Archives of Disease in Childhood 1994; 71: F224-F230

OBSTETRICS FOR PAEDIATRICIANS

Diabetes in pregnancy

Robert Fraser

IntroductionPerinatal mortality in association withmaternal diabetes in pregnancy, which were ashigh as 50% at the time of the Second WorldWar, have fallen dramatically. By the early1980s in most units offering some specialisedsupervision, perinatal mortality was no higherthan for non-diabetic women in the samehospital.The clinical problems which remain include:

maintaining the low level of perinatal mortalityin the face of changing obstetric practice andfashion; continuing attempts to reduce themortality and morbidity associated with theexcess of congenital malformations in the off-spring of women with diabetes; and the evenmore intractable problem of preventing thebirth of hyperinsulinaemic, macrosomicfetuses with the multiple problems they face asnewborns.

I will comment briefly on the vexed area ofscreening for, and diagnosis of, gestationaldiabetes mellitus. The other areas of interestto obstetricians and diabetic physicians caringfor these women - the effect of pregnancyon the progression of the disease and itscomplications, and the association betweengestational diabetes and subsequent develop-ment of adult diabetes - will not be con-sidered.

Clinical SciencesCentre, NorthernGeneral Hospital,Herries Road,Sheffield SS 7AUR Fraser

Correspondence to:Mr R Fraser.

Accepted 12 August 1994

Diabetes diagnosed before conception(pre-gestational diabetes)In the United Kingdom population insulindependent diabetes (IDDM), often of child-hood origin, predominates among cases ofdiabetes predating pregnancy. In populationswhere higher mean parity and correspondingtendency to reproduction at a later age aremore common, non-insulin dependent dia-betes (NIDDM) may predominate in numeri-cal terms. High rates ofNIDDM complicatingpregnancy are typically seen in emergingeconomies such as the Middle East and inimmigrant populations in the UnitedKingdom, particularly those from the Indiansubcontinent.

In the United Kingdom pre-gestationaldiabetes complicates two to 10 pregnancies ina thousand. Gestational diabetes, or carbo-hydrate intolerance diagnosed for the first timeduring pregnancy, may complicate 10 to 50pregnancies in every thousand. This latterfigure is only tentative and depends on variableascertainment rates.

PERINATAL MORTALITYWithout specialist care the perinatal mortalityassociated with maternal diabetes can beextremely high. There are many contributingfactors. Major congenital malformations arefour to six times more common in women whoare diabetic at the time of conception than innon-diabetic women. Many of these malfor-mations are lethal either in utero or in the new-born period. Diabetic pregnancy is associatedwith complications such as polyhydramniosand an excess of pre-eclampsia which may leadto spontaneous or iatrogenic pre-term birth.1The offspring of diabetic mothers have anexcess of hyaline membrane disease at allstages of gestation, and newborn respiratoryproblems have in the past been a major factorin the excess perinatal mortality rates.

Between four and 10 babies in every 1000may suddenly die in utero, particularly in thelater weeks of pregnancy. These deaths haveoften been in macrosomic fetuses. Why thesebabies die is still not known, although acute onchronic hypoxia is emerging as the most likelypossible cause. To reduce the high numbers ofdeath in utero elective early delivery becameestablished practice. This was advocated at 38,or more commonly, 37 weeks' gestation eitherby induction of labour or by elective caesareansection, although in some centres induction asearly as 35 weeks was recommended. Theresult of this practice, perhaps predictably, wasa number of deaths associated with pulmonaryimmaturity. These deaths may have exceededthe number of lives saved by the policy. Thependulum seems to be swinging back towardsa policy of allowing pregnancies to continueuntil the onset of spontaneous labour, butwhether this policy is any safer will be thesubject of later comment.

Other major contributing factors to perinatalmortality (and morbidity) were birth trauma,particularly asphyxiation, or fetal injury, whenshoulder dystocia obstructed delivery, andother complications of the newborn periodsuch as severe hypoglycaemia.

