acase-controlstudyof theaetiologyof cryptorchidismcryptorchidism were recorded in residents in the...

Journal of Epidemiology and Community Health, 1983, 37, 238-244

A case-control study of the aetiology of cryptorchidismA J SWERDLOW,' KATHRYN H WOOD,2 AND P G SMITH2

From the Oxford Regional Health Authority,' Oxford OX3 7Lf; and Department of Medical Statistics andEpidemiology,2 London School of Hygiene and Tropical Medicine, London WCIE 7HT, UK

SUMMARY A study of cryptorchidism was undertaken based on patients identified through theOxford Record Linkage Study. The cumulative lifetime incidence of archidopexy among males inthe study area was estimated to be 3 38%; 88% of orchidopexies were performed on males aged 5years and over and 32% on those aged 10 and over. In a case-control study with 146 casesascertained at orchidopexy and 146 matched controls there was a substantially increased risk ofcryptorchidism for those who had undergone breech labour. Raised risks of cryptorchidism were

found for boys born to mothers who were primigravidas or aged under 20, and a significantlyreduced risk was found for boys whose mothers were of B blood group. There was a gradient ofincreasing risk with decreasing birth weight and a significantly raised risk associated with clinicalinguinal hernia. Since cryptorchidism and testicular cancer share several risk factors, it may beprofitable to study newly identified risk factors for cryptorchidism in relation to testicular cancer.

Testicular maldescent is the most common disorderof sexual differentiation' and one of the mostcommonest congenital abnormalities in males.2 Thecondition is particularly important because of itsstrong associations with infertility2 3 andtesticular malignancy27 In nearly all cases theaetiology of cryptorchidism is unknown.'

This investigation used data extracted from thefiles of the Oxford Record Linkage Study (ORLS) toinvestigate the role of prenatal and perinatal factorsin the aetiology of cryptorchidism. The ORLS8 9 wasbegun in 1962, and since that time data have beencollected on all hospital and domiciliary births, allhospital admissions and operations, and all deaths ofresidents in Oxfordshire and, since 1966, WestBerkshire-that is, a total population of about800 000 and about 14 000 births a year. TheRegistrar General notifies ORLS of all deliverieswithin the area covered by the study; recording ofbirths is therefore probably virtually complete.Trained clerks abstract information about thepregnancy, delivery, and puerperium of the mother,and about the child at birth, from hospital notes or,for domiciliary births, midwives' notes.

Methods

The ORLS records of hospital admissions for theyears 1976-8 (the most recent years available) weresearched to identify all patients who had had a

238

diagnosis of cryptorchidism and the operationorchidopexy. The birth records of such patients whowere born in 1970-2 were then sought in the ORLSbirth files. The items of information included in theORLS birth records have been altered and expandedperiodically, and the period 1970-2 was chosen asthe most recent for which consistent data wereavailable and which would include a fairly largenumber of births of patients operated on in 1976-8.Singleton born patients whose birth records werelocated-that is, those singletons born within theORLS area-formed the cases for the study. For eachcase, one matched control birth was chosen randomlyfrom the file of birth records: each control wasmatched to the case for year of birth, hospital ofdelivery (or, for domiciliary deliveries, geographicalarea of domiciliary delivery), male sex baby,singleton birth, live birth, and not neonatal death.For each case and control, information about thepregnancy, birth, and puerperium, and about thebaby, was extracted from the records. Differencesbetween cases and controls with respect to thesefactors were tested for statistical significance byanalysing the data as matched pairs using McNemar'stest"0 and linear logistic regression.11 For cases, theORLS records were also searched for all hospitaladmissions between birth and the index orchidopexyoperation; all recorded diagnoses at theseadmissions and at the admission for orchidopexywere noted. The matched controls from the case-

Protected by copyright.

