physician described in the patient record that the effect of flumazenil was consistent with benzodiazepine intoxication, or if flumazenil was given continuously, we considered this an indication of a benzodiazepine overdose. We grouped other primary causes in the categories ‘infection’, ‘respiratory insufficiency’, ‘respiratory insufficiency and sedation’, ‘postoperative’, ‘electrolyte disturbance’ and ‘other’ after collecting the data. If ‘respiratory insufficiency’ was presumably caused by infection such as pneumonia, the cause was evaluated as ‘infection’.
2.4 Reporting and Statistical Analysis
We expressed continuous variables with normal distribution (tested by histograms and qnorm plots) as mean ± standard deviation, non-normally distributed data as median [25th–75th percentile] and categorical variables in percentages. We did not report results concerning groups of less than three individuals to ensure anonymity of the patients. Differences in characteristics between two groups were calculated using Student's t-test, Wilcoxon's rank sum test or χ2 test, where appropriate. We considered a p value < 0.05 as statistically significant. We used STATA 17, Stata Corp, Texas, for statistical analysis and did not impute missing data. We calculated demographics and comorbidities for the unique patients at their first admission during the study period as well as characteristics related to the individual admissions to ICU (i.e., doses of chlordiazepoxide, flumazenil and length of stay per admission).
2.5 Ethical Considerations
The Central Denmark Region approved the study (jr.nr.: 1-45-70-114-21), and Aarhus University Hospital granted access to patient records for the included patients (jr.nr.: 1-16-5-72-719-23). We registered the study at Aarhus University Hospital (jr.nr.: 1-16-02-17-22) and handled data in accordance with the general data protection regulation and the Danish Data Protection Act. The study was conducted in accordance with the Basic & Clinical Pharmacology & Toxicology policy for experimental and clinical studies [18]. According to the Consolidation Act on Research Ethics Review of Health Research Projects, Consolidation Act number 1083 of 15 September 2017, Section 14 (2), the project did not need approval from the Committees on Health Research Ethics.
3 Results
3.1 Patients
The number of unique patients receiving chlordiazepoxide in the emergency room in the study period was 1363. We identified 68 unique patients who had both an administration of chlordiazepoxide and a contact to the ICU. Twenty-eight unique patients were admitted to the ICU within 21 days of the administration and were included in the study. Four had two admissions during the period, rendering a total of 32 admissions. The characteristics of the unique patients are shown in Table 1. The majority of patients had a diagnosis of alcohol use disorder, and 25% had known liver disease at the time of admission (Table 1).
| Number of unique patients | (N) | 28 |
|---|---|---|
| Age | (Mean ± SD) | 57 ± 12 |
| Female | N (%) | 5 (18) |
| Height (cm) | (Mean ± SD) | 175 ± 11 (n = 27) |
| Weight (Kg) | (Mean ± SD) | 75 ± 19 (n = 27) |
| BMI (kg/m2) | (Mean ± SD) | 23 ± 6 (n = 27) |
| Comorbidities | ||
| Alcohol use disorder | N (%) | 25 (90) |
| Other | N (%) | 18 (64) |
| Liver disease | N (%) | 7 (25) |
| COPD | N (%) | 7 (25) |
| Diabetes | N (%) | 3 (11) |
3.2 Cause of Admission to ICU
We found chlordiazepoxide to be the most likely reason for admission to the ICU in five admissions (16%) (CDPX group) and as an unlikely reason in 15 admissions (47%). We found the role of chlordiazepoxide undeterminable in 12 of the admissions (38%). In the CDPX group, flumazenil was either given continuously or described as having a clear effect, and there was no competing cause of the ICU admission. Suspicion of aspiration, pneumonia or other infection was mentioned in four, which was evaluated as a consequence of the intoxication with chlordiazepoxide. In the 15 admissions determined to be unrelated to chlordiazepoxide, the primary cause of the admission to ICU was evaluated as infection in six (40%), electrolyte disturbance in four (27%) and postoperative in three (20%). In the 12 admissions, where we found the role of chlordiazepoxide undeterminable, we found respiratory insufficiency and respiratory insufficiency plus sedation as the most likely cause in four (33%) and three (25%), respectively. In the remaining admissions, other different causes applied.
