Archives of Disease in Childhood 1994; 71: 511-515

Pupil size in diabetes

K Karavanaki, A G Davies, L P Hunt, M H Morgan, J D Baum

AbstractSympathetic function was studied in 101diabetic children and 102 age and sexmatched control children, as part ofa longitudinal study of the evolution ofmicrovascular disease in the population ofdiabetic children and adolescents in AvonCounty. The median (range) age of thediabetic population was 13-5 (6-0-17.2)years, the duration of diabetes was 4 0(0.4-13.9) years, and glycated haemo-globin (HbAl) was 10-9 (7-0-18-1)%.

Pupillary adaptation in darkness, asan index of sympathetic neuropathy,was measured using a Polaroid portablepupillometer. Diabetic children had asignificantly smaller median pupillarydiameter, measured as the pupil/iris ratioand expressed as a percentage, thancontrol children (median (range) 62*9(50.3-72.1) v 65 9 (52.2-73-8)). Pupillarydiameter was significantly related todiabetes duration (r=-0.22), HbAl(r= -0.34), systolic blood pressure(r= -0.25), diastolic blood pressure(r=-0.49), and mean albumin/creatinineratio on random urine samples (r= -0 26).Pupillary diameter was not related to age(r=-0.1). Eight (7/9%) diabetic and four(3-9%) control children were identified ashaving abnormal pupillary dilation indarkness. In comparison with the rest ofthe diabetic population, these diabeticchildren had longer diabetes duration andpoorer glycaemic control.

Polaroid pupillometry has demon-strated subclinical autonomic neuropathyin a population of diabetic children andadolescents. These abnormalities wererelated to poor metabolic control, longdiabetes duration, and also to otherindices ofmicrovascular disease.(Arch Dis Child 1994; 71: 511-515)

Royal Hospital for SickChildren, Bristol,Institute of ChildHealthK KaravanakiA G DaviesL P HuntJ D Baum

Department ofNeurophysiologyM H Morgan

Correspondence to:Dr K Karavanaki, 7 AetideonStreet, Cholargos, Athens15561, Greece.

Accepted 17 August 1994

The Avon childhood diabetes study is a

prospective study of a geographically definedpopulation of diabetic children, designed todescribe the evolution of microvascular andneurological abnormalities.Autonomic nervous dysfunction in adults

with diabetes has been defined by small pupilsize' and impaired pupillary response todarkness2 or light.3 There are two previousstudies on pupillometry in children4 andadolescents,5 reporting reduced pupillaryadaptation in darkness, early in childhooddiabetes. Pupillary measurements have alsobeen used in conjunction with cardiovasculartests,6 7 as indices of autonomic neuropathy.

There are conflicting reports among adultswith diabetes on the correlations between

pupillary diameter in darkness and duration ofdiabetes or metabolic control: some investi-gators report negative correlation5 8 9 of pupilsize with these variables, whereas others showno correlation.4 10 Pupil size in darkness hasbeen shown in adults to be negatively related tovibratory sensation threshold," while patientswith frank diabetic retinopathy or nephropathyare reported to have smaller pupillary diameterthan patients without these complications.9The present study describes the prevalence

of sympathetic dysfunction in a geographicallydefined population of diabetic children andadolescents and the influence of such factors asdiabetes duration and glycaemic control.

Subjects and methodsSUBJECTSAt the time of initiating this study June 1986)196 children with insulin dependent diabetesmellitus (IDDM) were identified as living inAvon County. The following sources ofpopulation identification were used: consultantpatient lists, patient lists of all 542 generalpractitioners in Avon County, list of diabeticchildren under community medical officers,computerised data on hospital admissions withthe diagnosis of IDDM in hospitals of AvonCounty, and Bristol parents' support group.