Although it had been suggested for manyyears that improved diabetic control mightreduce rates of perinatal mortality, it was the1972 keynote publication of the retrospectivestudy of diabetic pregnancy by Karlsson andKjellmer which was instrumental in changingpractice.2 These authors performed a 10 yearanalysis of 179 Swedish women in whom asystematic attempt had been made to docu-ment the quality of diabetic control in the

F224

on February 19, 2021 by guest. P

rotected by copyright.http://fn.bm

j.com/

Arch D

is Child F

etal Neonatal E

d: first published as 10.1136/fn.71.3.F224 on 1 N

ovember 1994. D

ownloaded from

Diabetes in pregnancy

Table 1 Congenital malformations in infants of diabeticmothers*

Ratio of incidencest

Spina bifida 2Anencephaly 3Cardiac anomalies 4Low gut atresias 3Renal anomalies 5Situs inversus 84Caudal regression 252

*Modified form Mills et al3.tRatio of incidences: frequency of malformation in fetuses ofdiabetic women compared with that of similar malformationsin non-diabetics.

weeks before delivery. Their results showedthat there was a linear relation between qualityof diabetic control and absolute levels of peri-natal mortality, with the most poorly con-trolled women (mean blood glucose of >7 7mol/l) experiencing perinatal mortality of 24%,rates that were almost as bad as those reported20 years earlier. Those women with meanglucose concentrations below 5-5 mmolI had a3-8% rate of perinatal mortality which was nohigher than non-diabetic women studied overthe same time period. The effects of theseresults were profound. Indeed, if blood glucoseis controlled in the third trimester, and averagedaily glucose concentrations are below 6mmol, very low perinatal mortality is seen.The main benefits of the intervention areprobably in terms of reduced polyhydramnios,pre-eclampsia, and pre-term labour. Laterdelivery is associated with less serious respira-tory distress.The newborn of the diabetic mother has also

certainly benefited from the dramatic improve-ments which have been seen in neonatal careover the past 15 years, particularly in termsof respiratory support and prevention ofhypoglycaemia.

CONGENITAL MALFORMATIONSIn collected series major structural malforma-tions occur four to six times more often in thefetuses of diabetic women than in those ofnon-diabetics. The pattern of malformationscomprises an excess of all the common types ofstructural malformations, including neuraltube defects, gut atresias, and perhaps par-ticularly, serious cardiac malformations. Thesearch for a typical pattern of anomalies associ-ated with maternal diabetes has been incon-clusive apart from two rather rare syndromeswhich are numerically much more common indiabetics - caudal regression and situs inversus(table 1) (figure). Despite the relatively highratios of incidences for caudal regression, thisis an extremely rare malformation, occurring inonly two in every 1000 diabetics. Althoughmany hypotheses have been generated toexplain this finding, once again the majordeterminant of congenital malformation in thefetus of a diabetic mother seems to be relatedto the quality of the mother's blood glucosecontrol, not in late pregnancy but around thetime of conception and embryogenesis. Arather clever observation was reported byMiller and colleagues in the early 1980s,4 when

it became possible to measure glycosylatedhaemoglobin in the laboratory. It occurred tothem that glycosylated haemoglobin measuredin the first weeks of pregnancy in diabeticwomen would reflect the quality ofpericoncep-tional glucose control. Their study of 112women showed that almost all the babies sub-sequently found to have congenital malforma-tions were in the upper half of the distributionof glycosylated haemoglobin measurements.This information was converted into a pros-pective trial by Fuhrmann,5 among others. Hisuncontrolled studies performed in the formerEast Germany suggested that women sub-jected to a rigorous programme of tight peri-conceptional diabetic control had a rate ofcongenital malformation no higher than that ofnon-diabetic women. Those who did notparticipate in the regimen had congenital mal-formation rates of 7-5%, close to the expectedfigure. Other groups have explored thisapproach with somewhat less convincingresults. At present no formal randomised trialofgood control v 'intensive' control of diabetesin the periconceptional period has been under-taken. Clearly there are disadvantages to theintensive regimen, not least that the womanwho is planning a pregnancy would not knowfor how long she would have to continue it.While she was in intensive control she wouldbe at risk of the serious problem of insulininduced hypoglycaemia. Of interest, in thisregard, are the detailed studies of Mills andcolleagues.6 They showed that although inwomen controlling their diabetes scrupulouslyin the periconceptional period the congenitalmalformation rate was 4-9% compared with9% for diabetic women not involved in theprogramme, the malformation rate in the well

Multiple congenital malformations, including severe caudalregression, in fetus ofa woman with IDDM, identified byultrasound scan at 19 weeks. Pregnancy was terminated.

