Table 3.

Sexual behaviour change following VCT, from generalized linear regression models predicting each behavioural outcome

	HIV positive
	Males			Females
	Exp(Coef.)	SE	95% CI	Exp(Coef.)	SE	95% CI
Visited bar/beer hall in the last month	1.26	0.09**	(1.10–1.44)	0.38	0.16*	(0.17–0.85)
New sexual partner in past yeara	0.75	0.16	(0.50–1.12)	0.38	0.13**	(0.20–0.73)
Sexual partner in past monthb	1.19	0.33	(0.69–2.05)	0.92	0.19	(0.61–1.37)
Multiple concurrent partnersb	0.72	0.32	(0.31–1.70)	0.46	0.58	(0.04–5.35)
Consistent condom use in last 2 weeks
With regular partnersb	1.65	0.74	(0.68–3.98)	1.89	0.76	(0.86–4.16)
With non-regular partnersb	0.65	0.55	(0.12–3.45)	2.07	2.69	(0.16–26.42)

	HIV negative
	Males			Females
	Exp(Coef.)	SE	95% CI	Exp(Coef.)	SE	95% CI
Visited bar/beer hall in the last month	1.02	0.04	(0.94–1.11)	1.66	0.39*	(1.05–2.64)
New sexual partner in past yeara	0.87	0.08	(0.72–1.04)	0.69	0.12*	(0.48–0.97)
Sexual partner in past monthb	0.89	0.12	(0.68–1.17)	0.91	0.10	(0.73–1.12)
Multiple concurrent partnersb	1.11	0.25	(0.71–1.73)	0.98	0.66	(0.23–4.26)
Consistent condom use in last 2 weeks
With regular partnersb	0.96	0.38	(0.45–2.08)	1.39	0.38	(0.81–2.39)
With non-regular partnersb	1.08	0.41	(0.51–2.28)	0.01	0.02	(0.00–2.80)

Exponentiated coefficients describe the change in the outcome variable associated with receiving VCT, stratified by sex and HIV status

Results are adjusted for age, calendar year and marital status, see “Methods” for details

Exp(Coef.): Exponentiated coefficient. A value of 1 indicates no change

SE: Standard error

* P < 0.05, ** P < 0.01

^aThose who had tested within the year before the survey were categorized as not tested when defining the indicator for having received VCT

^bModelled as a binary variable