Altogether 150 diabetic children wereeligible and were asked to participate. Fortysix children were ineligible because ofadditional illness, specific requests from thechildren's doctors that they should not beasked to participate, or they did not attend theconsultant diabetic clinics in Bristol.From the 150 eligible diabetic children, 129

agreed to participate in the first (DI) and 114in the second study period (D2). One hundredand twenty nine best friend non-diabeticcontrol children participated at DI; 15controls declined to repeat the measurementsduring the second study period: consequentlyan additional 15 age and sex matched controlswere recruited for D2. Thus a total of 144control children participated in the two studyperiods.The 46 excluded diabetic children did not

differ in mean age, duration of diabetes, orglycaemic control from the participants in theAvon study. However, when comparing the 1 14diabetic children who participated in both studyperiods with those who refused to participatein D2, the percentage of children with long dia-betes duration (>5 years) or poor glycaemiccontrol (glycated haemoglobin (HbA1) >+1SD) was increased among the refusers.The diabetic children were asked to identify

a best friend who was of the same sex and bornwithin the same year. The best friend wasconsidered an appropriate control for the

511

on April 10, 2021 by guest. P

rotected by copyright.http://adc.bm

j.com/

Arch D

is Child: first published as 10.1136/adc.71.6.511 on 1 D

ecember 1994. D

ownloaded from

Karavanaki, Davies, Hunt, Morgan, Baum

following reasons: (1) the friend would be ofthe same age, sex, and probably of similarsocial class. (2) A friend is a more suitablecontrol than a sibling (as used in many

previous studies12); the sibling may havesimilar hereditary predispositions to theproband, will not be of the same age and may

or may not be the same sex.

Pupillometry was performed as part of thesecond set of measurements only. Childrenbelow 6 years of age were not studied as theyhad difficulty in cooperating with the studyprotocol. Altogether 101 of the diabeticchildren and 102 of the controls providedtechnically acceptable (see methods section)pupillary photographs.The median (range) age of the diabetic

children was then 13-5 (6-0-17-2) years andthe diabetes duration was 4 0 (0-413-9) years.The male to female ratio for both diabetic andcontrol children was 0O8.The children received no drugs apart from

insulin and none had symptoms of clinicalautonomic neuropathy.

METHODSStudies on pupillometry have generally usedinfrared television techniques,2 5 10 whichprovide direct and continuous measurement ofthe pupil size in complete darkness. Wehave used the simple Polaroid photographictechnique described by Smith and Dewhirst in1986,13 which has been shown to correlate wellwith the television method.

Pupillary dilation in darkness of bothdiabetic and control children was estimatedusing a Polaroid portable pupillometer with a

75 mm lens with an incorporated electronicring flash. The camera gave immediate colourprints. The left eye was photographed first.The right eye was photographed after an

interval of 15 minutes (time for the constric-tion caused by the first flash to disappearl3).The criteria for the definition of acceptable

pupillary photographs were the following: theiris should appear round, rather than ovalelliptical shape, and the upper eyelid should beclear of the upper pupil margin. The pupil andiris sizes were measured in the horizontal planeon the pictures, and the pupillary diameter,measured as the pupil/iris ratio, was estimatedfor each eye and expressed as a percentage.13The scoring of the photographs was

performed 'blindly' by a single observer. Nodifference was observed between light(blue/grey) and dark coloured irises. For eachchild, the mean pupillary diameter of the leftand right eye was used for analysis.

In addition, the following sets of measure-

ments were made at both study periods:(i) Heart rate variability, using an electro-

cardiograph attached to an oscilloscope and a

Commodore Pet microcomputer.'4(ii) Vibration sensation threshold on lower

limbs was estimated using a biothesiometer(Bio Medical Instrument).

(iii) Urinary albumin/creatinine ratios on

aliquots from all voidings of urine over the

previous 48 hours. Urinary albumin concen-

tration was estimated by an immunoturbidi-metric technique using the Cobas kit.'5

(iv) Blood pressure, using a random zerosphygmomanometer. Diastolic blood pressurewas taken as Korotkoff phase IV and V.

(v) The fall of systolic blood pressure uponstanding. A blood pressure fall of 30 mm Hgwas defined as abnormal.'6

Glycaemic control on the diabetic childrenwas assessed by:

(i) HbAl using the Coening electroendo-osmosis technique.17 The normal range forour laboratory was 5 5-7 5%. HbAl measure-ments during the time of each data collection,as well as the mean HbAl of seven threemonthly measurements over a period of 18months time, were used for analysis.