F225

on February 19, 2021 by guest. P

rotected by copyright.http://fn.bm

j.com/

Arch D

is Child F

etal Neonatal E

d: first published as 10.1136/fn.71.3.F224 on 1 N

ovember 1994. D

ownloaded from

Fraser

Table 2 Diabetic control and birth weight centiles

Mean pre-prandial Birth weight centlefor gestational ageblood glucose (mmol/l)(weeks 30-38 ofgestation) 1Oth-90th >90th

<4-0 3 1<5-0 1 3<6-0 3 1>6-0 2 0Total 9 5 (35%/o)

Fourteen consecutive cases ofIDDM had regular blood sugarprofiles from 30-38 weeks of gestation (R Fraser, unpublishedobservations.) Half of the women who achieved very tightcontrol (<5 0 mmolIA) also had large for gestational ageinfants.

controlled diabetics was still twice that of non-diabetic control women. Furthermore, thequality of control as reflected by glycosylatedhaemoglobin results in early pregnancy did notshow the clear correlation with malformationsreported in the earlier study by Miller.4Although studies in small animals have sug-gested a correlation between hypoglycaemiaand malformation, this is not reported inhuman beings, apart from a few anecdotalcases. In one recent series only one of 46women who had severe hypoglycaemia had amalformed baby.7On a practical note it should be borne in

mind that with advances in the antenataldiagnosis of structural and chromosomalmalformations and the widespread acceptanceof termination of pregnancy, it is becomingquite rare for diabetic women in the UnitedKingdom to give birth to babies with majorstructural malformations.

THE MACROSOMIC FETUSThe classic study of Karlsson and Kjellmerconfirmed the association between perinatalmortality and poor third trimester diabeticcontrol, but perhaps, surprisingly, failed tomake a similar connection between maternalcontrol in late pregnancy and birth weight.2The birth weight distribution was such thatsome of the largest babies were born to some ofthe best controlled mothers, and vice versa.

There is a paradox here, however, whichmight contribute to this statistical observation,in that longstanding complicated diabetes withmacrovascular pathology in the mother isassociated with intrauterine growth retarda-tion. It is generally felt that poor placentalperfusion results from inadequate expansion ofthe blood supply to the placental bed becauseof arteriosclerosis. The effect would mean,perhaps, that some growth retarded babieswould be born to the worst controlled diabeticmothers, but it would not explain why some ofthe largest babies were born to some ofthe bestcontrolled mothers. Our own observations inthis area are summarised by the results of aseries of diabetics shown in table 2 in whomattempts were made to provide optimal dia-betic control, aiming for a mean blood glucoseof less than 6 mmol/l during the day. In each ofthese cases regular profiles were obtained from30 weeks' gestation to about 38 weeks, whendelivery was undertaken electively. A third ofthe babies had a birth weight above the 90thcentile for gestational age; four of the five large

babies were born to mothers among the eightbest controlled. Our experience in this smallseries is typical of the experience of others inlarger series, and it is rare to find reportedmacrosomia rates below 30%.The finding of these high rates of macro-

somia, despite apparently good diabetic con-trol, has been the source of much speculation.For all practical purposes, although glucosecrosses the placenta freely by facilitated diffu-sion, insulin does not, and the insulin secretorykinetics of the mother and fetus, each on theirown side of the placenta, are independent ofeach other. In the case of the insulin dependentdiabetic where the insulin is exogenous theusual administrative difficulties obtain. Theessential problem is that exogenous sub-cutaneous insulin or even intravenous insulincannot be administered in a pattern whichmimics the in vivo situation. This means thatdespite what would appear to be almost perfectdiabetic control, judged by intermittent diur-nal profiling, very high concentrations ofplasma glucose may be experienced transientlyafter meals. Within the overall pattern of con-centration dependent glucose flux across theplacenta, there is evidence of a degree of idio-syncracy in fetal uptake. These two factors maymean that the hyperinsulinaemia in the fetus isa simple response to excess glucose transferacross the placenta. Many workers are unwill-ing to accept this simple cause andeffect hypothesis and have searched for othermechanisms.

Insulin is a primitive hormone whichinfluences all nutrient subclasses. Insulin defi-ciency not only leads to hyperglycaemia butalso to high concentrations of amino acids, freefatty acids, and a secondary rise in ketonebodies, all of which may become importantalternate fetal substrates and could thereforecontribute to macrosomia. A good deal ofeffort has been put into evaluating the possibleeffects of these alternate substrates on theassumption that they could be operating tocause macrosomia despite optimal bloodglucose control, but there is both theoreticaland experimental evidence to suggest that thesimple explanation is the correct one. Thetheoretical evidence has already been alludedto in that it is very difficult to know in anindividual case what the true diurnal pattern ofglycaemic excursion is and, more particularly,what the rate of glucose transfer of theindividual placental system might be in differ-ent people with the same level of glycaemia.The experimental evidence that glucosecontrol may be at fault in cases of macrosomiaarises from the work of Roversi's group in Italy.They developed a system of thrice daily insulinwhich was increased until clinical hypogly-caemia was obtained at each time point.8 Thedose which had not provoked hypoglycaemiawas referred to as the 'maximum tolerateddose' and this was administered with relativelyfixed calorie intakes for the remainder ofpregnancy. The levels of maternal glycaemiawere to all intents and purposes ignored, butwhere they were measured they were found tobe about 25% lower on diurnal profiles than

