Archives of Disease in Childhood 1995; 72: 362-368

PERSONAL PRACTICE

An assessment of growth hormone provocation

tests

P C Hindmarsh, P G F Swift

There are a number of tests available forassessment of anterior pituitary hormone func-tion. Recent concern has centred on the use ofprovocative tests for growth hormone releaseand in particular the glucagon and insulininduced hypoglycaemia or insulin tolerancetest (ITT). These tests have been associatedwith cerebral oedema and death of at least twochildren and, as a result, questions have beenraised as to the current management andinvestigation of children with potential abnor-malities of growth hormone secretion. I

This review describes the background andtheory behind the tests as well as an appraisalof some of the more common methods ofassessing growth hormone release. It does notprovide detailed information on the construc-tion of 12 or 24 hour profiles of growth hor-mone secretion, as it is not felt that these areapplicable to routine clinical investigation.

Central to the use of any test are its perfor-mance characteristics in terms of efficiency,sensitivity, and specificity. Attention must alsobe paid to the method used to measure growthhormone in the serum. This review attempts toassess the importance of these issues.

Endocrine Unit,Cobbold Laboratories,Middlesex Hospital,Mortimer Street,London WIN 8AAP C Hindmarsh

The Children'sHospital, LeicesterRoyal Infirmary,LeicesterP G F Swift

Correspondence to:Dr Hindmarsh.

BackgroundA comprehensive assessment of anteriorpituitary function requires the proper use andinterpretation of well described combined func-tion tests.2 Classical endocrine function testsconsist of the administration of a stimulusknown to produce a quantifiable response ofthehormone thought to be deficient. For assess-

ment of growth hormone and adrenocorti-cotrophic hormone (ACTH) insulin inducedhypoglycaemia (ITT) has been for many yearsthe 'gold standard'. The advantage of using theI=T is that a clearly described standardisedstimulus is applied, a quantifiable hypogly-caemia stimulus is produced, and the endocrineresponses have been defined in adults at least.As an assessment of the cortisol response to a

stressful stimulus it is the only test that has beenvalidated (against surgical stress).3 As withother provocation tests, the function of thegonadotroph and thyrotroph cells can be testedconcurrently by the administration of theirrespective releasing hormones.When should combined anterior pituitary

function tests be performed and when onlya growth hormone provocation test?

Epidemiological studies would suggest thatidiopathic isolated growth hormone deficiencyoccurs more frequently than multiple pituitaryhormone deficiency. It is not always possible,however, for the clinician to be sure that thereis no other associated anterior pituitaryhormone deficiency in addition to growthhormone insufficiency. Further, the loss ofanterior pituitary hormones in multiplepituitary hormone deficiency most often startswith growth hormone. This pattern of loss maybe of relevance in ascertaining the correctunderlying diagnosis and whether other inves-tigations such as a computed tomography ormagnetic resonance imaging are required.Thyroxine and cortisol are important in theregulation of the growth hormone gene so thatknowledge of thyroxine and cortisol status isessential before embarking on growth hormonetesting. Thyroxine can be easily measured on asingle blood specimen but the integrity of thepituitary-adrenal axis is more difficult to assesswith a single random morning blood samplebecause of the pulsatility of cortisol.4 Inpractice most clinicians before embarking ongrowth hormone assessment in a slowly grow-ing child would attempt to confirm adequatethyroid and cortisol concentrations as well asmeasuring a full blood count and assessingrenal function. Antigliadin antibodies might bemeasured and a karyotype examined particu-larly in girls. Any doubt about the cortisolstatus makes a more comprehensive assess-ment in the specialised endocrine centreabsolutely mandatory.Growth hormone release is under the con-

trol of the two hypothalamic peptides, growthhormone releasing hormone (GHRH) andsomatostatin. Recent work has suggested thatendogenous GHRH is extremely important indetermining the magnitude of the growth hor-mone secretory burst, whereas somatostatin isimportant in determining the timing of thatevent.5 The stimuli used to provoke growthhormone secretion probably act through eitheror both of these peptides. In man there is noclear evidence to determine upon which of thepathways the various stimuli act. Therefore, itis difficult to compare the results of individualtests, although it is generally accepted that acut off growth hormone peak response of7 5-10 pg/l (15-20 mU/l) is a reasonable valuefor determining whether an individual isgrowth hormone sufficient or insufficient.