(ii) Seven point blood glucose profile onfilter paper, collected on one day duringthe week before the day of the hospitalmeasurement. The mean of the sevenblood glucose concentrations was estimated.Capillary blood glucose concentration wasestimated using the glucose oxidase technique.The Bristol and Weston, Southmead, and

Frenchay Health Authority ethical committeesapproved the study and participants gavewritten consent.

STATISTICAL ANALYSISThe pupillary diameter, heart rate variables,and blood pressure variables were not normallydistributed according to the Kolmogorov-Smyrnov test,'8 therefore non-parametricstatistics were used.The between eye coefficient of variation of

pupillary diameter was estimated as follows.We estimated the pooled within individual SD:

Id2SD== _2n

(d=pupillary diameter difference between leftand right eye; n=numbers of cases). The meanof the left and the mean of the right eye wassummed and divided by two. The coefficient ofvariation (CVO/o) between individual eyes wasthen estimated as follows:

SDCV%= x 100

Mean

For comparisons between diabetic andcontrol children, Mann-Whimey U testwas used. For estimation of correlations,Spearman's correlations, together with partialcorrelation coefficients, were used.For comparison of pupillary adaptation

among groups of diabetic children with differentlevels of glycaemic control (good, moderate,poor) the Kruskal-Wallis one way analysis of

Table 1 Characterstics of the diabetic population

Median (range)

Age (years) 13-5 (60-17-2)Duration (years) 4 0 (0-4-13-9)Age at diagnosis (years) 8-6 (14-15-5)Dose of insulin (U/kg/day) 0-8 (0-2-1-4)Mean HbAl (O/o) 10-9 (7-0-18-1)

512

on April 10, 2021 by guest. P

rotected by copyright.http://adc.bm

j.com/

Arch D

is Child: first published as 10.1136/adc.71.6.511 on 1 D

ecember 1994. D

ownloaded from

Pupil size in diabetes

variance was used. The diabetic childrenwith good control has a mean HbAl over the18 months of the observation period which wasless than -1 SD below the mean HbA1 of thetotal diabetic population (that is, HbA1<8.90/o). Similarly the group of diabeticchildren with poor control had mean HbAl>+1 SD (that is, HbA1 >13 3%). A 5% levelof significance was used throughout the study.

Pupillary diameter was not age related. Aspupillary diameter was not normally distrib-uted, we used empirical centiles for theestimation ofreference limits, using the SPSS-Xcomputer package. The lowest 5% level wasidentified from the measurements of the controlgroup (cut off point of normality: 56 85).

ResultsA total of 101 diabetic children (88-5%)and 102 control children (89.5%) providedacceptable pupillary photographs in darkness.Their median (SD) (range) age was 13-5 (2-9)(6-0-17-2) years. The demographic andglycaemic characteristics of the diabeticpopulations are shown in table 1.The diabetic children, as a group, had

impaired pupillary adaptation in darkness,showing significantly (p=0-0001) smallermedian pupillary diameter (median (SD): 62-9(4 3), range: 50-3-72 1) than the control child-ren (median (SD): 65-9 (4-3), range:52-2-73-8). There was no difference inpupillary ratios between left and right eyes.All diabetic and control children had both eyesphotographed in order to test the reproduci-bility of the method. The between eyes CVO/oof pupillary diameter was 2-4%.

Pupillary diameter was unrelated to age inthe diabetic and control populations (table 2).In diabetic children, mean pupillary diameterwas negatively related to duration of disease(r=-0-22, p=0 025) and to indices ofglycaemic control (HbA1: r=-0 33, p=0-00 1)(fig 1). When performing the partial correla-tions of pupillary diameter with HbA1,controlling for diabetes duration, the strengthof the correlation was reduced but stillremained significant (r= -0-29, p=0 01).The role of diabetes duration and glycaemic

control in the development of sympatheticdysfunction were further investigated bystudying the pupillary adaptation amongdifferent groups of diabetic children, using theKruskal-Wallis one way analysis of variance.19When dividing the population according toHbAl levels into groups with good, moderate,or poor control, the group with the smallestmean pupillary diameter was the one with thepoorest control (HbA1 > +1 SD), whereas thegroup with the best adaptation in darkness wasthe one with moderate control.