F226

on February 19, 2021 by guest. P

rotected by copyright.http://fn.bm

j.com/

Arch D

is Child F

etal Neonatal E

d: first published as 10.1136/fn.71.3.F224 on 1 N

ovember 1994. D

ownloaded from

Diabetes in pregnancy

the target levels referred to earlier. The policyof maximum tolerated doses of insulinproduced a fairly normal birth weight distribu-tion, with only 8% of babies born above the90th centile of weight for gestational age. Theproblem, therefore, for those who cannotaccept that simple glucose transfer is theexplanation for hyperinsulinaemia or macro-somia is that they must invalidate rather thansimply ignore Roversi's results.Among those fetuses destined to be born

macrosomic, hyperinsulinaemia is an almostconstant feature. It can be detected antenatallyby measuring liquor insulin concentrations.Not only can fetal hyperinsulinaemia be con-firmed, but there is evidence that oncerecognised target orientated insulin treatmentcan reverse it before birth.9 So far these obser-vations have not been translated into generalclinical practice. Antenatal diagnosis of fetalhyperinsulinaemia can also be made by cordo-centesis, although the relative risks of this pro-cedure compared with simple amniocentesisdo not seem to be justified. There is someevidence that cordocentesis can be used toidentify chronic hypoxaemia in the hyper-insulinaemic fetus, and this might assistdecision making about elective delivery toprevent intrauterine death. Tissue hypoxia inthe fetus may also be reflected in erythropoetinconcentrations.'0 Erythropoetin concentra-tions in liquor or cord blood may turn out to bea marker of chronic fetal hypoxaemia. The lackof a widely available erythropoetin assay limitsclinical use at present.For those reluctant to undertake invasive

testing, or without access to an insulin assay,the sensitivity of a single measure of fetalabdominal circumference for predictingmacrosomia is 56%, with a specificity of96%." Again, the possibility of target orien-tated insulin treatment could be entertainedwith an antenatal diagnosis of macrosomiabased on ultrasound scans. A failure of growthcurves, between head circumference andabdominal circumference, to continue todivaricate might be evidence of therapeuticsuccess. This is because the macrosomiatypical of hyperinsulinaemia affects differentorgans selectively. The large fetus of a diabeticmother is different from the constitutionallylarge fetus of the same weight in that the liver,muscles, and subcutaneous fat organs are allconsiderably increased in the diabetic fetuswhereas brain growth is not excessive. In theconstitutionally large fetus, while head circum-ference and abdominal circumference are rela-tively large, their ratio is not distorted. It is ofclinical importance that because brain growthis not excessive, whereas somatic growth doesexceed normal centiles, the mechanism existsfor shoulder dystocia. The head can passthrough the birth canal unobstructed, but theshoulders become arrested at the pelvic inlet.Imaging techniques, both ultrasound scansand computed tomography, have now beenapplied in an attempt to predict shoulderdystocia, and therefore to avoid the potentiallyfatal complication caused by performing acaesarean section. Shoulder widths above

14 cm are associated with birth weights above4200 g and might be judged to be an indicationfor caesarean section. 12 The other obviousassociations with macrosomia, which mightbe detected by ultrasound examination, areincreased placental volume and increasedliquor volume. The ultrasound scan detectionof polyhydramnios depends on the finding ofat least one liquor pocket above 8 cm indepth.There is a debate in obstetric circles about

the timing of delivery in pregnancies compli-cated by maternal diabetes. With fallingperinatal mortality, some authorities have sug-gested that induction at 37 or 38 weeks is nolonger necessary and indeed is not in themother's interest because failed induction isassociated with a high risk of caesarean section.It is true that spontaneous labour is less likelyto result in caesarean section, but delayingdelivery in these cases is probably not justifiedbecause of our continuing inability to identifythe fetus at risk of intrauterine death. It seemsunforgivable to preside over the avoidabledeath of an apparently normal fetus, after sub-jecting diabetic women to the rigours of anantenatal care that includes intensified glucosecontrol and frequent hospital visits.During labour or in preparation for

caesarean section there is an argument formaintaining good diabetic control - usuallywith balanced glucose and insulin infusions.The fetus of the diabetic mother has an adulttype insulin secretory kinetic pattern andmay respond to maternal hyperglycaemia inlabour with fetal hyperinsulinaemia, leadingto hypoglycaemia in the first four hours oflife. 13The other current argument involves paedi-

