Revealing a safer sex option to reduce HIV risk: a cluster-randomized trial in Botswana

BMC Public Health

Table 2 Effect on Sexual Behavior - Comparing Intervention versus Control

	All	Rural	Junior	Primary
Pregnancy
aRRR (95% CI)	0.657 (0.433 - 0.997)	0.518 (0.323 - 0.831)	0.575 (0.394 - 0.841)	0.557 (0.0539 - 5.748)
p-value	0.049	0.006	0.004	0.623
n	15,335	11,483	9902	5433
More than 1 Sexual Partner
aRRR (95% CI)	0.953 (0.808 - 1.125)	0.916 (0.760 - 1.104)	0.793 (0.643 - 0.979)	0.953 (0.715 - 1.271)
p-value	0.568	0.358	0.031	0.745
n	14,130	10,595	9151	4979
Ever Had Sex
aRRR (95% CI)	0.885 (0.757 - 1.034)	0.856 (0.714 - 1.028)	0.786 (0.641 - 0.963)	0.996 (0.716 - 1.386)
p-value	0.123	0.096	0.020	0.982
n	14,113	10,587	9131	4982
Ever Tried to Use a Condom
aRRR (95% CI)	1.006 (0.958 - 1.057)	0.984 (0.935 - 1.037)	1.050 (0.990 - 1.114)	0.968 (0.893 - 1.049)
p-value	0.801	0.553	0.106	0.428
n	13,944	10,476	8993	4951

^aAdjusted Relative Risk Ratios (aRRR) are calculated clustering standard errors by school and includes region and school level fixed effects to account for stratification. Adjustments were generated using a vector of school-level baseline control variables to enhance statistical precision. This vector includes: enrollment, dropout, class-reported meaures of pregnancy and class-reported rates of girls dating older partners. Due to limited pregnancy in primary schools, we include a specification without controls or region-fixed effects to enable inclusion of the full sample. The results are robust in effect size and significance to specification

ISSN: 1471-2458