cryptorchidism and hypospadias in a cohort of 934538 danish

American Journal of Epidemiology

ª The Author 2012. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of

Public Health. All rights reserved. For permissions, please e-mail: [email protected].

Vol. 175, No. 9

DOI: 10.1093/aje/kwr421

Advance Access publication:

March 27, 2012

Original Contribution

Cryptorchidism and Hypospadias in a Cohort of 934,538 Danish Boys: The Role ofBirth Weight, Gestational Age, Body Dimensions, and Fetal Growth

Morten Søndergaard Jensen*, Allen J. Wilcox, Jørn Olsen, Jens Peter Bonde, AneMarie Thulstrup,Cecilia Høst Ramlau-Hansen, and Tine Brink Henriksen

* Correspondence to Dr. Morten Søndergaard Jensen, Perinatal Epidemiology Research Unit, Department of Pediatrics, Aarhus

University Hospital, Brendstrupgaardsvej 100, 8200 Aarhus N, Denmark (e-mail: [email protected]).

Initially submitted August 15, 2011; accepted for publication October 20, 2011.

Early delivery and low birth weight are strong predictors of the urogenital anomalies cryptorchidism (unde-scended testis) and hypospadias. Understanding these associations may lead to important etiologic clues. There-fore, the authors revisited the prevailing hypotheses regarding fetal growth restriction as a risk factor for urogenitalanomalies. They studied a population of 934,538 Danish boys born alive between January 1, 1980, and December 31,2008. Cryptorchidism and hypospadias were associated with low weight-for-gestational-age, an indicator of fetalgrowth restriction, and furthermore the authors observed strong interaction with early delivery. Low birth weight ina singleton compared with the mean birth weight of all singleton brothers in the family or in a twin compared with themale co-twin was associated with higher risk of urogenital anomalies, suggesting an effect of relative fetal growthrestriction within families. Contrary to previous reports, newborns’ body dimensions assessed independently ofbirth weight were not associated with urogenital anomalies. The hypothesis that shared factors cause both fetalgrowth restriction and urogenital anomalies was supported by comparison of urogenital anomaly risks in singletonsand twins and by patterns of cryptorchidism and hypospadias co-occurrence in individuals. These novel insightsmight also extend to other male reproductive conditions with prenatal etiology.

anthropometry; birth weight; cryptorchidism; fetal growth retardation; gestational age; hypospadias; siblings; twins

Abbreviations: CI, confidence interval; ICD, International Classification of Diseases; SGA, small for gestational age.

Early delivery and low birth weight are well-known andstrong predictors of the urogenital anomalies cryptorchi-dism (undescended testis) and hypospadias (1–8). Boys bornsmall for gestational age (SGA) face especially high risks(2–5, 9), which has directed attention to impaired fetalgrowth as a risk factor (9, 10). Risk estimates have been quitesimilar for cryptorchidism and hypospadias (2, 3). This facthas puzzled researchers because, until recently, these twoanomalies were thought to have different etiologic time win-dows (2, 9, 11). Emerging evidence from rat models suggestsa programming window for reproductive tract masculiniza-tion that relates to gestational weeks 8–14 in humans (12).Normal androgen action seems crucial during this time (13),and disruption leads to both cryptorchidism and hypospadiasin rats (12). It thus seems that this puzzle may have beenresolved, but several other questions regarding fetal growth

restriction and urogenital anomalies have as yet defied solu-tion: Is the association with low birth weight due to fetalgrowth restriction or due to a small constitution? In addition,do the associations indicate a continuous biologic link betweenlow birth weight and each urogenital anomaly? Further, areother aspects of a newborn’s body size or proportion associatedwith urogenital anomalies when assessed independently ofbirth weight? Finally, do causal factors shared by fetal growthrestriction and urogenital anomalies create the associations?Failure to properly address such questions may have led tomisconceptions about the etiology of urogenital anomalies.Advancements in our understanding may also extend to othermale reproductive conditions with prenatal etiology, includingtesticular cancer and poor semen quality (11). Therefore, werevisited the prevailing hypotheses regarding fetal growth re-striction as a risk factor for cryptorchidism and hypospadias.

917 Am J Epidemiol. 2012;175(9):917–925

Downloaded from https://academic.oup.com/aje/article-abstract/175/9/917/173346by gueston 17 February 2018

Newborns weighing below the 10th percentile for their ges-tational age are commonly defined as SGA and are regarded as‘‘growth-restricted’’ (14). The growth potential of each fetusis unknown, however, which limits the usefulness of SGA asa measure of poor fetal growth. Not all small newborns arein fact growth-restricted, and many larger newborns could begrowth-restricted from a heavier target weight (14). Whethergrowth restriction might predict urogenital anomalies inde-pendently of absolute birth weight is not known. Therefore,we studied the associations between the relative birth weightof male singletons compared with the mean of all singletonbrothers in the family and the relative birth weight of twinscompared with their male co-twins and urogenital anoma-lies. In addition, associations with SGA are usually reportedcollectively for all gestational ages, although the conse-quences may vary by gestational age. Two studies have foundassociations between male urogenital anomalies and lowbirth weight (or SGA) according to gestational age (1, 2).We explored whether associations between growth restriction,early delivery, and urogenital anomalies might be of a dose-response type, indicating a biologic continuum.

Growth restriction coinciding with the suggested program-ming window for masculinization is of particular interest. Sym-metric reduction in newborns’ body proportions has beensuggested as a way to identify early growth restriction (startingin the first trimester) (15, 16). Other authors have questioned this(17–19). Body proportions are also inherently related to bodyweight. To clarify the predictive value of newborns’ body pro-portions, we studied their association with the risk of the uro-genital anomalies when assessed independently of birth weight.