on January 30, 2020 by guest.http://jech.bm

j.com/

J Epidem

iol Com

munity H

ealth: first published as 10.1136/jech.37.3.238 on 1 Septem

ber 1983. Dow

nloaded from

http://jech.bmj.com/

A case control study of the aetiology of cryptorchidism

control study would not have been satisfactory ascontrols for this analysis because they were notknown to have remained in the study area after birth;the frequencies of diagnoses in cases were comparedtherefore with published data on the frequencies ofsuch diagnoses. Seasonal variations in the month ofbirth ofcases and in the month of last menstrual periodof mothers of cases were tested for significance by themethod of Walter and Elwood."2 For births,correction was made for monthly variation in thetotal numbers of births in the ORLS area during1970-2. For last menstrual periods, 1970-2 data forthe area were not available, and corrections werebased instead on last menstrual periods for all birthsin the ORLS area during 1973-8.

Results

A total of 813 admissions for orchidopexy forcryptorchidism were recorded in residents in theORLS area in 1976-8, mainly of boys aged 6 to 10years (fig). Fifty seven of the patients were admittedfor the operation twice within the period-that is,756 residents underwent orchidopexy in the area.The cumulative life time risk of archidopexy formales in the area, assuming 1976-8 rates ofadmission, was 3X38%.

In the ORLS area during 1976-8,221 boys born in1970-2 had orchidopexies. The birth records of 150(68%) of these boys were traced in the ORLS files:the other 71 could not be traced and presumably.were born outside the area. Four of the 150 boystraced were twins (3.38 cases born from multiplepregnancies would be expected on the basis of allbirths in the ORLS area during 1970-2) and thesewere excluded from the study. Thus 146 singletonborn cases and 146 controls were included in thecase-control study.

No otoperotions

120

100l

80-

60-

40O

20C

Age(years)X - ;

Number of orchidopexies for cryptorchidism by age atoperation: residents operated on within the Oxford RecordLinkage Study area, 1976-8.

DEMOGRAPHIC FACTORS AND MEDICALCHARACTERISTICS OF THE MOTHERTable 1 shows the distribution of cases and controlswith respect to various characteristics of theirparents. The social class distribution of cases andcontrols did not differ significantly. Nevertheless, therisk for cryptorchidism was lower for boys in socialclass I than for those in other social classes. Maternalage at birth was not significantly related to risk ofcryptorchidism, but the risk was highest for thosewhose mothers were under 20. There were nonotable differences between cases and controls withrespect to the marital state or religion of the mother(not shown in table 1). The sons of blood group Bmothers were at low risk compared with those whosemothers were in other blood groups (p<0.01). Onepossible explanation of this finding that needsconsideration is that an abnormally high proportionof blood group B mothers might have been included,by chance, in the control group. Nevertheless, whenthe distribution of blood groups among the mothersof the cases was compared with that of all mothersdelivering in the ORLS area in 1970-2 broadlysimilar results were obtained (relative risk (RR) forsons of blood group B mothers compared with sons ofall other blood group mothers = 0-29; p<0-025).The rhesus blood group of the mother showed nosignificant association with risk of cryptorchidism inthe son (table 1). The only notable differences in thefrequency of non-obstetric diseases recordedbetween mothers of cases and of controls were withrespect to diabetes mellitus: three cases but nocontrols had mothers with this condition recorded inthe obstetric notes; non-endocrine obesity: fourmothers of controls but no mothers of cases; andcongenital genitourinary abnormality: one motherof a case but no mothers of controls (the mother ofthe case had a double uterus and vagina). None ofthese differences was statistically significant.

OBSTETRIC HISTORYThere was no significant difference between casesand controls with respect to the parity of the mother,the number of previous stillbirths, or number ofprevious miscarriages (table 2); risk was highest forsons of nulliparous mothers compared with parousmothers, but this was not statistically significant, andthere was no gradient of risk with parity amongparous mothers.

SEASONALITYNeither month of last menstrual period (XI = 2- 13)nor month of birth )x2 = 2.27) of cases haddistributions significantly different from expectationsbased on the corresponding measures in all womendelivering in the ORLS area.

