South Africa is committed to a health system built on primary health care (PHC) and has introduced a number of reforms to strengthen this foundation.¹ These reforms include the employment of district clinical specialist teams to improve maternal and child health, strengthening of school health services and the introduction of ward-based PHC outreach teams. These outreach teams consist of community health workers (CHWs) with a nurse supervisor and are intended to address population health management as a vital component of effective PHC service delivery. Population health management, as defined by the PHC performance initiative, includes local priority setting, community engagement, registration of the population and pro-active population outreach.²

Cape Town has implemented ward-based outreach teams within a community-orientated primary care (COPC) approach.³ In 2017, the Metropolitan Health Services (MHS) adopted a definition of COPC that was derived from the work of Abramson:⁴

In their framework document, they explained that their intention was to:

This same framework document identified 10 key elements to guide the implementation of COPC (Box 1).³ Four learning sites were identified, one in each substructure of the metropole, to learn about the implementation of the new COPC approach. Within these sites, teams of 10–15 CHWs were created around delineated geographic areas and groups of households and linked to the local primary care facility. Each CHW team was led by a professional nurse and employed by a non-profit organisation (NPO) under contract with the MHS. Each primary care facility had a number of CHW teams linked to it in order to cover the whole geographic catchment area and to form a larger PHC team that included the primary care providers (nurse practitioners and doctors) at the facility.

The concept of COPC was originally developed in South Africa during the 1940s before being suppressed during the Apartheid era.⁵ The concept was exported and implemented successfully in many other countries such as Brazil, Cuba, Spain, United States of America and Turkey.^6,7 The family health care team approach in Brazil has had a particularly effective impact on health outcomes.⁸ In Africa, most recent evidence on COPC relates to its implementation, particularly in South Africa, although one study from Kenya has again shown the effectiveness of such an approach.⁹ During the response to the coronavirus disease 2019 (COVID-19) epidemic in Cape Town, the COPC approach to organising the health system has also shown its ability to rapidly adapt to new challenges in PHC.^10,11 A study commissioned by the department of health in South Africa suggested that there would be a strong case for investment in COPC.^7,12

In Cape Town, one of the key elements in implementation was to define the scope of practice of the CHWs in a comprehensive and generalist approach. Household registration and assessment was a central part of this approach. Community health workers were expected to visit every household in their area and to gather information on the people living there, the type of dwelling and basic services available, the number of people with chronic conditions and to screen for a variety of health risks such as Tuberculosis (TB), human immunodeficiency virus (HIV), missing immunisations, vitamin A prophylaxis, pregnancy, need for family planning, tobacco smoking and wound care. People with such risks could be counselled or referred to appropriate services.

Whilst many studies have investigated factors impacting on the implementation of ward-based outreach teams,⁹ there has been little focus on the quality and usefulness of household assessment data collected by CHWs. This study, therefore, aimed to look at the household assessment data collected over a period of 1-year at the learning sites. At all these learning sites data had been collected on paper and in some cases captured by the NPO on electronic spreadsheets. None of the sites had actually analysed their data. Key objectives were to evaluate these data in terms of the quality of the data collected, the contribution of these data to a community diagnosis and understanding of local priority health needs, and the performance of the CHWs in identifying individuals with health issues and assisting them.

In Eastridge, data were analysed from 3564 households with 16 852 individuals, in Mamre from 1338 households with 3895 individuals and from Nomzamo 321 households with 1008 individuals. In Eastridge, data were collected at the household level and only some items could be extrapolated to individuals. The data sets were otherwise very complete for the demographic and housing information. Table 1 presents the findings in terms of the population demographics and characteristics of the households.

Mamre appeared to be an older population on social grants (probably pensions), with fewer children, in a community with almost all formal housing and good access to water and flush toilets in the house. Eastridge also had a substantial proportion of households receiving a social grant, mostly formal dwellings and good access to water and flush toilets. Nomzamo on the other hand was a much poorer community with a slightly younger population profile, smaller houses with more crowding, relatively fewer social grants (probably less pensions), much higher levels of informal housing and much less access to running water and flush toilets in the house.

Table 2 shows the percentage of the population that was identified as being on medication for chronic conditions. Again, Eastridge is reported differently as the proportion of households with a person on medication. The accuracy of this information was significantly limited by the amount of missing data, where it was not clear if the CHW had screened the household. Nevertheless, trends are clear in the sense that HIV and TB were more prevalent in Nomzamo, whilst non-communicable diseases were a much larger problem in Eastridge and Mamre.

Table 3 shows the number of people identified as having TB symptoms during the household assessments. The numbers are generally very small and again the information was limited by large amounts of missing data, where it was not clear if screening was performed or not.