362

on March 24, 2022 by guest. P

rotected by copyright.http://adc.bm

j.com/

Arch D

is Child: first published as 10.1136/adc.72.4.362 on 1 A

pril 1995. Dow

nloaded from

An assessment ofgrowth hormone provocation tests

Table 1 Effect of applying different peak serum growthhormone values in response to ITTfor diagnosis ofgrowthhormone insufficiency using height velocity as standard

Growth hormoneconcentration Efficiency Sensitivity Specificity(mU/I)* (%) (%) (%)10 63 51 7913-5 66 64 701 5 67 69 6320 68 82 49

*By immunoradiometric assay'0; 1 ,ug/1=2 6 mU/l.

Interpretation of growth hormone testsThe aim of any diagnostic test is to comple-ment the clinical history and examination tothe point where false positive and false negativediagnoses are reduced to a minimum. Thesensitivity and specificity of the test in questiongive guidelines to the clinician as to how likelya correct answer confirming or refuting thediagnosis is likely to be. Sensitivity and speci-ficity values of over 80% are generally acceptedas indicating an acceptable test performance.Defining the gold standard is often difficult6 7and the accepted cut off point for growthhormone concentrations remains debatableparticularly when it is remembered that normalchildren may have very low peak values.8

If it is accepted that growth rate is the mostimportant index against which growthhormone measurements should be compared,then the test parameters could be defined asfollows: the efficiency of the test would be thepercentage of all results which were true posi-tive or negative - a negative result being whenan arbitrary (but agreed) concentration isexceeded and a child is shown to have a normalgrowth velocity. A positive result would be whenthe cut off value is not exceeded in a child witha subnormal height velocity. With these defini-tions, the sensitivity of the test would becalculated as the number of patients growingpoorly with a positive test divided by the totalnumber of children growing poorly. Specificitywould be the number of children growingnormally with a negative test, divided by thetotal number of children growing normally.Both values are expressed as percentages.Few comparative studies have used growth

rate as part of the definition for interpretingresults. What has usually happened is that theITT and other tests have been compared. Notsurprisingly the ITT has scored extremely wellbecause, being the standard, it is allocated avalue of 100%!9 Table 1 provides an analysis ofITT performance against growth rate10 and

Table 2 Performance characteristics of tests ofgrowthhormone secretoty status compared with the ITT using acut off value of 7-5 ,ug/l (15 mU/l)

Efficiency Sensitivity Specificity Reference(0) (0) (%) No

SleepMonitored byEEG 88 86 95 11

Not monitored 79 67 82 12Arginine 72 73 85 13Clonidine 70 70 85 14IGF-1 75 95 60 15IGFBP-3 96 97 95 15

EEG=electroencephalography.

table 2 shows the efficiency, sensitivity, andfalse negative rate of a number of tests pub-lished in the literature compared with the ITT.There is little to choose between any of thetests.A number of reasons have been proposed to

explain the discordance between the resultsobtained from growth hormone testing com-pared with prolonged observation of growthvelocity in children. First, a raised growth hor-mone concentration at time 0 during the testleads to difficulty in interpretation. Tests aredesigned to evaluate the secretory pathway,particularly the readily releasable pool ofgrowth hormone. If there has recently been apulse of growth hormone and the pool is low,the subsequent response will be attenuated.'3Ideally, before any test, there should be a runin period where samples are obtained for atleast 30 and preferably 60 minutes beforethe study in order that the response can beinterpreted in the light of possible previoussecretory episodes.

Second, children's growth varies with theseasons and there is evidence to suggest cyclesof periodicity greater than one year.16 Growthhormone secretion is related to growthrate17-19 and it is possible that some of the dis-cordance could be explained by the perfor-mance of the test during a period of relativelyfaster growth when growth hormone secretionwas greater whereas annualised growth ratemight indicate a different picture of overallgrowth failure. This highlights the difficulty ofcomparing one year of growth data with a oneday stimulation test. Further, the relationbetween growth rate and growth hormonesecretion is a continuum so any cut off valuechosen to separate normality from abnormalityis a compromise balancing sensitivity andspecificity.A third source of discordance is that a

particular stimulus might release pituitaryhormones when the problem is in the hypo-thalamus. It is important therefore to usestimuli which seem to act predominantlyon the hypothalamus (for example, ITT,clonidine) rather than the pituitary gland (forexample, GHRH).

