Selection criteria	1994 Birth cohort (ascertained as having ASD in 2002 ADDM)			2000 Birth cohort (ascertained as having ASD in 2008 ADDM)
	No. of sites	No. of counties	No. of cases	No. of sites	No. of counties	No. of cases
Selection of final study population (population to which results can be generalized)
ADDM sites included in both 2002 and 2008 surveillance years	13	175	2529	13	98	3609
Limit to: counties included in both 2002 and 2008*	13	94	1825	13	94	3285
Limit to: children residing in same county at birth and age 8†	10	69	951	10	69	1880
Limit to: ADDM site that provided all necessary study variables	8	48	786	8	48	1551
Limit to: singletons	8	48	762	8	48	1489
Limit to: race-ethnicity NHW, NHB, Hispanic	8‡	48	747	8	48	1406
Sample exclusions
Children excluded if they lacked data on any study variable			726			1386
Children in matching stratum with <20 controls/case excluded§			703			1339

^*Not all ADDM sites included the same counties in their surveillance area in 2002 and 2008.

^†In addition to limiting to children identified as ASD cases in ADDM who were linked to a birth record indicating birth residence was within the ADDM catchment area, this restriction also pragmatically limited our sample to ADDM sites that included the maternal county of residence indicator in their submitted ADDM-natality data set; three sites did not.

^‡Final study population included NHW, NHB, and Hispanic children born as singletons in 48 counties within eight ADDM sites.

^§Twenty controls per case were selected using a matched design in which cases and controls were matched on five sociodemographic factors. Thus, for control selection, the study population was subdivided into numerous matching strata and some included a small number of births. The final sample selection criterion was birth within a study-matching stratum including a minimum of 20 potential controls.