Shared causes of fetal growth restriction and urogenitalanomalies have been hypothesized (3, 9). Explaining theassociations by the alternative hypotheses, that fetal growthrestriction causes urogenital anomalies or vice versa, mightalso seem even less attractive. We compared urogenitalanomaly risks in twins and singletons, as well as the co-occurrence of cryptorchidism and hypospadias in singletons,to further investigate this suggested shared causal pathway.

MATERIALS AND METHODS

Study population

We studied the entire population of 934,538 Danish boysborn alive between January 1, 1980, and December 31, 2008.Each boy was identified in the Danish Civil RegistrationSystem by a unique personal identifier given at birth (20).The Civil Registration System automatically stores dates ofemigration and death for all citizens. The personal identifierfurther facilitated linkage to the Danish Medical Birth Registryand the Danish National Registry of Patients (21). The DanishMedical Birth Registry provided information on pregnancy anddelivery. Data on birth weight, crown-heel length at birth, ges-tational age at birth, and maternal age at birth were available forthe entire study period (1980–2008). Data on head circumfer-ence and abdominal circumference were available only from1997 onwards. The completeness of the obstetric and anthro-pometric data was approximately 95% during the study period.Midwives recorded the data just after delivery and reportedthem to the register. Gestational age was estimated from the

first day of the last menstrual period or from ultrasound. By1990, ultrasound dating was performed in approximately 80%of pregnancies, and it was used in case of a discrepancy of morethan 2 weeks between the two measures (22).

Information on cryptorchidism, hypospadias, other con-genital malformations, and surgical procedures was obtainedfrom the Danish National Patient Registry. This registrycontains information on all inpatient diagnoses and surger-ies performed during the follow-up period, and it covered allhospitals in Denmark. Boys with a diagnosis of cryptorchi-dism (International Classification of Diseases, Eighth Revi-sion (ICD-8), codes 75210, 75211, and 75219; InternationalClassification of Diseases, Tenth Revision (ICD-10), codesDQ53, DQ531, DQ531A, DQ532, DQ532A, and DQ539)who also underwent corrective orchiopexy (codes 55600,55640, KKFH00, KKFH01, and KKFH10 in the NordicClassification of Surgical Procedures) were consideredcryptorchidism cases. Orchiopexy indicates that the crypt-orchidism persisted until the time of surgery (mean age, 6.3years; interquartile range, 3.7–9.3). Boys with a diagnosis ofhypospadias were considered cases (ICD-8 codes 75220,75221, 75222, 75228, and 75229; ICD-10 codes DQ540,DQ541, DQ542, DQ548, and DQ549). Boys with other con-genital malformations (ICD-8 codes 74000–75999; ICD-10codes Q00–Q99) diagnosed during follow-up were excluded.

Statistical analyses

Cryptorchidism and hypospadias are considered congenitalanomalies, but some cases are diagnosed throughout childhood(23, 24). In most analyses, we estimated crude and adjustedhazard ratios and 95% confidence intervals by means of Coxregression models, using boy’s age as the time variable. Theboys entered the risk set at birth and were followed until theirage at first diagnosis, death, emigration from Denmark, or theend of follow-up (October 21, 2009), whichever came first. Weadjusted for calendar year of birth (7 categories) and maternalage at birth (6 categories) in all unpaired analyses.

Anthropometric characteristics at birth (crown-heel length(cm), head circumference (cm), and abdominal circumference(cm)) were recorded in whole numbers. We added a randomdecimal digit to enable categorization into quintiles. (Forexample, a recorded head circumference of 34 cm was as-signed a random value from 33.5 cm to 34.4 cm.) All anthro-pometric measures were strongly correlated with gestationalage (see Web Figure 1 (http://aje.oxfordjournals.org/)), andtherefore we present quintiles of weight, length, head cir-cumference, and abdominal circumference for gestationalage. Outlying pairs of gestational age and birth weight wereexcluded on the basis of Alexander’s criteria (25). Hazardratios for cryptorchidism and hypospadias were estimatedaccording to quintiles of each anthropometric-for-gestational-age measure, as well as by birth weight group, gestational agegroup, and weight-for-gestational-age group. We consideredpossible interaction between weight-for-gestational-age andgestational age at birth in the risk of urogenital anomalies.

We identified term-born (�37 weeks) singleton brothers toestimate the effects of fetal growth restriction among siblings.The mean birth weight of all singleton brothers served asa measure of each family’s constitutional growth potential.

918 Jensen et al.

Am J Epidemiol. 2012;175(9):917–925


http://www.aje.oxfordjournals.org/lookup/suppl/doi:10.1093/aje/kwr421/-/DC1

http://aje.oxfordjournals.org/

The ratio of each brother’s birth weight to this family mean wascalculated, and ratios were categorized into groups of <85%,85%–94.9%, 95%–104.9%, 105%–114.9%, and �115%.Families with at least 1 affected boy (cryptorchidism or hypo-spadias) and 1 unaffected boy were informative in a conditionallogistic regression analysis using the mother as an identifier.Unaffected boys with a longer follow-up period than theiraffected sibling’s age at first diagnosis served as controls inthis analysis. We estimated odds ratios for urogenital anomaliesaccording to the ratio categories using 95%–104.9% as thereference group and adjusting for gestational week of birth,mother’s parity, and calendar year of birth. For male twinpairs, we estimated differences in birth weight by discordance ofurogenital anomalies using the paired t test.