Finally, the reproducibility of any test needsto be considered. A number of studies havedemonstrated that the reproducibility of thephysiological and pharmacological methods ofassessing growth hormone secretion in anindividual is poor.20-22 Interindividual andintraindividual variation is considerable andmakes interpretation difficult (table 3).

Table 3 Within individual coefficients of variation forgrowth hormone tests (derivedfrom Saini et al22)

Coefficient ofvariation (%o)

Mean 24 hour growth hormoneconcentration 35-0

Urinary growth hormone 30-1 (prepubertal)43-9 (pubertal)

Clonidine 58-0Arginine 77-5Levodopa 89-2GHRH 60-0Thyroxine 13 0

363

on March 24, 2022 by guest. P

rotected by copyright.http://adc.bm

j.com/

Arch D

is Child: first published as 10.1136/adc.72.4.362 on 1 A

pril 1995. Dow

nloaded from

Hindmarsh, Swift

Growth hormone testsThese are classically divided into physiologicaland pharmacological tests. Practical details ofendocrine investigations are given by Hughes2and what follows is a commentary on the indi-cations for and recent ideas on testing forgrowth hormone insufficiency. As a generalprinciple these investigations should be per-

formed only in centres with personnelexperienced in and dedicated to the perfor-mance of such investigations and in labora-tories regularly performing internal and externalquality assessment of assay methodology.

(A) INDICATIONS FOR GROWTH HORMONETESTING1. Inadequate growth velocityThis is the prime indication for investigation.Short stature in itself is not an indication fortesting and it is, therefore, important not to use

growth hormone measurements or provocationtests as a screen for all individuals with shortstature. The decision to investigate the growthhormone axis should be based on the knowl-edge that growth velocity is inadequate, thatother explanations for growth failure have beenexcluded as far as possible, and that there is a

real possibility of hypothalamic or pituitarypathology.

In practice, this means careful observationof growth and calculation of growth rate at an

interval not less than six months and preferably12 months. Growth rates vary from year toyear'6 but the following rules can be applied. Ifheight velocity, measured over one year, is lessthan the third height velocity centile, investiga-tion is warranted because the chances of thenext year's growth rate being normal is only3%. For children whose growth rates are

between the third and 50th centiles observa-tion for a second year is advisable. The chanceof a normal child growing for two successiveyears with a height velocity less than the 25thcentile is 0-25X0v25=0-0625. In other words,only 6.25% of the children would be expectedto yield normal results on investigation. Thelonger a velocity remains below the 50th cen-

tile, the more likely it is that investigation willreveal an abnormality.23 The new nationalgrowth charts encompass this concept so thatany child who crosses from one of the new cen-

tile divisions to another during the course ofone year requires evaluation.24

2. Neonatal problemsNeonates or infants with prolonged jaundiceand hypoglycaemia that might be associatedwith multiple pituitary hormone deficiency are

candidates for investigation of growth hor-mone secretory status. This must be done onlywhen thyroxine and cortisol deficiencies havebeen corrected. Extreme care needs to betaken in the performance of the growth

hormone provocation tests and, if hypogly-caemia is present on the time 0 minute sample,no further stimulus is required. Hypoglycaemiashould be an adequate stimulus and all thatshould be done is the collection of two samples

separated in time by five to 10 minutes forassay of growth hormone.

3. Growth failure following cranial irradiation forbrain tumours or leukaemiaCranial irradiation for brain tumours is associ-ated with growth failure, growth hormoneinsufficiency (in approximately 90% ofchildren) and other anterior pituitary hormonedeficiencies may also be present. The preva-lence of growth hormone insufficiency aftercranial irradiation for leukaemia is less clear.Investigation should be based on clinicalobservation of growth after radiotherapy withor without chemotherapy.

