1.1.1. Introduction

Exogenous glucocorticoids are used for their anti-inflammatory and immunosuppressive properties across many conditions ranging from asthma, inflammatory bowel disease, polymyalgia rheumatica and organ transplantation. Re-occurrence of symptoms during glucocorticoid withdrawal may reflect the return of the original disease for which the steroids were started or adrenal insufficiency, unmasked by the reducing steroid dose.

Mild symptoms during withdrawal of exogenous corticosteroids are an expected and common occurrence and generally do not indicate unmasked adrenal insufficiency. However, if there is underlying adrenal insufficiency, either owing to adrenal suppression or because of medication use or intrinsic pituitary/adrenal disease it is potentially life-threatening and needs to be taken seriously.

This evidence review seeks to address the approach required in managing patients being withdrawn from glucocorticoid therapy, and establishing when someone should be referred to a specialist for investigation of adrenal insufficiency, related to HPA-axis suppression.

1.1.2. Summary of the protocol

For full details see the review protocol in Appendix A.

Table 1

PICO characteristics of review question.

1.1.3. Methods and process

This evidence review was developed using the methods and process described in Developing NICE guidelines: the manual. Methods specific to this review question are described in the review protocol in Appendix A and the methods document.

Declarations of interest were recorded according to NICE’s conflicts of interest policy.

1.1.4. Diagnostic evidence

1.1.5. Summary of studies included in the diagnostic evidence

Table 2

Summary of studies included in the evidence review.

See Appendix D for full evidence tables.

1.1.6. Summary of the diagnostic evidence

Table 3

Clinical evidence summary: diagnostic test accuracy for morning serum cortisol for diagnosing AI in people being assessed for HPA recovery following supraphysiological therapeutic doses of oral glucocorticoids.

Table 4

Clinical evidence summary: diagnostic test accuracy for morning salivary cortisol for diagnosing AI in people being assessed for HPA recovery following supraphysiological therapeutic doses of oral glucocorticoids.

Table 5

Clinical evidence summary: diagnostic test accuracy for salivary cortisone for diagnosing AI in people being assessed for HPA recovery following supraphysiological therapeutic doses of oral glucocorticoids.

1.1.7. Economic evidence

1.1.8. Economic model

This area was not prioritised for new cost-effectiveness analysis.

1.1.9. Unit costs

Relevant unit costs are provided below to aid the consideration of cost-effectiveness.

Table 6

Unit costs of tests and referral.

1.2.1. The outcomes that matter most

1.2.2. The quality of the evidence

One UK-based diagnostic study was included in this review. It reported the diagnostic accuracy of early morning serum cortisol, salivary cortisol and salivary cortisone in a specific adult population of people having long-term exogenous glucocorticoids withdrawn. The study specified that the withdrawal from the long-term therapeutic doses of oral corticosteroids involved tapering the steroid dose to 5 mg of prednisolone or equivalent and then switching to hydrocortisone. Assessment of the HPA axis was then performed with a short Synacthen test once the patient had been on the replacement dose for at least 4 weeks. The committee agreed that this study was relevant to the review question but in current practice, clinicians would not routinely recommend switching from prednisolone to hydrocortisone.

The reference standard used in this study was the 0.25 mg short Synacthen test and if the 30-min serum cortisol was ≥450 nmol/L it was defined as an adequate response.

This study provided sensitivity and specificity data for the three index tests at different cut points in order to maximise specificity when looking for HPA recovery and to maximise sensitivity when looking for HPA suppression.

The quality of the evidence was assessed by an adapted GRADE framework and rated moderate for the majority of outcomes. These were downgraded on the QUADAS-2 checklist for risk of bias due to unclear reporting of patient selection. Several outcomes were rated low quality and downgraded due to imprecision. This arose from the confidence interval crossing one of the decision thresholds (thresholds set at 60% and 90% for sensitivity; and 50% and 70% for specificity). Overall, the evidence available was very limited and based on one study with 47 participants.

