Table 2.

Differences in HIV test acceptance according to treatment assignment

Variables	Percentage point difference (95% CI); P value
	1: treatment effects	2: treatments and risk	3: treatments and risk with interactions	4: subsample with completed questionnaire	5: subsample with completed questionnaire, with interactions
Treatment assignment:
Active choice	13.3 (9.8 to 16.7); <0.001	13.5 (10.0 to 16.9); <0.001	15.9 (10.6 to 21.1): <0.001	13.6 (9.8 to 17.3); <0.001	17.1 (11.3 to 22.8); <0001
Opt-out	27.9 (24.4 to 31.3); <0.001	27.9 (24.5 to 31.2); <0.001	31.2 (25.8 to 36.5); <0.001	27.3 (23.5 to 31.1); <0.001	31.6 (25.8 to 37.5); <0.001
Risk of infection:
Intermediate risk		6.4 (3.4 to 9.3); <0.001	8.5 (3.6 to 13.5); 0.001	8.7 (5.4 to 12.0); <0.001	12.0 (6.5 to 17.5); <0.001
High risk		8.3 (3.3 to 13.4); 0.0013	16.3 (7.5 to 25.1); <0.001	13.0 (7.2 to 18.9); <0.001	23.6 (13.1 to 34.0); <0.001
Interactions:
Active choice × intermediate risk			−3.1 (−10.1 to 4.0); 0.39		−4.9 (−12.7 to 2.9); 0.217
Active choice × high risk			−8.4 (−20.9 to 4.1); 0.19		−11.2 (−25.6 to 3.3); 0.131
Opt-out × intermediate risk			−3.5 (−10.5 to 3.5); 0.332		−4.8 (−12.5 to 2.9); 0.225
Opt-out × high risk			−15.5 (−27.8 to −3.1); 0.015		−20.3 (−34.6 to −6.0); 0.005
Refused questionnaire		−6.7 (−10.1 to −3.3); <0.001	−6.7 (−10.2 to −3.3); <0.0001
Constant*	38.0 (35.5 to 40.5)	35.2 (32.2 to 38.4)	33.4 (29.5 to 37.2)	33.8 (30.5 to 37.2)	31.2 (27.0 to 35.4)
No of observations	4800	4800	4800	3860	3860

Dependent variable=acceptance of HIV test. Each column shows percentage point difference in HIV test acceptance estimated from ordinary least squares regression. Omitted categories for defaults and risk groups are opt-in testing and low risk, respectively. Columns 4 and 5 repeat columns 2 and 3 but excluding those with missing questionnaire data on risks (for whom data was imputed in columns 2 and 3). Standard errors are clustered at day-zone level.

*Test acceptance percentage under base case: opt-in testing.