Research design

We used an inductive, co-productive research design influenced by principles of participatory research and constructivist grounded theory [23]. The research process involved a three-day concept-building workshop in Blantyre, Malawi (June 22–24, 2022), [21] thematic ‘open coding’ of workshop discussions, and further iterative development of a conceptual model following the workshop. Our approach follows the growing interest in “collective experimentation” [24] to address issues of transdisciplinary concern in public and global health, [24–26] of which multimorbidity is arguably a paradigmatic example.

Participants and sampling

The workshop organising committee comprised an interdisciplinary group of public health researchers, clinicians, and social scientists (EB, CIRC, JD, RAF, FL, EM, BM, JR). Potential participants were identified through purposive and snowballing methods between June 2021-March 2022, which are described in greater detail elsewhere [21]. The collaborator group drew together 60 researchers, clinicians, health planners, and policymakers (HIC-based n = 19, LMIC-based (n = 41), together representing a wide range of disciplinary perspectives, including from (sub-)fields of epidemiology, public health, clinical medicine, the social sciences and community advocacy (Fig 1). Given the current concentration of multimorbidity research, and the location of the workshop in Malawi, the regional expertise among the collaborator group stemmed primarily from countries within Southern Africa, with the greatest concentration of experience from Malawi (n = 33), South Africa (n = 13), and Zimbabwe (n = 9). Participants, several of whom were working across multiple countries, also brought experience from Central, Eastern, and Western Africa, together representing experience from 10 African countries (Fig 1). For practical and ethical reasons, we did not directly include PLWMM in the workshop; however, we asked collaborators, including social scientists and community advocates present, to represent learnings from their interactions to give an understanding of different patient perspectives. Supporting information includes a reflexivity statement detailing the measures taken to promote equitable partnership within this collaboration (S1 Table), [27] an overview of participant demographics (S2 Table), and a detailed breakdown of participants’ institutional location and disciplinary and regional expertise (S3 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Regional* and disciplinary expertise represented by workshop participants.

*Base map produced using R[28] with shape files obtained from the World Bank (https://datacatalog.worldbank.org/search/dataset/0038272/World-Bank-Official-Boundaries) and Humanitarian Data Exchange (for Western Sahara only) (https://data.humdata.org/dataset/cod-ab-esh).

https://doi.org/10.1371/journal.pgph.0003434.g001

Workshop design

A detailed account of the workshop design is published elsewhere [21]. The workshop was designed to optimise opportunities for cross-disciplinary discussion. Sessions were organised around four provisional thematic ‘domains’: (1) concepts and framings of multimorbidity (led by EC and CB); (2) population-level health data (led by NT and TT); (3) risk, prevention, and sites of intervention (led by NP and APK); (4) health systems and care models (led by EB). Each session began with an ‘ignition’ talk by session leads which outlined current knowledge, gaps, and key questions within each domain. These questions were then addressed by participants through a plenary discussion, facilitated by session leads and workshop organisers, and smaller focus-group discussions. The latter were facilitated by a team of PhD student rapporteurs (GTB, SS, IGS, SAS, NMY), who were mentored by members of the organising committee. In a final session, the group reviewed and summarised the workshop sessions in plenary, collectively identifying core and cross-cutting themes, before breaking into working groups which advanced agreed themes for further development. This article was led by working group focused on developing a common concept of multimorbidity.

Analytical framework

This research was guided by the principles of constructivist grounded theory, which holds that there is no ‘right or wrong’ and that emergent concepts and grounded theories are interpretive descriptions rather than an ‘objective’ account of reality [23]. Accordingly, the assumption underlying the workshop was that there is no privileged disciplinary vantage point from which to conceptualise or frame multimorbidity. Rather, different concepts and understandings foreground different aspects of the challenge multimorbidity presents for current systems, tied to particular knowledge bases, while potentially backgrounding others. This position, which was emphasised throughout the workshop and sustained through the post-workshop concept development, encouraged a diversity of concepts and understandings to be put into conversation, some of which diverged from and exposed the limits of the prevailing biomedical model. At the same time, in embracing difference, we were also able to find synergies and shared commitments. The embrace of both diversity and commonality formed the basis of developing a more cross-cutting, holistic, and ultimately more useful understanding of multimorbidity as it relates to African contexts.