1.2.3. Benefits and harms

The committee considered the diagnostic accuracy data available for serum and salivary cortisol and salivary cortisone, in diagnosing HPA suppression while withdrawing from corticosteroids. They concluded that there were no real differences in terms of the accuracy data for serum cortisol at 170 nmol/L, salivary cortisol at 1.92 nmo/l and cortisone at 9.42 nmol/L, with 100% sensitivity reported across the tests (specificity was: 59% for serum cortisol, 51% for salivary cortisol and 54% for salivary cortisone). These thresholds were set to avoid any false negative results and to not miss any potential diagnoses. Alternate thresholds were reported to identify HPA recovery, and these were set to maximise specificity at 100% and limit false positive results. The corresponding sensitivity values for serum cortisol at 365 nmol/L were 26.7%, salivary cortisol at 25.4 nmol/L were 0% and salivary cortisone at 37.2 nmol/L were 26.7%.

As discussed in review D the committee decided to make a recommendation for serum cortisol over the other index tests due to its widespread availability, and cost-effectiveness. They agreed that while the evidence base for salivary cortisol and cortisone is growing, there are only 2 test centres in the UK able to interpret these tests. Therefore, due to resource implications and practical issues, the committee were not able to recommend these tests in the UK at present.

Ultimately, the committee agreed that a 9 am serum cortisol test should be considered for people who develop signs and symptoms of adrenal insufficiency while they are withdrawing from long-term glucocorticoids, providing they have attempted a slow taper and are down to physiological doses. The committee specified that a 9 am cortisol test should not be attempted prior to this as the patient would still be covered by their steroid doses.

If a patient is attempting a slow taper and becomes symptomatic, the committee recommended that corticosteroids be paused for a period, (12 hours from hydrocortisone, 24 hours from prednisolone and 72 hours from dexamethasone) so that a 9 am cortisol test can be carried out. Steroids can then be resumed at the physiological dose they were on previously. Signs and symptoms of when to suspect AI when tapering, are covered in review B.

The committee deliberated on, at which cut-off should a referral be made for further testing with a short Synacthen test. They discussed the cut-offs presented in the evidence, however, the available evidence was limited, the cut-offs reported lower than expected, and the evidence not robust enough to base recommendations on for the guideline. Therefore, the group decided to use their experience and consensus opinion to make a recommendation.

They took into account the specific population of people withdrawing from glucocorticoids but agreed that the cut-off points for 9 am cortisol tests should be the same as those used to screen for adrenal insufficiency in a general population, as recommended in review D (recommendation 1.2.6). The committee specified that if the 9 am cortisol test result comes back below 150 nmol/L then a referral to endocrinology or for a short Synacthen test should be arranged immediately. However, if test results are between 150 and 300 nmol/L (‘grey zone’) the 9 am cortisol test could be repeated. If it remains in this grey zone after a second test, then a referral to endocrinology should be made. The committee agreed that if the test is above 300 nmol/L adrenal insufficiency is very unlikely, and that the tapering regimen can continue.

Due to the limited evidence available and the uncertainty in the population size, the committee made a weaker ‘consider’ recommendation for when to refer for a 9 am cortisol test in people who become symptomatic while tapering from long term glucocorticoids.

1.2.4. Cost effectiveness and resource use

No health economic studies were identified for this review. Unit costs were presented and discussed with the committee to aid their consideration of cost-effectiveness.

Serum cortisol tests were estimated to be approximately £6 per test. The costs include the cost of a blood test taken either in the community at a GP practice by a health care assistant (ten minutes of a Band 3 health care assistant time £4.33) or in an outpatient hospital setting by a phlebotomist (£4.70) and the cost of laboratory analysis (£1.55 for clinical biochemistry).

Salivary cortisol testing includes the cost of the ‘Salivette’ to collect the saliva and then the laboratory and postage costs. The former was estimated to be £0.37 per unit, the latter £27.80. The committee noted that salivary cortisol testing is not widely used and there are only two testing centres in the UK with machines that enable the analysis of mass spectrometry assays. It was noted that there could be significant set-up costs if it were to be recommended due to the current lack of infrastructure available for analysing these samples nationally. The committee noted that reliance on the available laboratories may result in delays in results of up to 5 weeks which would be impractical and possibly clinically inappropriate.