Body dimensions (length, circumference) are inherentlyassociated with body weight, and estimated effects of bodydimensions can easily be confounded by birth weight. Wegenerated categories of each body dimension independent ofbirth weight, as follows. Within each gestational week (weeks22–45), we modeled the body dimension as the response vari-able with birth weight and squared birth weight as explanatoryvariables in a generalized linear least-squares regression(Figure 1). This was done with body dimension measuresand birth weight on a log scale (natural logarithm) to reducecurvature for a better fit. Body dimensions more than 5 standarddeviations from the mean were excluded from all analyses.Predicted cutoff values for the 20th, 40th, 60th, and 80th per-centiles of the body dimension were estimated at each birthweight, assuming a normally distributed variance around theregression line (Figure 1). Each original body dimension wasthen assigned to a quintile. As intended, the mean birth length,head circumference, and abdominal circumference varied byquintile, while the mean birth weight remained constant (WebTable 1). The hazard ratios for the urogenital anomalies accord-ing to body dimension quintiles were estimated with the middlequintile (40–59th percentiles) used as the reference group.

We estimated the risk of urogenital anomalies in twinscompared with singletons and then stratified by gestationalage at birth (or birth weight), allowing interaction with twinor singleton status. We compared the observed co-occurrenceof urogenital anomalies in individuals with the expected co-occurrence from the marginal distributions of cryptorchidismand hypospadias. In logistic regression analyses, we estimatedthe overall and stratified co-occurrence ratios (odds ratios) foreach possible 23 2 table in 4 groups of weight-for-gestational-age and by term and preterm birth. Co-occurrence ratios above1 indicate higher co-occurrence than could be explained fromthe marginal distributions of cryptorchidism and hypospadiaswithin each group.

We performed secondary analyses 1) restricting our sampleto the years 1997–2008, to ensure that associations with birthweight and gestational age were comparable to those presentedfor the entire study period and 2) excluding 1,067 (9%) crypt-orchidism cases with a diagnosis of inguinal hernia, because thiscondition is associated with early delivery, low birth weight, andcryptorchidism (26).

The Danish National Board of Health and the Danish DataInspectorate approved the study protocol. Statistical analyseswere performed using Stata 11 software (StataCorp LP,College Station, Texas).

RESULTS

Between 1980 and 2008, 934,538 boys were born alive inDenmark. For analyses of singletons, we excluded 27,507 boyswith missing or invalid birth weight or gestational age data(2.9%), 61,250 boys with other congenital malformations(6.6%), 25,885 twins (2.8%), and 785 higher multiples, leaving819,111 boys (87.6%). Of these, 459,657 were born between1980 and 1996, when only data on gestational age, birthweight, and birth length were available. The remaining359,454 singleton boys were born between 1997 and 2008,when the register also included data on head circumferenceand abdominal circumference. Of the latter, 350,258 (97%)had a valid measure of head circumference and 340,980(95%) had a valid measure of abdominal circumference. Dur-ing follow-up, 11,586 cryptorchidism cases (14.1/1000) and3,268 hypospadias cases (4.0/1000) were identified amongsingletons.

Earlier birth and lower birth weight were both stronglyassociated with urogenital anomalies (Table 1). Hypospadiasshowed a steady risk increase with decreasing birth weight,whereas cryptorchidism risk increased only at birth weightsbelow 3,000 g. Boys in the smallest weight-for-gestational-age quintile had the highest risks of cryptorchidism (hazardratio ¼ 1.4, 95% confidence interval (CI): 1.3, 1.5) andhypospadias (hazard ratio ¼ 1.9, 95% CI: 1.7, 2.1) comparedwith the largest quintile. These associations were stronglymodified by gestational age at birth (Figure 2). Boys in the2 heaviest weight-for-gestational-age quintiles (60%–100%)experienced limited (if any) increased risk of urogenitalanomalies with early delivery. In contrast, boys in the low-est weight-for-gestational-age quintile (0%–<20%) expe-rienced the highest risks with early delivery, reachinga hazard ratio of 7.8 (95% CI: 6.0, 10) for cryptorchidism

Figure 1. Example of how a body dimension (head circumference forboys born at 30 weeks’ gestation) is categorized into quintiles that areindependent of birth weight by fitting a generalized linear model with birthweight and squared birth weight as explanatory variables, Denmark,1980–2008. Head circumference and birth weight data were trans-formed by natural logarithm (ln). Dotted line, 20th percentile; short-dashed line, 40th percentile; long-dashed line, 60th percentile; solid line,80th percentile.

Fetal Growth Restriction and Urogenital Anomalies 919

Am J Epidemiol. 2012;175(9):917–925




Table 1. Risks of Cryptorchidism and Hypospadias According to Gestational Age, Birth Weight, and Anthropometric Characteristics at Birth

Among 819,111 Danish Singleton Boys, 1980–2008

Characteristic Total No.