Correlations were sought between pupillarydiameter and other measures of microvasculardisease. Pupillary diameter was negativelyrelated to systolic blood pressure (r=-0-25,p=0-01) and diastolic blood pressure phase IV(r=-0-49, p=0-001). This correlation wasunaffected, when controlling (partial correla-tion) for the effect of age. No such correlation

was found between pupillary diameter andblood pressure in control children.

Pupillary adaptation in diabetic children wasalso related to urinary albumin/creatine ratio.Thus a negative correlation was found betweenpupillary diameter and the mean ratio ofrandom urine samples (r=-0-26, p=0008);this was not present in the control population.When controlling (partial correlation) forHbA1 the correlation became insignificant,whereas when controlling for diabetesduration, the strength of the correlation wasreduced, but still significant.

In the diabetic children the pupillarydiameter correlated with blood pressureresponse to standing (r=0-20, p=004).Among the cardiovascular tests, it correlatedonly with heart rate variation while resting(r=-0-19, p=0 049). There were nosignificant correlations of pupillary diameterwith other heart rate parameters or indices ofautonomic and microvascular function in thecontrol subjects.From the measurements of pupillary

diameter on control children, the 95%reference limits were estimated. Thus thediabetic and control children with a meanpupillary diameter lower than the lowerreference limit were identified as havingabnormal pupillary dilation in darkness (fig 2).Eight diabetic children (7-9%) and fourcontrol children (3-9%) were found to have apupillary diameter below the reference limit.Table 3 shows the major differencesbetween the diabetic children with anabnormal pupillary diameter and the rest of thediabetic population. These eight children weresimilar in age but had longer diabetes duration(median duration: 9-2 years v 3-7 years;p=0004) and poorer glycaemic control(median HbAl: 12-6% v 10.9%; p=0028),compared with the remaining diabetic popula-tion. The control children identified as havingabnormal pupillary dilation were significantlyolder (median age: 16-1 years) than the rest ofthe control population.The diabetic children with impaired

pupillary adaptation to darkness also hadincreased diastolic blood pressure phase IV>+ 2 SD above the mean and their urinaryalbumin/creatine ratios were significantlyraised, compared with the rest of the diabeticchildren (median SD scores: + 1I4 v +0 5;p=0002).One of the eight diabetic children with

abnormal pupillary dilation also demonstratedabnormalities in three out of five heart ratetests during both study periods. This child wasclassified as having combined sympathetic andparasympathetic dysfunction.

DiscussionThis study has shown that abnormalities inpupillary adaptation to darkness present earlyin childhood diabetes and are related to othercandidate indices of early neuropathy andmicroangiopathy, such as reduced heart ratevariability, raised blood pressure, andincreased urinary albumin excretion.

513

on April 10, 2021 by guest. P

rotected by copyright.http://adc.bm

j.com/

Arch D

is Child: first published as 10.1136/adc.71.6.511 on 1 D

ecember 1994. D

ownloaded from

Karavanaki, Davies, Hunt, Morgan, Baum

Table 2 Correlation coefficients for pupillary adaptationin darkness for diabetic and control populations(Spearman)

Diabetic children Control children

r p Value r p Value

Age -0-09 NS -0-02 NSDuration -0-22 0-025 -

HbAl -0-33 0 001 - -Blood pressure

Systolic -0-25 0-010 0-11 NSDiastolic -0 49 0-001 0-08 NS

Albumin/creatinine ratioof random urine -0-26 0-008 -0-001 NS

Previous studies on pupillometry have beenperformed on adult subjects and have usedinfrared television pupillometers. The Avonstudy is the first to screen for autonomicneuropathy in a geographically defined popula-tion of diabetic children and their age and sexmatched controls. The population of AvonCounty was considered a suitable sample forlongitudinal study of diabetes complications asit involves children from all social classes andfrom urban and rural areas. The 46 non-eligiblediabetic children did not differ in mean age,diabetes duration, or glycaemic control fromthe parprevaleidiseasetheir exA sin

was usadaptateasy toage gr(cooper.equipmflash p(girl refexperie

Prevdiabetiin darkare tw4childre:first ussecondmeter.5Avon cdiabetipupillanot a fiThe

develolbeen sconflicsubject

80

70

60

50

40

Co

0)Coa-

.

