Co-design and co-production: the vocabulary outruns the evidence

Co-design is asked to be three things at once. It is a method a team can follow, a value about who ought to decide, and an intervention whose effect can be measured.

This unit, the sixth of the ten, argues that co-design changes the unit of authorship: responsibility moves from the designer to a collective that includes the people the design is for. The previous unit asked who decides; this one asks what their deciding has achieved. The field's own reviews find improved relevance, usability and acceptability, and until 2025 little demonstrated clinical benefit; the first pooled trials now find modest effects that fade after six months.

Co-design moves authorship, and the shift is older than its names

Sanders and Stappers (2008) fixed the vocabulary this field still argues over. Co-creation, for them, is any act of collective creativity shared by two or more people. Co-design is that creativity applied across the whole design process, which makes co-design the narrower term.

Co-design is "collective creativity as it is applied across the whole span of a design process" (Sanders and Stappers, 2008, p. 6).

The user-centred approach, in which the user is a subject, was largely driven by the United States. The participatory approach, with the user as a partner, was led by Northern Europeans (Sanders and Stappers, 2008).

Robert (2013) gives health services their own version. Experience-based co-design is participatory action research in which patients and staff are equal partners. It has spread internationally since a pilot in an English head and neck cancer service in 2006. The chapter states that independent evaluations have shown positive impacts on staff practice and patient experience, but it is a methods account rather than one of those evaluations, so the benefit claim sits at the level of the method while the measurement sits elsewhere. Patients and carers give filmed narrative interviews, which are edited into a trigger film. Staff and patients then meet, separately to name priorities and together to agree them, and small groups work on those priorities over several months.

The vocabulary multiplies faster than it is defined

Mattelmaki and Sleeswijk Visser (2011) took the confusion as their topic, reviewing the literature and discussing cases, and found the co-terms tangled across academic, professional and organisational use. Vargas and colleagues (2022) write for public health. They note that co-creation, co-design and co-production are used interchangeably, propose co-creation as the overarching principle, and locate the real differences in the role of stakeholders and in the timing of their engagement. If the distinctions concern who decides and when, a study that reports "co-design" without those choices has not described its intervention.

Brandt, Binder and Sanders (2012) give the practical version of that point, describing the field as a proliferating family of practices organised by telling, making and enacting. An intervention described as a family of activities has no single definition to be evaluated.

The early benefit claims were plausible, local and unverified

Steen, Manschot and De Koning (2011) give the earliest systematic statement of the benefit case. From a literature review and three service design projects they distinguish benefits for the design project, the service's users and the organisation, reporting gains in the creative process and in the fit between service and context. The paper closes by asking how to monitor whether those benefits are realised and how to weigh the costs and risks.

Trischler and colleagues (2018) responded with a controlled comparison: co-design, in-house professional and user-only teams produced concepts for a library service. The co-design concepts scored significantly higher on user benefit and novelty, but lower on feasibility, and the advantages appeared only in cohesive teams, where dominating individuals produced conflict and weaker innovation. It is the strongest positive experimental result here, and it remains about ideas rather than services in use.

The reviews find relevance, usability and acceptability, and little on outcomes

Slattery, Saeri and Bragge (2020) reviewed 23 reviews of research co-design in health. They found it widely used but seldom evaluated, with effectiveness rarely tested empirically. Sumner and colleagues (2024) screened 3,334 reports on co-designed remote patient monitoring, which assumes a connection at home, and found 17 projects, none evaluating the efficacy of its own technology. Their review of co-designed care for multimorbidity screened 14,376 reports and found 13 studies. Only two of those studies reported health and well-being outcomes, and four of 17 outcomes favoured the co-designed arm. Greene Barker and colleagues (2024) ran the meta-analysis the field lacked: 20 studies of in-person, co-designed mental health interventions, yielding a moderate but non-significant effect on depression and a small non-significant effect on anxiety. Only three used the term co-design, and confidence in the estimates was low by GRADE.

Anieto and colleagues (2025) pooled sixteen randomised trials of co-created lifestyle interventions for adults with non-communicable diseases. At under six months there was moderate-quality evidence of a moderate effect on health behaviour and small effects on physical and mental health; beyond six months the effects shrank and the physical-health estimate lost significance. The powered trial has now been run sixteen times over, and the pooled effect is modest.

