Roles, responsibilities and characteristics of lay community health workers involved in diabetes prevention programmes: A systematic review

Aim To examine the characteristics of community health workers (CHWs) involved in diabetes prevention programmes (DPPs) and their contributions to expected outcomes. Methods Electronic databases including PubMed-MEDLINE, EBSCOHost, and SCOPUS/EMBASE were searched for studies published between January 2000 and March 2016. All studies that used CHWs to implement DPP in ≥18-year-old participants without diabetes but at high risk for developing the condition, irrespective of the study design, setting or outcomes measured, were included. Results were synthesized narratively. Results Forty papers of 30 studies were identified. Studies were mainly community-based and conducted in minority populations in USA. Sample sizes ranged from 20 participants in a single community to 2369 participants in 46 communities. Although CHWs were generally from the local community, their qualifications, work experience and training received differed across studies. Overall the training was culturally sensitive and/or appropriate, covering topics such as the importance of good nutrition and the benefits of increased physical activity, communication and leadership. CHWs delivered a variety of interventions and also screened or recruited participants. The shared culture and language between CHWs and participants likely contributed to better programme implementation and successful outcomes. Conclusions The complexity of DPPs and the diverse CHW roles preclude attributing specific outcomes to CHW involvement. Nevertheless, documenting potential CHW roles and the relevant training required may optimise CHW contributions and facilitate their involvement in DPPs in the future.


Introduction
The rapid worldwide increase in type 2 diabetes (henceforth referred to as diabetes) has led to the development of a variety of different delivery models to prevent the development of this condition. To this end, a number of large randomised controlled trials (RCT) have demonstrated that lifestyle interventions reduce the incidence of diabetes between 29% and 58% in high-risk populations, and this can be maintained for well over 10 years [1]. However, these programmes when conducted under research conditions are usually resource intensive and thus, not practical or feasible to conduct in primary healthcare (PHC) or community-based settings. Particularly costly is employing professional healthcare workers to implement such interventions, it is not the best use of this resource, especially in developing regions with shortages of skilled healthcare workers.
The global need for efficient and cost-effective use of healthcare resources, particularly in low-income countries has led to the introduction of lay health workers or non-professional workers, lay health workers, non-professional health workers, Promotores de Salud, community health aids, peer advisors, community health advisors, village aids, community aids, lay counsellors, health promotores, community health promotores, village health volunteers, lay health educators diabetes, diabetes mellitus, type 2 diabetes, and prevention. Boolean operators, such as AND/OR/NOT were used to string terms together. Searches were limited to publications in English. For example in PubMed Central the search strategy was the following: ("community health workers"[MeSH Terms] OR ("community"[All Fields] AND "health"[All Fields] AND "workers"[All Fields]) OR "community health workers"[All Fields]) AND ("diabetes mellitus"[MeSH Terms] OR ("diabetes"[All Fields] AND "mellitus"[All Fields]) OR "diabetes mellitus"[All Fields] OR "diabetes"[All Fields] OR "diabetes insipidus"[MeSH Terms] OR ("diabetes"[All Fields] AND "insipidus"[All Fields]) OR "diabetes insipidus"[All Fields]) AND ("prevention and control"[Subheading] OR ("prevention"[All Fields] AND "control"[All Fields]) OR "prevention and control"[All Fields] OR "prevention"[All Fields]) Additional methods to identify studies included manually searching journals and conference proceedings, checking reference lists, and identifying unpublished data. One author and an independent assessor (JH, LM) independently identified potentially relevant studies by reviewing titles and abstracts retrieved from the aforementioned databases. The full texts of studies identified as potentially relevant were retrieved and screened in duplicate for inclusion. Consensus was achieved through discussion and, when needed, consultation with a senior author (APK).

Definition of community health workers
For the purpose of this review, a CHW is any lay or non-professional health worker involved with the delivery of a diabetes prevention programme, either as a volunteer or for a stipend. This accords with Norris's description of a CHW as an individual without formal healthcare training but trained to deliver context-specific healthcare to a community with whom s/he has a relationship [2].

Study inclusion criteria
All studies that used CHWs to implement DPP in !18-year-old participants without diabetes but at high risk for developing the condition, irrespective of the study design, setting or outcomes measured, were included. Studies that examined outcomes only among the CHWs, e.g. reports of CHW training interventions were also included because the aim of this systematic review is to establish the key characteristics of CHWs who contribute to successful DPP.
Eligible studies included those published in peer-reviewed journals from January 2000 until March 2016. Studies were excluded if they: 1) focused on diabetes management, 2) did not make use of CHWs, 3) comprised interventions of less than three months' duration, or 4) were narrative reviews, opinion pieces, letters to the editor or any other form of publication without primary data.

