South Africa has the world’s largest HIV and AIDS epidemic (UNAIDS, 2014). An estimated 90% of HIV-positive people are in the economically active age group (SABCOHA, 2015; Setswe, 2009). Yet, recent studies show that in 2012 only 44.8% of the estimated 6.4 million HIV-positive people living in South Africa knew their HIV status (Shisana, 2013); therefore, many may not seek timely treatment. To address the epidemic, the South African National Strategic Plan on HIV, STIs and TB, 2012–2016 (SANAC, 2012), and the Code of Good Practice on HIV and AIDS and the World of Work (Department of Labour, 2012) foresee that workplaces should address HIV, which would include the promotion of HCT that is considered to be one of the most effective and efficient instruments in HIV prevention and access to care in Africa (Corbett et al., 2006). HIV management in the workplace is becoming a priority for companies operating in countries with a high HIV prevalence (Arimoto et al., 2013; George & Quinlan, 2009) as HIV and AIDS are eroding corporate investments and companies are losing trained staff because of AIDS (Arimoto et al., 2013; George, Surgey & Gow, 2014). In the context of wide access to antiretroviral treatment (ART), HIV-positive workers unaware of their status may not receive timely support and treatment that will prevent ill-health and an untimely death, detrimental to both their families and companies. Consequently, many companies in South Africa have introduced HIV and AIDS WPP (Arimoto et al., 2013; Ellis & Terwin, 2005). These WPP interventions cover, for example, HIV and AIDS policies, HIV prevention, treatment and care education, provision of condoms, workplace HCT, ART or medical referral for treatment of employees (Mahajan et al., 2007). To ensure quality and standardisation of such WPP, the South African Bureau of Standards developed the SANS 16001:2013, a national standard for a wellness and disease management system including HIV and TB (Gansan, 2013). One of the key issues of these HIV and AIDS WPP is the focus on determining a reliable HIV prevalence for the company’s staff through workplace HCT. High workplace HCT participation allows defining a reliable company HIV prevalence rate, which in turn is a prerequisite for efficient and effective HIV and AIDS management (UNAIDS, 2002). This has also been acknowledged by the SANS 16001:2013 wellness and disease management system that aims to provide organisations with solid data and statistics on the health determinants and burden of disease within their workplaces necessary for proper strategic human resource planning (Gansan, 2013). Unfortunately, relatively low HIV testing uptake rates are reported by South Africa’s companies (Connelly & Rosen, 2006; George, 2006; George & Quinlan, 2009; SABCOHA, 2015). Offering incentives might be a useful tool to encourage workplace HCT (George & Quinlan, 2009; Lee et al., 2014; Mahajan et al., 2007; Weihs & Meyer-Weitz, 2014).

From a cultural perspective, competitions such as lotteries are extremely popular among South Africans and are especially present in people’s daily life (Van Wyk, 2012). In her study, Van Wyk (2012) found that although almost 70% of her study’s interviewees reported never winning anything in the lottery, 80% of them kept on ‘taking a chance’.

Lottery competitions have already been used for mass HIV testing in South Africa. For instance, the South African ‘Right to Know’ HIV testing competition tested more than 50 000 people in 2009 (Keeton, 2009). Supporters of the use of lotteries for encouraging HIV testing claim that lotteries help people to make the ‘right choice’ and are not coercive. Neville Koopowitz, CEO of Discovery Health that supported the ‘Right to Know’ competition, said that his organisation wanted to apply its ‘model of giving people incentives to participate in healthy behaviours … to our country’s most pressing and important health issue’ (Keeton, 2009, p. 2). Sunday Times editor Mondli Makhanya described how lotteries made it exciting to test: ‘By offering an incentive, we believe that the Right to Know campaign has contributed to making it hip to test’ (Keeton, 2009, p. 2). New Start’s technical adviser, Scott Billy, said:

I think it’s a great idea. If you test and win the prize, you’ve won twice. You’ve won some money and you’ve won the opportunity to take control of your health, your life and your family’s future by learning your HIV status. And if you don’t win the prize, you’ve still won because you got yourself tested. (Keeton, 2009, p. 2)