Cryptorchidism Hypospadias

No. ofCases

HRa 95% CINo. ofCases

HRa 95% CI

Gestational age, weeksb

<28 1,412 31 3.7 2.6, 5.3 5 1.9 0.8, 4.6

28–31.9 3,936 136 3.0 2.5, 3.6 37 3.0 2.1, 4.1

32–34.9 10,060 248 1.9 1.7, 2.2 92 2.5 2.1, 3.1

35–36.9 24,025 478 1.5 1.4, 1.7 133 1.5 1.3, 1.8

37–38.9 125,892 1,845 1.2 1.1, 1.2 552 1.2 1.1, 1.3

39–39.9 168,711 2,163 1.0 1.0, 1.1 699 1.1 1.0, 1.2

40–41.9 414,435 5,695 1.0 Reference 1,514 1.0 Reference

�42 70,640 990 1.0 1.0, 1.1 236 0.9 0.8, 1.0

Birth weight, gb

<1,500 4,538 221 5.5 4.8, 6.2 55 4.8 3.6, 6.3

1,500–1,999 5,997 179 2.4 2.1, 2.8 72 3.6 2.8, 4.6

2,000–2,499 17,999 478 2.1 1.9, 2.3 147 2.4 2.0, 2.8

2,500–2,999 78,426 1,383 1.4 1.3, 1.4 419 1.6 1.4, 1.8

3,000–3,499 244,253 3,471 1.1 1.1, 1.2 985 1.1 1.0, 1.3

3,500–3,999 291,688 3,657 1.0 Reference 1,050 1.0 Reference

4,000–4,499 141,071 1,746 1.0 1.0, 1.1 442 0.9 0.8, 1.0

4,500–4,999 30,581 382 1.0 0.9, 1.2 89 0.8 0.6, 1.0

�5,000 4,558 69 1.3 1.0, 1.6 9 0.5 0.3, 1.0

Percentile of weight for gestational ageb

<20 156,430 2,891 1.4 1.3, 1.5 875 1.9 1.7, 2.1

20–39.9 162,335 2,305 1.1 1.0, 1.2 649 1.3 1.2, 1.5

40–59.9 170,435 2,293 1.0 1.0, 1.1 683 1.3 1.2, 1.5

60–79.9 158,593 1,948 1.0 0.9, 1.0 526 1.1 0.9, 1.2

�80 171,318 2,149 1.0 Reference 535 1.0 Reference

Percentile of birth length for gestational ageb

<20 162,727 2,771 1.2 1.1, 1.3 856 1.7 1.5, 1.9

20–39.9 160,767 2,262 1.0 1.0, 1.1 630 1.2 1.1, 1.4

40–59.9 161,110 2,142 1.0 0.9, 1.0 610 1.2 1.1, 1.3

60–79.9 159,977 2,019 0.9 0.9, 1.0 584 1.1 1.0, 1.3

�80 165,295 2,204 1.0 Reference 535 1.0 Reference

Percentile of head circumference for gestational agec

<20 68,337 734 1.1 1.0, 1.2 294 1.5 1.3, 1.7

20–39.9 69,512 667 1.0 0.9, 1.1 340 1.2 1.1, 1.5

40–59.9 68,466 677 1.0 0.9, 1.1 306 1.1 1.0, 1.3

60–79.9 71,173 622 0.9 0.8, 1.0 309 1.1 0.9, 1.3

�80 72,770 750 1.0 Reference 288 1.0 Reference

Percentile of abdominal circumference for gestational agec

<20 70,474 768 1.1 1.0, 1.2 441 1.8 1.5, 2.1

20–39.9 67,294 628 1.0 0.9, 1.1 331 1.4 1.2, 1.7

40–59.9 68,422 654 1.0 0.9, 1.1 302 1.3 1.1, 1.5

60–79.9 66,438 590 0.9 0.9, 1.0 262 1.1 0.9, 1.3

�80 68,352 672 1.0 Reference 241 1.0 Reference

Abbreviations: CI, confidence interval; HR, hazard ratio.a Adjusted for maternal age at birth and calendar year of birth.b 1980–2008 (n ¼ 819,111 for gestational age and birth weight; n ¼ 809,876 for birth length).c 1997–2008 (n ¼ 350,258 for head circumference; n ¼ 340,980 for abdominal circumference).

920 Jensen et al.

Am J Epidemiol. 2012;175(9):917–925


and 10.4 (95% CI: 6.9, 16) for hypospadias at a gestationalage below 32 weeks.

Inferred growth restriction was studied using the relativebirth weights of singleton brothers in 3,939 families with atleast 1 cryptorchidism case and in 1,221 families with at least 1hypospadias case (Table 2). Lower relative birth weightshowed a dose-response-type association with higher risk ofhypospadias. For cryptorchidism, no such pattern was seen,but brothers with birth weights below 85% of the family meanhad an increased odds ratio of 1.6 (95% CI: 1.1, 2.4).

Among 9,801 male twin pairs, we identified 238 pairsdiscordant for cryptorchidism and 99 pairs discordant forhypospadias. Compared with the unaffected twin, the twinwith cryptorchidism was on average 136 g lighter (95% CI:70, 202) and the twin with hypospadias was 164 g lighter(95% CI: 45, 283).

Hypospadias was associated with short crown-heel length,head circumference, and abdominal circumference for gesta-tional age, while these associations were weaker or absent with

cryptorchidism (Table 1). However, we observed no associationbetween length, head circumference, or abdominal circumfer-ence independent of birth weight and the urogenital anomalieswhen using the constructed quintiles (Web Table 1). Stratifica-tion by prematurity did not modify the associations.

We compared the risk of urogenital anomalies in twins withthat in singletons. Twinning was associated with a hazard ratiofor cryptorchidism of 1.2 (95% CI: 1.1, 1.3), but this was re-duced to 0.9 (95% CI: 0.8, 1.0) when adjusted for gestationalage and to 0.8 (95% CI: 0.7, 0.8) when adjusted for birthweight. Likewise, twinning was associated with a hazard ratiofor hypospadias of 1.4 (95% CI: 1.1, 1.6), which was reduced to1.0 (95% CI: 0.8, 1.2) when adjusted for gestational age and to0.8 (95% CI: 0.6, 0.9) when adjusted for birth weight. Differ-ences between singletons and twins were most clear for birthweight when allowing interaction between singleton versustwin status and gestational age or birth weight (Figure 3).

Co-occurrence was rare by our case definitions (84 cases;1/10,000). The overall cryptorchidism and hypospadias co-occurrence ratio was 1.8 (95% CI: 1.5, 2.3) (Table 3). Termboys had a ratio of 1.6 (95% CI: 1.3, 2.1), while pretermboys had a ratio of 3.2 (95% CI: 2.0, 5.1). Co-occurrenceratios increased with lower weight for gestational age, andthis association seemed stronger among preterm boys thanamong term boys (Table 3).

