Article Text

Download PDFPDF

Cost-effectiveness analysis of human-centred design for global health interventions: a quantitative framework
  1. Chen Liu1,
  2. Jae Hyoung Lee2,
  3. Amanda J Gupta3,4,5,
  4. Austin Tucker6,
  5. Chris Larkin7,
  6. Patricia Turimumahoro4,
  7. Achilles Katamba4,8,
  8. J Lucian Davis4,5,9,10,
  9. David Dowdy2,4,11
  1. 1Department of Division of Pulmonary & Critical Care Medicine, Johns Hopkins School of Medicine, Baltimore, Maryland, USA
  2. 2Department of Epidemiology, Johns Hopkins University Bloomberg School of Public Health, Baltimore, Maryland, USA
  3. 3Department of Health Equity and Social Justice, Johns Hopkins School of Public Health, Baltimore, Maryland, USA
  4. 4Uganda Tuberculosis Implementation Research Consortium, Makerere University, Kampala, Central, Uganda
  5. 5Department of Epidemiology of Microbial Diseases, Yale University School of Public Health, New Haven, Connecticut, USA
  6. 6Department of Population Health Sciences, Harvard University T H Chan School of Public Health, Boston, Massachusetts, USA
  7., Palo Alto, California, USA
  8. 8Clinical Epidemiology Unit, Makerere University, Kampala, Uganda
  9. 9Department of Pulmonary, Critical Care and Sleep Medicine, Yale School of Medicine, New Haven, Connecticut, USA
  10. 10Center for Methods in Implementation and Prevention Science, Yale School of Public Health, New Haven, Connecticut, USA
  11. 11Department of Medicine, Johns Hopkins Medicine, Baltimore, Maryland, USA
  1. Correspondence to Dr Jae Hyoung Lee; jlee736{at}


Introduction Human-centred design (HCD) is a problem-solving approach that is increasingly used to develop new global health interventions. However, there is often a large initial cost associated with HCD, and global health decision-makers would benefit from an improved understanding of the cost-effectiveness of HCD, particularly the trade-offs between the up-front costs of design and the long-term costs of delivering health interventions.

Methods We developed a quantitative framework from a health systems perspective to illustrate the conditions under which HCD-informed interventions are likely to be cost-effective, taking into consideration five elements: cost of HCD, per-client intervention cost, anticipated number of clients reached, anticipated incremental per-client health benefit (ie, disability-adjusted life years (DALYs) averted) and willingness-to-pay. We evaluated several combinations of fixed and implementation cost scenarios based on the estimated costs of an HCD-informed approach to tuberculosis (TB) contact investigation in Uganda over a 2-year period to illustrate the use of this framework.

Results The cost-effectiveness of HCD-informed TB contact investigation in Uganda was estimated to vary from US$8400 (2400 clients reached, lower HCD cost estimate) to US$306 000 per DALY averted (120 clients reached, baseline HCD cost estimate). In our model, cost-effectiveness was improved further when the interventions were expected to have wider reach or higher per-client health benefits.

Conclusion HCD can be cost-effective when used to inform interventions that are anticipated to reach a large number of clients, or in which the cost of HCD is smaller relative to the cost of delivering the intervention itself.

  • tuberculosis
  • health economics
  • health policy

Data availability statement

Data are available in a public, open access repository. The full dataset and statistical code are available in a public Github repository. DOI: 10.5281/zenodo.5854794.

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See:

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • An increasing number of global health interventions involve human-centred design (HCD), a creative process that often involves large initial costs for technical experts.

  • The cost-effectiveness of HCD-informed global health interventions is largely unknown.


  • The authors developed a quantitative framework to illustrate the conditions under which HCD interventions would likely be cost-effective.

  • This framework was applied to an HCD-informed tuberculosis contact investigation intervention in Kampala, Uganda.

  • The cost-effectiveness of HCD-informed interventions is optimised by minimising up-front costs of HCD, increasing the number of clients reached, and increasing the per-client health benefit.


