Undiagnosed adrenal insufficiency as a cause of premature death in glucocorticoid users

Review of medical records

A total of 300 randomly selected medical records were reviewed to determine if the death was unrelated to, possibly related to, or probably related to GC-induced adrenal insufficiency. Past medical history, duration of, and indication for, GC use, symptoms and signs before death, and biochemical data were collected. One investigator conducted the initial review and estimated if it was unlikely, possible, or probable that the death had been related to GC-induced adrenal insufficiency. Following this initial review, 51 deaths were considered possibly or probably related to GC-induced adrenal insufficiency. Death probably caused by GC-induced adrenal insufficiency was defined as death in which GC-induced adrenal insufficiency was a major contributing factor to the cause of death (e.g. if GC stress dose had been given, the patient would most likely have survived). Death possibly related to GC-induced adrenal insufficiency is defined as death in which GC-induced adrenal insufficiency was a contributing factor but not a dominant factor (e.g. if GC stress dose had been given, the patient may have survived or died). Each of these 51 medical records was subsequently independently reviewed by two additional investigators in the following manner: one investigator reviewed all 51 cases, and two additional investigators reviewed 26 and 25 of these cases each. Thus, all 51 medical records were reviewed by three investigators. All investigators are experienced endocrinologists. During the review process, the investigators were unaware of each other’s grading. Two reviewers had to agree for a death to be considered as unlikely, possibly, or probably related to GC-induced adrenal insufficiency. However, in three cases, all reviewers disagreed (one said unlikely, one possible, and one probable). In these three cases, the death was classified as possibly related to GC-induced adrenal insufficiency. Since plasma cortisol was not measured before death, the grading ‘possibly or probably related to GC-induced adrenal insufficiency’, was not considered to confirm the diagnosis; it rather depicts the authors’ assumption based on information gathered from the medical records.

Ethics statement

The study was approved by the Regional Research Ethics Committee in Gothenburg, Sweden (reference number 773-14; approved March 9, 2015) and by the National Board of Health and Welfare, Sweden.

Statistical analysis

Age was defined as the age at death. Only patients with ongoing GC treatment on admission were included in the calculations of treatment duration and GC dosage. The prednisolone-equivalent dose was calculated as follows: prednisolone 5 mg = hydrocortisone 20 mg = dexamethasone 0.5 mg = betamethasone 0.5 mg.

Descriptive data are presented as mean ± s.d. and/or median (range). For the comparison of groups, Student’s t-test was used for continuous variables with a normal distribution, and the Mann–Whitney U test for non-normally distributed continuous variables. Fisher's exact test was used to estimate the differences between categorical variables. All analyses were conducted using IBM SPSS Statistics for Windows, version 25 (IBM Corp). A P-value < 0.05 was considered statistically significant.

Deaths where GC-induced adrenal insufficiency was probably involved

We identified 11 (3.7%) cases where undiagnosed and untreated GC-induced adrenal insufficiency probably contributed to the patient’s death (Fig. 1). In 5 of these 11 cases, long-term (>30 days) GC treatment was discontinued for unknown reasons during the hospital stay. One patient received an increased GC dose for unknown reasons several hours before death. The following four cases illustrate patients from this group.

View Full Size

Flowchart showing the study design.

Citation: Endocrine Connections 13, 4; 10.1530/EC-23-0535

• Download Figure
• Download figure as PowerPoint slide

Deaths where GC-induced adrenal insufficiency was possibly involved

Undiagnosed or undertreated GC-induced adrenal insufficiency was considered to be a possible contributing factor of death in 36 patients (12%). In 7 of these 36 cases, long-term (>30 days) GC treatment was discontinued for unknown reasons during the hospital stay. Eight patients received an increased GC dose during hospitalization. The dosage of GC was increased in three patients due to suspected GC-induced adrenal insufficiency, in two patients due to allergy, in one patient due to nausea, and in one patient for unknown reasons. The following two cases illustrate patients from this group.

Adrenal crisis following GC cessation

The period after GC treatment cessation is potentially life-threatening if the hypothalamic–pituitary–adrenal axis has not recovered (5). Up to 90% of patients have suppressed adrenal function for several days after short-term oral GC treatment (10, 11), and the recovery can take from several days to several months (10, 11). As an example, in case 1, the patient contracted an infection 4 days after cessation of GCs, which probably led to the adrenal crisis and death as no stress doses of GC were administered. Adrenal suppression has been described after short-time treatment (≤2 weeks) (11, 12), which may have been the case for this patient. In case 2, the long-term GC treatment was abruptly discontinued, most likely due to misunderstanding, and the patient did not receive a stress dose of GC during an infection. Moreover, the patient had abdominal pain and hypotension, which are common symptoms of adrenal crisis. In case 4, the patient should have received IV hydrocortisone when he was not permitted to take oral medications. These cases highlight that sudden discontinuation of GC treatment can lead to an adrenal crisis.

The lesson to be learned from these cases is the importance of obtaining a detailed drug history. First, on admission, it is important to inquire about current medications and medications that have recently been discontinued. Secondly, in long-term GC users, it is necessary to document the need to return to the normal dose of GC following a temporary increase in GC dosage (e.g. due to a chronic obstructive pulmonary disease exacerbation, allergic reaction, or gout attack). Lastly, when all oral medications are suddenly discontinued due to fasting, essential medications such as GC should be administered via alternative routes.