We repeated all analyses of gestational age, birth weight, andbirth length in the 1997–2008 subset (n¼ 340,091), and resultswere essentially unchanged. Exclusion of cryptorchidism caseswith inguinal hernia slightly attenuated some associations, butnot to a degree that affected our conclusions.

DISCUSSION

We revisited the roles of birth weight, gestational age, bodydimensions, and fetal growth in the risks of cryptorchidism andhypospadias using a large population-based cohort. We ob-served an association between low birth weight for gestationalage and urogenital anomalies that was strengthened by earlydelivery. Estimated fetal growth restriction, identified by lowbirth weight in a singleton compared with the mean of all sin-gleton brothers in the family and in a twin compared with themale co-twin, was associated with a higher risk of urogenitalanomalies. Body proportions, when assessed independently ofbirth weight, had no impact on risk of the anomalies. Twins hadhigher crude urogenital anomaly risk than singletons, but twin-ning became ‘‘protective’’ after adjustment for birth weight orgestational age. The co-occurrence of cryptorchidism and hy-pospadias was higher in boys with a high risk of fetal growthrestriction (low weight for gestational age). Taken together,these data strongly suggest that fetal growth restriction is asso-ciated with urogenital anomalies and, further, that this associa-tion is due to shared causes.

A few studies have explored the possible interaction betweengestational age and birth weight or SGA in the risk of urogenitalanomalies (1, 2). We extended this to show continuous effects ofweight for gestational age modified by gestational age at birth(Figure 2). Short gestation had little if any effect on the risk ofurogenital anomalies among normal-to-heavy boys.

Few studies have found higher risk of hypospadias in thesmaller of twin or singleton brother pairs (27, 28). We observed

Figure 2. Hazard ratios (HR) forA)cryptorchidismandB)hypospadiasin 3 birth-weight-for-gestational-age groups (dotted line, <20%; dashedline, 20%–59.9%; solid line, 60%–100%), according to gestational ageatbirth, Denmark, 1980–2008. Boys born in gestational weeks 40–41 andweighing 60%–100% of weight for gestational age were the referencegroup. Vertical lines represent 95% confidence intervals.


Am J Epidemiol. 2012;175(9):917–925



lower birth weight in the urogenital anomaly-affected twin thanin the unaffected twin among discordant pairs. We had no in-formation on zygosity, chorionicity, or twin-twin transfusionsyndrome with which to explain the weight differences. We

corroborated the reported association of lower birth weight inhypospadias-affected singleton brothers (27) and extended it tocryptorchidism (Table 2). However, unlike the case with hypo-spadias, brothers heavier than the family mean (the mean of all

Table 2. Odds Ratios for Cryptorchidism in 3,939 Families and Odds Ratios for Hypospadias in 1,221 Families

According to Family Birth Weight Ratioa Among Singleton Brothers Born at Term (�37 Weeks), Denmark, 1980–2008

Birth WeightRatio, %

Cryptorchidism Hypospadias

No. ofCases

Total No. ORb 95% CINo. ofCases

Total No. ORb 95% CI

<85 137 249 1.6 1.1, 2.4 47 87 1.2 0.6, 2.4

85–94.9 925 1,932 1.1 0.9, 1.3 299 578 1.2 0.9, 1.7

95–104.9 1,972 4,323 1.0 Reference 641 1,398 1.0 Reference

105–114.9 853 1,949 1.1 0.9, 1.3 224 577 0.9 0.6, 1.2

�115 90 254 1.1 0.7, 1.7 26 86 0.6 0.3, 1.3

P-trendc 0.11 <0.001

Abbreviations: CI, confidence interval; OR, odds ratio.a Ratio (%) of a singleton boy’s birth weight to the mean birth weight of all singleton brothers in his family. A ratio

below 100 represents a birth weight that was lighter than the mean.b Estimated by conditional logistic regression (mother was the identifier) and adjusted for gestational age at birth,

mother’s parity, and calendar year of birth.c P value estimate from regression trend test of the birth weight ratio modeled as a continuous variable.

Figure 3. Hazard ratios (HR) for cryptorchidism (top) and hypospadias (bottom) in singletons (solid line) and twins (dashed line) according togestational age (left) and birth weight (right), Denmark, 1980–2008. The reference group was the highest category of singletons in each graph.Vertical lines represent 95% confidence intervals. P values tested the null-hypothesis of no overall difference between singletons and twins.

922 Jensen et al.

Am J Epidemiol. 2012;175(9):917–925


singleton brothers’ birth weights) had no reduced risk of crypt-orchidism. This discrepancy may be due to the different riskpatterns of the anomalies at high birth weights. Hypospadiasshowed a steady risk decrease with increasing birth weight,whereas cryptorchidism risk remained constant above 3,000 g(Table 1). Using family controls removes the between-familyvariation in fetal growth and birth weight, including a compo-nent of constitutional birth weight (29). The remaining within-family variation in birth weight might thus be a better measureof true fetal growth restriction (i.e., not attaining a target birthweight). Our analyses of twin and singleton brothers clearlyshowed that attaining a lower birth weight than one’s brother,indicating relative growth restriction, is associated with higherrisk of urogenital anomalies.