239

1-



j.com/

J Epidem

iol Com

munity H


ber 1983. Dow

nloaded from


240 A J Swerdlow, Kathryn H Wood, and P G Smith

Table 1 Parental characteristics

No (%) ofcases with No (%) ofcontrols Chi squareRisk factor risk factor' with risk factor' Relative riskt (significance)t

Social class of father at index birth:I 10 (7-4) 18 (14-3) 1-00II 26 (19-1) 25 (19-8) 2-10 = 425 (NS)III 69 (50-7) 61 (48.4) 2-21IV 22 (16-2) 16 (12-7) 2-65 X2Trend = 2-71 (NS)V 9 (66) 6 (4.8) 2-34Total 136 (100) 126 (100)

Age of mother at index birth (years):<20 14 (9.6) 5 (3.4) 1-0020-24 47 (32-2) 54 (37-0) 0-29 X4 = 5 34 (NS)25-29 60 (41-1) 60 (41-1) 0-3330-34 17 (11-6) 17 (11-6) 0 33 X' Trend = 0-82 (NS):35 8 (5-5) 10 (6.9) 0-28

Total 146 (100) 146 (100)

Blood group of mother:ABO A 66 (46.8) 59 (40-4) 1-00

B 4 (2-8) 23 (15-7) 0-18 X' = 14-80 (p<0-01)AB 4 (2.8) 2 (1-4) 2-930 67 (47-5) 62 (425) 0 95

Total 141 (100) 146 (100)Rhesus + 124 (87-9) 118 (80.8) 1-00

17 (12-1) 28 (19-2) 0-58 X2 = 2-66 (NS)Total 141 (100) 146 (100)

Based on all individuals for whom the information was recorded.t Based on analysis of matched pairs.

PREGNANCYNo significant differences were found between the and nine cases and seven controls had an antenatalpregnancies involving the cases and controls with haemorrhage).respect to time from last menstrual period tobooking, number of antenatal attendances, numberof radiographs taken during pregnancy, maximum DELIVERYrecorded systolic and diastolic blood pressures of the Ten cases but only two controls presented in themother, albuminuria in pregnancy, rhesus breech position before delivery (RR = 5 O; p<O-O5).incompatability, and complications of pregnancy, (in Eight of the 10 cases were delivered vaginally andparticular, 24 cases and 24 controls had pre- two by caesarian section, ofwhom one underwent theeclampsia, eclampsia, or toxemia; one case and one operation during labour and the other was an electivecontrol were recorded with hyperemesis gravidarum; caesarian. The relative risk for those experiencing

Table 2 Obstetric history of mother

No (%) of cases with No (%) of controls Chi squareRisk factor risk factor* with risk factor* Relative riskt (significance) t

Parity:0 69 (47-3) 55 (37.7) 1-001 42 (28-8) 49 (33.5) 0-68 X' = 2-74 (NS)2 23 (15-7) 28 (19-2) 0-67;1:3 12 (8.2) 14 (9-6) 0-67 X' Trend = 1-88 (NS)Total 146 (100) 146 (100)

Previous stillbirths:0 144 (98-6) 143 (97-9) 1-00;W1 2 (1-4) 3 (2-1) 0-67 X' = 0-20 (NS)Total 146 (100) 146 (100)

Previous miscarriages:0 121 (82.9) 118 (80.8) 1-001 19 (13.0) 24 (16.4) 0°76 X, = 1-07 (NS)-- 2 6 (4-1) 4 (2.7) 1-46 x2 Trend = 0-01 (NS)Total 146 (100) 146 (100)

'Based on all individuals for whom the information was recorded.t Based on analysis of matched pairs.



j.com/

J Epidem

iol Com

munity H


ber 1983. Dow

nloaded from



breech labour was 4-5 (p = 0-065). Two of the fourcryptorchid boys who were twins born in the ORLSarea 1970-2 had undergone breech birth. Six of thenine mothers of cases undergoing breech labour werenulliparous; the birth weights of these cases rangedfrom 2 3 kg to 3 - 5 kg. Five cases but only one control(not statistically significant) underwent Keilland'srotation, and nine cases and six controls weredelivered by caesarian section. For significantlyfewer cases than controls was labour induced(RR= 0-55; p<005), the risk of cryptorchidismbeing reduced similarly for surgical induction (RRcompared with no surgical induction=0-52; p<0.05)and hormonal induction (RR compared with nohormonal induction = 0 58; NS). Cases did not differnotably from controls for complications of deliveryother than those mentioned above, use of forceps orvacuum extraction, operations on the mother afterdelivery, and maternal complications in thepuerperium.