.

.QCL

Figure 1

intillar

757p=o_oool

70 K

65 H

0-

E 60E

Xo 55._

a-50

45

400a

C

OC)ocx)co

cxccox

coxoo

c:mocccxxcoco

r

cmD0

ccxmoccax

oocx)

ocooococsoo

CocoOOOmx)mCCOO

-~ ~~~~~~~~~~~~~~~cx

Oc)0 0

- 0

Diabetic children Control childrenFigure 2 Pupillary diameter in darkness of the diabeticand control populations (horizontal bars=mean values).

ticipants of the Avon study; thus the decreased gradually with increasing durationnce of impaired indices of microvascular of disease.'9 In fact they showed a rate ofare unlikely to have been affected by diminution of pupil size of about 2% per year.clusion from the study. Conversely Smith et al showed no correlationnple Polaroid photographic technique'3 between pupillary abnormalities and diabetessed for the estimation of pupillary duration in a group of adult diabetic patients,tion in darkness. The method was aged 16-62 years.'0perform (except among the youngest In diabetic children and adolescents,

Dup (<6 years) who were unable to Schwingshandl et al reported a significantate with the procedure), used portable negative correlation between resting pupillarylent, and gave immediate results. The diameter and diabetes duration (r= -029,ower was strong but tolerable: only one p=00006).5 Our findings are similar. Clarkeused to repeat the picture after having et al found no correlation between pupillarynced the brightness of the flash. adaptation in darkness and diabetes durationious studies in adults have shown that in diabetic children.4 These conflicting resultsc patients have smaller pupil diameters may be attributed to the effect of age, or toness compared with controls.2 9 There genuine population differences.O previous studies on pupillometry in The effect of metabolic control on then and adolescents with IDDM.45 The development of autonomic neuropathy has,ed a Polaroid pupillometer,4 while the been studied in adults. In terms of pupil size inused infrared computerised pupillo- darkness in subjects with short duration of

5 In agreement with both studies, the disease (0-3 years),'9 autonomic damage to

:hildhood diabetes study has shown that the iris appears to be reversible with improvedc children and adolescents have reduced metabolic control. Moreover Hreidarssonry adaptation in darkness and that this is showed in adult diabetic subjects a strongunction of age. inverse relationship between long termeffect of diabetes duration on the glycaemic control and pupil size.9

pment of autonomic abnormalities has Schwingshandl et al showed in diabetic;tudied in adults, but the results are adolescents a significant negative correlation ofting. Hreidarsson showed in diabetic pupillary diameter with HbAl and with bloodts aged 125-43 years that pupil size glucose values.5 Our findings are similar:

pupillary diameter was significantly related toglycaemic control, expressed as the mean ofHbAl measurements over 18 months. The

- :";: correlations were unaffected when controllingfor diabetes duration. When dividing the

- * * *.dC^....V*A * diabetic population in to three groups differingin glycaemic control, the narrowest pupil

r=-O330, p=0-001 diameter was found in the group with thehighest HbAl levels. Clarke et al reported no

) 5 10 15 20 correlation between pupil diameter andHbAl (%) glycaemic control in diabetic children, albeit

Scatterplot of correlation between HbAI and based on single measurements of HbAl andv diameter of the diabetic population. fructosamine.