atricians in decisions as to whether routineadmissions to the special care baby unit arenecessary. Some modern series suggest thatneonatal complication rates are now so lowthat routine admission would be unjustified.Admission policy should therefore be based oncondition at birth and subsequent develop-ment of biochemical hypoglycaemia or respira-tory distress. Such a policy has obviousadvantages in terms of bonding and perhapswould contribute to successful continuationrates of breast feeding. A recently publishedseries from Sweden of national data on IDDMcomplicating pregnancy, in 491 cases studiedover a five year period, reports rates of idio-pathic respiratory distress syndrome of 1 6%,transient tachypnoea of the newborn of 3.3%,symptomatic hypoglycaemia of 8%, hyper-bilirubinaemia of 16.3%, polycythemia of2-2%, and an Apgar score of less than seven atfive minutes in 3.4%.14 This complication rateseems particularly low and indeed the expecta-tion of complications in our experience iscloser to that of Berk et al (table 3).15 Closescrutiny of this table is of interest, not leastbecause of the high rate of neonatal complica-tions in babies who are judged to be appropiatefor gestational age. 'Normal birth weight' hasoften been the justification for allowing suchbabies to go to the routine posmatal ward onthe first day of life.

F227

on February 19, 2021 by guest. P

rotected by copyright.http://fn.bm

j.com/

Arch D

is Child F

etal Neonatal E

d: first published as 10.1136/fn.71.3.F224 on 1 N

ovember 1994. D

ownloaded from

Fraser

Table 3 Neonatal morbidity in appropriate for gestationalage (AGA) and large for gestational age (LGA) neonates*

LGA AGA p Value

Polycythaemia 24% 7% <0-01Hyperbilirubinaemia 30% 13% <0 05Hypoglycaemia 60% 50% NSRDS 10% 12% NSBirth asphyxia 16% 35% NS

NS: differences not significant.*Modified from Berk et al 15.

Gestational diabetesThe uncertainty which still surrounds thissubject is surprising after the extraordinaryamount of scientific and clinical effort spent onit by obstetricians over the past 30 years. Theterm used to refer to carbohydrate intolerance,diagnosed during pregnancy, which resolvedafter pregnancy. This definition has beenupdated by the 3rd International WorkshopConference on gestational diabetes mellitus to'Carbohydrate metabolism of variable severitywith onset or first recognition duringpregnancy'.'6 The new definition of courseincludes IDDM, and NIDDM diagnosed forthe first time during pregnancy, as well as thereversible state referred to earlier. The problemwas a practical one - that reversibility could beconfirmed only by postnatal glucose tolerancetesting, and a considerable proportion ofwomen defaulted from such testing.

In brief, there are problems with ascertain-ment of gestational diabetes mellitus, compara-bility of different diagnostic tests for it, andindeed interpretation criteria. Most Europeanscurrently favour the 75 g oral glucose tolerancetest proposed by the WHO Expert Committee,with diagnosis based on a two hour value.Americans generally favour the AmericanDiabetic Association criteria based on a 100 gload and a three hour test in which any twovalues have to exceed a normal range for thediagnosis to be made. The WHO criteria havebeen criticised because they fail to takeaccount of physiological changes in glucoseconcentrations, both at fasting and at twohours which are seen in normal pregnancy, andmay overestimate the true incidence of gesta-tional diabetes mellitus. Despite this, they onlydiagnose about half the cases of gestationaldiabetes mellitus (or as the WHO styled it,impaired glucose tolerance) compared with theAmerican criteria.'7 Some authorities havereviewed the situation and concluded that thecondition does not exist as a pathological statejustifying screening,'8 but others have come tothe conclusion that the American requirementfor two abnormal values is too rigorousbecause macrosomia and perinatal morbiditycan be detected in women with only oneabnormal value.'9

In summary, it is likely that no two hospitalswill have an exactly identical programme ofscreening for, diagnosis of, or treatment ofgestational diabetes mellitus.

PERINATAL MORTALITY AND MORBIDITYAlthough the classic studies of O'Sullivanreported an excess perinatal mortality in

women with gestational diabetes, moreparticularly in those who were overweight andover 25 years of age,20 more recent series havereported no excess perinatal mortality ingestational diabetes mellitus. This may relateto increased awareness and supervision, orindeed treatment.There does not seem to be the same excess

of congenital malformation and this wouldmake sense if fetopathy was related to hyper-glycaemia in early pregnancy, a conditionwhich is usually absent in gestational diabetesmellitus.The problem of excess fetal growth or fetal