THE BABYCases tended to have lower birth weights thancontrols but this trend was not significant (table 3).Risk varied with gestational period (p<0025), witha high risk in those born under 36 weeks, and a lowrisk in those born 36-37 weeks (table 3). The bloodgroup of the baby was known for too few patients (12cases and 15 controls) for useful analysis. Cases andcontrols did not differ notably for Apgar score atbirth, nor for diseases noted at birth other than threehernias in cases (one inguinal, two umbilical) andnone in controls, and 12 cases but no controls withgenital anomalies (nine cases with cryptorchidism,three with other genital anomalies). Seven cases andeight controls were noted at birth to have othercongenital defects. The coding ofORLS birth records

allowed space for only one disease for each patient atbirth; therefore the above figures for specific diseasesmay underestimate the numbers actually diagnosed,although there is no reason to believe that theunderestimation is biased between cases and controlsexcept for those cases with cryptorchidism recordedat birth. Since 97 of the cases had no disease recordedat birth, at most 49 (34%) cases may have beendiagnosed cryptorchid at birth.Twenty-two (15%) of the 146 cases had had a

hospital diagnosis of inguinal hernia by 1978, in allbut one instance undergoing herniorrhaphy. Up to1978 six of the cases had a hospital diagnosis ofhydrocoele and one had a diagnosis of a renal disease(acute nephritis). The expected number of cases withthese associated diseases is difficult to determinebecause the presence of cryptorchidism, andoperation for it, may have led to the diagnosis ofassociated conditions that might otherwise haveremained undetected. Various studies suggest aprevalence of clinical hernia in boys in the generalpopulation in England of about 1-2%13 14; assumingthat all the hernias operated on in the present serieswere clinically significant, the relative risk ofcryptorchidism for boys with clinically significanthernia was about 11 (p<0-01).

Discussion

The estimated cumulative lifetime risk of operatedcryptorchidism in the ORLS area (3.38%) is muchgreater than the prevalence of cryptorchidism afterinfancy found in studies reported in the 1960s inEngland"5 16 and Denmark" (around 0.8%); earlierstudies had given widely varying estimates.218 Theactual prevalence of cryptorchidism after infancy inthe ORLS area may be greater than 3-38% because

Table 3 Birth weight and gestation of index child

No (%) of cases with No (%) of controls Chi squareRisk factor risk factor with risk factor Relative riskt (significance)tBirth weight (kg):<2-5 7 (4.9) 6 (4-1) 1-002-5-2-9 26 (18-1) 22 (15.2) 1-09 = 2-09 (NS)3-03-4 48 (33-3) 42 (29-0) 0-973-53-9 45 (31-2) 49 (33-8) 0-81>4-0 18 (12-5) 26 (17-9) 0-65 X Trend = 1-87 (NS)Total 144 (100) 145 (100)

Gestation (weeks):<36 6 (4-5) 1 (0-7) 1-0036-37 10 (7-5) 7 (5-6) 0-12 = 11-98 (p<0-025)38-39 37 (27-8) 35 (27-8) 0-5240-41 56 (42-1) 66 (52-4) 0-53-42 24 (18-1) 17 (13-5) 0-44 X,Trend = 0.74(NS)Total 133 (100) 126 (100)

Based on all individuals for whom the information was recorded.t Based on analysis of matched pairs.

