HIV testing preferences among long distance truck drivers in Kenya: a discrete choice experiment

Figures & data

Table 1. Attributes and levels used in the design the discrete choice experiment.

		Levels
		Level 1 (Baseline*)	Level 2	Level 3	Level 4
Attributes	Type of test	Finger-prick blood test	Oral mouth-swab test	–	–
	Type of counselling	In-person counselling	Telephonic counselling	–	–
	Who administers the test	Nurse-administered	Self-administered	–	–
	Location	At a roadside clinic	At a clinic near home	At the company office	At home
	Time	90 min	20 min	40 min	3 h
	Cost	Free	You pay 250 Kenyan Shillings (approx. US$2.50)	You pay 300 Kenyan Shillings (approx. US$3.00)	We pay you 350 Kenyan Shillings (approx. US$3.50)

*Note: baseline characteristics shown in the Level 1 column are used as the reference category in the regression analysis.

Table 2. Interaction variables used to stratify the analysis of the discrete choice experiment data including sexual behaviour and HIV testing history.

Individual characteristic	Dummy variable created from baseline questionnaire data
Ever tested for HIV	Ever tested  =  1; never tested  =  0.
Regular testing for HIV*	Regular tester  =  1; non-regular tester  =  0.
Multiple concurrent sex partners	Has one or more sex partner other than wife or main partner at home  =  1; has no regular sex partner other than wife or main partner at home  =  0.
Engages in sex with FSW	Paid for sex with money, gifts or a ride in the past six months  =  1; did not engage in sex for money, gifts or a ride in the past 6 months  =  0.
Consistent condom use	Always used a condom when having sex in the past 6 months  =  1; sometimes or never used a condom when having sex in the past 6 months  =  0.
Experience of testing at NSA roadside clinic	Ever tested at an NSA roadside clinic  =  1; never tested at an NSA roadside clinic  =  0.

*Note: In this paper we define regular testing as having tested more than once, and within the past 6 months in line with international guidelines for testing among high-risk populations (CDC, 2016; Mitchell & Horvath, 2013).

Table 3. Key sample characteristics of 305 truck drivers recruited from 2 North Star Alliance clinics in Kenya, October to December 2015.

Individual Characteristic	n	%
Age
20–29 years	58	19
30–39 years	145	48
Age 40–49 years	73	24
> 49 years	28	9
HIV testing
Tested for HIV before	279	92
Regular tester	161	53
Ever tested at NSA roadside clinic	152	50
Came to clinic specifically for HIV testing	131	43
Risk factors
Paid for sex with money, gifts or a ride in the past 6 months	179	59
Has one or more regular partners other than wife or main partner at home	142	47
Self-reported always using condoms (male or female) when having sex in the past 6 months	43	14

Table 4. Main effects conditional logit model (showing odds ratios), 305 truck drivers.

Attribute	Level (reference group shown in brackets)	OR	95% Confidence Interval		p-Value
Type of counselling	Telephonic counselling (in-person counselling)	0.881	0.809	0.959	0.003
Who administers the test	Self-testing (nurse-administered)	0.985	0.906	1.071	0.726
Type of test	Oral test (finger-prick blood test)	0.977	0.898	1.062	0.583
Location	Test at home (roadside clinic)	0.864	0.746	1.000	0.051
	Test at a clinic near home (roadside clinic)	0.907	0.767	1.073	0.257
	Test at work office (roadside clinic)	0.826	0.714	0.955	0.010
Time	20 min long test (90 min long test)	1.172	1.012	1.357	0.034
	40 min long test (90 min long test)	0.960	0.812	1.136	0.637
	3 h long test (90 min long test)	0.776	0.671	0.898	0.001
Cost	Test costs US$2.50 (no test cost)	0.561	0.485	0.650	0.000
	Test costs US$3.00 (no test cost)	0.351	0.297	0.415	0.001
	Receive US$3.50 (no test cost)	0.870	0.755	1.004	0.056
Number of observations	4828
Log likelihood	−1558.694
Pseudo R squared	0.068
LR Chi squared (12)	229.1

Figure 1. Conditional logit models stratified by HIV testing history. Point estimates are presented as odds ratios, and confidence intervals are estimated using 95% as a threshold for significance. When confidence intervals do not overlap, this indicates that there is a significant difference between the preferences across two groups.

CDC. (2016). HIV testing. Retrieved June 25, 2016, from https://www.cdc.gov/hiv/testing/

 Google Scholar

Mitchell, J. W., & Horvath, K. J. (2013). Factors associated with regular HIV testing among a sample of US MSM with HIV-negative main partners. JAIDS Journal of Acquired Immune Deficiency Syndromes, 64(4), 417–423. doi: 10.1097/QAI.0b013e3182a6c8d9

 PubMed Web of Science ®Google Scholar