Objective: We audited our practice of insulin tolerance testing (ITT) in terms of safety and technical success. We reviewed the results of those tests performed over a 12-month period. By relating peak cortisol response to 0900 h screening cortisol level, we determined whether we could reduce the number of tests performed.
Design: The results of all ITTs performed on our unit between 1 January and 31 December 1991 were reviewed.
Patients and measurements: Patients were prescreened by measurement of serum cortisol and thyroxine, and recording of an electrocardiogram. A subnormal serum thyroxine (< 58 nmol/l), 0900 h serum cortisol (< 100 nmol/l) or an abnormal ECG were taken as contraindications to the test. Minimum glucose and maximum cortisol levels were recorded, along with peak GH responses when measured. The peak cortisol response was compared to the screening and basal cortisol levels.
Results: A total of 161 tests were performed, 135 of which fulfilled our inclusion criteria. The test was technically successful in all but 5 of these; a significant adverse event occurred in one patient with full recovery after reversal of hypoglycaemia. Thirty-two patients had a suboptimal serum cortisol response to hypoglycaemia: screening cortisol level ranged from 128 to 493 nmol/l and 0900 h serum cortisol measured prior to the ITT from 97 to 431 nmol/l. The lowest level of screening cortisol above which all patients would be expected to achieve the normal peak cortisol of 580 nmol/l or over is therefore 494 nmol/l. If this cut-off level had been adopted, 10 (8%) ITTs need not have been performed if their only purpose had been to assess cortisol reserve. Altering the criterion for the necessary peak cortisol to 500 nmol/l did not affect the number of ITTs required. Our lower limit for testing could not be revised upwards from 100 nmol/l. Adequacy of cortisol reserve did not predict a normal GH response to insulin-induced hypoglycaemia.
Conclusions: When performed in an experienced endocrine unit with adequate supervision, the insulin tolerance test is a safe procedure. According to the current sample, fewer tests would be performed without detriment to patient care if those with a screening cortisol of greater than 500 nmol/l did not proceed to testing, unless the purpose of the test was also to exclude GH deficiency. A lower limit of 100 nmol/l appears reasonable and need not be revised upwards.