Hypospadias is one of the most commonly occurring birth defects, affecting four to six per 1000 male births (Paulozzi, 1999; Paulozzi and others, 1997; Porter and others, 2005). In hypospadias, the urethral opening is displaced to a certain degree, varying from placement just off the tip of the penis to a perineal meatus. Three phenotypic subgroups are identified and classified as first degree hypospadias when the urethral opening is located at the glans or corona, second degree hypospadias when it is located on the shaft of the penis, and third degree hypospadias when the urethra ends in the penoscrotal region, scrotum, or perineum. Patients often undergo surgery in the first two years of life, but may still encounter medical, social, and sexual problems later in life (Mieusset and Soulie, 2005; Schonbucher and others, 2008). Also, the complication rates of surgery are relatively high, especially with the more severe forms of hypospadias (Eassa and others, 2011; Nuininga and others, 2005).

So far, several risk factors for hypospadias have been proposed, including assisted reproductive technology (Brouwers and others, 2007; Brouwers and others, 2010; Carmichael and others, 2007), maternal hypertension during pregnancy (Akre and others, 2008; Caton and others, 2008), thyroid disease (Browne and others, 2009), high maternal age at delivery (Carmichael and others, 2012; Carmichael and others, 2003; Porter and others, 2005), and maternal obesity (Akre and others, 2008; Brouwers and others, 2010). Associations with low birth weight, preterm birth, and primiparity are well established (Porter and others, 2005; van der Zanden and others, 2012). Familial clustering points towards a genetic contribution to the etiology of hypospadias (Carmichael and others, 2012; Harris, 1990; Schnack and others, 2008).

In a study assessing phenotypic differences, Brouwers et al. (2010) found evidence for etiological heterogeneity between the different hypospadias phenotypes. Others found additional associations (van Rooij and others, 2013), but the few studies investigating risk factors for different hypospadias phenotypes were too small to draw firm conclusions. In the present study, data from the National Birth Defects Prevention Study (NBDPS), a large case-control study, were used to investigate etiological heterogeneity of second and third degree hypospadias.

The mothers of 2148 cases with second or third degree hypospadias and 5183 male controls were interviewed. Excluding 212 cases (123 second and 89 third degree) with multiple major anomalies left 1547 second degree and 389 third degree hypospadias cases for analysis.

Many descriptive characteristics differed among the three groups (table 1). Mean birth weight, gestational age, and SGA were statistically significantly different among all three groups (unpaired t-test, p<0.001), with the highest proportion of low or very low birth weight, preterm delivery, and SGA infants among third degree cases. Time to interview, study center, and year of EDD varied between both phenotypes and controls, as well.

Many risk factors were associated with one or both of the hypospadias subtypes in the crude analyses (table 2) and were included in table 3. After adjustment for confounding, the adjusted odds ratios (aORs) for 5 of the 15 potential risk factors were statistically significantly different between second and third degree hypospadias (table 3). Family history of hypospadias, increasing maternal age, and primiparity were associated with both phenotypes. The OR for family history was higher for second than third degree hypospadias, although the OR for 2^nd vs 3^rd degree hypospadias was not statistically significant. The associations for maternal age and primiparity were stronger for third than second degree hypospadias. Decreased odds ratios were observed for second degree hypospadias for all racial/ethnic categories except for non-Hispanic black mothers when compared to non-Hispanic white mothers, whereas for third degree hypospadias, elevated ORs were observed for non-Hispanic black and Asian/Pacific Islander mothers. Early exposure to antihypertensive medication was not associated with hypospadias, but exposure to medication later in pregnancy and untreated hypertension were associated with both phenotypes. These associations were stronger for third than second degree hypospadias. The association with thyroxin exposure was of borderline statistical significance in second degree hypospadias only. The associations for plurality and fertility treatments remained elevated when combined into a 6-level variable in the multivariable analysis. Compared to singleton infants conceived without assistance, increased aORs were observed for multiples conceived without fertility treatment for both second and third degree hypospadias. IVF/ICSI was associated with both hypospadias phenotypes, whereas treatment with clomiphene citrate prior to conception was only associated with third degree hypospadias, regardless of plurality. Results from a multinomial model with both subtypes and controls did not differ substantially from the results obtained through the binomial models described above.

Most risk factors evaluated were either associated with both second and third degree hypospadias or with neither phenotype, indicating that the underlying mechanism is partly similar for second and third degree hypospadias. Effect estimates were generally larger for third degree hypospadias than for second degree hypospadias, except for a family history of hypospadias which seemed to occur more often among second degree hypospadias.

The odds ratios for a family history of hypospadias were by far the strongest effect estimates observed in our study, with odds ratios of 25.9 (95% CI 13.1-51.3) for second degree hypospadias and 16.4 (95% CI 6.2-43.5) for third degree hypospadias. Most other studies, that also included first degree hypospadias cases, similarly showed lower risks for family history in third degree cases when compared to less severe subtypes (Brouwers and others, 2010; Fredell and others, 2002; van Rooij and others, 2013), but one U.S. study did not (Bauer and others, 1979).

