Thematic analysis of workshop proceedings revealed three core themes that are pertinent to the conceptualisation of multimorbidity in Africa: the disease-centricity of current definitions; the need to foreground what matters to PLWMM; and the importance of conceptual breadth and flexibility. Building on this framework, in the following, we place these core themes into conversation with current global health scholarship on multimorbidity, thereby developing an expanded model of multimorbidity that can underpin a coordinated but context-specific response to multimorbidity in Africa.

Our criticism of dominant constructions of multimorbidity emergent from Theme 1 resonates with several recent commentaries, reviews, and analyses. These have, in different ways, argued that a disease-centred framing of multimorbidity promotes a static, additive rendering of disease that treats illness as the sum of its parts.(16,20,29) One consequence of this disease-centred framing, as Blarikom et al.(16) have argued, is that the multimorbidity conversation has become preoccupied (and indeed, largely paralysed) by the search for standard biomedical definition and common core of conditions. Such a definition may be useful for drawing comparisons between populations and for identifying areas of met and unmet need. However, it may perpetuate a focus on diseases and individual behaviour and therefore ultimately fail to harness the critical potentials of multimorbidity to move beyond the single disease paradigm.(16,20,29) Against this backdrop, it is significant that attempts to adapt – and as one article puts it, “decolonise”(30) – multimorbidity for use beyond HICs highlight the need to move from a focus on diseases associated with ageing to recognise the ‘colliding epidemics’ of communicable and NCDs that characterise multimorbidity in LMICs.(3) While this may have important implications for integrating care across historically separated disease domains, the workshop proceedings suggest that ‘decolonising’ multimorbidity means more than adjusting its constituent diseases. Rather, beginning to decolonise the multimorbidity conversation means moving away from models of research and care that define and categorise people by their diseases.

The workshop concluded (as summarised in Theme 2) that there is a need to broaden our focus from the diseases that sit within multimorbidity to foreground what really matters to PLWMM in context. Calls for more holistic approaches resound in the multimorbidity literature from both HICs and LMICs and align with theoretical perspectives from multiple disciplines. This includes eco-social theory from epidemiology;(31) the syndemic framework,(32,33) the theory of recursive cascades(34) and burden of treatment theory(13) from the social sciences, and novel applications of complexity theory to multimorbidity within the primary care sciences.(29) Despite this wealth of theory and the recent proliferation of person-centred chronic care models,(14,35) in practice attempts to integrate care have not proven to be especially ‘person-centred’ in practice. Currently advocated-for person-centred care models are still generally designed using vertical disease programmes, for instance South Africa’s Integrated Chronic Disease Management (ICDM) system modelled on the country’s HIV programme.(15,36) Moreover, the common aim of such programmes to ‘activate’ PLWMM’s agency for better health decision-making emphasise individual responsibility, but often fail to respond to high rates of poverty and inequality that compromise the ability of PLWMM to control their exposure to health risks. A growing number of qualitative studies reporting PLWMM’s perspectives have been reported from countries including Malawi,(13) Ghana(37), Ethiopia,(38) and South Africa.(39) These show that PLWMM are unable to take control over their own health if they experience a ‘lack’ of health services, information, and basic necessities. Such lack in turn feeds into a cycle of precariousness that negatively impacts people’s ability to cope, often sending them down a slippery slope – or in complexity theory terms, over ‘tipping points’(29) – towards further disability and decline. Also appealed to but rarely prioritised in practice is the active involvement of caregivers and support networks in the provision of care for PLWMM.(15,40) Here, African social network theories such as Ubuntu – promoting mutual caring through compassion, reciprocity, dignity, and humanity – could be a useful concept, as they potentially cultivate resilience by creating a shared identity between PLWMM and carers, allowing them to flourish even when living in precarity.(40) Starting from where PLWMM are now – their priorities, values, needs, burden, and challenges navigating current healthcare systems – is a more promising way into the development of genuinely person-centred care models than the current retrospective ‘integration’ of different disease programmes. While an elaboration of what this might entail in practice is beyond the scope of this article, it would undoubtedly require careful reconsideration of what, where, how, and by whom care is delivered, likely involving the decentralisation of chronic care through healthcare workers trained in holistic, community-based care models – as elaborated, for instance, within the syndemic care framework.(33,41)