Associations between anthropometric factors at birth (such ascrown-heel length, head circumference, and abdominal circum-ference) and hypospadias have been reported (15, 24, 30–32).Symmetric reduction in all body dimensions has shown partic-ularly strong associations (15). Some authors have suggestedthat symmetric growth restriction begins in the first trimesterand hence spans a relevant time window for urogenital anoma-lies (15, 16). This use of symmetry to indicate early growthrestriction has been questioned (17–19). We studied whetherbody proportions had explanatory value when assessed inde-pendently of birth weight. That is, do a long and skinny boy anda short and fat boy with identical birth weights face the samerisks of urogenital anomaly? Our data showed no associationbetween body proportions (independent of birth weight) andurogenital anomalies. It thus seems that, when studying urogen-ital anomalies, newborns’ body dimensions are merely proxymeasures of the birth weight and hence might not contributefurther to our understanding of the etiology of these anomalies.

Shared causes of fetal growth restriction and urogenitalanomalies have been hypothesized (3, 9), and we provide evi-dence supporting this. Twins are generally born earlier and withlower birth weight than singletons (33, 34), often because ofuterine crowding by 2 healthy fetuses (35). Twins in our studyhad higher urogenital anomaly risk than singletons, but whenresults were adjusted for birth weight (or gestational age), twin-ning became ‘‘protective.’’ Twins had a lower urogenital anom-aly risk than singletons at most low birth weights (Figure 3),indicating that the causes of singleton growth restriction may

also be stronger risk factors for urogenital anomalies than twin-ning. Other data have supported this (3, 8, 36), and within theconceptual framework of causal graphs, this indicates colliderbias or intermediate-outcome confounding (37). This suggeststhat shared (and perhaps unknown) factors may cause both lowbirth weight (by fetal growth restriction) and urogenital anom-alies. This might also be the case with early delivery and uro-genital anomalies. A similar argument has been made to suggestshared causes of preterm birth and cerebral palsy (38). Theexistence of shared factors is further supported by a higherco-occurrence of urogenital anomalies among preterm boysand boys with low weight for gestational age (Table 3). Differ-ential reporting of co-occurrence by prematurity or low birthweight seems unlikely.

Placental dysfunction and androgen deficiency in early preg-nancy may be candidate shared factors due to their associationwith fetal growth and androgens’ importance for masculiniza-tion (13, 32, 39–41). Any such shared factors might also beshared by other male reproductive conditions associated withlow birth weight, including testicular cancer and poor semenquality (42–46). Data on low birth weight and semen qualityare as yet inconclusive (42–45), but testicular cancer has shownan association with low birth weight (46). As with cryptorchi-dism and hypospadias, twins had higher testicular cancer riskthan singletons when findings were unadjusted for birth weight(46, 47). However, we were unable to identify data evaluatingwhether twinning becomes ‘‘protective’’ after birth weightadjustment (47, 48), which would indicate shared causes offetal growth restriction, cryptorchidism, hypospadias, andtesticular cancer.

We used administrative health data from Danish registriescovering all boys born between 1980 and 2008. As a conse-quence, there was no selection in establishing the cohort, andloss to follow-up was minimal. Midwives routinely recordednewborns’ anthropometric characteristics and basic maternalcharacteristics. Any measurement error, including errors in ges-tational age, weight, length, and circumference measures,would most likely have attenuated associations. The presum-ably greater error in length and circumference measures than inbirth weight might have attenuated associations with dispropor-tionality. The register-based endpoint of cryptorchidism wasverified by orchiopexy to exclude transient cases (spontaneously

Table 3. Co-occurrence Ratios for Cryptorchidism and Hypospadias in Individuals According to Percentile of

Weight for Gestational Age and Preterm Birth Statusa Among 819,111 Danish Singleton Boys, 1980–2008

Percentile ofWeight for

Gestational Age

No. ofBoys With

Co-OccurrenceTotal No.

Co-occurrence Ratiob

All Births Preterm Births Term Births

OR 95% CI OR 95% CI OR 95% CI

Total 84 819,111 1.8 1.5, 2.3 3.2 2.0, 5.1 1.6 1.3, 2.1

<20 38 156,430 2.4 1.8, 3.4 2.7 1.5, 4.9 2.0 1.4, 3.0

20–39.9 16 162,335 1.8 1.1, 2.9 2.9 0.9, 9.5 1.6 0.9, 2.8

40–59.9 11 170,435 1.2 0.7, 2.2 2.7 0.6, 11.1 1.1 0.6, 2.1

�60 19 329,911 1.5 0.9, 2.3 1.1 0.2, 7.9 1.5 0.9, 2.4

Abbreviations: CI, confidence interval; OR, odds ratio.a Term birth, �37 weeks; preterm birth, <37 weeks.b A ratio above 1 indicates higher co-occurrence than would be expected from the marginal distributions of

cryptorchidism and hypospadias within each group.


Am J Epidemiol. 2012;175(9):917–925


descending testicles) caused by prematurity. A review ofmedical records from 43 hypospadias cases from the registerconfirmed 40 (93%) of the diagnoses (49). Lack of diagnosticspecificity would tend to attenuate associations.

We conclude that cryptorchidism and hypospadias areassociated with indicators of fetal growth restriction. Esti-mated relative growth restriction based on comparisons ofbrothers’ birth weights supported this. Newborns’ body pro-portions assessed independently of birth weight were notassociated with urogenital anomalies. Previous specula-tions about shared factors’ causing fetal growth restriction,low birth weight, and urogenital anomalies were stronglysupported by our data. Identifying shared causal factorsmight reveal new insights into the biology and etiology ofurogenital malformations.