4. Pubertal delayIn certain situations, it may be difficult toascertain whether a child who is growingslowly with constitutional delay of growth andpuberty is going to have an adequate pubertalgrowth spurt. In such a situation, it may bevaluable to perform a growth hormoneprovocation test following 'priming' withexogenous sex steroid before the start of theinvestigation. The aim of this investigation isto ascertain whether there is going to be suffi-cient growth hormone available during thepuberty growth spurt. Priming has usuallybeen performed in males by giving a singleintramuscular injection of testosterone(100 mg) three days before the investigationor in females by giving ethinyloestradiol orstilboestrol (1 mg twice a day for two days).The latter can actually be used quite satisfac-torily in both sexes.8 25 Using high doses ofethinyloestradiol is unnecessary and causesnausea and headache.

(B) PHYSIOLOGICAL TESTS OF GROWTHHORMONE SECRETIONSerum growth hormone concentrations duringthe daytime are low and because growthhormone secretion is pulsatile there is no placefor random serum growth hormone measure-ments. As a general principle growth hormoneresponses are blunted in all tests in the pres-ence of hypothyroidism, cortisol excess, andobesity. Four types of assessment of physio-logical secretion can be performed.

1. Exercise testThe exercise test should be performed with abicycle ergometer and relies on the generationof a standard amount of work. At best the testgives 20% false positive results2 but whenpoorly performed the diagnostic yield is verylow with at least 20-30% falsely low growthhormone concentrations 25 minutes aftercompleting the exercise.

2. Sleep studiesSleep studies have been used to investigatephysiological growth hormone secretion.These tests can be performed with or without

364

on March 24, 2022 by guest. P

rotected by copyright.http://adc.bm

j.com/

Arch D

is Child: first published as 10.1136/adc.72.4.362 on 1 A

pril 1995. Dow

nloaded from

An assessment ofgrowth hormone provocation tests

electroencephalographic monitoring, the aim isto obtain blood samples for growth hormone,at 15-20 minute intervals during stage 3 and 4sleep. Performance characteristics are shownin table 2. This type of study when performedregularly in specialised units is well acceptedby children but places heavy demands on

manpower. It should be stressed that the rela-tionship between sleep and maximum growthhormone release is one of association ratherthan causality which may be affected in a

number of disease situations, for exampledeprivation syndromes.26 27

3. Urinary growth hormone measurementUrinary growth hormone has been proposed as

a non-invasive method of assessing growthhormone secretory status.28 With improvedradioimmunoassay and enzyme linkedimmunosorbent assay (ELISA) techniques,increased assay sensitivity allows measure-

ments in small volumes of urine. Variability ishigh and two or three overnight collections are

advised to overcome this problem. Timedovernight samples are recommended but thereappears to be little to be gained by expressingvalues in terms of creatinine excretion, which isalso highly variable (coefficient of variation27%).29 Renal function needs to be assessedbefore the study is conducted. Normal rangeshave been described, but there is considerableoverlap between groups of children with differ-ent abnormalities of growth hormone secre-

tion.29-32 This reduces the specificity andsensitivity ofthe test to some extent but urinarygrowth hormone measurement in some cir-cumstances does perform well against conven-

tional tests with efficiency of 90%, sensitivity of80%, and specificity of 98%.29

4. Insulin-like growth factor-i and binding proteinsWith the increasing availability of specificradioimmunoassays for insulin-like growthfactor-i (IGF-1), the possibility ofusing a singleblood test to aid in the diagnosis of growthhormone insufficiency has been promulgated.Serum IGF-1 values are low in growth hormoneinsufficiency. However, normal IGF-1 valuesare low in early childhood resulting in poordiscrimination from subnormal values in thatage group. There is also considerable overlapof IGF-1 in patients with growth hormonesufficiency and insufficiency resulting in poorperformance of the test (table 2).The recent demonstration of IGF binding

proteins and their dependence on a number ofhormonal regulators has suggested a role forthese proteins in the assessment of growthhormone secretory status. IGF bindingprotein-3 (IGFBP-3) is largely regulated by thecirculating concentrations of growth hormoneand might be used as an indicator of growthhormone secretory status. Blum et al have

suggested that there is excellent specificity andsensitivity when IGFBP-3 measurements areused in a group of children with short stature.15This measure deserves further exploration as itmight reduce the investigation of growth

hormone insufficiency to a single blood testand allow sequential measures to be made overthe course of a year to compare with growthdata. Falsely low values might be expected instates of growth hormone resistance and inliver disease.

