An independent samples t-test was used to test the difference in age at menarche between study participants who developed T1DM before menarche and those who developed T1DM after menarche. Due to the complex sampling strategy used for NHANES 2001-2006, sample weights were used in PROC SURVEYMEANS (SAS version 9.2, SAS Institute, Cary, NC) to obtain sample estimates and confidence intervals. A one sample Z-test was used to compare the study sample to the NHANES data. An independent samples t-test was used to test the difference in age of menarche between study participants who had 21-hydroxylase autoantibodies compared to those without, and to compare age of menarche among study participants with a history of diabetic ketoacidosis (DKA) or severe hypoglycemia compared to those without. A Chi-square test was used to compare a history of menstrual irregularity between study participants with and without a history of DKA, and between adolescent females diagnosed with T1DM pre- vs. post-menarche.
Linear regression analysis was used to examine relationships of age at menarche with clinical factors, including body mass index (BMI) and A1c at the time of menarche, and to examine age at menarche between study participants who developed T1DM before menarche vs. study participants who developed T1DM after menarche, adjusting for age at visit.
Average age of menarche in adolescents with T1DM
Characteristics of study participants with T1DM are shown in Table , by diagnosis of T1DM before vs. after menarche. There was no difference in age at visit, A1c, BMI, BMI z-score, prevalence of overweight, race/ethnicity, Tanner Stage, or menstrual irregularity between these groups. Adolescent females diagnosed pre-menarche had a longer duration of diabetes and a higher daily insulin dose per kg than the group diagnosed with T1DM after menarche.
The overall mean age of menarche among females with T1DM was 12.69 ± 0.08 years. Adolescent females who developed diabetes before menarche had an average age of menarche of 12.81 ± 0.09 years compared to 12.17 ± 0.19 years, p = 0.0015 (Figure ) in those who developed diabetes after menarche. In multivariable linear regression analysis further adjusting for age at visit, the average age of menarche among adolescent females diagnosed with T1DM before menarche remained significantly older than the age of menarche among adolescent females diagnosed with T1DM after menarche (least square means 12.82 ± 0.09 years vs. 12.13 ± 0.18 years, p = 0.0006).
T1DM before and after menarche versus NHANES. T1DM before menarche versus NHANES without T1DM, P < 0.0001. T1DM after menarche versus NHANES without T1DM, P = 0.7676.
Comparison of T1DM to NHANES 2001-2006
NHANES 2001-2006 data from female participants interviewed using the Reproductive Health Questionnaire and who were between 12 and 24 years of age (mean ± SE age = 18.2 ± 0.09) showed that the average age of menarche among adolescent females without diabetes (n = 3667) was 12.27 ± 0.038 years, which was significantly earlier than adolescent females with T1DM diagnosed prior to menarche (12.81 ± 0.09, p < 0.0001), and similar to adolescent females diagnosed after menarche (12.17 ± 0.19 years, p = 0.6993) (Figure ). As the mean age of the NHANES cohort was older than the T1DM cohorts, we performed a linear regression analysis adjusting for current age, and found that the least square mean age of menarche among NHANES study participants at a current age of 15.8 was 12.27 ± 0.03, which remained significantly younger than the T1DM group diagnosed prior to menarche (12.81 ± 0.09, p < 0.0001), while the least square mean age of menarche among NHANES study participants at a current age of 16.5 was 12.28 ± 0.03, which was not significantly different from the T1DM cohort diagnosed after menarche (12.17 ± 0.19, p = 0.6922). A further analysis was conducted using only adolescent females who were at least 16 years of age (e.g. without primary amenorrhea), to ensure that differences in ascertainment of younger adolescents didn't explain the observed differences in age of menarche. The average age of menarche among adolescent females with T1DM diagnosed before menarche and age 16-24 at the visit (n = 101) was 13.0 ± 0.13 years, which was significantly older compared to the average age of menarche among female NHANES participants (n = 2272) of 12.5 ± 0.05 years (p = 0.001) and adolescent females with T1DM diagnosed after menarche (n = 30) of 12.32 ± 0.22 (p = 0.007).
Age at menarche and clinical factors
A1c at menarche was available for 165 of 185 subjects with T1DM prior to menarche. A1c was not associated with age at menarche, although mean A1c (9.3%) was well above the recommended target for optimal diabetes control. Out of the 165 girls, 30 (18%) had an A1c at or below the target (A1c ≤ 7.5%) at the time of menarche. However, there was no difference in age of menarche between those who achieved the target A1c (12.61 ± 0.21) and those who had not (12.91 ± 0.10, p = 0.20). When those with menarche within one year of diagnosis with T1DM were eliminated from analysis in order to avoid bias due to the 'honeymoon' period, 21 (15%), of the remaining 136 adolescent females had A1c ≤ 7.5%. Although those adolescent females who reached the target A1c in this subset tended to have an earlier age of menarche (12.63 ± 0.21) than those who did not (12.99 ± 0.10), the difference did not reach statistical significance (p = 0.12).
