Strengthening private-sector medicine systems to tackle the persistence of poor-quality medicines in Africa: a proof-of-concept study

Poor-quality medicines, containing little or no active ingredient - whether through deliberate fraud, poor manufacturing practice or post-manufacture deterioration - represent a major public health threat in low/middle-income countries (LMICs): responsible for more than 200,000 under-5 deaths each year in Africa and contributing to antimicrobial drug resistance. Efforts by governments and international agencies to curb the problem through improving detection rates, tightening regulation and public education have been hampered by the economic realities of medicine supply in resource-poor, high-need contexts, but also - we suggest - by a failure to apprehend fully the complex workings of medicine supply systems, particularly beyond the public sector.

Private-sector medicine systems can be characterised as 'complex systems' involving multiple dispersed actors with no central organising authority. Recent developments in the study of complex systems have revealed how the actions of individuals can combine to have non-intuitive effects on the system as whole. This has significant implications for well-intentioned policy interventions based on 'common sense' intuition, which may have unwelcome unanticipated consequences.

Our ultimate goal is to understand - and predict - the workings of complex medicine systems in order to inform effective interventions to minimise the penetration of poor-quality products in LMICs. This will require: mapping complete medicine supply chains; understanding the motivations/behaviours of buyers, sellers and regulators; developing sophisticated computational models to simulate the system; and engaging stake-holders to co-design evidence-based interventions. This is an ambitious project which needs careful groundwork through a proof-of-concept study with the following objectives and activities:

(1) To assess fieldwork feasibility in these contexts and validate research instruments:
Very few people have attempted to trace a full medicine supply chain in an under-regulated context. In order to assess feasibility, safety and ethical issues, we will map a limited number of medicine supply chains starting at retail outlets across Ghana and Tanzania and moving upwards to point of manufacture, obtaining as much relevant information as possible at each stage. Research instruments will be validated in each local context, and re-validated across contexts to ensure consistency.

(2) To develop our understandings of the structure and operation of private-sector medicine systems:
Geographically weighted analysis will be employed to describe the structure/organization of supply systems (length, number of transaction points, degree/level of vertical 'collapse', etc.) and investigate spatial dependencies in the data. Thematic analysis of ethnographic data and secondary sources will be used to understand actors' decision-making and behaviour at each point.

(3) To build Agent-Based Models (ABMs) simulating medicine systems, based on empirical data:
We will build a sequence of ABMs simulating medicine supply systems in Ghana and Tanzania as 'complex systems'. These models will allow us to understand, and ultimately predict, how individual behaviours might affect the system as a whole. We will develop 'user-friendly' models to use with policy-makers, highlighting potential unintended consequences of interventions.

(4) To develop and evaluate strategies for engaging relevant actors (market, regulatory, political) in the research and intervention design:
In each country, we will convene National Stake-holder Groups (NSGs), with policy, regulatory and high-level market actors and Project Working Groups (PWGs), comprising 'on-the-ground' supply-chain actors (buyers, sellers and regulators). Through a series of collaborative workshops, we will work with 'user-friendly' models to identify potential 'bottlenecks' or problematic behavioural logics that might underpin interventions.

Private sector medical supply chains in sub-Saharan Africa are not centrally planned. Their structure emerges from individual behaviour by actors with limited resources and information, and may be complex and counter-intuitive. Agent-Based Modelling (ABM) allows us to represent this heterogeneous behaviour in a supply system model. Ethnographic data on actors' motivations and decision-making behaviours can be incorporated into ABMs, enabling prediction of the emerging supply system structure. Predictions can be validated using empirical observation to confirm model assumptions. The validated model can then legitimately explore likely outcomes of changes relevant to policy makers.

We have already developed an ABM representing a simplified medicine supply chain using pilot work in Ghana and Tanzania (Ackland et al. 2019). Our goal for this project is to develop a geographically-accurate model using a systematic strategy for calibration and validation based on good statistically-derived estimates of retail outlet distribution, supply chain structure and transport connections. Using GIS data, we will overlay economic, geographic and socio-cultural information (including information on formal and informal regulatory mechanisms) onto each supply chain node to explore the system-level consequences. The ABM will incorporate theoretically and ethnographically informed algorithms for decision-making at each node. Independent data will be used for model calibration and validation. The model will be designed to scale so that (in a full proposal), with the supercomputing capability at Edinburgh University, we will be able to simulate millions of medicine transactions involving thousands of suppliers.

A second technical goal will be to develop prototype 'user-friendly' versions of the ABMs to share with policy makers, enabling them to design and explore virtual interventions and associated supply chain outcomes and thus promoting more systematic and informed policy debate.

This Foundation grant will provide a basis for a full-scale research project to support the design and delivery of sustainable strategies in Ghana and Tanzania to strengthen medicine supply systems and minimise penetration of poor quality products. Our overall impact goal is to improve the delivery of quality-assured essential medicines to private-sector outlets in Ghana and Tanzania, contributing directly to improving health outcomes. Private-sector outlets constitute the major source of essential medicines for resource-poor populations across the African continent; this work is thus explicitly pro-poor and represents an important step to achieving *effective* Universal Health Coverage.

There is strong local demand for our research. National/international stake-holders and market actors are only too aware of the threats that poor-quality medicines present to public health. There is a clear appetite for new, evidence-based approaches to address the problem systematically. This proposal was developed in close consultation with key stake-holders in each country (including Food and Drugs Authorities, (FDAs), Ministries of Health, National Medical Stores Departments (MSDs), Pharmacy Councils, and major commercial actors) and the WHO Global Surveillance & Monitoring Unit.

Our impact goal will be achieved through the following pathways and intermediate outcomes:

1) Advancement of scientific knowledge and methodology: Careful mapping and characterisation of 'real-world' medicine supply chains alongside multi-sited ethnographic research and agent-based modelling (ABM) will lead to a better understanding of system vulnerabilities and potential tipping points.

2) Assisting policy development through sustained stake-holder partnership: National Stakeholder Groups (NSGs) in each country will be established with Government and other key actors (see above). Through a series of participatory 'Theory of Change workshops', we will work with NSGs to map out the "missing middle" between research and expected impact in the short, medium and long term. A key focus will be identifying bottle-necks or problematic behavioural logics that might undermine interventions and working together to identify solutions. Prototype 'user-friendly' versions of agent-based models will be used to simulate the effects (intended and unintended) of possible interventions.

3) Improving market practices and transparency: The knowledge we generate about the workings and vulnerabilities of private-sector medicine supply systems will be made publicly available. Working with representatives from local manufacturers, importers, distributors, wholesalers, retailers and other private-sector actors through our Project Working Groups (PWGs), knowledge-sharing will help market actors to reflect on their own business models, reduce informational asymmetries and help align healthcare and business needs/opportunities.

4) Capacity building within Ghana and Tanzania: This truly collaborative project draws on long-term research partnerships. Ongoing capacity building will entail exchange of skills and expertise (GIS and spatial analysis, computational modelling, etc.) between all institutions (South-South, South-North and North-South). All publications will be jointly authored, with African Co-Is leading on publications from their country's data. These activities will mobilise the influence and capacity of African scholars to lead research and change policy and practice in their home countries and the Region.

5) Engaging international audiences: While the study findings will be immediately applicable in the two focus countries (Ghana and Tanzania), the methodological approach and principles will have wider relevance in under-resourced contexts vulnerable to penetration of poor-quality medicines. Communication/dissemination to international audiences will be via open-access journal publications, and conference presentations.