Associations between non-Hispanic white race/ethnicity and isolated hypospadias have been reported by several researchers (Carmichael and others, 2003; Porter and others, 2005; van der Zanden and others, 2012). For third degree hypospadias, however, the risk for non-Hispanic white mothers was similar to that for Hispanic mothers and those with other race/ethnicities, whereas elevated odds ratios were observed for non-Hispanic black and Asian/Pacific Islander mothers. To our knowledge, the latter associations have not been observed before, but suggest that several race/ethnicities have different risks for certain phenotypes of hypospadias.

A link between hypospadias and placental insufficiency was first published by Stoll et al. (1990), showing an association between hypospadias and low placental weight. Others found that hypospadias was more likely to occur in the smallest of monozygotic twins (Fredell and others, 1998) and that SGA newborns with hypospadias had a high rate of early onset intrauterine growth retardation due to placental insufficiency (Yinon and others, 2010). Placental human chorionic gonadotropin (hCG) stimulates fetal testicular steroidogenesis before the fetus' own production is started, and placental insufficiency may result in a failure to transfer hCG to the fetus in early pregnancy (Brouwers and others, 2007; Fredell and others, 1998). Therefore, placental insufficiency could result in both hypospadias and early onset intrauterine growth retardation. Our findings that boys with hypospadias were more likely to have a low birth weight or to be SGA support this. Since these associations seemed to be stronger for third degree than second degree cases, the role of placental insufficiency may be more prominent in the etiology of the third degree hypospadias phenotype.

Maternal hypertension is also associated with placental insufficiency and was found to be a risk factor for hypospadias in several studies (Akre and others, 2008; Brouwers and others, 2010; Caton and others, 2008; van der Zanden and others, 2012; van Rooij and others, 2013). An earlier NBDPS study that did not include a stratification by phenotype found that late exposure to antihypertensive medication was related to hypospadias and suggested that these cases could represent mothers with hypertension caused by severe placental insufficiency (Caton and others, 2008). In our analysis, which included approximately 3000 additional controls and over 1000 additional hypospadias cases, both phenotypes were associated with late use of antihypertensive medication and with untreated hypertension. The fact that treatment started late in pregnancy might indicate that these mothers received treatment for gestational hypertension or (pre)eclampsia, which have been linked to placental insufficiency (Mayhew and others, 2004). The aOR for late treatment was higher for third than for second degree hypospadias, which is a new finding, supporting a more prominent role for placental insufficiency in the most severe phenotype.

Our findings that primiparity and multiple gestations were associated with hypospadias in both phenotypes, with the highest odds ratios observed for third degree hypospadias, confirms earlier findings (van Rooij and others, 2013). We also found some evidence for associations between fertility treatments and hypospadias. The largest odds ratio was seen for multiples with third degree hypospadias born after use of clomiphene citrate. These results confirm the findings by Meijer et al (2006), but differ from earlier NBDPS findings based on a smaller amount of data (Reefhuis and others, 2011). It was not possible to evaluate whether these findings were due to the treatment itself or to the underlying subfertility, since women were not asked about time-to-pregnancy or a possible subfertility diagnosis.

Because consistent case finding and classification is difficult for first degree hypospadias, this phenotype is not included in the NBDPS. Patients with a subglanular and subcoronal meatus were classified as second degree hypospadias, whereas some other studies classify these phenotypes as first degree hypospadias, which limits comparability. Although the risk factors for first and second degree hypospadias were similar in a previous study (van Rooij and others, 2013), the results may not be generalizable to all phenotypes. In addition, rates of hypospadias varied considerably between study sites, indicating possible differences in case finding. We tried to adjust for this by including a random effect for study center in our multivariable model. Sub-analyses excluding centers with the lowest hypospadias prevalence rates showed no appreciable change in effect estimates. There is a possibility of some remaining ascertainment bias by race or fertility treatment, but we expect this to be minimal, and at least partially addressed by the random effect. Mothers were provided with pregnancy calendars before the interview to improve recall of timing of exposures, but due to the self-reported nature of our exposure data, recall bias cannot be ruled out. Due to the relatively large number of exposures studied, we cannot exclude the possibility that some of our findings are due to chance.

The large, population-based, multistate design of the NBDPS is one of the key strengths of this study. Strict case ascertainment and classification protocols ensured consistent classification of subtypes. The large amount of exposure data allowed us to study rare exposures and to adjust for many possible confounders in one multivariable model, while keeping sufficient statistical power. Information on timing of exposure was available, which created the opportunity to study exposures in early pregnancy. However, we could not assess the effects of exposure on the last stages of urethral closure which take place in the fourth month of pregnancy, as most of the data were collected by trimester after the third month of pregnancy.

The fact that most risk factors identified were associated with both phenotypes indicates that the embryological process is at least partly similar. Many effect estimates, however, were larger for third degree hypospadias than for second degree hypospadias, while our findings regarding race/ethnicity and fertility treatment also showed some differences between the two phenotypes. Since mild hypospadias is more common than severe forms, studies without stratification by phenotype may over- or underestimate the effects of exposures or the role of genetics in the more severe subtypes. Identifying etiological differences between hypospadias phenotypes could bring us closer to understanding the disease mechanisms and developing preventive strategies.