Data collection

Source material collected during the workshop included, most centrally, detailed notes of proceedings, including both focus groups and plenary discussions, taken by the rapporteur team. Rapporteurs were instructed to capture summaries of what speakers had said and, where points were made that the rapporteurs felt were especially significant, direct quotes were also taken. Where necessary, the rapporteurs went back to individuals for clarification on points made, which were absorbed into their notes. Also absorbed into the final account were the flip chart pages composed during focus groups and the Microsoft Word documents produced using a shared screen during plenary discussion. Finally, all source materials were reconciled to produce a unified account of both focus-group and plenary discussions. These were made available to all in unanonymised form through a secure online storage platform, before being anonymised in the final deposited dataset.

Data analysis

As this was an iterative, co-productive process, [23] analysis began during the workshop itself, with all participants in engaging in critical reflection during sessions and collaboratively drawing out major cross-cutting issues in the final session [21]. After the rapporteurs had produced the final account of proceedings, the joint first authors (JD, BM, MJN, AS, IGS, SAS, MVP) subsequently conducted iterative inductive thematic analysis on the data using a team-based, ‘open coding’ approach that is designed to optimise inter-coder consensus in the context of collaborative research [29] To facilitate this process, data were transferred onto a shared drive, which enabled the coding team to code collaboratively in real time. After an initial pass through the data, the team met to consolidate initial codes into a provisional framework, which was subsequently applied to the whole dataset by JD and MVP. A second meeting was used to further refine the framework, with an emphasis on agreeing higher-level themes that grouped and explained lower-level ones. The final framework, of which the (sub-)themes related to the conceptualisation and framing of multimorbidity are presented in the results section, was then taken to the larger working group and used to develop of an expanded model of multimorbidity. Developing and refining this model was itself an iterative process, involving the working group assembling a tentative visual model based on the coding framework and relevant literature, which was circulated by email to the wider collaborator group along with the draft manuscript. Feedback was used to refine the model before being sent back for a final round of comment and refinement.

Research ethics

Due to the collaborative research design employed, in which all investigators were participants and vice versa, and all are named co-authors, formal ethical review was not required for this research. This follows recent examples adopting similar co-productive research models to develop cross-disciplinary frameworks and agendas [25,30]. All participants provided either verbal consent or written email consent to taking part in the concept-building workshop, which was captured using an Excel spreadsheet. At the end of the workshop, participants jointly agreed upon research outputs and working groups to take forward prominent themes from the discussions, and all consented to being named as co-authors on this particular output. All participants in the initiative have reviewed the contents of the manuscript and have approved its final version. The concept-building workshop was subject to the safeguarding mechanisms of the host organisation (Malawi-Liverpool-Wellcome Programme), which included anonymous reporting procedures.

Theme 1: Multimorbidity is currently centred on constituent diseases

Across the countries represented at the workshop, a key challenge expressed by participants was that populations are living longer and facing increasingly complex disease burdens, including the colliding epidemics of chronic infectious diseases and NCDs. Yet, health systems remain built around siloes of expertise based on single diseases. This is perpetuated by ‘vertical’ funding models adopted by Northern donors, which work against care for multimorbidity:

“Most of the chronic diseases in sub-Saharan Africa countries such as HIV and TB are managed through vertical programs, which inhibits care for multimorbidity” (Session 1, plenary, rapporteur notes)

Vertical approaches to single diseases have come to shape almost all aspects health system functioning: from health policy and planning, to research and surveillance, to training and care delivery, to the monitoring and evaluation of health systems performance. Disease-centred thinking is, as one family physician argued, so entrenched within current systems that it is extremely challenging to step beyond this frame of reference:

“The challenge is that the system we work in does not accommodate a transformation in thinking; it keeps bringing us back to a focus on diseases” (Session 1, plenary, rapporteur notes)

Throughout the workshop, participants grappled with the tension that, on the one hand, multimorbidity foregrounds disease concentrations and interactions rather than diseases in isolation. This makes it a potentially powerful concept for re-aligning priorities and systems with the increasingly complex disease burdens affecting many African countries. On the other hand, the concept of multimorbidity, participants noted, is generally that of a compound disease category, most commonly defined as the presence or absence of two or more long-term conditions. As a result, it is still a disease-centred concept. Participants therefore questioned whether this definition was suited for moving us from a scenario in which people are defined by their diseases, to what was often referred to as a more ‘person-centred’ approach:

“The definition of two or more conditions may have limitations, and may continue to perpetuate a disease specific approach” (Session 5, plenary, Word document on shared screen)

“We are still using a disease-focused lens to define multimorbidity–tensions versus a person-centred approach” (Session 5, group 5, flip chart excerpt)

