Table 3.

Sex-stratified GEE models examining potential interactions between survey and ethnicity for ATP-III MetS and MetS severity Z-score

	ATP-III MetS		MetS severity Z-score
	OR (95% CI)	P value	β (95% CI)	P value
Women
Survey
SAMINOR 2 vs SAMINOR 1	1.08 (0.99 to 1.18)	0.095	0.02 (0.01 to 0.04)	0.010
Ethnicity
Sami vs non-Sami	1.16 (1.03 to 1.30)	0.011	0.02 (−0.01 to 0.04)	0.14
Survey × ethnicity
SAMINOR 2 × Sami	—		0.03 (0.00 to 0.05)	0.024
Age (per 10 years)	1.37 (1.30 to 1.45)	<0.001	0.09 (0.08 to 0.10)	<0.001
Men
Survey
SAMINOR 2 vs SAMINOR 1	1.43 (1.29 to 1.58)	<0.001	0.06 (0.01 to 0.10)	0.021
Ethnicity
Sami vs non-Sami	1.00 (0.89 to 1.13)	0.95	−0.02 (−0.07 to 0.04)	0.62
Survey × ethnicity
SAMINOR 2 × Sami	—		0.14 (0.07 to 0.21)	<0.001
Age (per 10 years)	1.06 (1.00 to 1.12)	0.034	−0.04 (−0.06 to 0.02)	0.001

We tested whether the change in ATP-III MetS and MetS severity Z-score differed by ethnicity, by using interaction terms (ethnicity × survey) in GEE logistic or linear models that included age, survey and ethnicity as covariates. Analyses were not stratified by ethnicity. The interaction term was excluded from a model if p≥0.05. In women, the MetS severity Z-score was log-transformed. When interpreting the coefficients for survey and ethnicity in the models for MetS severity Z-score, one should be aware that these must be interpreted together with the interaction term.

ATP-III, Adult Treatment Panel III; GEE, generalised estimating equation; MetS, metabolic syndrome.