Data extraction and synthesis
Two data extraction tables summarise the data from the included studies. Table 1 shows the study-specific details such as the authors' names, demographic data, methodology and outcomes. Table 2 describes the CHW-specific characteristics including gender, age, education level, the training and support they received, and their role in the intervention. JH entered the summary data in the tables and LM then checked these. Given that the studies included in this systematic review were not sufficiently similar for a meta-analysis, data were synthesised Community health workers involved in diabetes prevention programmes: A systematic review narratively. We compiled thematic summaries relevant to the review objective and research questions stated in the systematic review protocol (unpublished).

Overview of the searches
Database searches identified 18906 entries; after removing duplicates and screening titles and abstracts, 182 publications were selected for further full-text evaluation (Fig 1). Forty one papers fulfilled the inclusion criteria and form the basis of this review. These 41 papers reported on 30 studies with companion papers comprising protocol papers, pilot studies, cost analyses and process evaluation papers.

Study settings, participants and interventions
The sample sizes varied widely from 20 participants in a single community to 2369 participants in 46 communities. Intervention settings varied but were mainly community-based such as churches, homes and community centres. A study each was conducted at senior centres [10,11] and worksites [25][26][27].

Profile of community health workers
CHWs were generally from the local study community and shared the same race/ethnic and language backgrounds as the participants. Two exceptions were a worksite-specific study which made use of peer coaches [25][26][27] and a senior centre study of whom some of the lay health educators or coaches were staff members [12]. Nevertheless, these CHWs were likely very familiar with the communities under study. Language was a specified criteria for studies conducted in Hispanic [13,28,31] and Korean-American [19] communities in the US, Aborigines in Australia [44], and in Gujarat, India [33].
The qualifications and work experience of CHWs varied widely across studies. Qualifications included: 1) completing only high-school [13,22,33,34] or 2) a CHW-certification programme [43] 3) previous training on measuring automated blood pressure (BP) [31] 4) and 5) post-high school education and some undergraduate education in nutrition [28]. Instead of qualifications, other studies focused on work experience as a CHW [22], prior or current experience in working in various areas of health [51], experience as a community organiser, personal trainer or caregiver [41,46,48], or having worked with the investigative team previously [34], as well as being involved in developing and adapting programme curriculum [48]. A qualification or experience in the education, social welfare or health sectors was a pre-requisite in an Australian study [44].
A single study used CHWs with well-controlled diabetes and a history of healthy eating, regular physical activity, weight loss and group leadership experience [14,18]. In contrast, a study in Thailand recruited CHWs who were simply interested in developing a DPP [32]. Many studies emphasised soft skills in the selection of their CHWs. These included possessing leadership qualities [13,33,34,40,41], good communication skills and the ability to listen [37], and a dedication to the community [13,34,37].

Community health worker training
The duration of the training varied considerably among studies, from a single one-hour session [25,27] to 40 hours [36], and even 100 hours or more [21,22,30]. Overall, the training was culturally sensitive and/or appropriate (Table 2). Topics covered included the importance of good nutrition [32,43,44] and the benefits of increased physical activity [14,38,43,46]. Communication [23,24] and leadership [38,40,41] skills to run group sessions, and motivational interviewing to facilitate behaviour change [19,24,28,30,36,37] were also emphasised. Most studies utilised interactive training and role-playing to impart skills during the training sessions. Four studies did not describe CHW training [32,45,47,50]. One study put specific emphasis on intervention fidelity/following study protocol [12].
Post-training support ranged from daily supervision to monthly meetings. These included ongoing supervision, monitoring and support [13,14,23,38], weekly technical conference calls [10], weekly staff meeting [43], and monthly meetings with programme leaders and evaluators [41].
Some studies also provided continuous or booster training. These included ongoing training with quarterly refresher training sessions as well as local and national conference attendances [29,42], biannual booster training sessions [21,28] and a two-day refresher training mid-way through the intervention programme [23].

Community health worker responsibilities
As summarised in Table 2, most CHWs in these studies delivered the intervention activities, i.e. led or facilitated the group sessions. These included developing and organising activities to promote healthy diets and physical activity such as a weekly senior citizen walking club, diabetes awareness and prevention events, cooking demonstrations, and a monthly fruit and vegetable market [10, 11, 13-24, 28-30, 32-39, 42, 43, 47, 51]. In Buffalo City (US), CHWs led community "living diabetes well" conversations and established "diabetes resource libraries", which provided information on diabetes, healthy living, healthcare providers, and recipe cards and cookbooks [40,41]. CHWs linked African-American men with primary healthcare providers and other supportive community resources [39], and in migrant farmworkers screened and referred those at high risk for diabetes and cardiovascular disease (CVD) [31].
Three of the included studies, however, used occupational [27], homecare [50] and rural health nurses [45], instead of CHWs, to deliver the interventions. CHWs only recruited [45] and maintained regular contact with participants providing ongoing support and information [27], and set-up meeting venues and coordinated walking groups [50] in these studies.
CHWs also recruited programme participants in four of the studies where they led the interventions [21,32,39,49] and provided follow-up support between the intervention programme/sessions in three studies [30,35,49]. Other CHW tasks included data collection [39] and entry [14,17], and recording attendance [18,49] and interactions between participants and project staff [23].

