- Top of page
We compared the effects of single doses of etomidate and thiopentone on adrenocortical function in a randomised controlled clinical trial involving 35 critically ill patients who needed a general anaesthetic. Just before induction of anaesthesia, a baseline blood cortisol sample was taken. Twenty-four hours later we performed a short adrenocorticotrophic hormone stimulation test. No patient had a low cortisol level (< 160 nmol.l−1) at any time during the study. Baseline, pre-ACTH and post-ACTH cortisol levels were similar in the two groups. However, significantly more patients in the etomidate group had an ACTH-stimulated cortisol increment < 200 nmol.l−1. The clinical significance of these findings is not clear, but we conclude that single doses of etomidate may interfere with cortisol synthesis for at least 24 h in the critically ill.
- Top of page
The study was approved by the local ethics committee and conducted according to the principles of studies in humans established in Helsinki. Wherever possible informed consent was obtained from the patient or assent was obtained from a relative.
Thirty-five critically ill adults (age > 16 years) who needed a general anaesthetic were studied. Inclusion criteria were that subjects had to be ASA grade III or worse, have two or more organ system failures and be considered to require admission to the intensive care unit (ICU). Exclusion criteria were known adrenal impairment, known contraindications to etomidate or Synacthen, and a history of steroid supplementation in the previous 6 months.
Once enrolled in the study, patients were randomly allocated to receive either etomidate or thiopentone for induction of anaesthesia. Thiopentone has no known effect on adrenal function. The dose of induction agent, and all other aspects of the peri-operative management and care on the ICU were left to the discretion of the attending anaesthetists. Management was routine in all aspects except that midazolam infusions for sedation were discouraged as these have been shown to interfere with the adrenal responses to stress .
Individual patient data collected included age, sex, APACHE II score, diagnosis and relevant medical history.
Evaluation of adrenal function
Each patient had three blood samples taken for cortisol assay. The first (baseline) was taken shortly before induction of anaesthesia. Studies in healthy subjects have shown up to 6 h of adrenocortical suppression after a single dose of etomidate [2–5], and so to assess whether this effect persists in the critically ill a short ACTH stimulation test was performed after 24 h. This involved taking two further blood specimens for cortisol assay — one just before (pre-ACTH), and the other (post-ACTH) 30 min after administering 250 μg of synthetic ACTH1–24 (Synacthen®, CIBA-Geigy). On receipt of the samples by the laboratory they were separated, frozen and stored. Later they were assayed in batches by automated chemiluminescence assay (ACS: 180, Chiron Diagnostics). The assay sensitivity is 5.5 nmol.l−1, and the coefficient of variation covered is in the range 4.6–7.8%, dependent on cortisol concentration. For individual cortisol assays we used our laboratory's 8 a.m. reference range of 160–600 nmol.l−1 as the ‘normal range’ for our patients, for three reasons. First, it has been shown that many critically ill patients lose the diurnal cortisol rhythm (the midnight cortisol trough is lost) . Second, in a recent study, 17% of healthy subjects had no diurnal rhythm . Finally, the peak cortisol, the most commonly used criterion for evaluating the result of the short ACTH stimulation test , has been shown to be the same regardless of when in the day the test is performed. This was found in patients with asthma  and rheumatoid arthritis , and in healthy subjects .
As we made no assumptions about the data distribution, all tests for statistical significance were performed using nonparametric methods. For continuous variables we used the Mann–Whitney U-test and for categorical variables the Chi-squared or Fisher's exact tests. The criterion for significance was p < 0.05 and all statistical analyses were performed using the statistical package statistica for Windows (release 4.5).
- Top of page
Thirty-five patients were studied. There were no statistically significant differences between the groups in terms of age, sex and APACHE II score (Table 1). The median APACHE II scores were 18.5 in the etomidate group and 17 in the thiopentone group. The indications for general anaesthesia are summarised in Table 2. The most common reasons were surgical, but several patients had respiratory failure and needed a brief anaesthetic to facilitate tracheal intubation.