There are many examples in South Africa where lotteries are claimed to have increased HIV testing rates and thus promoted efficiency: a little increase in expenditures for lotteries nevertheless created a large health benefit. In November 2011, the Western Cape Government implemented a pilot lottery to encourage people in communities to get tested for HIV. Preliminary indications show that the incentive scheme significantly increased the number of people tested (Mbhele, 2011). When the National Department of Health launched its HIV testing campaign ‘First Things First’ among tertiary institutions in the country, prizes such as a car, laptop computer and book vouchers were offered in the lottery to attract students to test. As an incentive to participate, first-year students could enter a lucky draw at each campus. Approximately 97% of students counselled and tested for HIV (Cameron & Van der Merwe, 2012). The ‘First Things First’ campaign was overall described as a great success and was awarded by the African Network for Strategic Communication in December 2011. The non-governmental organisation Bulungula Incubator managed to triple the HIV testing uptake in four Eastern Cape villages by using lotteries to encourage villagers to check their status (Malan, 2013). In various discussions held with senior staff of the automotive industry in the Nelson Mandela Bay, South Africa, one of the authors learned that they were convinced that lotteries increased workplace HCT rates, that staff highly appreciated lotteries and that it created a positive atmosphere around workplace HCT. In an interview, Sean Jelly, CEO of a South African company specialising in providing HIV and AIDS medical support to corporate employers, reported that lotteries greatly increased workplace HCT uptake among their clients (George & Quinlan, 2009).

While Rebecca Hodes, Director of Policy for the Treatment Action Campaign welcomed initiatives to promote HIV testing, some of the Treatment Action Campaign leaders and members were critical of lotteries and said that ‘it is a great enough incentive to know your status and to access anti-retrovirals if you need them’ (Keeton, 2009, p. 2). Lotteries are described to be coercive as they put undue pressure on people to participate in HIV testing (Cameron & Van der Merwe, 2012), especially on a person in desperate need for extra financial income. The Office of the ANC Chief Whip argued with regard to the 2011 Western Cape’s Government lottery that the HIV and AIDS testing lottery down played the gravity of HIV and AIDS and that it generated the impression that HIV and AIDS concern only the poor, who are easily tempted by prizes (Office of the ANC Chief Whip, 2011). A frequent criticism is that lotteries to encourage HIV testing are a poor use of public resources that should rather be used for more urgent needs: ‘The battle against HIV/AIDS must not be a game where people’s participation is motivated by winning the jackpot’, and that the money might be better spent if handed over to the National AIDS Council and the Department of Health (Office of the ANC Chief Whip, 2011, p. 1). Lotteries are accused of being paternalistic as they undermine individuals’ autonomy. Lotteries used in the 2011 Western Cape’s Government lottery were described to undermine the importance of HIV and AIDS education that ensures understanding of HIV testing, counselling and the advantages of knowing one’s status (Office of the ANC Chief Whip, 2011).

In conclusion, many claims for and against the use of lotteries to influence workplace HCT in South Africa exist, but very little scientific evidence about their influence on employees’ workplace HIV testing behaviour has been published.

The quantitative results of a study done in South African automotive companies in which lotteries were used to encourage workplace HCT showed that the announcement of lotteries significantly increased employees’ workplace HCT behaviour intention and perceptions of support by the company, colleagues and family to participate in HIV testing. The results reflected employees’ strong conviction that such a collective approach to workplace HCT would be successful. The lotteries helped to address various existing HCT barriers at a ‘social-cultural-collective’ level rather than at an individual level (Weihs & Meyer-Weitz, in press). Furthermore, a qualitative study done in the same companies after the HIV testing to explore participants’ experiences of the lottery and all aspects in this regard showed that in the 2 weeks prior to the HIV screening event, lotteries brought excitement to the workplace and enhanced shop-floor workers’ personal interest in HIV testing (Weihs & Meyer-Weitz, 2014). This enabled social interactions and collective support among the staff with regard to participating in HIV testing and helped in mitigating HIV stigma. Lotteries were described as being a supportive and innovative company initiative. Employees expressed social support and encouragement for testing but did not experience this as coercive (Weihs & Meyer-Weitz, 2014). A qualitative research study done among students who participated in the South African ‘First Things First’ campaign, showed similar findings. Most of the participants supported the lottery. Students’ reported that the reasons for getting HIV tested were the excitement created by the campaign and friends encouraging them. Although many participants were encouraged by the prospect of winning a prize, no student reported feelings of coercion. The students reported that the decision to participate in HIV testing was based on ‘their own best interests’ (Cameron & Van der Merwe, 2012).

Against the above background, this research posed the following question: How do lottery incentives influence permanent and non-permanent workers’ workplace HIV testing behaviour intention, when lottery entry is conditional on HIV testing (setting 1) compared to when HIV testing is not conditional for entry into the lottery (setting 2)? If workers’ workplace HCT behaviour intention in setting 1 was not found to be significantly higher than that in setting 2, this would suggest that workers’ intention to participate in workplace HIV testing was not significantly influenced by the prizes of the lottery.