(C) PHARMACOLOGICAL TESTS OF GROWTHHORMONE SECRETIONAs a first principle all these tests require goodvenous access. A nurse and a doctor should bepresent throughout the procedure and facilitiesfor resuscitation need to be available. All of thetests were standardised in adults after anovernight fast. Fasting from midnight is theideal but young children should be encouragedto have a late night snack and infants shouldnot starve for more than six hours. Drinks ofwater are allowed.

After insertion of an intravenous cannula aperiod of acclimatisation should be allowed.Whenever possible, the test should be startedby drawing a sample of blood 15-30 minutesbefore the application of the stimulus. Thissample together with the 0 minute samplegives an idea of the growth hormone secretorystatus after the cannula is inserted before thetime that the study is carried out. High growthhormone values which are declining duringthis time period are often associated with apoor response to the stimulus.

1. ITT (often combined with thyrotrophinreleasing hormone and luteinising hormonereleasingfactor tests)The ITT (0-15 U/kg soluble insulin) shouldonly be used in specialist units when there is aneed to test the hypothalamo-pituitary-adrenalaxis as well as the growth hormone axis.33 Itshould only be performed under strict condi-tions of surveillance utilising safety precautionsdetailed previously.1 The IT-l is contraindi-cated in children with a recent history ofepilepsy or hypoglycaemic convulsions andspecial precautions must be taken in childrensuspected of having hypopituitarism.The growth hormone response may be

impaired if hypoglycaemia is inadequate. Anadequate hypoglycaemic stimulus is when theblood glucose falls to 2-2 mmolI or less, if thechild is symptomatic, or if there is a 50%reduction in the blood glucose concentra-tion.34 Blood samples need to be drawn at 30minute intervals for two hours.Lunch should be provided for the child at

the end of the test and the child should not besent home until an adequate meal has beentaken and observed for at least 90 minutes afterthe meal to make certain that there is novomiting. The intravenous cannula should beleft in situ until lunch has been assimilated. Ifthere is any doubt about the child's wellbeingovernight observation is recommended.

2. Glucagon stimulation testThis test is often used in younger children.Although the usual recommended dose of

365

on March 24, 2022 by guest. P

rotected by copyright.http://adc.bm

j.com/

Arch D

is Child: first published as 10.1136/adc.72.4.362 on 1 A

pril 1995. Dow

nloaded from

Hindmarsh, Swift

glucagon is 05 mg problems can be encoun-tered. Rebound hypoglycaemia may resultbecause glucagon is a potent insulin secreta-gogue. The timing of this is variable and ifdelayed beyond the end of the test at 186minutes it is particularly dangerous. This testshould not be considered as any easier to per-form than the ITT and careful monitoring ofblood glucose concentrations is essential.Current work suggests that a glucagon dose of15 pug/kg is adequate to stimulate growth hor-mone and cortisol (M 0 Savage, personalcommunication).

Care needs to be exercised in interpretingthe cortisol response to the glucagon test. PeakACTH concentrations are significantly lessduring glucagon than during the ITT and arealso delayed.35 Higher and more consistentcortisol levels are obtained after intramuscularadministration than after the subcutaneousroute.36 37

Problems with hypoglycaemia may ariseparticularly in young children if glycogen storesare depleted by an overnight fast before theadministration of glucagon and as recom-mended above for the ITT, lunch must be seento be absorbed after the test before the intra-venous cannula is removed. The test iscontraindicated when cortisol deficiency issuspected as glycogen stores are likely to be low.

3. Clonidine stimulationThis is a commonly used stimulus in a dose of0a 15 mg/m2 orally. The Health Service HumanGrowth Hormone Committee compared theclonidine test with the ITT. 14 Oral administra-tion of clonidine is associated with a repro-ducible peak serum clonidine concentrationabout 60 minutes after administration. Sideeffects include drowsiness, pallor, and some-times a marked fall in blood pressure. Theeffects seem to be related to the age of the childand are particularly marked in adults.Therefore, care needs to be taken with this testand hypotension with the prolonged drowsi-ness may necessitate the child being observedfor the remainder of the day. Despite thesereservations this test is commonly used inpaediatric practice. Performance characteris-tics are shown in table 2.

