Discussion
Our findings suggest that opioid-induced adrenal insufficiency is uncommon when applying a diagnostic cutoff of <405 nmol/L (14.7 μg/dl). Borderline stimulated cortisol levels (peak cortisol between 405 and 500 nmol/L [14.7–18.1 μg/dl]) were found in 15% of patients on chronic opioids. To our knowledge, this is the first study of opioid-induced adrenal insufficiency using assay-specific diagnostic cutoffs. Our finding that 64% of hospitalized patients with a morning serum cortisol <83 nmol/L (3 μg/dl) passed a cosyntropin stimulation test suggests that this cut-off is not an accurate diagnostic tool for adrenal insufficiency in hospitalized patients, regardless of whether they are receiving opioids.
Among patients receiving chronic opioids, estimates of the prevalence of hypocortisolism have varied with a recent systematic review reporting 15%–24% based on five studies that included dynamic HPA axis testing.[10] A recent study from Li et al. reported a prevalence of 9% in 102 patients treated with opioids for noncancer pain[7] based on an unstimulated morning serum cortisol level in six of the nine patients diagnosed and peak cortisol levels >440 nmol/L (15.9 μg/dl) in the other three patients. Similar to the findings in our cohort, when cosyntropin stimulation tests are abnormal in the setting of chronic opioid exposure, the results appear to be borderline with the categorization of the result as normal or abnormal depending on the cutoff used. Notably, chronic pain itself has been associated with low cortisol levels and inadequate responses to ACTH stimulation, although findings have been variable.[27–29] The decision to treat abnormal test results requires clinical judgement which may be complicated in patients treated with opioids, as the side effects of opioids overlap with some symptoms of adrenal insufficiency.
We hypothesized that the acute suppressive effects of opioids on serum cortisol levels and the disrupted circadian rhythm observed in patients on opioids would make the use of morning serum cortisol levels for the diagnosis of adrenal insufficiency unreliable. The sensitivity and specificity of low morning cortisol levels for diagnosis of adrenal insufficiency were similar in inpatients with and without opioid exposure in this study. However, when interpreting the specificity of these tests, it is important to consider the low prevalence of adrenal insufficiency in the study sample. Our data demonstrated that morning cortisol levels <83 nmol/L (3 μg/dl) were more commonly found in hospitalized patients resulting in a high percentage of false positive tests (64%), suggesting that this cut-off does not accurately diagnose adrenal insufficiency in hospitalized patients. The higher frequency of low morning cortisol levels may be due to the impact of hospitalization on cortisol circadian rhythm. This finding is consistent with a retrospective study of 128 hospitalized patients who underwent cosyntropin stimulation testing that found that morning serum cortisol showed no statistical difference compared to random serum cortisol for predicting adrenal insufficiency.[30] A time-adjusted morning cortisol cutoff was shown to reduce the need for cosyntropin stimulation tests by Henley et al. who found that median cortisol decreased by ~30 nmol/L (1.0 μg/dl) per hour after 7 AM.[31] However, a time-adjusted morning cortisol cutoff may be considered an additional challenge in an already-complex diagnosis. Given the risks associated with misdiagnosis of adrenal insufficiency, clinicians should use caution when relying on morning cortisol values to diagnosis adrenal insufficiency in hospitalized patients.
Lower morning unstimulated cortisol cutoffs to rule-out adrenal insufficiency may be appropriate with new assays. Currently, the Endocrine Society proposes a cutoff of 414 nmol/L (15 μg/dl), while Sbardella et al proposed levels as low as 336 nmol/L (12.2 μg/dl) in a study of three cortisol assays.[32] In hospitalized patients included in our study, a morning cortisol level of <348 nmol/L (12.6 μg/dl) had a sensitivity of 100% for the diagnosis of adrenal insufficiency. When specific cortisol assays are used this may serve as a useful clinical cutoff for inpatients to rule out adrenal insufficiency and avoid further testing.
Very few patients with opioid prescriptions (N = 13) underwent outpatient HPA axis testing with a morning cortisol collected in this cohort. Therefore, we are unable to draw conclusions about the validity of a low morning cortisol for the diagnosis of adrenal insufficiency in outpatients taking opioids. The small number of outpatient tests in opioid-exposed patients may reflect low suspicion for adrenal insufficiency and/or this group of patients may experience barriers to accessing specialty care. It is also possible that there is not general awareness among healthcare providers that opioid exposure is a risk factor for the development of adrenal insufficiency.
A limitation of this study is the retrospective nature. Conclusions cannot be made about the prevalence of adrenal insufficiency because only patients who underwent HPA axis testing due to clinical suspicion for the condition were included in the study. Our data suggest that a similar level of suspicion for adrenal insufficiency, indicated by presence of symptoms attributable to the condition, prompted testing in hospitalized patients taking opioids and in those not receiving opioids. This suggests a lack of selection bias from a higher suspicion for adrenal insufficiency in patients on opioids. However, prospective studies in which subjects are tested for adrenal insufficiency, regardless of symptoms, are needed, to establish whether the prevalence of adrenal insufficiency in patients taking opioids approximates that reported here. The use of morphine milligram equivalent doses may also contribute to the inconsistent findings in studies of opioid-induced adrenal insufficiency as there is no universal method for converting opioid doses, and considerable variability in the calculations has been reported.[33] Methadone conversion has been found to be especially variable with a range of morphine equivalents used in the MEDD calculations. Adrenal insufficiency was diagnosed by the 250-μg cosyntropin test, which is the standard diagnostic test within MGB. Low-dose (1-μg) cosyntropin tests may be more sensitive for the diagnosis of secondary adrenal insufficiency, although metanalyses have provided inconsistent conclusions.[34–36] A strength of the study of was the number of subjects; prior studies evaluating morning serum cortisol for the diagnosis of adrenal insufficiency have been smaller and limited to outpatient testing for adrenal insufficiency.[37,38]
In conclusion, in this cohort undergoing adrenal axis testing as part of clinical care and highly enriched for symptoms of adrenal insufficiency, opioid-induced adrenal insufficiency was rare when applying a diagnostic cutoff of <405 nmol/L (14.7 μg/dl). Nearly one in six patients would be reclassified applying the society guideline recommend cutoff of <500 nmol/L (18.1 μg/dl). Moreover, morning serum cortisol should be used with caution to diagnose adrenal insufficiency in hospitalized patients, whether or not taking opioids; however a basal cortisol level of ≥348 nmol/L (12.6 μg/dl) may be used to rule out adrenal insufficiency. Given the profound health implications of chronic opioid use and adrenal insufficiency, prospective studies using assay-specific diagnostic cutoffs to investigate the prevalence of adrenal insufficiency are needed to guide screening recommendations.
Data Availability Statement
Some or all data sets generated during and/or analysed during the current study are not publicly available but are available from the corresponding author on reasonable request.
Clin Endocrinol. 2022;96(1):21-29. © 2022 Blackwell Publishing
