Article Text

Download PDFPDF

Safety of the insulin tolerance test
  1. P J Galloway1,
  2. E McNeill2,
  3. W F Paterson2,
  4. M D C Donaldson2
  1. 1Department of Biochemistry, Royal Hospital for Sick Children, Yorkhill, Glasgow, UK
  2. 2Department of Child Health, Royal Hospital for Sick Children
  1. Dr P Galloway, Consultant Clinical Biochemist, Department of Biochemistry, Royal Hospital for Sick Children, Yorkhill, Glasgow G3 8SJ, UK;


Concerns have been raised about the hazards of the insulin tolerance test (ITT), used to measure growth hormone secretion. In Glasgow, we continue to use this test, adhering to a strict protocol. A review of outcome over a 10 year period (1989–99), during which 550 ITTs were performed, was undertaken. No serious adverse events occurred; in particular, no child fitted or required intravenous glucose. Fewer tests were done during the latter five years, with a higher yield of growth hormone (GH) deficiency, reflecting our increasingly conservative approach to paediatric GH therapy during this period. We conclude that the ITT is safe and reliable in a paediatric setting provided that a strict procedure is followed.

  • insulin tolerance test
  • growth hormone secretion
  • hypoglycaemia
  • GH, growth hormone
  • ITT, insulin tolerance test

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

In 1991, following several fatalities in the UK,1 the Department of Health issued a circular2 warning of the hazards of the insulin tolerance test (ITT), stating that: “the insulin tolerance test should not be used in children when only growth hormone reserve needs to be tested”.

Since then many paediatric centres have abandoned the ITT as a means of assessing growth hormone (GH) reserve although it remains the standard diagnostic test for GH deficiency in adults, in view of its sensitivity and reproducibility.3,4 We have continued to use the ITT as our first line test, reserving the arginine test for children with epilepsy or cardiac disease and those under 5 years.

We have reviewed retrospectively our experience of performing ITTs over a 10 year period (1989–99), with particular attention paid to the last five years, during which a dedicated nurse specialist has been overseeing the tests. Our principal aim was to assess the safety of the ITT, defining the key morbidity measures as: (a) hypoglycaemia requiring emergency administration of intravenous glucose; (b) hypoglycaemic convulsions; and (c) death. We have also examined the timing of the glucose nadir following insulin administration as this has practical implications for oral glucose administration. Finally, in the light of a change in the pattern of GH prescribing in Scotland over the past 10 years,5 we have noted the annual number of ITTs performed.


The following strict ITT protocol was adhered to:

  • Child must be fasted overnight.

  • Blood glucose must be >3.0 mmol/l at time −30 minutes and 0 minutes.

  • Oxygen, glucose, and hydrocortisone must be available.

  • A doctor must be present for 45 minutes following insulin administration and a nurse specialist throughout. (Protocol since 1995, when an endocrine nurse specialist was appointed. During the first five years of the study period, endocrine tests were carried out on a paediatric inpatient ward by junior medical staff.)

  • Child must eat and remain on ward for one hour before cannula removal and discharge home.

  • Administration of insulin. An insulin concentration of 1 unit/ml is used and a dose of 0.15 units/kg is given (0.1 units/kg if panhypopituitarism is suspected). The doctor must sign for the insulin and the nurse must check the dosage.

  • Blood sampling. Glucose and growth hormone concentrations are measured in 4 ml samples drawn at −30, 0, 15, 30, 60, 90, and 120 minutes. The rationale of the −30 minute sample is to capture any stress related GH surge, as this may be followed by a refractory period, giving the misleading interpretation of GH deficiency.

  • Management of hypoglycaemia. If the blood glucose drops below 2.2 mmol/l and/or the child is symptomatic, 30–40 ml of Lucozade (up to 10 ml/kg body weight) is given. If the child becomes unrousable or is fitting, 0.5 g/kg dextrose is given intravenously, and blood glucose rechecked. Thereafter 50–100 mg hydrocortisone, followed by a 10% dextrose infusion at 0.1 ml/kg/min is given. A further dextrose bolus is administered only if blood glucose remains <6.0 mmol/l.

Statistical analysis

The relative risk of a future adverse event occurring as a result of the procedure, assuming that none had occurred to date, was calculated according to a simple formula, first described by Hanley in 1983,6,7 which states that: MathMath where n = number in series (i.e. number of past, uneventful procedures).


