Community health workers are lay workers who have no professional training but have some training in the context of the intervention and are responsible for delivering services related to health care (Lewin et al. 2005).

Non-communicable diseases such as diabetes and hypertension have contributed significantly to the burden of disease globally, including in South Africa (Mayosi et al. 2009). According to the International Diabetes Federation (IDF), 2.28 million South Africans had diabetes in 2015 (IDF 2015), while the prevalence of hypertension is on the increase (Gomez-Olive et al. 2013; Lloyd-Sherlock et al. 2014; Thorogood et al. 2007). According to the World Health Organization in 2015, three quarters of all NCD-related deaths now occur in low- and middle-income countries (LMIC) (WHO 2016), where diabetes prevalence quadrupled between 1980 and 2014 (NCD Risk Factor Collaboration 2016). This rise in NCDs has implications for health systems and populations, affecting the most economically active age groups (Islam & Biswas 2014; Peck et al. 2014). In many regions of the world, health systems are ill-prepared to deal with the problem, and the prevention and management of NCDs remain a significant challenge. One of the constraints to managing NCDs is the poor availability of adequately trained human resources (Islam & Biswas 2014; Peck et al. 2014).

Community health workers have been proposed as a strategy to mitigate the shortage of health care professionals (Lehmann & Sanders 2007; Liu et al. 2011; O’Brien et al. 2009) such as nurses and doctors. CHWs’ relationship with the community, as well as their understanding of the context, culture and language, puts them in an ideal position to assist communities with health-related issues (Lehmann & Sanders 2007; Lehmann et al. 2009). Furthermore, CHW programmes provide a link with the formal health care system, thereby ensuring a continuum of care (Liu et al. 2011), vital for longevity and the prevention of complications of NCDs.

Studies have shown that CHWs can play a valuable role in the prevention, management and care of chronic lifelong (CLL) conditions such as human immunodeficiency virus (HIV) and acquired immunodeficiency syndrome (AIDS) (Callaghan, Ford & Schneider 2010; Mwai et al. 2013; Schneider, Hlophe & Van Rensburg 2008). The role of the CHWs has also been cited in the prevention and management of diabetes and hypertension in LMIC (Farzadfar et al. 2012; Jafar et al. 2010). In these contexts, CHWs have shown to be effective in providing education as well as support to people with NCDs (Bradley & Puoane 2007; Gaziano et al. 2015; Ndou et al. 2013). There is also considerable evidence on the role and effectiveness of CHWs for the prevention and management of NCDs in high-income countries, where roles similarly include health education, adherence support and counselling (Brownstein et al. 2005, 2007; Cherrington et al. 2008). However, for CHWs to perform these tasks, they require appropriate training (Abrahams-Gessel, Denman & Montano 2015b; Lopes, Cabral & De Sousa 2014), supportive supervision (Källander et al. 2015), materials and equipment. Despite the potential role of CHWs in NCDs, studies indicate that CHWs often have poor knowledge about NCDs and their risk factors (Abrahams-Gessel, Denman & Montano 2015a; Puoane, Bradley & Hughes 2006).

Community health workers in many parts of South Africa are employed by non-governmental organisations (NGOs) intermediaries, which are often contracted by the government to render services to communities (Van Pletzen et al. 2014). These NGOs are responsible for the capacity development of the CHWs, including their training needs, orientation into roles and supervision. Overall, community-based approaches to NCDs are underdeveloped, inadequately defined and fragmented. However, in South Africa, a diverse community-based care and support infrastructure based on lay health workers is being reorganised into a more comprehensive system of outreach. Under a set of initiatives referred to as PHC Re-engineering (Naledi, Barron & Schneider 2011), a formalised CHW cadre has been piloted with expanded roles. These roles extend beyond HIV and tuberculosis (TB) to include maternal child health interventions and integrated approaches to chronic lifelong conditions (involving both communicable and NCDs).

This cadre of workers will be organised into ward-based outreach teams (WBOTs) supervised by a nurse and supported by health facilities. In the first phase of primary health care (PHC) re-engineering, the focus has been on developing the capacity for maternal child health (Whyte 2015).

Khayelitsha is a large informal settlement in Cape Town. In 2011, this township had an estimated population of 391 749 people (Statistics South Africa 2012). At the time of the study (2013), community-based services were provided by a network of 5000 CHWs employed by 45 NGOs, which were predominantly funded by government departments such as health and social development (Van Pletzen et al. 2014). However, only three NGOs were supported and funded by the Ministry of Health to provide NCD care in the community. All three NGOs were purposively selected and approached for the study, two of which consented to participate. All 160 CHWs employed in the two NGOs were then approached for an interview.

Trained research assistants using mobile phones administered a questionnaire in isiXhosa (local language). Data were collected in 2013 for 2 months.

The development of the questionnaire involved an initial phase where CHWs were observed conducting their daily activities. The observations of practice provided insights about CHWs scope of practice and work organisation among other things. This formative, exploratory phase, which is reported elsewhere (Tsolekile et al. 2014), provided the basis for designing the questionnaire for this study. Findings from this formative phase resulted in the development of four critical areas (constructs) within the CHWs’ sphere of activity related to diabetes and hypertension: roles, training and induction, supervision and support, and knowledge. In an iterative process, the co-authors developed a questionnaire to assess these activities quantitatively.