There was a significant negative correlation between age of menarche and BMI z-score at menarche (r = -0.23, p = 0.0029). In multivariable linear regression adjusted for age at T1DM onset and A1c at menarche, BMI z-score remained significantly associated with age of menarche (β = -0.32, p = 0.005), as was age at T1DM onset (β = 0.06, p = 0.02), but not A1c (β = 0.08, p = 0.15)
Age at menarche by race/ethnicity for the BDC sample and NHANES data are shown in Figure . In the BDC sample, females of Hispanic origin (n = 34) reported earlier menarche than non-Hispanic white (NHW) (n = 185) females (12.25 ± 0.20 vs. 12.77 ± 0.09, p = 0.0232). In multivariable linear regression analysis adjusting for BMI z-score and A1c at menarche, this difference in age of menarche persisted for Hispanic vs. NHW adolescent females (least square means 12.33 ± 0.24 vs. 12.94 ± 0.09, p = 0.019)
Age of menarche in NHW and Hispanic patients. Age of menarche in NHW BDC patients versus NHW NHANES sample (P < 0.0001) and Hispanic BDC patients versus Hispanic NHANES sample (P = 0.0653). NHW BDC versus Hispanic BDC (P = 0.0232).
In the NHANES sample, NHW (n = 1114) females without diabetes had an average age at menarche of 12.42 ± 0.05 years, which was significantly younger than the age of menarche in NHW girls in the BDC sample (p < 0.0001). In the NHANES sample, those of Hispanic origin (n = 1290) without diabetes had an average age of menarche of 11.95 ± 0.09 years, which was not significantly different from the age at menarche among Hispanic adolescent females in the BDC sample (p = 0.0653).
In the BDC sample, NHW girls with premenarchal presentation of T1DM (n = 151) had a later age of menarche than those who developed T1DM after menarche (n = 34) (12.92 ± 0.09 vs 12.32 ± 0.18, p = 0.0062). Subjects in the BDC sample who were of Hispanic origin and who developed T1DM before menarche (n = 25) also had later menarche than those who developed T1DM after menarche (n = 7) (12.25 ± 0.23 vs. 11.91 ± 0.64, p = 0.5461), although this did not reach significance. NHW girls who had a premenarchal presentation of T1DM (n = 151) reported a later age of menarche than Hispanic girls who developed T1DM prior to menarche (n = 25) (12.92 ± 0.09 vs 12.25 ± 0.23, p = 0.0083).
21 Hydroxylase autoantibodies
21-Hydroxylase (OH) autoantibody titers were obtained through retrospective review of the chart. Results were available on 212 subjects and of those only 8 subjects had positive titers. The age of menarche in those negative for 21-OH antibodies (12.69 ± 1.20 years, p = 0.4405) was not different from those with positive 21-OH antibodies (13.02 ± 0.85 years, p = 0.4405). Since only 8 subjects had positive 21 OH antibodies the sample size did not allow for any further analysis.
Severe hypoglycemia and/or episodes of DKA
There was no difference in age at menarche among girls with a history of at least one episode of DKA (n = 20) compared to girls without any DKA episodes (12.9 ± 1.1 vs. 12.8 ± 1.2 years of age, p = 0.78) or among girls with at least one episode of severe hypoglycemia (n = 11) compared to girls without any hypoglycemic episodes (12.7 ± 0.9 vs. 12.8 ± 1.2 years of age, p = 0.72) occurring within the two years prior to the start of menarche. There was a trend towards more irregular menses among the adolescent females who had one or more episodes DKA before menarche compared to those with no episode of DKA (53% vs. 32%, p = 0.08)
Analysis of menstrual regularity was limited to adolescent females who were at least 2 years past menarche. Among the 181 adolescent females analyzed, 63 reported usually or always irregular cycles (35%) and the remaining 118 reported regular or very regular cycles. There was no difference in age of menarche, age of T1DM diagnosis, total physical activity, typical physical activity per/day, total insulin units kg/day, A1c, or BMI at the most recent visit between those who reported regular (always or usually) compared to irregular (always or usually) cycles. Total physical activity (OR = 0.973 [0.849-1.116], p = 0.6949) and typical physical activity per/day (OR = 0.971 [0.845-1.116], p = 0.6801) were not associated with irregular periods.