Alternative mechanisms for delivery of medication in South Africa: A scoping review

Background The number of people in South Africa with chronic conditions is a challenge to the health system. In response to the coronavirus infection, health services in Cape Town introduced home delivery of medication by community health workers. In planning for the future, they requested a scoping review of alternative mechanisms for delivery of medication to patients in primary health care in South Africa. Methods Databases were systematically searched using a comprehensive search strategy to identify studies from the last 10 years. A methodological guideline for conducting scoping reviews was followed. A standardised template was used to extract data and compare study characteristics and findings. Data was analysed both quantitatively and qualitatively. Results A total of 4253 publications were identified and 26 included. Most publications were from the last 5 years (n = 21), research (n = 24), Western Cape (n = 15) and focused on adherence clubs (n = 17), alternative pick-up-points (n = 14), home delivery (n = 5) and HIV (n = 17). The majority of alternative mechanisms were supported by a centralised dispensing and packaging system. New technology such as smart lockers and automated pharmacy dispensing units have been piloted. Patients benefited from these alternatives and had improved adherence. Available evidence suggests alternative mechanisms were cheaper and more beneficial than attending the facility to collect medication. Conclusion A mix of options tailored to the local context and patient choice that can be adequately managed by the system would be ideal. More economic evaluations are required of the alternatives, particularly before going to scale and for newer technology.


Introduction
The South African health system is challenged by a number of colliding epidemics of chronic conditions. 1 The antiretroviral therapy (ART) programme for people with HIV is the largest in the world, and the epidemic of tuberculosis (TB) is closely related. 2 At the same time, there is a growing problem of non-communicable diseases, particularly diabetes, hypertension, asthma, chronic obstructive pulmonary disease and post-TB structural lung damage. 1 Diabetes is now the leading cause of death for women and affects 1 in 4 South Africans over the age of 45 years. 3,4 Human immunodeficiency virus on the other hand mainly affects younger people in the age range 20-35 years. 2 This means that millions of South Africans require chronic medication, often for life, while at the same time being part of the working-age population.
at home in order to decongest facilities and protect people with chronic conditions from COVID-19. 8 As the COVID-19 epidemic began to subside the MHS considered whether to sustain this intervention and commissioned an evaluation of home deliveries by CHWs. In addition, they wanted a rapid review of alternative approaches to delivery of medication in South Africa to guide future policy.
The aim of this study was to conduct a scoping review of alternative mechanisms for delivery of medication to patients within the South African primary health care setting. In particular, the review addressed the following questions: • What alternative mechanisms have been attempted in South Africa to deliver chronic medication to patients in primary care? • What do we know about the implementation of these alternative mechanisms? • What is the effect of these alternative mechanisms on patients, service delivery and clinical outcomes?

Study design
This was a scoping review of published and grey literature in South Africa to address the three questions outlined above.

Search strategy
The search looked at English articles published in the last 10-years (2010-2020) in PubMed, Google Scholar and Sabinet.
The following search terms were used: South Africa, medication systems, primary health care, primary *care, medication adherence, chronic disease, decongest* and *congest, and home delivery of medication. These terms were combined in a number of search strings as shown in Box 1. Delivery was defined as the mechanism by which medication is delivered to or collected by the patient after it has been dispensed by a pharmacist according to the prescription. Any patient related effects, service delivery or clinical outcomes were included, and there was no exclusion of articles on this basis.
Relevant articles were selected using first the title and then the abstract from each search. Duplicate articles were then removed and the full text of the remaining articles was obtained. Relevant articles were included or excluded after reading the full text. Two researchers performed the searches and any uncertainty about including or excluding an article was resolved through discussion. The reference lists of included full text articles were also examined for any additional eligible literature.
Projects to deliver medication via alternative means in primary health care, which were not identified in the published literature, were also looked for via Google, key individuals at academic departments of family medicine, the department of health, funders such as the Bill and Melinda Gates Foundation, and consultancies such as Percept. Additional initiatives were approached for any unpublished reports or information on websites that could be included in the review.

Extraction of data
Data was extracted into a standardised Microsoft Excel template according to the following fields: • Authors.
• Year of publication.
• Location of study or initiative.
• Aim/purpose of the publication.
• Methods (if relevant, a summary of the methods used).
• Description of the medication delivery mechanism/model. • Goals of the delivery mechanism.
• Evidence on the feasibility/fidelity of implementation (key influencing factors, modifications to the design to make it work). • Evidence on the cost of the delivery mechanism. • Evidence on the coverage or scalability of the delivery mechanism. • Evidence on the effect or effectiveness of the delivery mechanism. • Limitations of the work as reported by the authors.