hyperinsulinaemia does seem to be presentin women diagnosed as having gestationaldiabetes mellitus, although of course the preva-lence of this complication depends to someextent on the diagnostic criteria. The matter iscomplicated further by the fact that womenwith potential diabetes tend to have an excessof heavy birth weight infants, despite normalglucose tolerance testing.2' In collected seriesthere is an approximate doubling of birthweight over the 90th centile in women withuntreated gestational diabetes mellitus, andthis excess can be reduced by diet, treatment,and insulin treatment. The effect of insulin inreducing birth weight does not seem to bedirectly related to mean plasma glucose con-centrations.The real concern here is that because of

inadequate screening a proportion of hyper-insulinaemic fetuses may be unrecognised untilthey develop complications either duringdelivery or in the early newborn period. Ourown work in this area is consistent with that ofothers and suggests that among unselectedbabies born over the 90th centile of weight forgestational age, about a quarter are hyper-insulinaemic and therefore at risk.22 We havenoted, however, that hyperinsulinaemia alsooccurs rarely in appropriate for gestational agenewborns with no predisposing maternalhistory or glucose tolerance abnormality.

Summary of obstetric and medicalmanagement ofpregnant diabetic womenPRE-PREGNANCY COUNSELLINGIt goes without saying that any known diabeticshould have the benefit of pre-pregnancycounselling, including contraception or sterili-sation, as appropriate, or investigation andtreatment of infertility. There is a concern thatfor adolescent girls this type of counselling isnot as readily available as it should be. It maybe that they fall between two stools in thetransfer between paediatric diabetic clinics andadult diabetic clinics, each thinking that sexeducation and pre-pregnancy counselling is theprovince of the other.At pre-pregnancy counselling we advise con-

traception to be continued, with intensificationof insulin treatment to achieve glycosylatedhaemoglobin concentrations in the non-diabetic range. This type of supervision is oftenbest undertaken in diabetes centres under thesupervision of diabetes specialist nurses.Women with NIDDM planning a pregnancy

F228

on February 19, 2021 by guest. P

rotected by copyright.http://fn.bm

j.com/

Arch D

is Child F

etal Neonatal E

d: first published as 10.1136/fn.71.3.F224 on 1 N

ovember 1994. D

ownloaded from

Diabetes in pregnancy

who are maintained on oral hypoglycaemicsshould be carefully reviewed and the institu-tion of insulin treatment before pregnancyconsidered. Fundoscopy should be performedby a suitably experienced practitioner and anypre-proliferative or proliferative retinopathybe treated with photocoagulation, preferablybefore discontinuing contraception. Estab-lished nephropathy should be well docu-mented both in terms of average amounts ofproteinuria and creatinine clearance. Withsevere grades of nephropathy (creatinine clear-ance of <40 ml/minute) pre-pregnancy coun-selling might raise the question of whetherchildbearing should be deferred until after suc-cessful transplantation, if that option seemsimminent.

CARE IN EARLY PREGNANCYIt is important to try and confirm pregnancy assoon as possible in diabetic women; sensitivepregnancy tests can be followed up by earlyabdominal or vaginal ultrasound scanning. Indiabetics a condition known as 'early growthdelay' may operate and prejudice subsequentmeasurements ofmaturity which are importantfor antenatal diagnostic screening and fortiming of elective delivery. The possibility thatearly growth delay is an artefact induced bydelayed ovulation does not seem to have beensatisfactorily excluded.23

Early scanning will confirm fetal viabilityand gestational age. Diabetic women should beoffered all available antenatal diagnosticscreening for neural tube defects and chromo-somal abnormalities, with recourse to amnio-centesis if these tests show abnormal results.More important is to offer a detailed structuralscan at 19 to 20 weeks; the ultrasonographershould be informed that the woman is diabeticand make sure that experience in cardiacscanning and limb assessment is available.

Antenatal care is almost always managed onan outpatient basis, although most expertcentres prefer the visits to be to hospital clinicsrather than general practitioner or communitymidwives' surgeries. Obstetric assessmentincludes surveillance for development of pre-eclampsia, polyhydramnios, and abnormalitiesof fetal growth, the latter measured by serialultrasound scans. Diabetic -care consists ofinsulin adjustments which aim to maintainmean pre-prandial blood glucose (usuallyassessed on sticks with meters at home) below6 mmol/l if this can be achieved withoutclinical hypoglycaemia.

Admission to hospital may be indicated ifoutpatient control fails to achieve satisfactoryprofiles.The management of pre-term labour

presents problems because c sympathometicsare almost certainly contraindicated due to theacute loss of diabetic control which has some-times been associated with their use. There is asimilar although probably quantitatively lesserrisk associated with the mother taking steroidswhen pre-term delivery seems likely. Onbalance it seems reasonable to administersteroids when indicated but it should ideally be

done with the knowledge of the consultantresponsible for diabetic care so that appro-priate adjustment in insulin dose can be made.