Diabetes prevention programme outcomes
Outcomes of the DPPs are reported in Table 1. Most studies (20) included a measure of adiposity such as weight or waist circumference with weight loss ranging from more than 7% of body weight [38] to modest weight loss of 1.1kg [28]. Unexpectedly, a study reported a significant difference between groups because of weight maintenance in the intervention group, but weight gain in the control group [27].
Of the nine studies that assessed various food and dietary behaviours, six reported significant improvements. These dietary variables included general food behaviour [22,42], consumption of sweetened beverages [48], and fruit and vegetable [30]; dietary fat intake [36], and caloric intake and proportion derived from protein sources [37].
Eight studies assessed knowledge on diabetes and/or CVD with all reporting significant improvements [19,22,30,39,41,43,44,48]. Improved mental health outcomes were noted in two studies; the Personal Health Questionnaire (PHQ-2) and the Generalized Anxiety Disorder Scale (GAD-2) were used in a US Korean community [19] and the Patient Health Questionnaire (PHQ-8) in a US Hispanic population [37]. A single study that examined fatalistic and cultural diabetes beliefs, measured by the Power Fatalism Inventory, showed significant reductions in fatalistic beliefs about diabetes manageability and endorsement of culturally driven diabetes beliefs [36]. Although two studies collected data on quality of life outcomes, neither study reported these findings [14,23].

Cost analyses of diabetes prevention programmes
Two studies, both from the US, conducted cost analyses for their DPPs [11,18]. In the study conducted in senior centres, total estimated cost for the CHW delivered lifestyle intervention was $2731 per senior centre or $165 per participant or $45 per kilogram weight lost [11]. These costs were almost half those of a health professional delivered DPP which cost $300 per participant or $88 per kilogram lost [52]. Direct medical costs per participant in the HELP-PD lifestyle intervention programme conducted over two years were $850 compared to $2361 for the first two years in the original DPP conducted in the US [18].

The effect of community health workers on programme outcome
Studies did not specifically compare the use of CHWs vs. health professionals on programme outcomes, except for Thompson et al. [31] who found that CHWs perform as well as registered nurses in the use of non-invasive risk screening tools. Therefore, it is difficult to quantify the specific advantages of using CHWs to implement DPPs. However, DPPs that targeted vulnerable/underserved communities highlighted the importance of CHWs in contributing to their acceptability and appropriateness (Table 3) [21,22,24,30,35]. Additionally, studies have also emphasised the importance of community participation per se as a major contributor to ". . .that CHWs can provide advice on lifestyle modification and improve awareness of diabetes and CVD similar to allied health professionals in earlier studies. . . The CHWs were able to successfully empower the women to speak up at their meetings and instil them with confidence in their decision making abilities as related to their health and well-being. Using CHWs strengthened the links among project personnel, the community and existing community networks" Philis-Tsimikas et al., 2014 [36] "Participants appreciated the convenient community location, social support received from other participants and the promotoras" programme effectiveness [32,33,38,[46][47][48][49]. The shared culture and language between CHWs and participants likely contributed to better programme implementation and outcomes [19,21,30,36,44]. In essence, using CHWs in DPPs that target culturally and linguistically diverse groups seem to be a credible strategy [44].