Table 1. Patient characteristics. Values for age and APACHE II score are medians Table 2. Summary of indications for anaesthesia (n)
The raw cortisol data, the sedative infusion regimens used on ICU after the general anaesthetic and whether an epidural infusion was administered are shown in Table 3. Despite our protocol discouraging midazolam infusions, four patients (two in the study group and two controls) were inadvertently prescribed these infusions during the night before the ACTH stimulation test. The two patients in the etomidate group had poor cortisol increments after ACTH, whereas the two in the thiopentone group responded normally.
Table 3. Cortisol assay results (nmol.l−1) before induction of anaesthesia (baseline), just before ACTH administration (pre-ACTH) and 30 min after ACTH administration (post-ACTH), and sedative and epidural infusion regimens used * Missing data.
No patient had evidence of absolute adrenal failure (plasma cortisol < 160 nmol.l−1) during the period of the study. One post-ACTH sample from the thiopentone group was lost, so that we were only able to analyse the response to ACTH in 17 control patients. The cortisol assay results ranged from 221 to 2200 nmol.l−1, the median results and interquartile ranges are shown in Table 4. There was a tendency for the baseline and pre-ACTH plasma levels to be higher in the etomidate group than in the thiopentone group, but these differences failed to reach statistical significance. The changes from the baseline to the pre-ACTH cortisol levels in the two groups were also similar. The median (interquartile) change in the control group was − 217 (− 694 to − 44) nmol.l−1 compared with − 240 (− 783 to + 11) nmol.l−1 in the etomidate group, a statistically insignificant difference. Post-ACTH cortisol concentrations tended to be higher in the thiopentone group, but this difference narrowly failed to reach significance (p = 0.052).
Table 4. Results of cortisol assays. Values given are median (interquartile range) plasma cortisol concentrations (nmol.l−1). No significant differences between groups (Mann–Whitney U-test)
Two methods of interpreting the response to ACTH were used. Table 5 contains the results and associated outcomes. The first was the cortisol increment, i.e. the difference between the pre- and post-ACTH levels. Fifteen patients (88%) in the study group and five (29%) in the control group had an impaired response according to the criterion of an increment of < 200 nmol.l−1 (p < 0.0006). The median (interquartile range) increment in blood cortisol after ACTH injection was 137 (− 2 to 161) nmol.l−1 in the etomidate group, as opposed to 233 (113–280) nmol.l−1 in the thiopentone group (p < 0.004). The other method used to evaluate the response was to judge the response as adequate if the post-ACTH cortisol was > 550 nmol.l−1. Two patients (12%) in the thiopentone group and five (29%) in the etomidate group had an impaired response by this criterion (p < 0.24).
Table 5. ICU outcome — values are the numbers of subjects who died in ICU or survived to discharge from ICU, grouped according to the result of the ACTH stimulation test
The total number of deaths was similar in the etomidate and the thiopentone groups (Table 5). There was a tendency for those that died on ICU to have a poor cortisol increment, but a normal peak level. The number of subjects was considered too small to warrant calculation of the predictive value of the two interpretations of the ACTH stimulation test (with regard to mortality). Fisher's exact tests on the differences between the numbers of ICU survivors meeting or failing to meet the two different criteria (control and study groups separately and combined) did not show any significant associations.
When the cortisol results of the study and control groups were combined there was a tendency for plasma cortisol concentrations to be higher in subjects who died in ICU than in those who were discharged from ICU to the general wards (Fig. 1). However, this difference was not statistically significant.
Figure 1. Plasma cortisol concentrations of study and control groups combined, classified according to ICU survival. Symbols represent the median cortisol values, boxes represent interquartile ranges and whiskers represent the range at the times shown.
Download figure to PowerPoint
- Top of page
Etomidate, a carboxylated imidazole, interferes with adrenal function by reversibly inhibiting 11β-hydroxylase and cholesterol side-chain cleavage enzyme . In healthy patients single doses have been shown to cause short-lived suppression of cortisol synthesis, which has generally been considered clinically insignificant [2–5]. However, a study of the time-course of adrenocortical suppression in men undergoing cardiac surgery showed effects that lasted 24 h . In the latter study, patients who received a single dose of etomidate had significantly raised 11-deoxycortisol concentrations and smaller ACTH-stimulated cortisol increments at 24 h compared with patients who had not received the drug. In addition, the subjects who received etomidate had significantly elevated ACTH levels after 8 h, and markedly reduced 24-h urinary-free cortisol excretion. To date these studies have not been repeated in the critically ill. We therefore compared adrenocortical function in critically ill patients randomised to receive a single induction dose of either etomidate or thiopentone.