Table 4 shows the proportion of the population identified as having a number of specific health needs. Mamre’s data suggested a large number of children were not immunised or received vitamin A prophylaxis, whilst in Nomzamo very small numbers of children were identified. This reflects a substantial amount of missing data where it is not clear if children were actually screened. At Eastridge, a substantial portion of data were identified at household level and suggested high rates of immunisation and children receiving vitamin A prophylaxis. Very small numbers of women needing assistance with pregnancy and family planning were identified. The majority of households in Eastridge had a member that smoked tobacco, whilst 27% of people in Mamre were tobacco smokers. The lower proportion in Nomzamo reflects a large amount of missing data, where it is not known if people were asked about smoking.

In Mamre, only 6 out of 272 (2.2%) of those identified as having a health risk had a record of referral in the household assessment form. In Eastridge, a different referral system was used, which was not captured on the electronic database. In Nomzamo, 45 out of 162 (27.8%) of those identified were recorded as referred.

At all learning sites, CHWs had visited a large number of households and completed household registrations and assessments during the first year of the COPC initiative. There was thus a huge potential for useful information. The findings demonstrated significant issues with the quality of the data and possibly with the performance of the CHWs. These issues impaired the usefulness of the data and the ability to accurately identify health needs of the communities. The findings also illustrate the strengths and weaknesses of the data in terms of giving a comprehensive view of health needs in these communities.

In all learning sites, there was a problem with missing data for many of the items. It was impossible to determine if the CHWs had asked everyone the relevant questions and only recorded positive responses or whether the questions had not been asked at all. With many items there were very few people identified (e.g. TB symptoms, missing immunisations and chronic medication), suggesting that CHWs had not consistently collected data from all household members. The household assessment form itself did not always require a definite answer and encouraged CHWs to only record a ‘yes’ answer by ticking. It is also possible that the household members who assisted the CHWs were not able to answer all of the questions for other household members who were absent. The extremely low number of referrals suggested that CHWs were using other mechanisms to record referrals and were not completing this aspect of the household assessment to save time and avoid duplication of tasks. Nevertheless, the low number of referrals raised a question as to whether CHWs were competently recognising health risks and responding to them appropriately. This question could not be answered from this data.

The need to conduct a study of this nature to analyse data that had been collected at least a year ago illustrates the flawed nature of the health information system for COPC and household assessments. None of the NPOs had the capacity to analyse the data themselves and in one case were just storing the forms in a cupboard. In addition, none of the data had been captured or analysed by the department of health. It was clear that an electronic m-health solution would potentially solve many of the problems. If data were collected via cell phone, then CHWs could be prompted to enter data more completely and accurately. Data would automatically be captured in a standardised way and analysed in terms of pre-determined indicators. Results could be immediately available to users at different parts of the system from the NPOs, primary care facilities to managers at substructure and district level. Data could be easily entered into the provincial data centre and combined with other individualised data to contribute to a personal electronic medical record. A locally developed m-health solution called Catch and Match was introduced at two of the learning centres during the implementation of COPC but has not yet delivered on its full potential.

The household assessment form does not provide a comprehensive view of the health risks or the CHWs’ scope of practice. In terms of health promotion and disease prevention, there is a focus on TB and HIV, but little attention to non-communicable diseases, other than tobacco smoking. Local research on CHWs has shown the ability to assess cardiovascular disease risk using a non-laboratory approach.¹⁴ For example, identifying people with a high risk (> 20% of an event in the next 10 years) could lead to useful counselling and referral for further assessment. Apart from wound care there is no identification of the need for physical care. For example, people with disabilities, such as stroke survivors and their caregivers, may need substantial help,¹⁵ as well as people needing palliative care.¹⁶ In terms of child health, there is no space for assessment of growth and malnutrition. Injuries and trauma are also a major contributor to the burden of disease¹⁷ and there is no focus on suspected alcohol or substance abuse, intimate partner violence, other forms of abuse or disability as a consequence of injuries. Maternal and women’s health is covered to some extent but need for cervical and breast cancer screening could be another useful addition. Finally, there is no prompting to consider social services and social grants are not disaggregated into different types (e.g. old age pension, disability and child support). It is clear that as a more comprehensive scope of practice is clarified that the household assessment form will need to be revised and aligned with this. Some of the missing data may be sensitive and the CHW would have to gain the trust of household members and maintain confidentiality.

Ultimately, the purpose of the CHWs is to assist the population at risk to live a healthier life. In the household, they need to have generalist skills to enter, form a relationship, identify any health risks, and respond helpfully and appropriately. There is a risk that the household assessment becomes an administrative exercise in obtaining data or filling out a form for its own sake. Completion of the form is used to evaluate CHW performance. Community health workers must retain a person centred approach to their work and use the household assessment form as a prompt and a guide to useful action in service of the community and household members.

Household assessment visits have great potential to identify individuals at risk and to help appropriately as well as to contribute to a community diagnosis that identifies local health priorities. Data collected is currently of poor quality, inconsistent or not captured, infrequently analysed and not comprehensive. There is a need to introduce an electronic m-health solution to assist the health information system, to revise the contents of the household assessment form and to ensure that CHWs are competent to identify risks and respond appropriately.