3.3 Characteristics of the Admissions
The CDPX-group consisted of male only and did not differ from the rest of the patients in terms of age and BMI. Table 2 illustrates the doses of chlordiazepoxide and flumazenil given, duration of the ICU stay, need for mechanical ventilation and other medications given prior to the admittance to ICU according to the assumed role of chlordiazepoxide. In the five CDPX group admissions, the accumulated dose of chlordiazepoxide was significantly higher than in the rest of the admissions (425 [325–625] mg vs. 150 [75–250] mg, p = 0.01). Furthermore, the median duration of the ICU stay was significantly longer (8 [6–11] vs. 2 [1–6] days, p = 0.01), and the median dose of flumazenil was higher (p = 0.04) (Table 2). In the group where the role of chlordiazepoxide was deemed undeterminable, flumazenil was administered in 3 of 12 admissions. Dosage of flumazenil in these admissions were less than 1.5 mg, which was significantly lower compared to the CDPX group. Treatment with opioids or other sedatives prior to admission occurred in less than three of the five admissions in the CDPX group (Table 2). In-hospital deaths occurred in all groups (Table 2). Not all patients in the CDPX group had known liver disease.
| CDPX group | Unlikely | Undeterminable | ||
|---|---|---|---|---|
| Admissions | N | 5 | 15 | 12 |
| Accumulated dose of chlordiazepoxide | mg | 425 [325–625] | 100 [50–200] | 167.5 [113–475] |
| Age (unique patients) | Years | 63 [61–65] | 61.5 [55–66] | 54 [39–62] |
| Female (unique patients) | % | 0 | 21 | 20 |
| Time from last chlordiazepoxide administration to ICU admittance | Days | 3.1 [2.5–8.0] | 0.3 [0.11–8.4] | 5.5 [1.7–6.7] |
| Duration of ICU stay | Hours | 202 [146–270] | 44 [14–135] | 43 [17–120] |
| Mechanical ventilation | N (%) | 3 (60) | 5 (33) | 7 (47) |
| Duration of mechanical ventilation | Days | 3 [1–17] | 4 [2–10] | 4 [2–18] |
| Death in hospital | N (%) | < 3 | 4 (27) | 4 (33) |
| Opioids | N (%) | < 3 | 8 (53) | 5 (67) |
| Antibiotics | N (%) | < 3 | 4 (27) | 3 (25) |
| Antidepressants | N (%) | < 3 | < 3 | 4 (33) |
| Antipsychotics | N (%) | < 3 | < 3 | 4 (33) |
| Other benzodiazepines/barbiturates | N (%) | 0 | 3 (20) | 3 (25) |
| Flumazenil administered | N (%) | 5 (100) | 0 (0) | 3 (25) |
| Accumulated dose of flumazenil | mg | 26.7 [2.7–47] | — | < 1.5 |
- Note: Results concerning groups of less than three individuals are not shown to ensure anonymity of the patients. Unlikely = admissions where chlordiazepoxide was most like not the cause of the admission to intensive care unit. Undeterminable = admissions in which the role of chlordiazepoxide for the admission to ICU was not determinable.
We could not obtain data on vital and biochemical parameters from all admissions (Table 3). Based on the available data, we observed no significant differences between groups in vital parameters whether chlordiazepoxide was evaluated as causing the ICU stay or not (Table 3). CRP was significantly higher in the CDPX group, compared to the rest (145 [142–164] vs. 71 [30–111], p < 0.01) Apart from this, we detected no clear differences in biochemical parameters (Table 3).
