A meta-analysis was performed of the research published from 1985 to 1998 e
xamining the effect of in utero exposure to cocaine on infant neuobehaviora
l outcome. The initial search for articles to include in the meta-analysis
identified 18 studies with potentially meta-analytic variables. Of the stud
ies originally retrieved, 13 failed to meet all of the inclusion criteria a
nd were excluded from the meta-analysis. A total of 14 meta-analyses were p
erformed comparing cocaine-exposed infants to nonexposed infants on NBAS cl
uster scales at birth and at 3-4 weeks of age. While the meta-analytic comb
ination of studies produced a large enough sample size to drive statistical
significance in a small majority of the tests of difference between the co
caine-exposed and nonexposed infants both at birth and soon after, the magn
itude of all effects was small. The largest reliable differences appeared f
or the motor performance and abnormal reflexes clusters. Both also demonstr
ated a slight trend for increasing standard differences from birth to measu
res obtained at 3-4 weeks. The orientation and autonomic regulation cluster
s produced small, significant effects at both time periods, but the trend w
as for reduced effect sizes over time. All other effects appear truly negli
gible. (C) 1999 Elsevier Science Inc. All rights reserved.