The main conclusion from Theme 3 was the need for a broad, flexible understanding of multimorbidity for it to be genuinely cross-cutting, folding in the possibility that in some scenarios it may not be so useful at all. This conclusion aligns with a growing body of multimorbidity scholarship running against the grain of the dominant drive to standardise and harmonise a biomedically-defined multimorbidity concept. A recent workshop exploring lessons from South Africa notably concluded that the ‘minimalist’ definition of multimorbidity as two or more chronic conditions, while appealing for its simplicity, is too simple. Any definition, it was argued, needs to take into consideration the causes and consequences of multimorbidity and will need to remain flexible enough to be responsive to different research questions, (inter-)disciplinary contexts, and scenarios.(42) Building on these observations, our findings suggest that a focus on common consequences for PLWMM holds particular promise for bringing different perspectives together around a common frame of reference. If, indeed, the needs and treatment burdens of PLWMM are often not disease specific, such a disease- agnostic focus could bring together disciplines and specialities in a way that has eluded the abstract and seemingly irresolvable attempts to pin down multimorbidity and its causal pathways in pathophysiological terms. The broader point, perhaps, is that the distinction between causes and consequences itself begins to break down once we recentre what matters to PLWMM, from whose perspective ‘consequences’ of a living with MM today may be ‘causes’ of further problems tomorrow.(34) By centring PLWMM and their ability to lead healthy, fulfilling and independent lives, we may better align medical and lay models of multimorbidity, such that it may not only be more cross-cutting and useful but also, perhaps, carry fewer negative connotations as a label. Figure 2 presents an expanded model of multimorbidity that folds in Themes 1, 2 and 3.

This model draws on the multiple theoretical influences of the participant group, including the syndemics framework,(32,33) burden of treatment theory,(13) and complexity theory.(29) It retains within it the basic idea of multimorbidity as involving more than one primary condition, in recognition of the utility of diagnostic categories within public health, epidemiology, and clinical medicine. But it remains agnostic about which kinds and combinations of condition ‘count’ as multimorbidity, instead expanding and bringing into the foreground the consequences of multimorbidity and its distributed burden on PLWMM, families, social networks, health providers, and the healthcare system. The model highlights factors identified by the collaborator group that currently overwhelmingly pull towards the right of the model, that is, that set PLWMM on a slippery (but not inevitable or linear) slope towards secondary complications and potentially catastrophic burdens that are felt across the system. It also casts our attention to the factors that might pull back people towards the left of the model. This includes seemingly small changes to a person’s circumstances that may make multimorbidity more manageable for the person affected and their wider networks, even in cases of advanced or complex disease. Given that many secondary complications cannot be reversed, the model suggests the emphasis should be on primary and secondary prevention. More than pharmacological or lifestyle interventions, this demands and rather interventions that take into account the interdependency between PLWMM and family, social network, healthcare providers, the health system, and wider social, structural, and environmental context.

While the ‘maximalist’ orientation of this multimorbidity model may lack some of the appeals of a minimalist definition, its dynamic and multidimensional nature enables different questions and disciplinary perspectives to be brought to bear on multimorbidity, including more refined definitions suited for specific questions. At the same time, the model suggests an overarching shift in the kinds of question that we ask and how we answer them. First, a focus on the complex bio-social interactions and impacts of multimorbidity moves us away from the disease-centred, cross-sectional designs from which most knowledge about multimorbidity to date originates (and that have tended to favour pharmacological and behavioural interventions) to disease-agnostic, richly contextual cohort research designs focused on burden, outcomes, and quality of life across the life course. Second, recognition of the multilevel factors at play implies that multimorbidity cannot be fully appreciated through any one disciplinary lens alone. Inter- and trans-disciplinary approaches, as well as holistic stakeholder engagement are needed to understand how biological, socio-cultural, political, economic, and environmental factors intertwine to co-produce multimorbidity; as well as to design holistic, person-centred, and systems-directed interventions. Third, the model implies a more context-specific approach to decide which disciplines and fields of expertise are relevant within the multimorbidity arena. Multimorbidity may result in similar needs and treatment burdens across different condition clusters, but its manifestation may also vary considerably across African settings given the tremendous heterogeneity of social, structural, environmental, and health system contexts. Thus, the knowledge bases and skillsets required to understand and address multimorbidity need to be similarly adaptive.(36) Whilst challenging, this is precisely the kind of context-sensitive approach that multimorbidity compels, offering a more promising pathway towards holistic person-centred care than universal, disease-centred interventions.

This model and the approach underlying it have several strengths. This was the first workshop to systematically bring together a multidisciplinary group of academic and applied actors to critically consider the meaning and utility of multimorbidity specifically within African settings. This was a bold exercise, not only in multidisciplinary experimentation,(23) but in elevating perspectives from a range of African contexts to reframe the multimorbidity conversation. This work provides a conceptual infrastructure to undergird the North-South and, more importantly, South-South partnerships that have been explicitly recognised as necessary for responding to multimorbidity in the region.(7) This research also had several limitations. First, while the collaborator group advocated for a person-centred perspective on multimorbidity, we were unable to directly include the voices of PLWMM. Further research is needed to gauge the extent to which this model resonates with PLWMM, whose perspectives remain under-represented within the multimorbidity conversation in Africa. The collaborator group was also biased towards Southern Africa, with the largest representation of expertise from Malawi, Zimbabwe, and South Africa, and thus our work disproportionately reflects views from these countries. Finally, our thematic analysis may be biased towards the views of the core working group. Whilst we endeavoured to represent the wider group’s perspectives objectively through a rigorous thematic analysis of workshop proceedings, the fact that we were both participants (in the workshop) and observers (analysing proceedings) means that the model may be biased towards our own values and viewpoints.