Table 1

Descriptions of the different sex hormone, asthma and confounder variables included in the observational analyses, with percentage missingness in the original variables out of a total subsample size of 512 males in the ALSPAC cohort

Age at measurementnMean (SD)Missing (%)*Imputation method
SHBG
 9.944192.33 (43.67)13.9PMM
 11.848274.96 (37.47)5.9PMM
 13.841645.41 (26.03)18.8PMM
 15.546330.82 (13.33)9.6PMM
 17.842726.26 (12.39)16.6PMM
Total testosterone
 9.94400.82 (0.09)14.1PMM
 11.84811.54 (1.23)6.1PMM
 13.84149.05 (4.66)19.1PMM
 15.546314.8 (2.72)9.6PMM
 17.841216.48 (2.62)19.5PMM
Bioavailable testosterone
 9.94410.17 (0.07)14.1PMM
 11.84820.42 (0.47)6.1PMM
 13.84143.84 (2.57)19.1PMM
 15.54647.33 (1.97)9.6PMM
 17.84139.01 (1.97)19.5PMM
Asthma
 10.74678.8Logit
 13.14767.0Logit
 13.845511.1Logit
 16.54737.6Logit
 22.925051.2Logit
Maternal confounders
 Maternal education4776.8Mlogit
 Maternal smoking4816.1Mlogit
 Parity4747.4Logit
 Maternal age492363.52 (52.52)3.9Regress
 Gestational age48839.42 (1.86)4.7Regress
  • *As a percentage of total size of male subsample that had at least two hormone measures at two time points (n=512). Missingness estimate does not take into account non-overlap between asthma responses and hormone measurements. For example, although there are n=440 with testosterone at 9.9 years and n=467 with asthma data at 10.7, individuals with data do not fully overlap.

  • ALSPAC, Avon Longitudinal Study of Parents and Children; Logit, logistic regression; Mlogit, multinomial logistic regression; PMM, predictive mean matching; Regress, linear regression; SHBG, sex hormone-binding globulin.