4. Arginine stimulationArginine is a recognised stimulus of growthhormone release. Arginine (arginine mono-hydrochloride) provokes release of growthhormone when infused intravenously. Givenover a 30 minute period (0-5 g/kg) the sideeffects of nausea and irritation at the infusionsite are minimised. Performance characteris-tics of this test are shown in table 2 and care isalso required as amino acids such as arginine,while having a direct action on the hypothala-mus, are also insulin secretagogues.

5. LevodopaLevodopa may be administered orally tostimulate growth hormone release. The side

effect of nausea limits its use, althoughconcomitant administration of propranololminimises this. A peak response is notexpected until 90 minutes after administrationso that a long sampling period of 150-180minutes is required for full assessment.

6. GHRH stimulation testThis tests the readily releasable pool of growthhormone contained within the somatotroph.There is, however, some uncertainty regardingthe amount of growth hormone released for aparticular GHRH given dose. Supramaximaldoses of GHRH (1-2 ,ug/kg) have been givenas an intravenous bolus, probably in excess ofthe requirement for maximal release. This testis not recommended for routine clinical prac-tice. It does not test the hypothalamo-pituitaryaxis and it is not, therefore, a useful 'screening'test.

Growth hormone assaysThere has been a recent increase in the numberof growth hormone assays available for themeasurement of serum growth hormoneconcentration. These methods range from thetraditional polyclonal methodology usingradioimmunoassay to sophisticated mono-clonal radioimmunometric assays. Sensitivityhas been greatly improved by the use ofenzyme linked or chemiluminescent systems.The variety of assays introduces furtherproblems in the interpretation of growthhormone responses.

Assays employ different reaction matricesand are almost certain to use antisera with dif-fering epitope specificity so that universal cutoff values, for example 7 5 ,ug/l (15 mU/l), maynot be appropriate using different assaysystems. Care must be exercised in interpretingresults from different laboratories. The clini-cian needs to be aware of the method incurrent use, the precision profile of that assay,and how the laboratory performs within theUK external quality assurance scheme. Table1 shows the results of using the Hybritechsystem for measuring growth hormone.

ConclusionsGrowth hormone provocation tests should beused only after careful clinical assessment hasconfirmed inadequate growth velocity, otherpotential causes of growth failure have beenexcluded, and that there is a real possibility ofhypothalamic or pituitary hormone deficiency.These tests should be performed only incentres where there is experience in performingand interpreting them. The clonidine test is themost useful for paediatric practice. ITT andglucagon tests should only be used in specialistcentres.There remains the question of whether any

growth hormone test should be performed inthe assessment of the slowly growing child.Although idiopathic isolated growth hormoneinsufficiency is one of the commonestendocrine causes of poor growth the clinical

366

on March 24, 2022 by guest. P

rotected by copyright.http://adc.bm

j.com/

Arch D

is Child: first published as 10.1136/adc.72.4.362 on 1 A

pril 1995. Dow

nloaded from

An assessment ofgrowth hormone provocation tests 367

features are poor indicators of the diagnosis orwhether there is an isolated or multiplepituitary deficiency. In the neonatal period andafter cranial irradiation there are clearer indica-tions for testing.The authors consider that growth hormone

measurements remain a valid investigation inorder to provide biochemical support for sub-sequent growth hormone treatment. Despitethe statistical constraints of the tests the like-lihood that a slowly growing child is growthhormone insufficient is increased twofold inthe presence of a positive test but onlydecreased by half in the presence of a negativeone. Improving the probability of the presenceof a growth hormone secretory disorder is par-ticularly important if general practitioners areto be asked to prescribe and, therefore, to takesome shared responsibility for the prescription.Also, detailed evaluation would seem to be aprerequisite for outcome audit and researchpurposes. Advances in cellular and molecularbiology will gradually clarify the diagnosis ofidiopathic isolated growth hormone insuffi-ciency but these techniques will only be asgood as the defined phenotype.We thank the following paediatric endocrinologists for theirhelp and advice in the preparation of this manuscript: C G DBrook, M D C Donaldson, CJ Kelnar, M 0 Savage, S Shalet,and R Stanhope.

