Logistic regression with Effect Modification or Mantel-Haenszel

[robert seru]

Hello.
I'm trying to find an association between Sex of child and Exposure to or having a Defect from babies delivered in different sites
Sex 0 "Male" 1 "Female"
Defect 1 "Yes" 0 "No"
Site 1 "Site1" 2 "Site2" 3 "site3" 4 "Site4"
Table of Sex and defect by site

Site and Defect
	Site1		Site2		Site3		Site4
Sex	No	Yes	No	Yes	No	Yes	No	Yes
Male	9	1	9	0	20	0	259	4
Female	2	0	5	0	11	0	112	13
Unclear	0	0	0	0	0	0	8	1

What would be the best measure of association taking site as an effect modifier, or should I use the Mantel-Haenszel odds ratio estimate.

xi: logistic Sex Anencephaly_1##i.site, or
vs
cc Sex Anencephaly_1, by(site)

And how would one interpret the results if Males are taken as the reference?

Thanks
Robert