The data for setting 1 were collected among shop-floor workers (permanent and non-permanent workers) in two independent automotive component manufacturing companies of the NMBM in November and December 2010. Similarly, the data for setting 2 were collected among shop-floor workers in two other independent automotive component manufacturing companies of the NMBM in July and November 2012. Participating companies had never offered incentives in the past to encourage workers to test for HIV in the workplace. None of the companies offered ART to their employees. Workers in all companies had been involved in a minimum of 6 months of HIV and AIDS WPP activities of similar standard. These comprised the implementation of HIV and AIDS policies and education activities directed at HIV prevention, treatment and care. All workers were also informed that a wellness event would soon be held on the premises of the company where four health-screening tests would be offered for free, that is, hypertension, blood sugar level, body mass index and HIV.

The study population included all permanent and non-permanent shop-floor workers working in two different shifts, one shift receiving the experimental intervention, the other shift being the control group. In all companies, the two shifts were set up by production managers unaware of the research study when planning the shifts. The sample size of setting 1 was 203 workers, of whom 198 completed the questionnaires (control group [N = 88] and experimental group [N = 110], response rate was 97.5%). The sample size of setting 2 was 356 workers, of whom 316 completed valid questionnaires (control group [N = 147] and experimental group [N = 169], the response rate was 88.8%).

For the larger study, a questionnaire was compiled in accordance to the Theory of Planned Behaviour (Ajzen, 1991) to measure predictors of workplace HCT behaviour intention for use in both settings. In addition, various socio-demographic information were sought, that is age, gender, marital status, highest education reached and employment status. However, for this research only data that measured workplace HCT behaviour intention and socio-demographics were used to answer the research question. Intention to participate in workplace HCT (BIntention) was investigated with the statement ‘If the company would organise its on-site Wellness Day tomorrow, I would go testing for HIV tomorrow’. BIntention was assessed using a four-point Likert scale.

A post-test only, quasi-experimental study design was used. A structured, self-administrated questionnaire in English was administered to the control and experimental groups in both settings. English was the official language used in the workplaces and the large majority of workers had national senior certificates. The experimental intervention was a leaflet informing shop-floor workers about data and venue of the wellness event and the conditions to be entered into the lottery (shopping vouchers worth R2000 [first prize], R500 [second prize], and 10 extra R100 prizes). In setting 1, the experimental intervention informed that while four health-screening tests were offered, that is, hypertension, blood sugar level, body mass index and HIV, workers had to test for HIV to be entered into the lottery. In setting 2, the experimental intervention informed workers that in the context of the same four health-screening tests, they had to participate in three tests out of the four offered, to be entered into the lottery; so it was made clear that they did not necessarily have to participate in HIV testing to be entered in the lottery. The first prize was close to half a month of a worker’s normal wage.

Peer educators supported the data collection by explaining and clarifying questions and misunderstandings. They were not informed about the experiment but were told that the survey was done to improve the HIV and AIDS WPP. The data were captured considering quality assurance for completeness and consistency.

Ethical approval for the study was obtained from the University of KwaZulu-Natal Ethics Committee (reference number HSS/1263/010).

The study data were analysed using SPSS 22 software. Frequencies were calculated and the central tendencies of the data investigated. Chi-square tests were used to compare demographics of setting 1 with demographics of setting 2, distinguishing between permanent and non-permanent workers. For better distribution, BIntention was re-coded into a dichotomous variable because of a negatively skewed response rate. To determine whether workers’ intention to test for HIV was solely based on a chance to win a lottery prize, chi-square tests were used to measure the difference in behaviour intention between settings 1 and 2 for the control and experimental groups, and distinguishing between permanent and non-permanent staff.

A chi-square test for independence (with Yates Continuity Correction) indicated no significant difference between the control groups of permanent workers in both settings (not depicted in a table) for workplace HCT behaviour intention ( χ² = 0.059, p = 0.808, phi = 0.037).

As depicted in Table 3, a chi-square test for independence (with Yates Continuity Correction) indicated no significant difference for workplace HCT behaviour intention between setting 1 and setting 2 for the experimental group of permanent workers ( χ² = 1.145, p = 0.285, phi = -0.097).

A chi-square test for independence (with Yates Continuity Correction) indicated no significant difference between the control groups of non-permanent workers in both settings (not depicted in a table) for workplace HCT behaviour intention ( χ² = 0.231, p = 0.631, phi = -0.075).

Table 4 shows that for the ExpGroup among non-permanent workers, a significant difference was measured between setting 1 and setting 2 ( χ² = 8.04, p = 0.005). The effect size (phi = -0.279) using Cohen’s (1988) criteria, was small.