1 Shah A, Stanhope R, Matthew D. Hazards of pharmaco-logical tests of growth hormone secretion in childhood.BMJ 1992; 304: 173-4.

2 Hughes IA. Handbook of endocrine investigations in children.London: Wright, 1986.

3 Plumpton FS, Besser GM. The adrenocortical response tosurgery and insulin induced hypoglycaemia in corti-steroid treated and normal subjects. Br J Surg 1969; 56:216-9.

4 Crowley S, Hindmarsh PC, Honour JW, Brook CGD.Assessment of cortisol secretion in children with asthmatreated with inhaled steroids. Pediatr Res 1993; 33 (suppl):S13, 61.

5 Hindmarsh PC, Brain CE, Robinson ICAF, Matthews DR,Brook CGD. The interaction of growth hormone releas-ing hormone and somatostatin in the generation of a GHpulse. Clin Endocrinol (Oxf) 199 1; 35: 353-60.

6 Sox HC Jnr. Probability theory in the use of diagnostic tests.Ann Intern Med 1986; 104: 60-6.

7 Pauker SG, Kopelman RI. Interpreting hoofbeats: canBayes help clear the haze. N Engl J Med 1992; 327:1009-13.

8 Marin G, Domene HM, Barnes KM, et al. The effects ofestrogen priming and puberty on the growth hormoneresponse to standardized treadmill exercise and arginine-insulin in normal girls and boys. J Clin Endocrinol Metab1994; 79: 537-41.

9 Rose SR, Ross JL, Uriarte M, Barnes KM, Cassorla FG,Cutler GB Jnr. The advantage of measuring stimulated ascompared with spontaneous growth hormone levels in thediagnosis of growth hormone deficiency. N Engl J Med1988; 319: 201-7.

10 Dattani MT, Pringle PJ, Hindmarsh PC, Brook CGD.What is normal stimulated growth hormone concentra-tion?JEndocrinol 1992; 133: 447-50.

11 Hindmarsh PC, Smith PJ, Taylor BJ, Pringle PJ, BrookCGD. Comparison between a physiological and pharma-cological stimulus of growth hormone secretion: responseto stage IV sleep and insulin-induced hypoglycaemia.Lancet 1985; ii: 1033-5.

12 King JM, Price DA. Sleep-induced growth hormone release- evaluation of a simple test for clinical use. Arch Dis Child1983; 58: 220-2.

13 Youlton R, Kaplan SL, Grumbach MM. Growth andgrowth hormone. IV. Limitations of the growth hormoneresponse to insulin and arginine in the assessment ofgrowth hormone deficiency in children. Pediatrics 1969;43: 989-1004.

14 Health Services Human Growth Hormone Committee.Comparison of the intravenous insulin and oral clonidinetolerance tests for growth hormone secretion. Arch DisChild 1981; 56: 852-4.

15 Blum WF, Ranke MB, Keitzmann K, Grauggel B, Zeisel H,Bierich JR. A specific radioimmunoassay for thegrowth hormone (GH)-dependent somatomedin-bindingprotein: its use for diagnosis of GH deficiency. J ClinEndocrinolMetab 1990; 70: 1292-8.

16 Butler GE, McKie M, Ratcliffe SG. The cyclical nature ofprepubertal growth. Ann Hum Biol 1990; 17: 177-98.

17 Hindmarsh PC, Smith PJ, Brook CGD, Matthews DR. Therelationship between growth velocity and growth hormonesecretion in short prepubertal children. Clin Endocrinol(Ox]) 1987; 27: 581-91.

18 Albertsson-Wikland K, Rosberg S. Analyses of 24-hourgrowth hormone profiles in children: relation to growth. JfClin Endocrinol Metab 1988; 67: 493-500.

19 Spadoni GL, Cianfarani S, Bernardini S, et al. Twelve hourspontaneous nocturnal growth hormone secretion ingrowth retarded patients. Clin Pediatr (Phila) 1988; 27:473-8.

20 Donaldson DL, Hollowell JG, Pan F, Gifford RA, MooreWV. Growth hormone secretory profiles. Variation onconsecutive nights. Jf Pediatr 1989; 115: 51-6.