| CDPX group | Unlikely | Undeterminable | ||
|---|---|---|---|---|
| (N) | 5 | 15 | 12 | |
| Glasgow coma score | < 10 (n < 3) | 15 [14–15] (n = 9) | 11.5 [10–15] (n = 6) | |
| Heart rate | /min | 92 ± 14 (n = 5) | 101 ± 31 (n = 14) | 109 ± 16 (n = 11) |
| Mean arterial blood pressure | mmHg | 78 ± 26 (n = 5) | 83 ± 17 (n = 15) | 93 ± 19 (n = 10) |
| Respiratory rate | /min | < 22 (n < 3) | 18 ± 5 (n = 15) | 24 ± 10 (n = 10) |
| Peripheral saturation | % | 89 ± 3 (n = 4) | 93 ± 6 (n = 15) | 83 ± 16 (n = 11) |
| Body temperature | °C | 36.5 ± 1.0 (n = 4) | 37.2 ± 1.2 (n = 13) | 37.3 ± 0.6 (n = 11) |
| Biochemical values | ||||
| Sodium | mmol/L | 136 [135–140] (n = 5) | 135 [129–139] (n = 15) | 138 [130–141] (n = 12) |
| Potassium | mmol/L | 3.9 [3.3–4.1] (n = 5) | 3.7 [3.5–4.3] (n = 15) | 3.9 [3.5–4.5] (n = 12) |
| Creatinine | mmol/L | 81 [49–104] (n = 5) | 88 [46–107] (n = 15) | 57 [43–72] (n = 12) |
| Bilirubin | mmol/L | 16 [13–19] (n = 4) | 14 [10–61] (n = 15) | 20 [8–106] (n = 12) |
| Alanine aminotransferase | U/L | < 101 (n = 3) | 43 [19–88] (n = 14) | 36 [26–64] (n = 12) |
| Alkaline phosphatase | U/L | < 80 (n = 3) | 82 [62–106] (n = 15) | 144 [88–198] (n = 12) |
| Ammonia | mmol/L | < 40 (n < 3) | < 30 (n = 3) | 63 [36–85] (n = 4) |
| Haemoglobin | mmol/L | 7.3 ± 1.0 (n = 5) | 7.1 ± 1.2 (n = 14) | 7.4 ± 1.7 (n = 12) |
| Glucose | mmol/L | 7.36 ± 2.6 (n = 5) | 8.3 ± 3.3 (n = 14) | 9.9 ± 2.8 (n = 12) |
| INR | 1.2 [1.2–1.4] (n = 5) | 1.1 [1.0–1.3] (n = 12) | 1.3 [1.1–1.7] (n = 11) | |
| C-reactive protein | Umol/l | 145 [142–164] (n = 5) | 58 [27–109] (n = 15) | 76 [43–115] (n = 12) |
| Leucocytes | 109 | 11 [10–16] (n = 5) | 14 [8–18] (n = 15) | 11 [7–24] (n = 12) |
- Note: Results concerning groups of less than three individuals, including mention of sex in groups of 5 or less, are not shown to ensure anonymity of the patients. Results mentioned as less than a certain value indicates that all obtained values in the group was below this.
4 Discussion
To our knowledge, this is the first study to estimate the prevalence of chlordiazepoxide poisonings in a hospital setting. We identified chlordiazepoxide as the likely cause of five ICU admissions over a 2-year study period, during which more than 1300 patients received the drug in the emergency room. This suggests that such incidents are relatively rare (approximately 0.35%). However, in a further 12 cases, the role of chlordiazepoxide as a contributor to the admission could not be completely excluded. As we considered the use of flumazenil infusion in evaluating chlordiazepoxide's role, true cases of chlordiazepoxide intoxication could be overlooked if the attending physicians did not suspect this diagnosis and therefore did not administer flumazenil. Given the latency between treatment and symptom onset, we find it plausible that suspicion may not arise. Thus, the number of cases found in the present study may be under-estimated.
