. 2016 Dec 14;3:162–171. doi: 10.1016/j.ssmph.2016.12.004

Table 2.

Longitudinal mixed-effect and structural marginal models for secondary years of education as predictors of adolescent fertility rate, mortality and HIV prevalence in 15 to 24 year olds.

		Adolescent fertility rate			Male mortality						Female mortality						HIV prevalence
		log births per 1000 women 15-19 years			log deaths per 100,000 pyo						log deaths per 100,000 pyo						% population aged 15-24 years
Mixed effect models
Country N			137		141						141						93
Country-years			1702		141						4456						2133
Country-years			1702		4456						4456						2133
					15-19 years			20-24 years			15-19 years			20-24 years			males			females
		B	ExpB	p	B	ExpB	p	B	ExpB	p	B	ExpB	p	B	ExpB	p	B	ExpB	p	B	ExpB	p
Constant		3.375			5.429			5.931			5.651			6.374			-3.240			-2.928
Time (years)	linear	−0.014	0.986	0.02	0.000	1.000	0.9	0.002	1.002	0.5	−0.011	0.989	<0.001	−0.001	0.999	0.8	0.092	1.097	<0.001	0.006	1.006	0.5
	quadratic																−0.004	0.996	<0.001	−0.003	0.997	<0.001
Education
Primary education per capita (years)		0.027	1.027	0.3	−0.011	0.989	0.4	0.004	1.004	0.9	0.003	1.003	0.9	0.009	1.009	0.7	0.024	1.024	0.8	0.335	1.398	0.08
Secondary education per capita (years)		−0.088	0.916	0.002	−0.026	0.974	0.03	−0.039	0.962	0.003	−0.047	0.954	0.001	−0.050	0.951	0.02	−0.281	0.755	<0.001	−0.563	0.569	<0.001
GDP (log $ per annum)		0.184	1.202	0.006	−0.033	0.968	0.04	−0.053	0.948	0.05	−0.088	0.916	<0.001	−0.141	0.869	<0.001	0.253	1.288	0.07	0.282	1.326	0.05
Country status	Low income	0			0			0			0			0			–			–
	Lower middle	−0.539	0.583	0.003	−0.255	0.775	0.001	−0.192	0.825	0.08	−0.593	0.553	<0.001	−0.650	0.522	<0.001	–			–
	Upper middle	−1.194	0.303	<0.001	−0.139	0.871	0.3	−0.037	0.963	0.9	−0.702	0.495	<0.001	−0.773	0.462	<0.001	–			–
	High income	−2.203	0.110	<0.001	−0.255	0.621	<0.001	−0.415	0.661	0.007	−1.138	0.321	<0.001	−1.240	0.289	<0.001	–			–
Interaction of income with time		−0.005		0.04	−0.005		<0.001	−0.005		<0.001			0.2	−0.003		0.06	–			–
Inverse probability weighted (IPW) models
		B	ExpB	p	B	ExpB	p	B	ExpB	p	B	ExpB	p	B	ExpB	p
Secondary education		−0.158	0.854	<0.001	−0.121	0.886	<0.001	−0.092	0.912	<0.001	−0.185	0.831	<0.001	−0.160	0.852	<0.001

Table notes

IPW models: models weighted for time-varying variables (log of lagged GDP, log of country population in each year, proportion not in education by country per year) and country income status, separately by sex.

IPW models were not estimated for HIV due to inclusion of a significant quadratic term in the models.

Income status not significant in mixed effects HIV models in either sex thus not included.

Pyo: person years of observation.

B: coefficient.

ExpB: exponentiated coefficient.