514

FVF**>#_ @^^*t-jJ ^wr,.-~~~~

on April 10, 2021 by guest. P

rotected by copyright.http://adc.bm

j.com/

Arch D

is Child: first published as 10.1136/adc.71.6.511 on 1 D

ecember 1994. D

ownloaded from

Pupil size in diabetes

Table 3 Comparison between the eight diabetic children with abnormal pupillaryadaptation and the remaining diabetic children in terms of demographic, glycaemic, andmicrovascular variables (Mann-Whitney U test); values are median (range)

Rest of the diabeticAbnormal (n=8) children (n=93) p Value

Age (years) 14-3 (9-6-16-4) 13-8 (6-0-17-1) NSSex (boys:girls) 4:4 45:48Duration (years) 9-2 (2-6-12-0) 3-7 (0-4-13-9) 0-004Height SD score -0-2 (-1 9 to 1-2) 0-2 (-2-1 to 2 3) NSBody mass index (kglm') 19-5 (15-0-25-4) 19-1 (14-3-27-2) NSHbA1 (%/6) 12-6 (10-9-16-3) 109 (7-4-18-1) 0-028Diastolic blood pressure (SD score) 2-5 (1-5-4-4) 1-8 (-0 5 to 5 4) NSAlbumin/creatinine ratio of daytime urine 1-4 (0 2-3 4) 0-5 (-1 0 to 2 8) 0-002Pupil diameter (%) 54-8 (50 3-56-5) 63-4 (56-9-72-1) 0 0001

Pupillary dilatation in darkness correlatedwith other candidate tests of latent diabetescomplications. Thus, in keeping with previousstudies4 13 pupil diameter was positivelyrelated to the blood pressure response tostanding.

Relationships of pupillary variables withcardiovascular tests have been found inadults,2 but not in children and adolescentswith IDDM.45 In keeping with the abovestudies, no correlation between pupillarydiameter and cardiovascular tests was found inthe Avon study. The different results betweenthe above studies could be attributed to thecomplex involvement of sympathetic andparasympathetic pathways in different tests ofautonomic function.

Overall, eight (7-9%) diabetic children andadolescents and four (3-9°/O) controls were

found to have a pupillary diameter below thereference limit. These children had longerdiabetes duration and poorer glycaemiccontrol when compared with the rest of thediabetic population. Schwingshandl et al alsoidentified 14/142 (9-8%) diabetic adolescentswith pupillary adaptation below the referencerange,5 with similar characteristics to the ones

reported in the Avon study. Clarke et alreported an increased prevalence (19%)of abnormal pupillary adaptation in a group ofdiabetic children with age range, diabetesduration, and glycaemic control similar to theAvon study and in 3-7% of the controlchildren.4 Thus, abnormalities of pupillaryadaptation in darkness as an index ofsympathetic neuropathy are not uncommon inchildhood diabetes.

Previous studies have shown that para-sympathetic impairment is also present inchildhood diabetes. Mitchell et al reportedabnormality in three heart rate tests in 7-8%diabetic children,20 while Schwingshandl et alreported abnormality in two heart rate tests in3-5% of diabetic adolescents.5 In the Avonstudy, 8-7% of diabetic children wereabnormal in two heart rate tests and 4 3% inthree heart rate tests (unpublished data).Thus, although there are no clinical symptomsof autonomic neuropathy in children withIDDM, there is evidence that indices of bothparasympathetic and sympathetic neuropathyare impaired.

Adult IDDM patients with symptomaticautonomic neuropathy have an increasedmortality rate compared with the generaldiabetic population and are at risk of cardio-respiratory arrests with anaesthesia during

minor operations.21 It is not known whetherasymptomatic neuropathy in diabetic childrencarries similar risks.

It has been shown in adults with diabetesin the early stages of neuropathy thatabnormalities are reversible with improvedglycaemic control.22 The recent reports fromthe Diabetes Control and Complications Trialstudy underline the benefit of intensive therapyin preventing the appearance of clinicalneuropathy.23 Screening for autonomic dys-function can therefore be justified in paediatricclinical practice, even among diabetic childrenwith relatively short duration of disease.Identifying those with sympathetic impairmentraises the prospect of targeting attempts toimprove glycaemic control.

We thank Dr Charles Pennock and the staff of the chemicalpathology department, for measuring the urinary albuminand creatinine concentrations. We are also grateful to theconsultants Dr D L Savage and Dr R Chambers for allowing usto study their patients. We also thank the young people whoparticipated in the study. Dr Karavanaki was an InternationalScholar of the Juvenile Diabetes Foundation USA.