Data analysis
The extracted data was collated in an Excel sheet template where all data from each field could be reviewed in a single column. The characteristics (e.g. year of publication, location, type of article, study design) of published articles were analysed quantitatively and reported as frequencies. The other data was analysed qualitatively in order to address the study questions. Key themes in the data were identified and reported in text or tables.

Findings
The search strategies yielded 4253 publications, and after screening by title and abstract 71 were relevant to the review ( Figure 1). Once duplicates (31) were removed, the remaining 40 full text publications were assessed and a further 14 (13 found to be irrelevant and 1 being duplicate) were excluded, leaving 26 in the review.

BOX 1:
Search strings used in scoping review. "South Africa" AND "medication systems" AND "primary health care" "South Africa" AND "medication systems" "South Africa" AND "chronic disease" AND "medication adherence" AND "primary *care" "South Africa" AND "chronic disease" AND "medication adherence" "South Africa" AND "chronic disease" AND "*congest" AND "primary *care" "South Africa" AND "chronic disease" AND "*congest" "South Africa" AND "chronic disease" AND "decongest*" AND "primary *care" "South Africa" AND "chronic disease" AND "decongest*" "South Africa" AND "home delivery of medication" AND "primary *care" "South Africa" AND "home delivery of medication" http://www.safpj.co.za Open Access

Characteristics of the included articles
Altogether, 26 published articles were included in the review ( Table 1). Most of the 26 articles were published in the last 5-years and came from the Western Cape (Table 2). A number of studies were also conducted in Gauteng, KwaZulu-Natal, Limpopo and North West. A substantial number (n = 17) focused on HIV and evaluated the use of adherence clubs.
The most common type of study was qualitative, descriptive and exploratory using interviews with key informants or mixed methods. There was however quite a range of methods that also included experimental, observational and action research studies. Table 3 outlines the different alternative mechanisms. In all these approaches, criteria are used to select stable patients with chronic conditions who do not have to be seen more frequently by a clinician. How stable is defined will depend on the condition, but in HIV, for example, it typically means someone who had been on ART for at least 12 months, has a current cluster of differentiation 4 (CD4) count > 200 cells/mL, and is virologically suppressed. 9 Medication must also be prepackaged, and this can be done centrally by a dispensing unit or by the pharmacy at the local primary care facility. Going to scale requires a high-volume centralised system of prepackaging the medication.

Typology and description of alternative mechanisms
Smart lockers and pharmacy dispensing units have been piloted in Gauteng, but no published studies or scientific papers evaluating the innovations were found. 34,35 Smart lockers are marketed under the name Pelebox and automated teller-like machines (ATM) are trademarked as pharmacy dispensing units.
Regardless of the system the goals were very similar. Goals included: • Decongesting the facility and reducing the workload so that more attention can be given to new, unstable or sick patients. Specially in 2020, to also reduce the risk of exposure to COVID-19. • Task-shifting to CHWs in order to address human resource shortages at the primary care facility. • Providing a more efficient, convenient and patient-centred system for those who are doing well and just need medication. Reducing the burden on patients in terms of travel, costs and time spent at primary care facilities. • To improve retention in care, adherence to medication and therefore clinical outcomes. • To improve access to chronic medication and reduce stock-outs by having a more predictable demand that can be planned for. Table 4 compares the different mechanisms in terms of the feasibility of implementation, scalability and cost. Smart lockers and workplace outreach were excluded as there was insufficient evidence. Several issues are cross-cutting across all the alternative mechanisms:

Implementation of alternative mechanisms
• The mechanism must be supported by clear unambiguous policy, guidelines or standardised operating procedures and consistent leadership. • There must be a steady supply of medication with minimal out-of-stock items with procurement linked to quantified demand from registered patients. • Patients must be motivated and educated on how the system works, and supported to change their behaviour. • Health services must understand patient or health service related barriers that prevent people from effectively utilising these alternative systems. • Facility-and community-based health workers must be trained on how to use the system (e.g. selection criteria, scripting). • There must be an electronic communication system to inform patients on key issues (e.g. date of pick up or delivery, venue) and allow patients to ask questions. • There must be a reliable electronic audit trail to monitor the safety, status and location of the parcel. • Complex systems require good communication and feedback between all role players such as private sector centralised dispensing units, district management, facility pharmacists and CHW teams. • There must be a mechanism to link patients back into care if they develop a problem and to ensure they attend the 6-monthly review. • Consideration must be given to issues of inadvertent disclosure with stigma or discrimination, particularly for those who are HIV-positive.
A number of more specific issues around the feasibility of implementation are listed in Table 4. All mechanisms have the possibility of going to scale; although the feasibility of this will vary by context. Furthermore, clubs and home delivery have opportunity costs for health workers who if they support the  system for alternative delivery will not then be able to perform other duties in the time available. Central Chronic Medicines Dispensing and Distribution has gone to scale nationally and is a flagship project of the National Department of Health, combining centralised dispensing and alternative PuPs. Pharmacy dispensing units have been piloted but remain underutilised. They may be more efficient if linked to a centralised medication supply rather than a facility.

Cost and effectiveness of alternative mechanisms
These alternative mechanisms had many benefits for patients and were positively received. Patients saved time and money in not having to attend the primary care facility; and obtaining medication was less disruptive to work and other responsibilities. 14,15,27 Attending the facility implied long waiting times during restrictive opening hours as well as the cost of transport and lost income. 19 Facilities reported a decrease in congestion, reduced workload and alleviation of staff shortages. 27 Stock-outs were less frequent with the alternative PuPs because the quantification of patients on different items was improved and procurement was more effective in a centralised dispensing system. 20 Most mechanisms demonstrated improved adherence in terms of collecting monthly medication and for HIV this also translated into improved retention in care. At 40-months there was 97% retention in adherence clubs vs 85% in usual care for similarly eligible patients. 33 There was some evidence that retention in care was higher in men, but less in younger patients (< 25 years) and in facilities with larger cohorts of patients. 15 Home deliveries were also associated with improved retention in care with more patients receiving their

Type of delivery mechanism Description
Alternative PuP Patients collect pre-packaged medication from a more convenient PuP. A variety of PuP is used such as community halls, schools, private general practitioners, private pharmacies or fast-lanes at the primary care facility. Medication may be prepackaged by a central dispensing unit. The largest example of this mechanism is the Central Chronic Medicines Dispensing and Distribution (CCMDD) programme which operates in eight provinces in South Africa. An equivalent mechanism, the Central Dispensing Unit (CDU) operates in the Western Cape only. In some projects, medication was pre-packaged by the facility's pharmacy.
Workplace outreach The primary care facility performed an outreach clinical service to remote farming communities, which included the delivery of chronic medication.

Mpumalanga 2
Northern Cape 1 Free State 1

Mixed methods 5
Cost-effectiveness 1 Observational cross-sectional descriptive 2 Observational cross-sectional analytical 2 Observational cohort 1 Experimental before-and-after 1 Experimental randomised controlled trial 2 Qualitative exploratory descriptive 6 Qualitative case study 1 Action research 1 Protocol 1 Short report 2

Type of delivery system
Alternative pick-up points 14 Adherence or support clubs 17 Home delivery 5 Workplace outreach 1

Type of chronic condition
All chronic conditions 8

HIV 17
Non-communicable diseases 2 HIV, human immunodeficiency virus. medication (8.9% non-delivery vs 12% non-collection at facilities). 36 Forty-two percent of patients on home delivery also reported taking their medication more often. 36 In addition, adherence clubs showed a follow through to viral suppression (67% improvement in club vs usual clinic care). 30 Although we only had one cost-effectiveness study on one mechanism, 24 less robust evidence alludes to alternative mechanisms being cheaper and more beneficial than attending the facility to collect medication.
Pharmacy dispensing units have high start-up costs and are not yet cost-effective because of underutilisation. 35 There is a need for more comprehensive economic evaluations of different systems, particularly at scale.