Regarding their roles, CHWs were asked the following question (without further probing): ‘What are the services that you provide to clients with hypertension and diabetes?’ The responses were entered into a pre-coded list by fieldworkers. Knowledge was similarly assessed through a series of closed-ended questions covering risk factors, complications and prevention of hypertension and diabetes. There were 26 and 27 possible knowledge responses about diabetes and hypertension, respectively. These were basic and appropriate to the scope of CHWs.

The data collection tool was piloted to ensure the appropriateness and understanding of questions and to test its content validity. The questionnaire was piloted in another township 15 km away from the study site with 17 CHWs who also provided NCD care to community members.

Data analysis was performed using SPSS software, version 24 for Windows (Microsoft, USA). Univariate analysis was conducted on socioeconomic characteristics, NCD-related training, knowledge about NCD (including risk factors, complications, preventive measures) and roles of participants. Multivariate analysis was conducted to establish the relationship between the independent variables (various socioeconomic factors, training) and the dependent variables (knowledge scores for diabetes and hypertension). In all analyses, statistical significance was set at p < 0.05.

For each domain of knowledge, responses were scored. A score of zero was given to wrong or missing responses, and a score of one for correct answers. The means, medians and 95% confidence intervals for knowledge scores were calculated for each disease.

Permission (No. 11/4/4) to conduct the study was obtained from the Research and Ethics Committee at the University of the Western Cape, the Provincial Department of Health and the two NGOs that took part in the study.

Table 1 shows that CHWs were mostly female (n = 144) rather than male (n = 6). The mean age of CHWs was 35.4 years; 88% had some secondary schooling and 36% had completed grade 12. More than half (53%, n = 79) of the CHWs had been in employment for 4 years and more. Seventy (47%) CHWs reported to have at least one NCD, with hypertension (33%) being the most common. Nearly one quarter of the 150 CHWs (n = 34, 23%) had more than one NCD referring to a combination of diseases or conditions such as diabetes, hypertension, arthritis, heart diseases, asthma and strokes.

Community health workers offered a diversity of responses on their roles in the management and care of NCDs (Figure 1). The four most reported roles performed included the distribution of medication (84%), advising about diet (72%), measuring of blood pressure (63%) and conducting physical activity sessions (53%). Only 21% conducted pill counts as part of their routine practice.

Of the 150 CHWs interviewed, only 79 (52%) reported having received formal NCD-related training, and of those, less than half (n = 35; 44%) received refresher training after the initial training.

A wide variety of training experiences and providers were reported, the duration of which ranged from 1 to 270 days (9 months) (Figure 2). More than half of the 79 respondents (n = 46; 58%) reported training of 14 days or less, mostly 1 or 2 days (28%), while the remainder indicated more extended periods. Training providers were a mix of higher education institutions (often as part of research), the Department of Health and independent trainers.

Although 48% (n = 71) of the CHWs had no NCD-related training, they reported being orientated into their NCD-related roles by supervisors (nurses) (31%), NGO coordinators (34%) and fellow CHWs (25%).

The frequency of supervision of CHWs varied from engaging the supervisor once a month (25%) to more than three times a month (n = 93; 62%) (Table 2). Observed supervision (supervisor accompanying CHWs on the job) was commonly reported, with only 10% of CHWs not remembering when last they were observed performing their tasks. A vast majority (91%) of CHWs viewed their supervisors as being very supportive or supportive.

Knowledge of diabetes and hypertension among CHWs was poor, with mean scores being one third of the expected knowledge scores (Table 3). Scores for preventive measures were slightly better than for the risk factors and complications.

Multiple regression analyses were performed to ascertain the combined effects of any NCD training, age, years of schooling, NCD status (absence vs. presence of NCDs), duration of employment (≤ 3 years vs. > 3 years), and frequency of supervision (< 1 per month vs. ≥ 1 per month) on diabetes and hypertension knowledge scores (as a continuous variable).

Having an NCD and regular supervision (≥ 1 month) were strongly associated with both diabetes and hypertension knowledge scores (Table 4). The odds of knowing about diabetes were 5.4 and 4.9 times for those with an NCD or supervised more than once a month, respectively. Similar findings were observed for hypertension knowledge. Training in NCDs or years of basic schooling was not associated with improved scores.

Although this study provided noteworthy findings, the results should be interpreted with caution, especially about the generalisability of the findings. It was conducted in only two NGOs that deal with NCDs such as diabetes and hypertension as a part of a comprehensive package of care for a range of diseases, leaving out NGOs that may predominantly work with diseases such as HIV. The Western Cape, where the study was undertaken, is well resourced compared to other provinces in South Africa; this could be reflected in the quality of the supervision and the ability to access training opportunities from other institutions. Finally, the sample size was relatively small, limiting the possibility of disaggregated analyses. Nevertheless, the study provides insight into community-based NCD care, as well as training, supervision, knowledge and roles of CHWs who are responsible for providing NCD services in communities.