ACKNOWLEDGMENTS

Author affiliations: Perinatal Epidemiology Research Unit,Department of Pediatrics, Aarhus University Hospital, Aarhus,Denmark (Morten Søndergaard Jensen, Tine Brink Henriksen);Department of Occupational Medicine, Aarhus UniversityHospital, Aarhus, Denmark (Morten Søndergaard Jensen,Ane Marie Thulstrup); Epidemiology Branch, National In-stitute of Environmental Health Sciences, Durham, NorthCarolina (Allen J. Wilcox); Department of Epidemiology,School of Public Health, University of Aarhus, Aarhus,Denmark (Jørn Olsen, Cecilia Høst Ramlau-Hansen); andDepartment of Occupational and Environmental Medicine,Bispebjerg Hospital, University of Copenhagen, Copenhagen,Denmark (Jens Peter Bonde).

Aarhus University, Faculty of Health Sciences, supportedthis work by providing Dr. Morten Søndergaard Jensen witha PhD scholarship. Additional financial support was providedby the Dagmar Marshall Foundation; the Aarhus UniversityResearch Foundation; the Foundation for Advancement ofMedical Science; Knud Højgaard’s Foundation; Jacob Madsenand Olga Madsen’s Foundation;Christian and Ottilia Brorson’sScholarship for Young Scientists; the Danish Ramazzini Cen-tre; C. C. Klestrup and Henriette Klestrup’s Memorial Foun-dation; and Carl and Ellen Hertz’s Scholarship for DanishMedical Science.

The sponsors played no part in the study design, data col-lection, analysis, or preparation of the manuscript, and they arenot responsible for the scientific content of this article or theconclusions expressed.

Conflict of interest: none declared.

REFERENCES

1. Kallen B, Bertollini R, Castilla E, et al. A joint internationalstudy on the epidemiology of hypospadias. Acta Paediatr ScandSuppl. 1986;324:1–52.

2. Akre O, Lipworth L, Cnattingius S, et al. Risk factor patterns forcryptorchidism and hypospadias. Epidemiology. 1999;10(4):364–369.

3. Weidner IS, Møller H, Jensen TK, et al. Risk factors for crypt-orchidism and hypospadias. J Urol. 1999;161(5):1606–1609.

4. Jones ME, Swerdlow AJ, Griffith M, et al. Prenatal risk factorsfor cryptorchidism: a record linkage study. Paediatr PerinatEpidemiol. 1998;12(4):383–396.

5. Pierik FH, Burdorf A, Deddens JA, et al. Maternal and paternalrisk factors for cryptorchidism and hypospadias: a case-controlstudy in newborn boys. Environ Health Perspect. 2004;112(15):1570–1576.

6. Biggs ML, Baer A, Critchlow CW. Maternal, delivery, andperinatal characteristics associated with cryptorchidism:a population-based case-control study among births inWashington State. Epidemiology. 2002;13(2):197–204.

7. Ghirri P, Ciulli C, Vuerich M, et al. Incidence at birth and naturalhistory of cryptorchidism: a study of 10,730 consecutive maleinfants. J Endocrinol Invest. 2002;25(8):709–715.

8. Damgaard IN, Jensen TK, Petersen JH, et al. Risk factors forcongenital cryptorchidism in a prospective birth cohort study.Nordic Cryptorchidism Study Group. PLoS ONE. 2008;3(8):e3051. (doi:10.1371/journal.pone.0003051).

9. Møller H, Weidner IS. Epidemiology of cryptorchidism andhypospadias. Epidemiology. 1999;10(4):352–354.

10. Gatti JM, Kirsch AJ, Troyer WA, et al. Increased incidence ofhypospadias in small-for-gestational age infants in a neonatalintensive-care unit. BJU Int. 2001;87(6):548–550.

11. Main KM, Jensen RB, Asklund C, et al. Low birth weight andmale reproductive function. Horm Res. 2006;65(suppl 3):116–122.

12. Welsh M, Saunders PT, Fisken M, et al. Identification in rats ofa programming window for reproductive tract masculinization,disruption of which leads to hypospadias and cryptorchidism.J Clin Invest. 2008;118(4):1479–1490.

13. Macleod DJ, Sharpe RM, Welsh M, et al. Androgen action inthe masculinization programming window and development ofmale reproductive organs. Int J Androl. 2010;33(2):279–287.

14. Wilcox AJ. Birth weight and fetal growth. In: Wilcox AJ, ed.Fertility and Pregnancy. New York, NY: Oxford UniversityPress; 2010:211–229.

15. Hussain N, Chaghtai A, Herndon CD, et al. Hypospadiasand early gestation growth restriction in infants. Pediatrics.2002;109(3):473–478.

16. Villar J, Belizan JM. The timing factor in the pathophysiologyof the intrauterine growth retardation syndrome. Obstet GynecolSurv. 1982;37(8):499–506.

17. Vik T, Vatten L, Jacobsen G, et al. Prenatal growth in symmetricand asymmetric small-for-gestational-age infants. Early HumDev. 1997;48(1-2):167–176.

18. Hemachandra AH, Klebanoff MA. Use of serial ultrasound toidentify periods of fetal growth restriction in relation to neonatalanthropometry. Am J Hum Biol. 2006;18(6):791–797.

19. Rasmussen S, Kiserud T, Albrechtsen S. Foetal size andbody proportion at 17–19 weeks of gestation and neonatalsize, proportion, and outcome. Early Hum Dev. 2006;82(10):683–690.

20. Pedersen CB, Gøtzsche H, Møller JO, et al. The Danish CivilRegistration System. A cohort of eight million persons. DanMed Bull. 2006;53(4):441–449.

21. Andersen TF, Madsen M, Jørgensen J, et al. The Danish NationalHospital Register. A valuable source of data for modern healthsciences. Dan Med Bull. 1999;46(3):263–268.

22. Jørgensen FS. Epidemiological studies of obstetric ultrasoundexaminations in Denmark 1989–1990 versus 1994–1995. ActaObstet Gynecol Scand. 1999;78(4):305–309.