241



j.com/

J Epidem

iol Com

munity H


ber 1983. Dow

nloaded from


242

the present study was confined to operated cases, andbecause possible coding errors were interpretedconservatively when validation was not possible.Recent data from the Hospital In-patient Enquiry ona 10% sample of NHS hospital patients in Englandand Wales"' also suggest a far higher prevalence ofcryptorchidism than in England in the 1960s:estimation of cumulative lifetime incidence oforchidopexy in England and Wales from publishedestimated operation rates for 197819 would give afigure of 3 07%. This figure requires correctionbecause it probably omits some patients whounderwent another operation as well as orchidopexyduring the same hospital admission, and because theoperation rates are based on unlinked data-that is,patients admitted for the operation more than onceduring the year are counted on each occasion. Thelatter correction would probably be small: in theORLS linked data over 90% of orchidopexyoperations in a year were on different patients.Estimates of the prevalence of cryptorchidism basedon ORLS and the Hospital In-patient Enquiry mustbe treated cautiously because the data are foroperated cases only and rely on ascertainment andcoding in information systems whose data areextracted from routine hospital records.Nevertheless, prevalence of cryptorchidism inEngland has probably risen substantially, whichrequires further investigation by clinical screeningstudies. There is some evidence for an increasingsecular trend in the prevalence of cryptorchidism inthe United States also.7The distribution of age at operation in the present

study is similar to that in recently reported series inEngland20 and Scotland21 and to that fororchidopexies in England and Wales overall where,in 1978, for the broad age groups published, 9% oforchidopexies were before age 5 years and 78% fromage 5 to 14 years.19 In these series most -of theoperations were performed at an older age than thatusually recommended.2023 There seems to be astrong case for more systematic screening of youngboys for undescended testis to allow earlierorchidopexy.A major problem in studying epidemiologically the

aetiology of cryptorchidism is the difficulty ofdiagnosing the condition. The present study wasconfined to operated cases, which has the advantagethat operation is likely to be the most reliable point ofdiagnosis of cryptorchidism in records of routineclinical practice. Another advantage of studyingoperated cases is that since such patients are almostall over 1 year of age, spontaneous descent of theircryptorchid testes is unlikely to occur.15 17 It is forsuch "permanently" cryptorchid patients that risksfor testicular cancer and infertility have been shown.

A J Swerdlow, Kathryn H Wood, and P G Smith

Study of operated cases, however, introduces somecase selection that may, for example, lead to socialclass bias.The most striking finding of the study was the high

risk of cryptorchidism for those presenting in thebreech position. Breech delivery is known to causesevere injuries that may be fatal, but the implicationsof lesser injuries for future health are not clear.24Prolapse of the scrotum of a male fetus through apartially dilated cervix may cause severe congestionand necrosis of the scrotal skin.24 Rali's conductedpostmortem examinations on 49 boys who diedduring or soon after breech delivery and found thattrauma to the testes was frequent and that in twocases there was a severely damaged testis within theinguinal canal.25 There have been reports of scrotalbruising and changes in the size, consistency, andsensation of testes of infants after non-fatal breechbirth;26 27 with abnormalities sometimes persisting forup to a few years,26 and reports of otherwiseunexplained testicular atrophy in two olderindividuals who had been breech born.28 It has notbeen shown previously, however, whether breechbirth may be a cause of serious long term testicularabnormality.

Breech birth is associated with several majorcongenital abnormalities2930 for which it is unlikely tobe aetiological. Such associations are thought tooccur because the abnormal morphogenesis orfunction, or both, associated with the abnormalitiesreduces the chance that the fetus will assume thevertex position.30 It seems improbable, however, thatthis explains the association of cryptorchidism withbreech birth, and it seems more likely that theassociation results from trauma from the breechposition. If this is so other birth and postnataltraumas might also cause cryptorchidism. If futurestudies confirm the association between obstetrictrauma and cryptorchidism the possible protectiverole of caesarian section will require consideration.