We found that patients in the CDPX group had age of 60 years or above, that the cumulative dose of chlordiazepoxide was high and that there was a latency between the last administration of chlordiazepoxide and admittance to ICU. Furthermore, we found that they needed a long stay in ICU and a high cumulative dose of continuous administered flumazenil in the absence of administration of other benzodiazepines. In a series of 4 case reports on long-term sedation caused by chlordiazepoxide [1], age was between 54 and 71 years, the cumulative chlordiazepoxide dose ranged from 100 to 800 mg, the stay in ICU was between 14 and 27 days, latency from chlordiazepoxide to start of mechanical ventilation was 4–6 days, and days with flumazenil treatment was 11–42. Four other reported cases concerns persons between 55 [8] and 80 years [7], who needed mechanical ventilation or flumazenil infusion after cumulative doses between of 300 [8] and 1250 mg [6]. Thus, our cases in the CDPX group resemble other reported cases of chlordiazepoxide-induced toxicity [1, 5-8], which supports our evaluation of chlordiazepoxide as the cause of admission to ICU.
In our study, known liver disease was not present in all patients likely admitted to the ICU due to chlordiazepoxide treatment. This contrasts previous reports, in which nearly all patients had liver cirrhosis or other liver conditions [1, 5-8], with the exception of one patient in the case series [1]. However, liver parameters were not available in all patients in our study, and furthermore, assessment of liver enzymes and INR are not sufficient to diagnose liver cirrhosis [19]. Thus, although diagnosed liver disease was not evident in all our cases in the CDPX group, we cannot conclude whether toxicity can occur in patients without liver disease. The World Health Organization recommends long-acting benzodiazepines for alcohol withdrawal symptoms, but short-acting benzodiazepines in people with liver disease or the elderly [20]. Perhaps clinicians at our study site are already attentive to patients with known liver disease and avoid using chlordiazepoxide in these cases. This could explain why we found few instances of chlordiazepoxide toxicity in patients with known liver disease.
The median value of CRP in the CDPX group was higher compared to the rest of the patients, which might be explained by aspiration due to loss of consciousness associated with chlordiazepoxide toxicity. We could not find CRP values to compare with in the previously reported cases [1, 5-8]. The number of cases was too small to determine whether or not drug–drug interactions (e.g., opioids, other benzodiazepines and inhibitors of CYP2C19 or CYP3A4) [1] contributed to chlordiazepoxide toxicity.
A strength of our study is the use of the BI portal. The system contains data on all patients admitted to our hospital, thus allowing us to identify the complete population of patients relevant for the study. Another strength is our manual review of all the patient records to obtain a deeper understanding of the admissions. On the other hand, ICU stays are complex and influenced by many factors, making it difficult to interpret the role of chlordiazepoxide in several cases, which must be considered a limitation. Furthermore, the concentrations of chlordiazepoxide and its metabolites, as well as CYP3A4 status, were not measured. This could have supported our evaluations and informed diagnosis of future intoxications. Additionally, we only looked at patients receiving chlordiazepoxide in the emergency room as this is the common gateway to all admissions to this hospital. As treatment for abstinence is often initiated here, we have most likely included the vast majority of relevant patients. However, some patients could have started treatment with chlordiazepoxide after being transferred to the ICU from another department. In that case, the patient is not included in our study. This would tend to underestimate the occurrence of chlordiazepoxide toxicity. Finally, the modest number of patients that we found with both chlordiazepoxide administrations and ICU stays with a relevant temporal association is a limitation.
5 Conclusion
Our results suggests that long-term toxicity requiring intensive care occurred in less than five per 1000 cases of patients treated with chlordiazepoxide, although this may be under-estimated. Admission to ICU due to chlordiazepoxide seemed associated with dose, age of 60 years or above or increased CRP and not necessarily with known liver disease or treatment with other sedatives. Thus, increased caution concerning elderly or frail patients in addition to those with liver disease may be warranted. However, it is worth discussing the ethical acceptability of using chlordiazepoxide, with its risk of serious long-term toxicity, when shorter-acting benzodiazepine alternatives are available.
Acknowledgements
We warmly thank Charlotte Olesen and Jakob Redder for their help with the BI portal.
Conflicts of Interest
The authors declare no conflicts of interest.
Open Research
Data Availability Statement
The data that support the findings of this study are available from Central Denmark Region. Restrictions apply to the availability of these data, which were used under license for this study. Data are available from the author(s) with the permission of Central Denmark Region.
