Skip to main content
. Author manuscript; available in PMC: 2015 Sep 8.
Published in final edited form as: Ann Epidemiol. 2014 Jan 15;24(4):260–266. doi: 10.1016/j.annepidem.2013.12.014

Table 4. Average and summary ASD PAFs for PTB, SGA, and CD.

Risk factor(s) 1994 Birth cohort 2000 Birth cohort Change in PAF (95% CL*)


No. of cases Average PAF, % (95% CL*) No. of cases Average PAFs, % (95% CL*)
PTB 703 4.2 (0.0 to 8.7) 1339 2.0 (0.0 to 4.0)
SGA 703 0.9 (0.0 to 3.5) 1339 3.1 (1.0 to 5.0)
CD 703 7.9 (6.2 to 9.5) 1339 6.7 (3.0 to 10.0)
PTB or SGA or CD (summary measure) 703 13.0 (7.7 to 19.5) 1339 11.8 (7.5 to 15.9) 1.2 (−9.3 to 5.4)
Subgroup analyses: summary PAFs
 ASD with ID 234 12.3 (1.8 to 23.0) 377 17.1 (8.5 to 25.9) 4.8 −9.1 to 18.6)
 ASD without ID 305 9.6 (0.8 to 18.6) 680 12.2 (6.2 to 18.5) 2.6 −8.3 to 13.1)
 ASD, ID Unknown 164 20.5 (9.0 to 31.7) 282 3.4 (0.0 to 12.7) −17.1 (−31.3 to −2.3)
 Males 580 13.6 (6.9 to 20.2) 1116 10.8 (6.2 to 15.4) −2.8 (−11.2 to 4.8)
 Females 123 10.2 (0.0 to 24.8) 223 16.3 (6.1 to 26.8) 6.1 (−12.0 to 24.3)
*

Bootstrap CLs based on 1000 replications. A 95% CL was estimated as the 2.5 and 97.5 percentiles of the 1000 replica estimates; each resampled data set comprised case-control sets from the original study sample randomly selected with replacement.

For average PAF estimates, lower bound of CL is reported as 0.0 for all instances in which lower bound estimate was <0.0.

Bootstrap CL could not be directly estimated due to small sample sizes. A hybrid method was used in which information gleaned from the 2000 birth cohort variance estimation was applied to the 1994 cohort data.