Several lines of evidence make it reasonable tosuppose that hormonal factors are important innormal testicular descent and that abnormalities ofthem might be a cause of cryptorchidism,1 2but thereis no convincing evidence that hormonalabnormalities are commonly a cause ofcryptorchidism in man. Seasonal variation innumbers of births of cryptorchid boys in Hungary hasbeen adduced as evidence for a hormonal aetiology31;however, as that study included no correction fornormal seasonal variation in births it is difficult tointerpret. The present study found no evidence ofseasonal variation in births of cryptorchid boys.The small rise in risk for those of low birth weight

(table 3) accords with previous data,1' 17 suggestingthat some raised risk persists beyond infancy in this



j.com/

J Epidem

iol Com

munity H


ber 1983. Dow

nloaded from



group, even though this risk is probably less than thatfor such boys at birth.5 32A raised risk in boys of lowbirth weight might reflect the absence of maternalhormonal stimulus for descent in those already born atthe normal time of testicular descent (mainly aroundeight months' gestation"8).

If a hormonal abnormality of the mother or baby isresponsible for a substantial proportion of cases ofcryptorchidism the abnormality must be a subtleone-neither the case mothers nor the cases in thestudy showed an excess of hospital diagnoses ofconditions likely to be associated with endocrineabnormality, nor have such abnormalities been notedpreviously other than for the small minority of casesof cryptorchidism in patients with rare disorders thatinclude abnormal gonadotrophin production,androgen secretion, or androgen action.1The low risk of cryptorchidism for boys born after

induction has not been reported previously, andneeds further study. Risk of cryptorchidism by ageand parity of the mother has been reported upon byCzeizel et al,31 who found no substantial excess inyoung mothers or primiparae. The analysis of socialclass in the present study is potentially biased, firstly,because greater migration out of the area after birthby higher social class cases could have left an excessof lower social class native cases resident in thearea by 1976-8 and, secondly, because presentationand ascertainment might be worse in the lowerclasses; however, in another case-control study ofcryptorchidism31 the cases had significantly less welleducated parents and mothers of cases hadsignificantly less well qualified occupations than thecontrols, suggesting that the social class gradient inthe present study may well be real.An association of cryptorchidism with inguinal

hernia and hydrocoele is to be expected onanatomical grounds.2 In clinical series inguinal herniahas been reported in most cases ofcryptorchidism. 3 In very few cases, however,does the hernia cause symptoms requiringoperation.2 34 This low clinical importance may bethe reason for the low prevalence of hernia in thepresent study compared with other series, since thepresent data reflect routine recording of diagnoses byclinicians not engaged in special study of theassociations of cryptorchidism.

Cryptorchidism is the main known risk factor fortesticular cancer.6 The cause of the association isunknown: cryptorchidism might be 'a cause oftesticular cancer or the two conditions might havecauses in common, or both. The raised risk oftesticular cancer in normally descended testesopposite cryptorchid testes suggests some degree ofcommon aetiology.6 Around 90% or more of cases oftesticular cancer occur in men who are not

cryptorchid.57' 36 Therefore, even if cryptorchidismis a cause of testicular cancer, factors that increase therisk of cryptorchidism may well not be detected instudies based on all cases of testicular cancer. If thetwo conditions share common causes, however, theymight be expected to have some epidemiological riskfactors in common. Several risk factors have beenfound for both testicular cancer and cryptorchidism,which supports the latter hypothesis: inguinal herniais probably associated with testicular cancer,7'"and its association with cryptorchidism has alreadybeen discussed; abnormalities of the urinary tractmay well be associated with both testicular cancer4' 38and cryptorchidism39; there is some evidence thatprenatal hormones may be a risk factor both fortesticular cancer"7 40 and cryptorchidism41; bothtesticular cancer4" and cryptorchidism46 are rare inUnited States negroes compared with United Stateswhites; sons of mothers of blood group B were atmuch reduced risk of cryptorchidism in the presentstudy, and at reduced risk of childhood testicularcancer in one study.33 It would be of value toinvestigate in future studies of testicular cancerwhether breech birth and the other prenatal factorsleading to a raised risk of cryptorchidism in thepresent study are also risk factors for testicularcancer.

