Objective: We tested the hypothesis that the frequency of growth discordanc
e among twins is not related to the uterine capacity for carrying twins.
Method: We counted and compared the frequencies of birth weight discordance
of more than 25% in an unlike-sexed twin cohort (n = 1244) and in a popula
tion-based twin cohort (n = 7570) across the deciles of the total twin birt
h weight (twin A + twin B) distribution. The birth order of the heavier twi
n was noted.
Results: Similar frequencies of discordant pairs were found in both cohorts
(11% and 12%, respectively; Mantel-Haenszel chi(2) test: P = .131, odds ra
tio (OR) 0.9, 99% confidence interval (CI) 0.67, 1.11; Woolf test for heter
ogeneity: two-tailed P = .472). In the discordant pairs, twin A was conside
rably more often the heavier twin in all birth weight deciles (unlike-sexed
cohort: P < 10(-8), OR 5.9, 99% CI 3.0, 11.7; population-based cohort: P <
10(-8), OR 3.1, 99% CI 2.3, 4.0), and in both cohorts (inter-cohort differ
ence: P = .109, OR 1.4, 99% CI 0.83, 2.32). Both cohorts showed a similar n
onlinear trend: given that X = decile order, discordance decreased as a fun
ction of 22.0 - 6.54 In[X] for the unlike-sexed twins cohort and 23.0 - 8.1
8 In[X] for the population-based cohort, with r values of 0.967.
Conclusion: The more favorable the uterine milieu for carrying twins, the s
maller the likelihood of discordant twin growth. Birth order of the heavier
twin appears to be an integral part of the discordance phenomenon. The sim
ilarity of the cohorts suggests that these conclusions are valid for both l
ike and unlike-sexed twins. (C) 1999 by The American College of Obstetricia
ns and Gynecologists.