Concerns raised about a disease-centred lens included that this may promote an overly simplistic, additive view of multimorbidity as the sum of interacting (but ultimately discrete) disease conditions. Relatedly, reducing multimorbidity to its constituent diseases was argued to narrow consideration social context to ‘modifiable lifestyle factors’ associated with particular conditions, a particularly prominent discourse in the context of NCDs (i.e., smoking, poor diet, sedentism, and substance abuse). As several participants noted, targeting individual behaviour and lifestyles does not amount to a ‘person-centred’ approach to multimorbidity and may lead to patient shaming and stigmatisation:

“Labelling some of the non-communicable conditions related to multimorbidity, such as hypertension and diabetes, as ‘lifestyle diseases’ creates stigma and shaming” (Session 3, group 2, rapporteur notes)

In turn, this behavioural and lifestyle focus deemphasises structural factors that constrain people’s lifestyle ‘choices’ in ways that make them vulnerable to multimorbidity.

Theme 2: Potential for revised concepts to centre the priorities, needs, and social context of PLWMM

Continuous with theme 1, a particular concern expressed with the disease-based definition of multimorbidity is that it also fails to capture what matters to PLWMM. Several points about patients’ needs and priorities were highlighted. First, a concern voiced particularly by clinicians was that it often matters far less to people what the diagnoses and their causes are than their consequences and impacts, which become especially complex when it comes to secondary medical complications (e.g.. diabetic nephropathy). Symptom and treatment burden, functionality, and quality of life were highlighted as important to fold into our understanding of multimorbidity:

“Multimorbidity is likely to be useful if it includes burden and function, for instance pain, disability, and sleep” (Session 1, group 1, rapporteur notes)

While there may be considerable heterogeneity in the lived experience of multimorbidity across different condition combinations, it was also observed that the symptoms, treatment burden, and needs of PLWMM are often not disease-specific and may share similar profiles regardless of condition or aetiology. Tellingly, when it was put to a vote whether the concept of multimorbidity should draw a distinction between ‘communicable’ and ‘non-communicable’ diseases, a significant majority said it should not. Participants were not suggesting that we cease thinking about causes and determinants of multimorbidity, which remain important especially within epidemiology, public health, and clinical medicine, but rather that we ground our concept of multimorbidity in the experiences and priorities of PLWMM.

Second, participants emphasised that when considering the impacts of multimorbidity in African contexts, it is vital to expand the focus beyond the affected person to the pivotal role of families, informal caregivers and larger support networks in navigating the health, social, and economic burdens of multimorbidity:

“Family members/community are very important for patients’ improvement because they have an influence on the treatment given, in the same way that nutrition post-delivery is mostly influenced by relatives” (Session 1, group 3, rapporteur notes)

Accordantly, participants including community advocates argued that while person-centred care is important, we need to be talking about family-centred care in this setting:

“We need to think about family centred care–patient-centred care is important, but there is also a family who has to be involved in the process” (Session 5, plenary discussion, Word document on shared screen)”

A related consideration when designing multimorbidity care models for African health systems is not to ignore current social realities such as medical pluralism and the existing role of traditional healers. In the absence of such recognition, care models risk recreating Northern global health policies which fail to account for local experiences and African-centred knowledge systems and values.

Third, widening the lens further still, participants argued that social, structural and environmental factors are crucial for understanding the patterning, impacts, and consequences of multimorbidity. The concepts of syndemics and the social and commercial determinants of health were recognised as useful and well-established frameworks for capturing ‘upstream’ factors that socially pattern multimorbidity along lines of poverty and inequality:

“Social determinants of health are important–syndemics, the commercial determinants of health, and others. For example, for TB patients, food insecurity, biological predisposition, and access to care have been used to identify patients/potential patients with multimorbidity” (Session 3, group 2, rapporteur notes)

Social scientists reported from qualitative research that daily socio-economic struggles preceded and exacerbated bodily ailments and symptoms, and that within such fragile arrangements, interruptions to or changes in circumstance could have huge impacts on overall ability to cope, as was evidenced during the COVID-19 pandemic. It was further observed that specialised care organised around single diseases compounds the burden placed on PLWMM: it shapes which diseases are diagnosed and prioritised, the time and resources that are needed to (self-)manage different conditions, and the added care burden on families and carers:

“Most care in urban areas is specialized. This leads to fragmentation, adding a burden to the family and caretakers” (Session 1, group 3, rapporteur notes)

“Individual disease treatment compounds the burden. Patients start prioritising certain conditions and some may go untreated” (Session 1, group 2, rapporteur notes)

Challenges commonly observed during qualitative research included reaching secondary and tertiary health facilities to receive care, especially for NCDs; struggles to mobilise funds to spend on multiple medicines and appointments; and losing income when spending time at the clinic or balancing medical appointments with family responsibilities.