Biochemical tests for lung maturity basedon the lecithin:sphingomyelin ratio are notori-ously unreliable in diabetic women below 37weeks' gestation. In current practice theindications for this type of testing even in dia-betics have become few indeed, perhapsbecause of improvements in ultrasound datingof pregnancy.

TIMING OF DELIVERYFor the reasons reviewed we still favour elec-tive delivery at around 38 weeks' gestation,although practice varies and in some units 40weeks is becoming the norm. For induction oflabour cervical favourability can be induced bylocal prostaglandin treatment and indeed thiscan shorten the length of the latent phase oflabour, particularly in the patient pregnant forthe first time. This seems worthwhile asprolonged labour in a woman prone to ketosismay be dangerous to both mother and fetus.Modem insulin and glucose regimens formanagement of labour correspond to standardregimens for intraoperative and postoperativemanagement of the diabetic. Either dextroseinfusions with fixed insulin and potassiumcontents or a piggy-back arrangement using adextrose drip and a separate insulin infusionpump (adjustable) can be used. Glucoseconcentrations in labour are generally held tobe between 5 and 8 mmol/l.

There is a low threshold to opt for caesareansection for the reasons described earlier andthis may be particularly the case where therehas been a previous caesarean section, orwhere there is a breech presentation, or anultrasound diagnosis of macrosomia. Whenvaginal delivery is anticipated this should besupervised by an experienced midwife and it isa sensible precaution to have a registrar orother experienced obstetrician in the room toassist should there be unexpected difficultieswith the shoulders. In most units the on-callpaediatric team will be happy to attend, both incase resuscitation should be needed and toexamine the newborn to provide reassuranceabout normality for the new parents. Policyabout indications for transfer to the specialcare baby unit for observation or intravenousdextrose should preferably be agreed with theparents in the antenatal period.

MANAGEMENT AFTER DELIVERYDiabetic women having caesarean section willnormally be given prophylaxis with antibioticsand routine subcutaneous heparin and anti-embolism stockings. They should be managedwith intravenous dextrose insulin regimensuntil they return to a normal diet. The insulinregimen can almost immediately be restartedat the pre-pregnant dose. Non-insulindependent diabetics treated with insulin andgestational diabetics will generally not requireinsulin after the delivery and can be managedwith serial measurements of the blood glucose

F229

on February 19, 2021 by guest. P

rotected by copyright.http://fn.bm

j.com/

Arch D

is Child F

etal Neonatal E

d: first published as 10.1136/fn.71.3.F224 on 1 N

ovember 1994. D

ownloaded from

Fraser

by sticks and a meter. If the newborn baby of adiabetic mother is returned to the postnatalward a regimen for supervision of both feedingand glucose monitoring should be clearlyunderstood and enforced by paediatricians andmidwives or nurses in charge of the ward.

ConclusionsThere are sufficient problems with the obstet-ric management of the pregnant diabeticpatient to justify continuing clinical research. Itis possible, however, to identify modificationsof practice which may improve outcome.

1. Improved pre-pregnancy counselling anddissemination of information about the effectsof diabetes to women at risk.

2. Attention to diabetic control, particularlywith the involvement of diabetes centres anddiabetes nurse specialists, may reduce theincidence of congenital malformations andfetal hyperinsulinaemia and macrosomia.

3. Improved antenatal screening and theuse of selective termination, where acceptable,can 'reduce' perinatal mortality and morbidity.

4. Improved diagnostic testing using ultra-sound scanning and invasive techniques maybecome available to identify the hyper-insulinaemic fetus and also secondary hypox-aemia. Coupled with steroid treatment inreducing the risk of respiratory distresssyndrome, it may be possible to rationalisepolicies for timing of delivery. Such a futurescenario would have a significant impact onshort term neonatal morbidity and may verywell improve long term outcomes of the new-bom of diabetic mothers.

1 Greene MF, Hare JW, Krache M, Phillippe M, Barss VA,Saltzman DH, et al. Prematurity among insulin-requiringdiabetic gravid women. Am 7 Obstet Gynecol 1989; 161:106-11.

2 Karlsson K, Kjellmer I The outcome of diabetic pregnan-cies in relation to the mother's blood sugar level. Am JfObstet Gynecol 1972; 112: 213-20.

3 Mills JI, Baker L, Goldman AS. Malformation in infants ofdiabetic mothers occurs before the seventh gestationalweek. Diabetes 1979; 28: 292-3.