21 Albertsson-Wikland K, Rosberg S. Methods of evaluatingspontaneous growth hormone secretion. In: Ranke MB,ed. Functional endocrinologic diagnostics in children andadolescents. Mannheim: Verlag, 1992: 76-101.

22 Saini S, Hindmarsh PC, Matthews DR, et al.Reproducibility of 24-hour serum growth hormoneprofiles in man. Clin Endocrinol (Oxf) 1991; 34: 455-62.

23 Brook CGD, Hindmarsh PC, Healy MJR. A better to detectgrowth failure. BMJ 1986; 293: 1186.

24 Cole TJ. Do growth chart centiles need a face-lift? BMJ1994; 308: 641-2.

25 Lippe B, Wong SR, Kaplan SA. Simultaneous assessmentof growth hormone and ACTH reserve in children pre-treated with diethylstilbestrol. Jf Clin Endocrinol Metab1971; 33: 455-62.

26 Taylor B, Brook CGD. Sleep EEG in growth disorders.Arch Dis Child 1986; 61: 754-60.

27 Stanhope R, Adlard P, Hamill G, Jones J, Skuse D, PreeceMA. Physiological growth hormone (GH) secretionduring the recovery from psychosocial dwarfism: a casereport. Clin Endocrinol (Ox]) 1988; 28: 335-9.

28 Sukegawa I, Hizuka N, Takano K, et al. Measurement ofnocturnal urinary growth hormone values. Acta Endocrinol(Copenh) 1989; 121: 290-6.

29 Skinner AM, Clayton PE, Price DA, Addison GM, MuiCYW. Variability in the urinary excretion of growthhormone in children: a comparison with other urinaryproteins. 7 Endocrinol 1993; 138: 337-43.

30 Albini CH, Quattrin T, Vandlen RL, MacGillivray MH.Quantitation of urinary growth hormone in children withnormal and abnormal growth. Pediatr Res 1988; 23:89-92.

31 Walker JM, Wood PJ, Williamson S, Betts PR, Evans AJ.Urinary growth hormone excretion as a screening test forhuman growth hormone deficiency. Arch Dis Child 1990;65: 89-92.

32 Phillip M, Chalew SA, Sten MA, Kowarski AA. The valueof urinary growth hormone determination for assessmentof growth hormone deficiency and compliance withgrowth hormone therapy. Am J Dis Child 1993; 147:553-7.

33 Department of Health. Anterior pituitary function tests inchildhood. London: DoH, 1992.

34 Gale EAM, Bennet T, MacDonald IA, Holst JJ, MatthewsJA. The physiological effects of insulin-induced hypogly-caemia in man: responses at differing levels of bloodglucose. Clin Sci 1983; 65: 263-71.

35 Littley MD, Gibson S, White A, Shalet SM. Comparison ofthe ACTH and cortisol responses to provocative testingwith glucagon and insulin hypoglycaemia in normalsubjects. Clin Endocrinol (Ox]) 1989; 31: 527-33.

36 Spathis GS, Bloom SR, Jeffcoate WJ, et al. Subcutaneousglucagon as a test of the ability of the pituitary to secreteGH and ACTH. Clin Endocrinol (Ox]) 1974; 3: 175-86.

37 Rao RH, Spathis GS. Intramuscular glucagon as a provoca-tive stimulus for the assessment of pituitary function;growth hormone and cortisol responses. Metabolism 1987;36: 658-63.

CommentaryIt is now more than 30 years since the induc-tion of hypoglycaemia was described as a wayof manipulating growth hormone secretion.1The ITT, which can also provide an assess-ment of ACTH reserve, has been an estab-lished method to measure growth hormonesecretion primarily in short children and stressrelated cortisol secretion primarily in adultswith pituitary disorders. The Health ServicesHuman Growth Hormone Committee was agovernment quango which was remarkablyeffective in maintaining a satisfactory standardof investigation for short children who mightbenefit from growth hormone treatment.2When the government in its wisdom disbandedthe quango, an internationally recognised UKdatabase of adequately investigated childrenreceiving growth hormone treatment soon dis-appeared. To its credit, the Kabi Pharmacia

on March 24, 2022 by guest. P

rotected by copyright.http://adc.bm

j.com/

Arch D

is Child: first published as 10.1136/adc.72.4.362 on 1 A

pril 1995. Dow

nloaded from