In our subjects, cortisol concentrations just before anaesthetic induction were high. This is in keeping with other studies of adrenocortical function in the critically ill [14–16]. In these patients, levels remain high for several days, despite the fact that ACTH tends to decrease after 2 days . The range of blood cortisol levels found in healthy individuals probably does not apply to the critically ill, in whom a higher normal range has been suggested . In intensive care patients, single plasma cortisol concentrations have been shown to correlate positively with the severity of illness and negatively with outcome [15, 19]. Some authors argue that serial measures have a greater predictive value than single levels . There are many reasons why the critically ill may have high basal cortisol concentrations. These include: loss of the circadian rhythm , activation of the hypothalamo-pituitary-adrenal axis, decreased cortisol extraction from the blood, reduced binding of cortisol to transcortin, and an increased half-life of cortisol . Endothelin, elevated markedly in the critically ill , may also play a role. It has been shown to stimulate the hypothalamo-pituitary-adrenal axis in rats  and humans . Other investigators have shown that endothelin directly stimulates cortisol secretion by the adrenal cortex and potentiates the effects of ACTH .
None of the patients studied had evidence of absolute adrenal failure. There was also no statistically significant difference, at any of the measurement points, between the plasma cortisol levels of those who had been given etomidate and those who had been given thiopentone. ACTH assays were not performed, but in studies where they have been performed on subjects given a single dose of etomidate, ACTH levels were up to 100 times higher than in controls [3, 13]. ACTH assays may have given an indication of whether the patients given etomidate had cortisol levels similar to those in controls (after 24 h) because etomidate had ceased to interfere with cortisol synthesis, or because of maximal ACTH secretion to overcome incomplete inhibition of 11β-hydroxylase.
There is controversy over whether a single cortisol measurement is sufficient in the assessment of adrenocortical function. Recently, it was suggested that a single value > 700 nmol.l−1 was sufficient to exclude adrenal insufficiency . The interpretation of lower values remains difficult. It has been suggested that low levels are not necessarily indicative of adrenocortical insufficiency and that in such a situation an ACTH stimulation test is necessary to confirm the diagnosis . In addition, a syndrome of relative or functional adrenal insufficiency has been postulated. Proponents have suggested that in this syndrome the cortisol level, despite being normal or high, is still inadequate for the current physiological stress, and that the patient may be unable to respond to any additional stress [21, 25–27]. For this reason, the ACTH stimulation test is used to diagnose the relative adrenal insufficiency, but this is an area of great controversy.
The two main areas of debate are the significance of the ACTH stimulation test, and the interpretation of the results. This test has been shown to be a reliable indicator of adrenal function and glucocorticoid requirements in patients with pituitary tumours [28, 29] and after hypophesectomy [30, 31], but is reported to be less reliable in assessing adrenal function during withdrawal of long-term steroid therapy [31, 32]. The situation in the critically ill is less clear-cut. Many workers have used it in critically ill patients to assess adrenocortical function and some have shown an association between a poor response and higher mortality [19, 21, 25–27, 33]. However, Bouachour et al. did not find this link, and concluded that the ACTH stimulation test was unable to estimate impairment in adrenocortical function (and thus the need for steroid therapy) in patients with septic shock . In a subsequent study of the response to ACTH within 24 h, and then 48 h after the onset of septic shock, there was no correlation between the cortisol increment on days 1 and 2, whereas there was a positive correlation between the peak cortisol concentration .
There are no agreed criteria for evaluating the changes in cortisol concentrations after ACTH stimulation in the critically ill. In 1985, May and Carey wrote of a ‘bewildering variety of criteria for normalcy of rapid ACTH test results’ and of the need for a consensus on the subject . They suggested that a peak cortisol value > 550 nmol.l−1 was a satisfactory single criterion for normal adrenocortical function, but unfortunately there remains an array of criteria. Other minimum peak levels that have been suggested include: 400 nmol.l−1 , 500 nmol.l−1  and 600 nmol.l−1 . Yet others support the use of the cortisol increment. Criteria that have been used include: an increment of 200 nmol.l−1 at 30 min after ACTH injection , an increment of 200 nmol.l−1 at 30 or 60 min , and an increment of 250 nmol.l−1 at either 30 or 60 min . Others have defined a normal response as an increase of 200 nmol.l−1 and/or a peak of 500 nmol.l−1 .
