Data on the prognosis of twins from tertiary institutions may not repr
esent the general incidence of adverse infant outcomes. We retrospecti
vely investigated the infant outcome in 32 twin gestations referred to
us at 29.9 +/- 3.7 weeks of gestation (range, 21 to 38 weeks) (late r
eferral group) and 237 twin gestations that had been monitored by us s
ince < 20 weeks' gestation (control group). The physical and neurologi
cal status of infants was assessed at 1 year of corrected age. Deliver
y occurred 3.1 weeks earlier in the late referral group than in the co
ntrol group (32.4 +/- 3.9 vs 35.5 +/- 2.4 weeks, p < 0.001). Deaths by
1 year of age and disabilities such as cerebral palsy, mental retarda
tion, and epilepsy occurred in 16 (25%) of 64 infants in the late refe
rral group compared with 24 (5.1%) of 474 infants in the control group
(p < 0.001). Thus, the inclusion of data on women who were referred l
ate increased the incidence of adverse infant outcomes in our tertiary
hospital from 5.1% to 7.4% (40/538). It was not known whether an earl
y referral to a tertiary hospital would have improved the outcome in t
he late referral group. These findings suggest that data from tertiary
institutions on the prognosis of twins may be affected adversely by t
he inclusion of data on women who are referred because of complication
s.