4 Miller E, Hare JW, Cloherty JP, Dunn PJ, Gleeson RE,Soeldner JS, et al. Elevated maternal HbA1, in early preg-nancy and major congenital anomalies in infants of dia-betic mothers. N Engl _J Med 1981; 304: 1331-4.

5 Fuhrmann K, Reiher H, Semmler K, Fischer F, Fischer M,Glockner E. Prevention of congenital malformations ininfants of insulin dependent diabetic mothers. DiabeticCare 1983; 6: 219-23.

6 Mills JL, Knopp RH, Simpson JL, Iovanovic-Peterson L,Metzger BE, Holmes LB, et al. Lack of relation ofincreased malformation rates in infants of diabeticmothers to glycemic control during organogenesis. N EnglJf Med 1988; 318: 671-6.

7 Steel JM, Johnstone FD, Hepburn DA, Smith AF. Canprepregnancy care of diabetic women reduce the risk ofabnormal babies? BMJ7 1990; 301: 1070-4.

8 Roversi GD, Gargiulo M, Nicolini U, Pedretti E, Marini A,Barbarani V, et al. A new approach to the treatment of dia-betic pregnant women. Am _7 Obstet Gynecol 1979; 135:567-76.

9 Weiss PAM. Prophylactic insulin in gestational diabetes.Obstet Gynecol 1988; 71: 951.

10 Salvesen DR, Brudenell JM, Snijders RJM, Ireland RM,Nicolaides KH. Fetal plasma erythropoetin in pregnanciescomplicated by maternal diabetes mellitus. Am ] ObstetGynecol 1993; 168: 88-94.

11 Bochner CJ, Medearis AC, Williams J, Castro L, Hobel CJ,Wade ME. Early third trimester ultrasound screening ingestational diabetes to determine the risk of macrosomiaand labour dystocia at term. Am ] Obstet Gynecol 1987;157: 703-8.

12 Kitzmiller JL, Mall JC, Gin GD, Hendricks SK, NewmanRB, Scheerer L. Measurement of fetal shoulder widthwith computed tomography in diabetic women. ObstetGynecol 1987; 70: 941-5.

13 Miodovnik M, Mimouni F, Tsang RC, Skillman C,Siddigi TA, Butler JB, et al. Management of the insulin-dependent diabetic during labor and delivery: Influenceson neonatal outcome. Am 7 Perinatol 1987; 4:106-14.

14 Hanson U, Persson B. Outcome of pregnancies complicatedby type I insulin-dependent diabetes in Sweden: Acutepregnancy complications, neonatal mortality andmorbidity. AmJ Perinatol 1993; 10: 330-3.

15 Berk MA, Mimouni F, Miodovnik M, Hertzberg V, ValuckJ. Macrosomia in infants of insulin dependent mothers.Pediatrics 1989; 83: 1029-34.

16 American Diabetes Association. Summary and recommen-dations of the 3rd International Workshop-Conference onGestational Diabetes Mellitus. Diabetes 1991; 40 (suppl2): 1-201.

17 Li DH, Wong VCW, O'Hoy KM, Ma HK. Evaluationof the WHO criteria for the 75 g oral glucose tolerancetest in pregnancy. Br . Obstet Gynaecol 1987; 94:847-50.

18 Jarrett RJ. Gestational diabetes: a non-entity? BM3 1993;306: 37-8.

19 Langer 0, Brusman L, Anyaegbunam A, Mazze R. The sig-nificance of one abnormal glucose tolerance test value onadverse outcome of pregnancy. Am ] Obstet Gynecol 1987;157: 758-63.

20 O'Sullivan JB, Charles D, Mahon CM, Dandrow RV.Gestational diabetes and perinatal mortality rate. Am .Obstet Gynecol 1973; 116: 901-4.

21 Ford FA, Bruce C, Fraser RB. Fetal macrosomia inpotential diabetics with normal oral glucose tolerance: acase-control studv. Br . Obstet Gynaecol 1990; 97:957-62.

22 Stanley KP, Fraser RB, Milner M, Bruce C. Cord insulinand C-peptide - Distribution in an unselected population.Br3 Obstet Gynaecol 1992; 99: 512-5.

23 Steel JM, Corrie JET, Johnstone FD. Assessment of gesta-tion by salivary progesterone and its relevance to the con-cept of early growth delay in diabetic pregnancy.Diabetologia 1984; 27: 334A.

F230

on February 19, 2021 by guest. P

rotected by copyright.http://fn.bm

j.com/

Arch D

is Child F

etal Neonatal E

d: first published as 10.1136/fn.71.3.F224 on 1 N

ovember 1994. D

ownloaded from