1 Martin MM. Diabetic neuropathy. A clinical study of 150cases. Brain 1953; 76: 594-624.

2 Smith SA, Smith SE. Reduced pupillary light reflexes in dia-betic autonomic neuropathy. Diabetologia 1983; 24:330-2.

3 Hreidarsson AB, Gundersen HJG. The pupillary responseto light in type I diabetes. Diabetologia 1985; 28: 815-21.

4 Clarke CF, Piesowicz PT, Spathis CS. Pupillary size inchildren and adolescents with type I diabetes. Diabet Med1989; 6: 780-3.

5 Schwingshandl J, Simpson JM, Donaghue K, Bonney MAs,Howard N, Silink M. Pupillary abnormalities in type Idiabetes occurring during adolescence. Diabetes Care1993; 16: 630-3.

6 Clarke BF, Ewing DJ, Campbell IW. Diabetic autonomicneuropathy. Diabetologia 1979; 17: 195-212.

7 Bennet T, Fentem PH, Fitton D, Hampton JR, Hosking DJ,Riggott PA. Assessment of vagal control of the heart indiabetes. Measures of R-R interval variation under differ-ent conditions. BrHeartJ 1977; 39: 25-8.

8 Gundersen HJG. An abnormality of the central autonomicnervous system in long-term diabetes: absence of hippus.Diabetologia 1974; 10: 366.

9 Hreidarsson AB. Pupil size in IDDM. Relationship withduration, metabolic control and long-term manifestationsof diabetes. Diabetes 1982; 31: 442-8.

10 Smith SE, Smith SA, Brown PM, Fox C, Sonksen PH.Pupillary signs in diabetic autonomic neuropathy. BMJ1978; ii: 924-7.

11 Hreidarsson AB. Pupil motility in long-term diabetes.Diabetologia 1979; 17: 145-50.

12 Rowe DJF, Hayward M, Bagga H, Betts P. Effect of gly-caemic control and duration of disease on overnight albu-min excretion in diabetic children. BMJ 1984; 289:957-9.

13 Smith SA, Dewhirst RR. A simple diagnostic test of pupil-lary abnormality in diabetic autonomic neuropathy.Diabet Med 1986; 3: 38-41.

14 O'Brien IAD, Corrall RMJ. Cardiovascular autonomicfunction testing: an automated method for measuringheart rate variation. Diabet Med 1985; 2: 143-4.

15 Uloyd DR, Hindl EJ, Marples J, Gatt JA. Urinary albuminmeasurement by immunoturbidimetry. Ann Clin Biochem1987; 24: 209-10.

16 Sundkvist G, Almer LO, Lilja B. Respiratory influence onheart rate in diabetes mellitus. BMJ 1979; i: 924-5.

17 Menard L, Dempsey ME, Blankstein IA, Aleyassine H,Wacks M, Soeldner JA. Quantitative determination ofglycosylated hemoglobin Al by agar gel electrophoresis.Clin Chem 1980; 26: 1598-602.

18 Siegel S, Castellan NJ Jr. Nonparametric statistics for thebehavioral sciences. 2nd Ed. New York: McGraw-Hill,1988, 4.3: 51-5.

19 Hreidarsson AB. Acute reversible autonomic nervous systemabnormalities in juvenile insulin-dependent diabetes. Apupillographic study. Diabetologia 1981; 20: 485-81.

20 Mitchell EA, Wealthall SR, Elliot RB. Tests for autonomicneuropathy in diabetic children. Australian PaediatricJoumal 1985; 21: 105-9.

21 Ewing DJ, Campbell IW, Nielson JMW, Clarke BF.Heart rate changes in diabetes mellitus. Lancet 1981; i:183-6.

22 Jakobsen J, Christiansen J, Kristoffersen C, et al. Autonomicand somatosensory nerve function after 2 years of contin-uous subcutaneous insulin infusion in type I diabetes.Diabetes 1988; 37: 452-5.

23 Santiago JV. Lessons from the Diabetes Control andComplications Trial. Diabetes 1993; 42: 1549-54.

515

on April 10, 2021 by guest. P

rotected by copyright.http://adc.bm

j.com/

Arch D

is Child: first published as 10.1136/adc.71.6.511 on 1 D

ecember 1994. D

ownloaded from