Clearly the incidence of adrenal insufficiency in any population will depend on the criteria used to evaluate the response to ACTH. There are strong arguments in favour of the use of the peak cortisol level after ACTH stimulation. In one study, several normal subjects failed to achieve a cortisol increment of > 200 nmol.l−1 . The cortisol increment has also been shown to be inversely proportional to the baseline value , so that those with the highest baseline values are less likely to achieve a ‘normal increment’. A recent review of adrenal insufficiency supported the use of the peak value .
In our study, the difference between the peak cortisol levels of the two groups narrowly failed to reach statistical significance. Using the criterion of a peak cortisol > 550 nmol.l−1 there were five patients in the study group, and two in the control group with relative adrenal insufficiency, a difference that was also statistically insignificant. On these grounds it is difficult to conclude that the etomidate group were unable to respond appropriately to exogenous ACTH. However, it is difficult to ignore the fact that, despite having similar pre-ACTH cortisol levels, there was a significant difference in the cortisol increment in the two groups, and a highly significant difference between the numbers of patients with increments < 200 nmol.l−1.
When looking at the mortality rates for the two groups combined it is interesting to note that patients with peak cortisol levels < 550 nmol.l−1, despite fulfilling a criterion for adrenocortical insufficiency, had a lower mortality rate (14%) than those with peak levels > 550 nmol.l−1 (26% mortality). In contrast, the mortality rate tended to be higher in those with an inadequate response to ACTH when judging the test by the increment. In subjects with cortisol increments < 200 nmol.l−1, the mortality was 35%, whereas an increment > 200 nmol.l−1 was associated with a mortality rate of 7%. It is natural to assume that those who died on ITU were more ill, and stressed, and thus had higher pre-ACTH cortisol levels, and so were likely to have poorer increments, but still have higher peak concentrations. Indeed there was a tendency for nonsurvivors to have higher pre-ACTH cortisol levels, but this difference was not statistically significant.
If a poor response to synthetic ACTH (and hence relative adrenal insufficiency) is a significant problem in the critically ill, then these patients may benefit from steroid supplementation. In general, clinical trials of the use of corticosteroids in patients with sepsis have failed to show a reduction in morbidity and mortality [41–43]. However, there are several anecdotal reports of critically ill patients with relative adrenal insufficiency and septic shock resistant to conventional therapy, who have improved dramatically after the administration of pharmacological doses of steroids [14, 27, 44–46]. It is possible that patients with relative adrenal insufficiency represent a subgroup of critically ill patients who could benefit from steroid therapy, but there are very few reports of controlled clinical trials in this group. In the study by Soni et al. , steroid supplementation in patients with relative adrenal insufficiency appeared to improve short-term survival, but there were no control patients. The only prospective randomised, controlled trial of steroid therapy for adrenal insufficiency of which we are aware is that performed by McKee and Finlay . They studied steroid replacement in 18 critically ill patients with a basal cortisol < 350 nmol.l−1 and a poor response to ACTH, and showed that hydrocortisone therapy was associated with a reduction in mortality.
In conclusion, we have studied adrenocortical function in a group of critically ill subjects, and found that a single dose of etomidate was not associated with a significantly lower plasma cortisol after 24 h, either before or after ACTH stimulation, when compared with subjects given thiopentone. However, etomidate was associated with a smaller cortisol increment after ACTH stimulation. This suggests that there was still some residual biochemical effect on the ability of the adrenal cortex to respond to ACTH 24 h after a single dose of etomidate. The clinical significance of these findings is not clear. Interpretation of the results, and of those by other workers is hampered by the lack of a clear understanding of what the response of the ‘normal’ adrenal cortex is to critical illness, and thus the lack of a consensus of how to evaluate adrenocortical function in these patients .