A total of 550 ITTs were performed between 1989 and 1999. No serious adverse events occurred. In particular, no child lost consciousness, fitted, or required intravenous glucose following hypoglycaemia. The upper limit of the confidence interval for relative risk from the procedure is therefore 3/550 = 0.005 = 0.5%. Glucose profiles were documented for the 223 ITTs performed since 1995. The glucose nadir occurred at 15 minutes in 118 (53%) cases (range 0.6–2.2 mmol/l) and at 30 minutes in 105 (47%) cases (range 1.1–2.3 mmol/l). Between 1989 and 1992 the annual number of ITTs performed ranged from 57 to 77. During this period GH values of 10–20 mU/l were recorded in 27%. From 1993 to 1999, the annual number of ITTs performed decreased to 37–52 and the number of cases with GH values of 10–20 mU/l rose to 43%. Over the 10 year period GH values <10 mU/l were recorded in 14%. GH concentrations were >20 mU/l at time −30 minutes in 16%, indicating a normal GH response to stress, thus precluding GH deficiency.


In this assessment of the safety of the ITT in childhood we report no serious adverse events in the 550 cases studied. We acknowledge that this finding cannot preclude adverse events occurring in future patients but the probability, based on the numbers of patients studied, is, at most, 5 per thousand. Indeed no test that alters glucose homoeostasis can be regarded as safe. Even the simple act of fasting a severely hypopituitary child may cause symptomatic hypoglycaemia. However, our experience indicates that the ITT is relatively safe provided that this endocrine investigation is carried out in a well resourced, specialised setting by experienced, competent staff. A similar conclusion was reached by Jones et al, who carried out an audit of 161 ITTs performed in adults during a 12 month period.8 In this study one serious adverse event (a grand mal fit) occurred, but the patient recovered rapidly with no neurological sequelae following administration of intravenous dextrose and hydrocortisone.

It was not possible, in a retrospective study of this nature, to evaluate the degree of distress which may have been caused by the process of venepuncture, and by symptoms of insulin induced hypoglycaemia such as drowsiness, sweating, and malaise. Nor was it our remit to compare the symptoms of insulin induced hypoglycaemia with those caused by other pharmacological tests of GH secretion, such as clonidine (for example, drowsiness, hypotension),9–11 or glucagon (for example, nausea, headache, sweating, vomiting).12

During the 10 year study period an endocrine specialist nurse was appointed (1995) and a ward dedicated for day case investigation of children was established (1997). While it would be difficult to measure objectively an improvement in service resulting from these changes, we have no doubt that the standard of day care is much improved. Moreover, in the light of the recent changes to the working practice of junior medical staff , we consider that our endocrine day case service would be untenable without the consistent input from a specialist nurse.

The glucose nadir occurred at 15 minutes in just over half of our patients and by 30 minutes in virtually all. This finding has led us to favour the pre-emptive use of Lucozade (a readily available 20% glucose drink) at 15–20 minutes to avoid more profound hypoglycaemia developing.

Since 1993, we have been more selective in our approach to growth hormone therapy.5 This is reflected in a decrease in the number of children tested and a higher yield of subjects with peak GH values of <20 mU/l, the biochemical measure of GH deficiency, and indicates improved targeting of children likely to benefit from investigation.

Not only are we satisfied as to the relative safety of the ITT when performed by experienced staff using a strict protocol, but we would also argue that it is in many ways preferable to other tests of GH reserve. The stimulus of hypoglycaemia, a naturally occurring phenomenon, is arguably more physiological than that caused by pharmacological agents. Moreover, insulin is easily administered and the resultant hypoglycaemia is readily reversible. The elective use of Lucozade ameliorates the symptoms of hypoglycaemia without affecting the counter regulatory response.

Other methods of GH stimulation are not without their drawbacks. While arginine administration does not result in hypoglycaemia—the increase in insulin release is balanced by a concomitant increase in glucagon, resulting in a net rise in plasma glucose13—we have encountered problems with venous occlusion caused by the viscosity of the infusion, sometimes necessitating replacement of the intravenous cannula. We have no experience of other agents, but it is well recognised that clonidine causes unpleasant symptoms such as drowsiness and hypotension which are not rapidly reversible.9,10 In addition, clonidine has recently been shown to be associated with hypoglycaemia, although the mechanism of action is unclear.11 Glucagon administration causes nausea and vomiting,12 and may also result in hypoglycaemia by stimulating insulin release.14

Our observation that no child required intravenous dextrose in the treatment of hypoglycaemia emphasises that in virtually all cases oral glucose administration is all that is required. Indeed, fatalities associated with the ITT have involved the administration of large quantities of intravenous dextrose. There is no place for large quantities of either hypotonic (for example, 5%) or hypertonic (50%) dextrose in insulin induced hypoglycaemia. If necessary, the child should merely receive 0.5 g/kg of dextrose as a 10% solution, followed by 10% dextrose with 0.45% saline at maintenance rates.

In highlighting the favourable safety record of the ITT in our hospital we are not advocating the widespread reintroduction of this method of anterior pituitary function testing, especially in departments where relatively small numbers of children are involved. Indeed, given the controversies surrounding the selection of children for investigation and treatment of GH deficiency,4,15 we believe that the decision to embark on growth hormone stimulation testing should only be made by paediatricians who are experienced/trained in paediatric endocrinology, and/or working in close collaboration with a regional centre.