Discussion
The current review provides evidence from 33 studies of the increasing involvement of CHWs in implementing DPPs. The majority of the studies, however, were undertaken in high income countries, particularly the US. Only three studies were conducted in developing countries, where the incorporation of CHWs in resource-limited settings is likely to have a greater impact.
Most studies targeted the underserved minority such as African-American and Hispanic communities in the US, Aborigines in Australia and the Maori people in New Zealand [24]. The same was true of developing regions where rural communities were the focus in India [23,33] and Thailand [32]. DPPs that target minority and other vulnerable groups are likely suited to use CHWs who share a common culture, belief system (tradition), and language with the programme participants. This accords with The Centers for Disease Control and Prevention's Policy Brief (2015) which produce strong evidence for the use of trained lay people i.e. CHWs as a best practice for reducing CVD risk and improving outcomes in high-risk minority populations [53]. Unfortunately, the existing evidence is unclear, inconsistent and insufficient to inform the scaling up of DPPs in diverse settings using CHWs. The complexity of the programmes precluded attributing any specific benefit to the use of CHWs. Similar to these findings, Shah and colleagues noted that despite the large body of literature on CHWs and diabetes care, the wide range of CHW roles and differing outcomes made it difficult to draw conclusions on their overall effectiveness [5].
The outcomes of interest in the included studies were mostly intermediate, such as changes in behaviour or body weight, with no study reporting diabetes incidence. Nevertheless, CHWs are regarded as adding value to programmes by fulfilling a gap/an unmet need [54]. CHW-led interventions are optimally suited to programmes where there is greater community involvement, as shown by a number of studies in this review, which used community-based participatory research approaches [19,20,22,30,32,33,35,37,38,[46][47][48].
The profile, scope of roles played, and training received by CHWs varied substantially across programmes. Most DPPs provided information on the training/curriculum to varying degrees [32,45,47,50]. The duration of training varied significantly with shorter training sessions  Thompson et al., 2014 [31] "This quasi-experimental study supports the hypothesis that Latino CHWs can use non-invasive diabetes and CVD screening tools with similar accuracy as a registered nurse" Coppell et al., 2009 [45] "A key factor underpinning the program was early involvement of CHWs, community members and local organizations and the use of local resources and talents so that the program would become embedded into everyday community life and sustainable in the long term.
Cohen & Ingram, 2005 [47] "The curriculum for the family component, for example, was developed collaboratively between academics and promotoras who had never worked together in the past" Whittemore et al., 2014 [50] ". . .better understanding of the role of the CHW in the delivery of health programs is needed" https://doi.org/10.1371/journal.pone.0189069.t003 Community health workers involved in diabetes prevention programmes: A systematic review generally scheduled in programmes with fewer CHW responsibilities. While a systematic review reported that the performance of CHWs might improve with regular supervision and continuous training, an optimal model was not suggested [55]. Additionally, CHWs would benefit from clearly defined roles and clear processes for communication [55]. Guidelines that describe the potential roles of CHWs, the appropriate training required for each task/role and the type of programmes that would benefit from CHW involvement may be useful. While such guidelines may need to be adapted for different populations, they may encourage greater utilisation of CHWs in in appropriate healthcare programmes, including DPPs, and lead to better outcomes. The frequency and duration of the DPP interventions varied, with higher versus lower intensity programmes having the greater impact. For example, CHWs in the HELP-PD programme delivered the intervention sessions, which comprised weekly group sessions for six months, and a monthly session for the following 18 months [16]. This study achieved significant improvements in blood glucose and insulin levels, insulin resistance, weight and BMI with the effects sustained over a two-year period [15][16][17]. Identifying the ideal programme intensity and duration may be useful/of value for optimal outcomes; however, this may possibly vary depending on the outcomes targeted. In addition, longer-term studies are required to ascertain the need to provide intermittent support beyond the 2-year period, which was the maximum duration of the studies included in this review.
The high concentration of studies from developed countries, especially the US, limits the generalisability of our findings. However, most populations studied were underserved or minority communities, which may be of relevance to developing region settings. The inclusion of only published studies in English language in this review may have missed lessons being learnt from potential ongoing CHW-led DPPs, particularly from developing regions, or studies published in other languages. Additional information was not obtained from the study authors to provide greater insight on CHW education, training, experience and supervision and attrition rates. Therefore, the potential for publication bias is evident in this review.

Conclusions
In view of the labour-intensive nature of community-based healthcare programmes, particularly DPPs, and the high cost of professional healthcare staff, pragmatic cost-effective solutions are required for optimal outcomes. The utilisation of adequately trained CHWs with ongoing supervision to perform clearly defined tasks is likely to be most beneficial, as noted in a World Health Organization report on the state of evidence of CHW programmes, ". . .they must be carefully selected, appropriately trained and-very important-adequately and continuously supported" [3]. In addition to the aforementioned, the success of CHW-led programmes lies in the commonalities such as language, culture and tradition shared between the CHWs and participants. These commonalities facilitate communication and dialogue, which is crucial, and may play a key role in the success of such programmes. However, considering the complexity of DPPs and the diverse roles played by CHWs, it is difficult to disentangle the specific contribution of CHWs to these programmes. Nevertheless, developing guidelines for potential CHW roles and determining the appropriate level of training required may help identify the optimal CHW contribution to DPPs, which is currently lacking in the literature. Furthermore, this may perhaps encourage the wider uptake of CHW-led DPPs and lead to the development of better programmes.