Discussion
Alternative PuPs and adherence clubs have already been widely implemented in South Africa with support from the National Department of Health. The COVID-19 pandemic put home delivery of medication in the spotlight as it was an appropriate response to decongesting facilities and national lockdown. Home delivery may be through CHWs in Ward Based Primary Health Care Outreach Teams or in some areas by local entrepreneurs. 38 Going to scale requires a chronic dispensing unit to pre-package medication centrally. Smart lockers and pharmacy dispensing units are more recent technologically driven mechanisms that have been piloted on a smaller scale.
The goals of these alternative delivery systems were broadly similar and the need for them was driven by the HIV epidemic and unprecedented numbers of people on longterm ARTs that were overwhelming health facilities. Decongesting facilities to allow the available health professionals to focus on new and unstable patients was a central goal. Access to medication was also limited by the unacceptable waiting times at facilities and it was hoped that alternative mechanisms would improve access, adherence and ultimately clinical outcomes. The growing prevalence of non-communicable chronic diseases has highlighted the need to extend these alternative mechanisms to all people with chronic conditions. This need was further emphasised by the COVID-19 pandemic where HIV and diabetes were both significant risk factors for severe disease. 39 Patients were generally very positive about these alternative mechanisms once they learnt to trust the new systems. They substantially reduce the burden of collecting medication, saving both time and money, while decongesting facilities and capacitating health workers. These initiatives can be seen as supporting key aspects of effective primary health care, 40 particularly accessibility and to some extent the personcentredness of services. There is good evidence that they improve collection of medication and therefore adherence and retention in care, but very limited evidence of the effect on clinical outcomes.
Although the cost per dispense is relatively low and better than the cost of dispensing at the facility level, 19,24,30 the costs aggregate to substantial amounts when taken to scale. For example, the cost of home delivery of medication by CHWs in Cape Town could be 1.3% of total expenditure for Metro Health Services if sustained (R3 800 000.00 in 2020), although given the potential benefits this might be regarded as costeffective. 36 The national cost of CCMDD (with a hybrid of alternative PuPs, adherence clubs and fast-lanes) has been predicted (2016) as between R1.5 billion and R3bn per year with 9.1 million patients registered. 41 At this point the start- up and ongoing costs of PDUs are less competitive, and there is little data available on smart lockers.
The review only looked at literature from South Africa to limit its scope and to enable a rapid, but relevant review process. The combination of databases should have identified articles in relatively high impact journals (PubMed), articles in lower impact journals not in PubMed (Google Scholar) as well as studies published on local university and institutional websites (Sabinet), although databases such as Scopus were not included. It is likely therefore that all relevant articles were included, particularly as effort was also made to obtain grey literature and study the references of included articles.
The evidence cannot support definitive recommendations on what combination of mechanisms will be most cost-effective, but does support the conclusion that these alternative systems mostly achieved their goals. Each mechanism has a number of pros and cons, and may suit different types of patients and different regional contexts throughout the country. A hybrid system therefore that offers a range of alternatives to match the needs of different patients seems ideal, while avoiding unmanageable complexity. Such a system might include: • Alternative PuPs for the majority of patients at locations that are convenient and accessible, particularly for working people. Consider installing smart lockers at key locations that are accessible 24-h a day. PuPs in the community rather than fast-lanes in facilities may decongest facilities more effectively. • Adherence or support clubs for a limited number of patients that can be managed by the available human resources at a given facility, particularly for those that need additional support, for example, more isolated, recently diagnosed or borderline in terms of control and stability. • Home delivery of medication by CHWs, particularly for those that are frail, disabled or housebound. Ensure that CHWs have sufficient capacity to still perform a comprehensive service. Home delivery may also suit working people who pay a local entrepreneur to deliver at home.
To remain person-centred there should be a process of mutual decision-making, with health workers identifying eligible patients, but negotiating which option will work best for them. Being respectful of patient choice and control is likely to improve patient utilisation and satisfaction. 42 The whole system should be supported by a centralised dispensing unit that automates the packaging and dispensing of medication in order to manage high volumes.

Conclusion
Six alternative medication delivery systems were identified in South Africa: alternative PuPs, adherence clubs, workplace outreach, home delivery, smart lockers, and automated pharmacy dispensing units. Alternative PuPs and adherence clubs combined with centralised dispensing units have already been widely implemented. Home delivery was emphasised as an appropriate option in response to the COVID-19 pandemic. More technologically driven options have been piloted. All of these options appear to improve accessibility of medication and are acceptable to patients with the likelihood of improved adherence, retention in care and possibility of improved clinical outcomes. A hybrid system that offers a manageable range of options to eligible patients in a person-centred approach is probably the way forward. Further research is needed into the most cost-effective combination of delivery systems to meet the diverse needs of patients as well as the effect on clinical outcomes.