Where reviewers look for a designed service, co-design delivers better relevance, usability and acceptability. Where they look for a clinical outcome, the effect is small, uncertain or unmeasured. The reviews' own explanation is that the projects were not built to be measured, and a weekday workshop in the working language selects for the participants who can reach it. The multimorbidity review names passive participation and power imbalances among the common challenges of the process (Sumner et al., 2024).

Contested question: is the missing evidence a failure of co-design or a failure of measurement?

The first position holds that the evidence is not yet in, and that what exists points the right way. Its statements are Steen and colleagues (2011), for whom the gains are real but located in the project, the users and the organisation; Trischler and colleagues (2018), whose controlled comparison found a significant advantage in novelty and user benefit; and Robert (2013), whose account reports positive impacts on staff practice and patient experience. Clinical scales are insensitive to relevance, usability and acceptability, so a null result is not evidence of no effect.

The second position holds that the reviews measured what matters and found little. Its statements are Slattery and colleagues (2020), the two Sumner reviews (2024) and the Greene Barker meta-analysis. Together they screened tens of thousands of reports to find a handful of outcome evaluations. A field that ran for two decades without a powered trial made a claim it could not support, and the first pooled trials, in 2025, find effects that are modest and fade with follow-up.

The third position rejects the terms of the disagreement. Farr (2018) followed co-design and co-production in two public-service settings and argues that the mechanism at work is power: collaborative techniques can challenge existing power relations, but only with constant critical reflection and dialogue. If the point of co-design is a redistribution of who decides, an outcome trial that leaves power unmeasured has tested the wrong thing.

What would settle it is a study the pooled trials lack. It would use a co-designed intervention with a pre-registered clinical outcome, a fidelity measure of what participants actually did, and a separate measure of decision power. The three positions would then make predictions that can fail. The first predicts benefit on the co-design-relevant measures where the clinical effect is small. The second predicts that the participatory and clinical measures move together or not at all. The third predicts that transferring more power produces larger effects than the same activities with less power. The reviews come closest by demanding uniform terminology and measurement (Slattery et al., 2020; Greene Barker et al., 2024), but no source here ran the four measures together.

The three positions carry evidence of unequal strength. The positive position rests on a literature review with three projects, one controlled comparison that scored concepts rather than services in use, a methods account that cites evaluations it does not report, and, since 2025, a meta-analysis of sixteen randomised trials reporting moderate-quality evidence on health behaviour. The null position rests on two systematic reviews and a meta-analysis of 20 studies whose confidence the authors grade as low. The power position rests on a qualitative longitudinal study of two settings, which can name the mechanism but cannot size an effect. Until a trial reports all four, the designer should report the intervention against a standard, naming who decided, when, and with what power, and should claim the relevance, usability and acceptability gains that the reviews do support while treating any clinical effect as unproven. The strongest objection to the unit's thesis is that the reviews pool heterogeneous interventions under a term only three of twenty studies used, so the null result may measure a label rather than a method.

Where the argument is heading

Named work points one way, from definition to reporting. Vargas and colleagues (2022) try to settle the vocabulary, making co-creation the overarching principle and locating the distinctions in who decides and when. The 2024 reviews and the 2025 meta-analysis push the same way from the evidential side. Greene Barker and colleagues (2024) call for uniform terminology and measurement instruments, and Sumner and colleagues (2024) for standardised terminology and a stronger commitment to clinical evaluation, having found that none of 17 co-designed remote patient monitoring projects evaluated efficacy. Farr (2018) supplies the direction the measurement may have to take: participation judged by the power it moves rather than the outcomes it is credited with. The likely next stage is a reporting standard for co-design and better-powered trials.