23. Jensen MS, Olsen LH, Thulstrup AM, et al. Age at cryptorchi-dism diagnosis and orchiopexy in Denmark: a population based

924 Jensen et al.

Am J Epidemiol. 2012;175(9):917–925


study of 508,964 boys born from 1995 to 2009. J Urol. 2011;186(suppl 4):1595–1600.

24. Boisen KA, Chellakooty M, Schmidt IM, et al. Hypospadiasin a cohort of 1072 Danish newborn boys: prevalence andrelationship to placental weight, anthropometrical measurementsat birth, and reproductive hormone levels at three months of age.J Clin Endocrinol Metab. 2005;90(7):4041–4046.

25. Alexander GR, Himes JH, Kaufman RB, et al. A United Statesnational reference for fetal growth. Obstet Gynecol. 1996;87(2):163–168.

26. Depue RH. Maternal and gestational factors affecting the risk ofcryptorchidism and inguinal hernia. Int J Epidemiol. 1984;13(3):311–318.

27. Fredell L, Kockum I, Hansson E, et al. Heredity of hypospadiasand the significance of low birth weight. J Urol. 2002;167(3):1423–1427.

28. Fredell L, Lichtenstein P, Pedersen NL, et al. Hypospadias isrelated to birth weight in discordant monozygotic twins. J Urol.1998;160(6):2197–2199.

29. Lunde A, Melve KK, Gjessing HK, et al. Genetic and envi-ronmental influences on birth weight, birth length, headcircumference, and gestational age by use of population-based parent-offspring data. Am J Epidemiol. 2007;165(7):734–741.

30. Fujimoto T, Suwa T, Kabe K, et al. Placental insufficiency inearly gestation is associated with hypospadias. J Pediatr Surg.2008;43(2):358–361.

31. Calzolari E, Contiero MR, Roncarati E, et al. Aetiologicalfactors in hypospadias. J Med Genet. 1986;23(4):333–337.

32. Hughes IA, Northstone K, Golding J. Reduced birth weight inboys with hypospadias: an index of androgen dysfunction?ALSPAC Study Team. Arch Dis Child Fetal Neonatal Ed.2002;87(2):F150–F151.

33. Luke B. The changing pattern of multiple births in the UnitedStates: maternal and infant characteristics, 1973 and 1990.Obstet Gynecol. 1994;84(1):101–106.

34. Gardosi J, Mul T, Francis A, et al. Comparison of second tri-mester biometry in singleton and twin pregnancies conceivedwith assisted reproductive techniques. Br J Obstet Gynaecol.1997;104(6):737–740.

35. Garite TJ, Clark RH, Elliott JP, et al. Twins and triplets: theeffect of plurality and growth on neonatal outcome comparedwith singleton infants. Am J Obstet Gynecol. 2004;191(3):700–707.

36. Carmichael SL, Shaw GM, Nelson V, et al. Hypospadias inCalifornia: trends and descriptive epidemiology. Epidemiology.2003;14(6):701–706.

37. Glymour MM, Greenland S. Causal diagrams. In: Rothman KJ,Greenland S, Lash TL, eds. Modern Epidemiology. 3rd ed.Philadelphia, PA: Lippincott Williams & Wilkins; 2008:183–209.

38. VanderWeele TJ, Hernandez-Diaz S. Is there a direct effect ofpre-eclampsia on cerebral palsy not through preterm birth?Paediatr Perinat Epidemiol. 2011;25(2):111–115.

39. de Zegher F, Francois I, Boehmer AL, et al. Androgens and fetalgrowth. Horm Res. 1998;50(4):243–244.

40. Scaramuzzo RT, Boldrini A, Bertelloni S, et al. Low testosteronelevels in pre-term newborns born small for gestational age.J Endocrinol Invest. 2010;33(4):215–217.

41. Miles HL, Gidlof S, Nordenstrom A, et al. The role of androgensin fetal growth: observational study in two genetic models ofdisordered androgen signalling. Arch Dis Child Fetal NeonatalEd. 2010;95(6):F435–F438.

42. Francois I, de Zegher F, Spiessens C, et al. Low birth weightand subsequent male subfertility. Pediatr Res. 1997;42(6):899–901.

43. Olsen J, Bonde JP, Basso O, et al. Birthweight and semencharacteristics. Int J Androl. 2000;23(4):230–235.

44. Ramlau-Hansen CH, Hansen M, Jensen CR, et al. Semenquality and reproductive hormones according to birthweightand body mass index in childhood and adult life: two decades offollow-up. Fertil Steril. 2010;94(2):610–618.

45. Cicognani A, Alessandroni R, Pasini A, et al. Low birth weight forgestational age and subsequent male gonadal function. J Pediatr.2002;141(3):376–379.

46. Cook MB, Akre O, Forman D, et al. A systematic review andmeta-analysis of perinatal variables in relation to the risk oftesticular cancer—experiences of the son. Int J Epidemiol.2010;39(6):1605–1618.

47. Neale RE, Carriere P, Murphy MF, et al. Testicular cancer intwins: a meta-analysis. Br J Cancer. 2008;98(1):171–173.

48. Ramlau-Hansen CH, Olesen AV, Parner ET, et al. Perinatalmarkers of estrogen exposure and risk of testicular cancer:follow-up of 1,333,873 Danish males born between 1950and 2002. Cancer Causes Control. 2009;20(9):1587–1592.

49. Pedersen L, Skriver MV, Nørgaard M, et al. Maternal use ofLoratadine during pregnancy and risk of hypospadias in off-spring. Int J Med Sci. 2006;3(1):21–25.


Am J Epidemiol. 2012;175(9):917–925


cryptorchidism and hypospadias in a cohort of 934538 danish

Documents