Theme 3: Need for revised concepts to respond to varied conceptual priorities amongst stakeholders

The third theme was that, if context is taken seriously, a one-size-fits-all definition of multimorbidity trying to pin down multimorbidity to a specific number or combination of conditions is neither possible nor desirable:

“There is no one-size-fits-all concept or framing of multimorbidity–context matters” (Session 5, group 1, flip chart)

The diversity of academic and applied perspectives present at the workshop underscored how counterproductive, even harmful, a narrow definition favouring one discipline or perspective can be. The concept, it was agreed, needs to remain broad, flexible, and able to foreground different priorities depending on the question being asked, the problem or perspective driving the question, the geographical, epidemiological, and health system context, and indeed the level of scale. Depending on the context, multiple definitions may be needed:

“Some flexibility / ambiguity in definition–or multiple definitions–may be needed” (Session 5, plenary discussion, Word document on shared screen)

Participants considered the utility of multimorbidity to different disciplinary and stakeholder groups, and how the priorities might shift in different contexts. As highlighted in Theme 2, the primary concern at the clinical and community level is functionality, quality of life, and the burden and complexity shouldered by PLWMM, their family and social networks:

“For academics, policymakers, and public health, the label or definition of multimorbidity is likely to be useful, whereas, for patients, wellbeing and function are more important” (Session 1, group 1, rapporteur notes)

Public health researchers, while similarly recognising the need for greater emphasis on function and quality of life, were often more focused on the societal impact of multimorbidity, for which well-defined (though not necessarily disease-centric) measures are needed to facilitate comparison:

“Well-defined concepts are useful in academia. The label [of multimorbidity] is useful for prevalence/mapping clusters, being able to study interactions between drugs and conditions” (Session 1, group 1, rapporteur notes)

For policymakers and health planners, also deploying multimorbidity at the population level, the concept was viewed as useful for reconfiguring funding streams (e.g., from ‘vertical’ to ‘horizontal’ models), for developing new care delivery models (e.g., from disease- to person- and family-centred care), for training and deploying the health workforce, and managing risks among the population. Social scientists and historians pushed for an expansive concept, one connecting the intricacies of lived experience to the ‘upstream’ structural and systemic factors that socially pattern multimorbidity and exacerbate the burden:

“You can’t do multimorbidity and focus only on medical conditions, there are a lot of things are going on–social, financial, medical. These factors work together and need to be addressed together, including clinical interventions, upstream solutions, downstream solutions, and community interventions.” (Session 4, ignition talk, rapporteur notes)

Finally, while PLWMM were not represented directly, the group noted that multimorbidity was unlikely to translate well into lay models:

“It is important to distinguish the medical framework and the patient model. Biomedical and societal framings of symptoms translate poorly into lay terminology–there is not necessarily a term for multimorbidity” (Session 5, plenary discussion, Word document on shared screen)

Because of these different concerns, we may need to foreground (and background) different aspects of multimorbidity, and recognise scenarios when it is not useful, to maximise its potential and minimise its harms:

“Part of the ‘art’ of multimorbidity may be centring diseases, people, and systems at different times and in different places and situations” (Session 5, group 1, rapporteur notes)

While recognising the impossibility of a one-size-fits-all definition given that multimorbidity is, by nature, heterogenous and context-specific, the collaborator group was nonetheless optimistic about folding the multiple perspectives represented at the workshop into a broadly shared frame of reference to facilitate cross-disciplinary working. Following theme 2, it was proposed a focus on common consequences of multimorbidity and associated needs for PLWMM could be one pathway towards conceptual alignment. In the final session of the workshop, a working definition was proposed as:

“A clustering of needs [added emphasis] and conditions that need to be addressed holistically, rather than in isolation” (Session 5, plenary discussion, Word document on shared screen).

Whilst only a starting point, it did draw together major points of agreement during the workshop: the foregrounding of clusters of needs rather than just medical conditions; the importance of a holistic purview; and the critique of compartmentalised approaches. Refining this concept was highlighted as a key aim moving forward.