Table 3.

Crude and adjusted analyses of effects of independent variables on prevalence of female adolescents’ early sexual initiation. Pelotas, Southern Brazil, 1982 to 2004-5

Variable	Crude analysis			Adjusted analysis^*
Variable	PR	95% CI	p	PR	95% CI	p
Skin color			1.00^**			0.50^**
White	1	-		1	-
Black or mixed	1.00	0.62;1.62		0.84	0.51;1.40
Family income - 1982 (MMW)			0.001^***			0.003^***
≤1.0	4.25	1.03;17.58		4.19	0.98;18.00
1.1 to 3.0	3.25	0.80;13.16		3.19	0.77;13.18
3.1 to 6.0	2.20	0.51;9.54		2.19	0.50;9.61
6.1 to 10.0	1.48	0.25;8.68		1.44	0.25;8.25
>10.0	1	-		1	-
Change in income (1982 → 2004-5)			<0.001^***			<0.001^***
Always poor	2.70	1.64;4.44		2.71	1.61;4.57
Not poor → poor	2.32	1.38;3.89		2.28	1.36;3.83
Poor → not poor	2.17	1.24;3.80		2.31	1.28;4.14
Never poor	1	-		1	-
Adolescent’s level of education (years)			<0.001^***			<0.001^***
0-4	10.99	5.15;23.43		11.24	4.48;28.23
5-8	3.51	1.65;7.49		3.44	1.36;8.70
9-11	1.44	0.67;3.11		1.34	0.57;3.17
≥12	1	-		1	-

MMW: Monthly minimum wage

Variables on the first level (ethnicity and family income in 1982) were adjusted among themselves and remained in the analysis model, if p<0.2. Change in income was adjusted for ethnicity; adolescent’s level of education was adjusted for ethnicity and family income.

^**

Wald test for heterogeneity.

^***

Wald test for linear trend.