Interpreting the Odds Ratio in proc surveylogistic

[sstoline]

Dear All:

I need some help interpreting the odds ratio in proc surveylogistic for the following two independent variables.

Dependent Variable: Smk_3last ,  Did you smoke at all in last 3 months of pregnancy?   1 = no, 2 = yes

Independent Variable: Drk_3Before   ------ 1= no, 2 = yes

proc surveylogistic data = have;

strata STRATUM1;

weight WTANAL;

class Drk_3Before(param=ref ref = "1");

model Smk_3last (event='1') = Drk_3Before;

run;

Odds Ratio: Drk_3Before 2 vs 1  OR = 1.966

Independent Variable: PREV_LB ------ 1= no, 2 = yes

proc surveylogistic data = have;

strata STRATUM1;

weight WTANAL;

class PREV_LB(param=ref ref = "1");

model Smk_3last (event='1') = PREV_LB;

run;

Odds Ratio: PREV_LB 2 vs 1  OR = 0.563

Thank you very much in advance

Steven

[ballardw]

You should include the confidence limits on these as well. If the interval includes the value 1 there is no statistically significant difference and you pretty much stop there. Add /clodds to the model statement after the Prev_lb and Drk_3Before variables.