We thank Professor M C Pike for stimulating us toundertake this study, the late Dr J A Baldwin,medical director of the Oxford Record LinkageStudy, whose generous cooperation made the studypossible, Dr J Golding and Ms P Adelstein for advice,Mrs Y Dolphin and Mrs R Whitehouse for clericalhelp, and Mrs N Cross and Mrs C McBride for typing.

References

Rajfer J, Walsh PC. Testicular descent. In: Blandau RJ,Bergsma D, eds. Morphogenesis and malformation ofthe genital system. Birth defects: original article series,Vol III, No 2. New York: Alan R Liss Inc, 1977:107-22.

2Whitaker RH. Management of the undescended testis. BrJ Hosp Med 1970; 4: 25-37.

'Fonkalsrud EW. Complications of cryptorchidism. In:Fonkalsrud EW, Mengel W, eds. The undescended testis.Chicago: Year Book Medical Publishers Inc, 1981;108-17.

4Li FP, Fraumeni JF Jr. Testicular cancers in children:epidemiologic characteristics. J Natl Cancer Inst 1972;48: 1575-82.

5Morrison AS. Cryptorchidism, hernia, and cancer of thetestis. J Natl Cancer Inst 1976; 56: 731-3.

'Henderson BE, Benton B, Jing J, Yu MC, Pike MC. Riskfactors for cancer of the testis in young men. IntJ Cancer1979; 23: 598-602.

7Schottenfeld D, Warshauer ME, Sherlock S, Zauber AG,Leder M, Payne R. The epidemiology of testicularcancer in young adults. Am J Epidemiol 1980; 112:232-46.

243



j.com/

J Epidem

iol Com

munity H


ber 1983. Dow

nloaded from


244

.8Acheson ED. Medical record linkage. London: OxfordUniversity Press, 1967.

9Baldwin JA. The Oxford record linkage study as a medicalinformation system. Proceedings ofthe Royal Society ofMedicine 1972; 65: 237-9.

°Armitage P. Statistical methods in medical research.Oxford: Blackwell Scientific Publications, 1971.

Holford TR, White C, Kelsey JL. Multivariate analysis formatched case-control studies. Am J Epidemiol 1978;107: 245-56.

Walter SD, Elwood JM. A test for seasonality of eventswith a variable population at risk. Br J Prev Soc Med1975; 29: 18-21.

Knox G. The incidence of inguinal hernia in Newcastlechildren. Arch Dis Child 1959; 34: 482-6.

"National Child Development Study (1958 cohort).11 000 seven-year-olds. London: Longmans, 1966.

"Scorer CG. The descent of the testis. Arch Dis Child1964; 39: 605-9.

16Cour-Palais IJ. Spontaneous descent of the testicle.Lancet 1966; i: 1403-5.

17Villumsen AL, Zachau-Christiansen B. Spontaneousalterations in position of the testes. Arch Dis Child1966; 41: 198-200.

Scorer CG, Farrington GH. Congenital deformities ofthetestis and epididymis. London: Butterworths, 1971.

1Department of Health and Social Security, Office ofPopulation Censuses and Surveys, Welsh Office.Hospital In-patient Enquiry: main tables 1978. SeriesMB4 No 12. London: HMSO, 1981.

20Bishop MC, Whitaker RH. Cryptorchidism: a renewedplea. Br Med J 1979; i: 407-8.

21Jones PF. Cryptorchidism: a renewed plea. Br Med J1979; i: 616.

22Mitchell RG. Paediatric aspects of cryptorchidism andhypogonadism. Dev Med ChildNeurol 1977; 19: 673-6.

23Scorer CG. Early operation for the undescended testis. BrJ Surg 1967; 54: 694-8.

24Anonymous. Birth trauma in vaginal breech delivery. BrMed J 1978; i: 320-1.