The unit cost of referral to secondary endocrinology care was presented, this was £293 for adults and £418 for paediatrics (a weighted average of consultant-led, non-consultant led and multi-professional cost of first endocrinology face-to-face appointments). The unit cost of a short Synacthen test was presented as a day case cost (£398). The cost of Synacthen would be included in the day case cost as it is not listed as a high-cost drug. This is a broad HRG unit cost which will cover many interventions so may not provide an exact estimate of the cost of this test. In addition, the committee noted that the unit cost of Synacthen in hospitals currently is higher than that listed in the drug tariff / BNF (£45 versus £38). Insulin tolerance tests are likely to be bundled in the same day case cost as a short Synacthen test, but the committee noted it is likely to be more expensive as it is a more resource-intensive test. The cost of an insulin tolerance test was calculated as part of the NICE Medtech innovation briefing [MIB320] on ‘Macimorelin for diagnosing growth hormone deficiency’ at £470.

The committee decided to set a cut-off for onwards referral at ≤200 nmol/L or for repeat testing if the value is between 201 – 300 nmol/L. This was based on the recommendations made for testing for adrenal insufficiency in a general population The committee deliberated that higher cut-offs would increase the risk of false positive results and lead to increased cost to the NHS associated with unnecessary referrals and short Synacthen testing.

The committee agreed that in people who develop signs and symptoms of adrenal insufficiency after slow tapering of long-term glucocorticoids, an 8–9 am serum cortisol should be considered providing they are down to physiological doses. The committee recommended 9 am serum cortisol tests over the other index tests (for example salivary cortisol) as they are readily available, relatively inexpensive and would ultimately lead to cost savings if referrals for unnecessary short Synacthen tests can be avoided. The timing of the test and the importance of tapering glucocorticoids to physiological doses were stressed to ensure the serum cortisol test measures accurate peak cortisol levels. Tests carried out later in the day (i.e., 10 am) or without tapering would not be of any clinical use and could lead to unnecessary referrals.

The exact size of the population at risk is difficult to accurately estimate. The committee noted that the number of people who are potentially at risk of HPA-axis suppression following long-term glucocorticoid use is uncertain due to a lack of suitable data. Furthermore, although there are one million glucocorticoid prescriptions per year in England, it is unclear the number of individuals receiving glucocorticoid prescriptions or what proportion are receiving them for over 4 weeks. Due to the uncertainty in the population size and to minimise the resource impact to the NHS, the committee were keen to restrict testing to those who develop signs and symptoms after attempting a slow taper, as opposed to all those withdrawing from long-term glucocorticoids and made a weaker ‘consider’ recommendation.

Of note a research recommendation looking at the diagnostic accuracy and cut-offs for referral (including cost-effectiveness) for salivary cortisol tests has been made as part of the review question looking at testing for adrenal insufficiency in a general population.

1.2.5. Recommendations supported by this evidence review

This evidence review supports recommendation 1.9.8 to 1.9.10.

B.1. Clinical search literature search strategy

Searches were constructed using a PICO framework where population (P) terms were combined with Intervention (I) and in some cases Comparison (C) terms. Outcomes (O) are rarely used in search strategies as these concepts may not be indexed or described in the title or abstract and are therefore difficult to retrieve. Search filters were applied to the search where appropriate.

Download PDF (211K)

B.2. Health Economics literature search strategy

Health economic evidence was identified by conducting searches using terms for a broad Adrenal Insufficiency population. The following databases were searched: NHS Economic Evaluation Database (NHS EED - this ceased to be updated after 31st March 2015), Health Technology Assessment database (HTA - this ceased to be updated from 31st March 2018) and The International Network of Agencies for Health Technology Assessment (INAHTA). Searches for recent evidence were run on Medline and Embase from 2014 onwards.

Download PDF (170K)

I.1. Clinical studies

I.2. Health Economic studies

None.