Administration of GC stress dose

Stress dose administration is essential to prevent adrenal crisis and death during intercurrent illness (5, 13). In case 3, the patient should have received stress doses of GC directly upon admission. Similarly, in case 5, the patient needed a higher dose than 2.5 mg of prednisolone per day during his intercurrent illness.

The awareness of GC-induced adrenal insufficiency continues to grow. During the last few years, three review articles have been published (7, 13, 14), and the European Society of Endocrinology has published recommendations concerning stress doses for GC users during infection with the COVID-19 virus (15). Such guidance is important and necessary: a survey showed that 71% of physicians and specialist nurses changed their management of adrenal insufficiency after being directed to the guidelines (16).

In 2020, Simpson et al. (17). published guidance on the prevention of adrenal insufficiency and the emergency management of patients presenting with the condition. As reported in these guidelines, two deaths and six severe harms caused by adrenal insufficiency were identified in a database of patient safety incident reports (the National Reporting and Learning System) in the UK (17). Multiple themes were presented, including failure to increase GC doses in surgical stress, inadequate practices regarding medication at admission and discharge, delayed administration of prescribed GC doses, and missed or delayed alternative administration routes in fasted patients (17). Our findings illustrate these themes well.

In the current study, we decided to choose sepsis as the recorded cause of death because sepsis and adrenal crisis share several clinical symptoms and signs, and infections are also one of the leading causes of adrenal crisis (1). Moreover, GC users have increased mortality due to infections. A previous analysis of our study group showed increased mortality in GC users from sepsis with an adjusted hazard ratio (HR) of 2.1 (95% CI 1.9–2.3) compared to non-users (9). The present study indicates that GC-induced adrenal insufficiency may contribute to the increased mortality observed in our previous study.

Other studies indicate that GC withdrawal or GC-induced adrenal insufficiency is underdiagnosed. Incidence rates of hyponatremia, hypotension, gastrointestinal symptoms, and hypoglycemia are increased during the withdrawal period (1 month before and after cessation of GC treatment) compared to before GC treatment was started, according to a Danish population-based study (18). The 47 patients in our study had significantly lower GC doses prior to admission, lower blood pressure on admission, and higher rates of gastrointestinal symptoms and hypoglycemia compared to the patients whose death was likely not related to adrenal insufficiency. All of these clinical features are common in patients with adrenal crisis (1). Low GC dose prior to admission could be a risk factor since these patients had an inadequate GC dose during intercurrent illness. Moreover, 48% of patients receiving long-term treatment with low-dose (5 mg) prednisolone treatment have adrenal insufficiency (19). This highlights the importance of stress dose administration in patients receiving low-dose GC.

Our previous study showed that of all individuals who were dispensed oral GCs in Western Sweden between 2007 and 2014, 55.6% were women (8), whereas in the current, study 60% were men. The reason is that the mean age of patients in this study is 76 years, which is when GC prescriptions become more common in men than in women (8).

Due to various comorbidities and high age, most patients in our study had short life expectancies. However, adrenal insufficiency caused by GC should not contribute to premature mortality in this patient population. Despite the increased rate of adrenal crisis incidence with age, with the highest incidence in patients over 60 years of age (20), GC-induced adrenal insufficiency does not just affect patients with high morbidity. There have been alarming case reports of children who have died or nearly died due to GC-induced adrenal insufficiency (21). Awareness and knowledge of GC-induced adrenal insufficiency are essential and lifesaving.

The main limitation of our study is that it is based on a subjective analysis of medical records without a clear confirmation of adrenal insufficiency. Conducting a study to determine if the recorded cause of death is accurate is a methodological challenge. Our hypothesis was that GC-induced adrenal insufficiency is an overlooked cause of death in GC users. As we did not have blood samples from the patients, we were unable to measure the levels of cortisol in plasma. Instead, we examined medical records for patients where the recorded cause of death was sepsis, since sepsis and adrenal crisis share many symptoms, and infection is one of the most common causes of adrenal crisis. This method has limitations since it is impossible to be 100% certain that adrenal insufficiency was the cause of death. Nevertheless, a retrospective study is probably the only method to investigate this topic. Another limitation is that we only included patients treated with oral GCs. Therefore, patients using other routes of administration, such as GC inhalation, topical, and injection, were not included. The number of individuals at risk of GC-induced adrenal insufficiency-related death is therefore probably underestimated. The main strength of our study is that three experienced investigators reviewed all suspected medical records in a blinded process.

It is necessary to conduct further research in order to determine whether GC users and health-care providers have sufficient knowledge about GC stress doses during intercurrent illnesses. The need for clinical guidelines for the management of GC-induced adrenal insufficiency is urgent and the evidence regarding when hydrocortisone treatment is indicated is lacking (14).

There are unanswered questions regarding how to improve the safety of patients taking GCs. Long-term GC users might benefit from a steroid emergency card. Moreover, a warning signal in an electronic medication management system could be useful if a physician were to accidentally interrupt long-term GC treatment.