25Ralis ZA. Birth trauma to muscles in babies born bybreech delivery and its possible fatal consequences.Arch Dis Child 1975; 50: 4-13.

26Tiwary CM. Testicular damage following breech delivery.Clin Res 1973; 21: 47.

27Dunn PM. Testicular birth trauma. Arch Dis Child 1975;50: 744-5.

2 Lips U. Francke C, Prader A. Testicular volume andfunction in adolescence after breech delivery. HelvPaediatr Acta 1979; 34: 437-42.

29Brenner WE, Bruce RD, Hendricks CH. Thecharacteristics and perils of breech presentation. Am JObstet Gynecol 1974; 118: 700-12.

30Braun FHT, Jones KL, Smith DW. Breech presentation asan indicator of fetal abnormality. J Pediatr 1975; 86:419-21.

A J Swerdlow, Kathryn H Wood, and P G Smith

31Czeizel A, Erodie E, T6th J. An epidemiological study onundescended testis. J Urol 1981; 126: 524-7.

32Buemann B, Henriksen H, Villumsen AL, Westh A,Zachau-Christiansen B. Incidence of undescended testisin the newborn. Acta Chir Scand 1961; 283 suppl:289-93.

33Waaler PE. Clinical and cytogenetic studies inundescended testes. Acta Paediatr Scand 1976; 65:553-8.

34Kark W. The problem of cryptorchidism and malignancy.Medical Proceedings 1962; 8: 83-92.

35Gross RE, Jewett TC Jr. Surgical experiences from 1222operations for undescended testis. JAMA 1956; 160:634-41.

36Miller A, Seljelid R. Histopathologic classification andnatural history of malignant testis tumors in Norway,1959-1963. Cancer 1971; 28: 1054-62.

3"Swerdlow AJ, Stiller CA, Kinnier Wilson LM. Prenatalfactors in the aetiology of testicular cancer: anepidemiological study of childhood testicular cancerdeaths in Great Britain, 1953-73. J EpidemiolCommunity Health 1982; 36: 96-101.

38 Sakashita S, Koyanagi T, Tsuji I, Arikado K, Matsuno T.Congenital anomalies in children with testicular germcell tumor. J Urol 1980; 124: 889-91.

"9Hendren WH, Ginsburg HB. Associated anomalies inundescended testis. In: Fonkalsrud EW, Mengel W, eds.The undescended testis. Chicago: Year Book MedicalPublishers Inc, 1981: 118-34.

40Loughlin JE, Robboy SJ, Morrison AS. Risk factors forcancer of the testis. N Engl J Med 1980; 303: 112-3.

41Rothman KJ, Louik C. Oral contraceptives and birthdefects. N Engl J Med 1978; 299: 522-4.

12Gill WB, Schumacher GFB, Bibbo M. Genital and semenabnormalities in adult males two and one-half decadesafter in utero exposure to diethylstilbestrol. In: HerbstAL, ed. Intrauterine exposure to diethylstilbestrol in thehuman. Proceedings of symposium on DES. WashingtonDC: American College of Obstetrics and Gynecology1978: 53-7.

43McLachlan JA, Newbold RR, Bullock B. Reproductivetract lesions in male mice exposed prenatally todiethylstilbestrol. Science 1975; 190: 991-2.

44Nomura T, Kanzaki T. Induction of urogenital anomaliesand some tumors in the progeny of mice receivingdiethylstilbestrol during pregnancy. Cancer Res 1977;37: 1099-104.

45 Waterhouse J, Muir C, Correa P, Powell J, eds. Cancerincidence in five continents. Vol III. Lyon: InternationalAgency for Research on Cancer, 1976.

46 Heinonen OP, Slone D, Shapiro S. Birth defects and drugsin pregnancy. Littleton, Massachussets: PublishingSciences Group Inc, 1977.



j.com/

J Epidem

iol Com

munity H


ber 1983. Dow

nloaded from


acase-controlstudyof theaetiologyof cryptorchidismcryptorchidism were recorded in residents in the...

Documents