The sources this unit rests on

• Sanders, E. B.-N. and Stappers, P. J. (2008). Co-creation and the new landscapes of design. CoDesign. doi:10.1080/15710880701875068 Sanders and Stappers define co-creation as collective creativity and co-design as that creativity applied across the whole design process, so co-design is the narrower term. It is a manifesto, not an evaluation.
• Steen, M., Manschot, M. and De Koning, N. (2011). Benefits of co-design in service design projects. International Journal of Design. https://www.ijdesign.org/index.php/IJDesign/article/view/890 Steen, Manschot and De Koning analyse three service design projects to identify co-design's benefits, finding gains in idea generation and in fit between service and context. It is a small multiple-case study.
• Mattelmaki, T. and Sleeswijk Visser, F. (2011). Lost in Co-X: interpretations of co-design and co-creation. Proceedings of IASDR 2011, Delft. https://research.tudelft.nl/en/publications/lost-in-co-x-interpretations-of-co-design-and-co-creation/ Mattelmaki and Sleeswijk Visser review the literature and discuss cases to clarify the relationship between co-design and co-creation, finding the co-terms tangled across academic, professional and organisational use. It measures no intervention.
• Brandt, E., Binder, T. and Sanders, E. B.-N. (2012). Tools and techniques: ways to engage telling, making and enacting. In Routledge International Handbook of Participatory Design. Routledge. doi:10.4324/9780203108543.ch7 Brandt, Binder and Sanders survey participatory design's proliferating tools and techniques, organised under telling, making and enacting: a richness of toolboxes rather than a canon of methods. A handbook chapter.
• Trischler, J., Pervan, S. J., Kelly, S. J. and Scott, D. R. (2018). The value of codesign: the effect of customer involvement in service design teams. Journal of Service Research. doi:10.1177/1094670517714060 Trischler and colleagues compared co-design, in-house professional and user-only teams producing concepts for a library service. Co-design concepts scored higher on user benefit and novelty, lower on feasibility, and services in use were untested.
• Robert, G. (2013). Participatory action research: using experience-based co-design to improve the quality of healthcare services. In Understanding and Using Health Experiences. Oxford University Press. doi:10.1093/acprof:oso/9780199665372.003.0014 Robert describes experience-based co-design as participatory action research in which patients and staff are equal partners, and recounts its spread since a 2006 pilot. It states that evaluations found benefits without reporting them.
• Slattery, P., Saeri, A. K. and Bragge, P. (2020). Research co-design in health: a rapid overview of reviews. Health Research Policy and Systems. doi:10.1186/s12961-020-0528-9 Slattery, Saeri and Bragge reviewed 23 reviews of research co-design in health, finding it widely used but seldom evaluated, with effectiveness rarely tested empirically. A rapid overview with few outcome data.
• Vargas, C., Whelan, J., Brimblecombe, J. and Allender, S. (2022). Co-creation, co-design, co-production for public health: a perspective on definitions and distinctions. Public Health Research and Practice. doi:10.17061/phrp3222211 Vargas and colleagues distinguish co-creation, co-design and co-production for public health, noting the terms are used interchangeably and that this obscures differences in who decides what. A position piece, with no data.
• Farr, M. (2018). Power dynamics and collaborative mechanisms in co-production and co-design processes. Critical Social Policy. doi:10.1177/0261018317747444 Farr followed co-design and co-production in two public-service settings and argues the mechanism at work is power: collaborative techniques can challenge existing power relations only with constant critical reflection. It is a qualitative study of two settings. Only two settings.
• Greene Barker, T., O'Higgins, A., Fonagy, P. and Gardner, F. (2024). A systematic review and meta-analysis of the effectiveness of co-designed, in-person, mental health interventions for reducing anxiety and depression symptoms. Journal of Affective Disorders. doi:10.1016/j.jad.2023.12.080 Greene Barker and colleagues meta-analysed 20 studies of in-person co-designed mental health interventions, finding a moderate but non-significant effect on depression and a small one on anxiety; only three used the term co-design.
• Sumner, J., Tan, S. Y., Wang, Y., Keck, C. H. S., Lee, E. W. X., Chew, E. H. H. and Yip, A. W. (2024). Co-designing remote patient monitoring technologies for inpatients: systematic review. Journal of Medical Internet Research. doi:10.2196/58144 Sumner and colleagues screened 3,334 reports and found 17 co-designed remote patient monitoring projects, none evaluating the efficacy of its own technology, reporting usability and feasibility instead. It measures no efficacy.
• Sumner, J., Ng, C. W. T., Teo, K. E. L., Peh, A. L. T. and Lim, Y. W. (2024). Co-designing care for multimorbidity: a systematic review. BMC Medicine. doi:10.1186/s12916-024-03263-9 Sumner and colleagues screened 14,376 reports on co-designed care for multimorbidity, found 13 studies, and only two reporting health and well-being outcomes. A systematic review.
• Anieto, E. M., Dall, P. M., Abaraogu, U. and colleagues (2025). The effectiveness of co-created lifestyle interventions in improving health behaviour, physical and mental health in adults with non-communicable diseases: a systematic review with meta-analysis. Public Health. doi:10.1016/j.puhe.2025.105929 Anieto and colleagues pooled sixteen randomised trials and found moderate-quality evidence of a moderate effect on health behaviour at under six months, shrinking later. It is the unit's first pooled trial evidence. Trials are heterogeneous.