  • HCD-informed global health interventions can be cost-effective if they reach enough clients and provide enough benefit to offset the up-front costs.

  • Decision-makers can use this simple framework to obtain initial estimates of cost-effectiveness when considering whether to employ an HCD-informed approach to a global health intervention.


The past two decades have seen unprecedented investment in research and development of new health innovations for low and middle-income countries (LMICs). However, very little of this investment has been directed towards design and delivery innovations.1 One example of this discrepancy is in the field of tuberculosis (TB) research, where less than 13% of TB research funding is allocated to operational research (or epidemiology).2 The predominant biomedical research model often implicitly assumes that interventions proven to work in certain settings will be transportable to others. Unfortunately, this assumption often does not hold—and for a given intervention to be effective in a new setting, it must be tailored to the specific context.3 One potential approach to this problem is human-centred design (HCD), a creative process aimed at improving efficiency and facilitating innovation of a product or intervention by putting people at the centre of the design focus.4 HCD allows one to gain insight into the needs of the beneficiaries, create innovative approaches to meet these needs and deliver solutions tailored to specific socioeconomic contexts.5 Early definitions of HCD “presumed an intended and predetermined use for each item/service as well as a static ‘user’ envisioned by the designer/engineer”.6 This definition has evolved to emphasise the ‘understanding of needs, desires and experiences’ of users and the techniques needed to reach this understanding through communication, interaction, empathy, and stimulation.7

While HCD was first implemented in industrial design and engineering, it has been increasingly used in the global health context, especially given the strong focus of HCD on people and their communities and its emphasis on human interactions from design to implementation.7 8 Examples of HCD-informed contributions to global health range from smaller ‘incremental’ innovations to bolster existing interventions, such as the development of a marketing strategy for Nigerian health insurance,9 to larger ‘disruptive’ innovations that result in completely new interventions such as the design of new services in Zambia to decrease unplanned pregnancy among girls aged 15 to 19.10

As the role of HCD in global health expands, a major consideration is whether HCD can be cost-effective in resource-limited contexts. Alternatives to HCD exist, including implementing without a specific strategy (which entails no additional costs) and tailoring interventions to address determinants of practice based on behavioural theory (which may be less expensive but more time-consuming than HCD).11 12 Until local capacity can be more broadly built, HCD often involves sourcing technical guidance and expertise through partnerships with design firms based in high-income countries, resulting in large up-front costs. The return on this investment (in terms of health benefit) varies, so decision-makers must generally assess whether an HCD-informed intervention would be cost-effective before initiating the HCD process. We, therefore, sought to develop a quantitative framework to help answer the question, ‘What level of anticipated incremental health benefit is needed to justify investment in an HCD-based global health intervention?’


Patient involvement

As part of the HCD process, the parent study engaged patients with TB, community members, healthcare workers, and implementing partners in the development and refinement of the proposed TB contact investigation implementation strategy.


Our primary objective was to identify the conditions under which HCD can be cost-effective, when applied to health interventions in LMICs. To accomplish this objective, we first defined a range of reasonable HCD costs in consultation with the non-profit design studio The costs included fixed costs from the design phase and variable implementation costs that comprised of personnel, travel, and equipment costs. Using these costs, we then estimated the incremental cost-effectiveness of a hypothetical HCD-informed intervention using an analytical framework with five elements: (1) cost of HCD, (2) incremental per-client intervention cost (HCD-informed minus non-HCD-informed), (3) anticipated number of clients reached, (4) anticipated incremental per-client health benefit (ie, DALYs averted) and (5) willingness-to-pay.13

Motivating example

To provide a benchmark example, we projected the potential cost-effectiveness of an HCD-informed intervention for TB contact investigation in Kampala, Uganda.14 This intervention targeted urban and peri-urban communities surrounding the health facilities previously identified based on their geographic proximity to the Uganda-based implementing partner, Uganda Tuberculosis Implementation Research Consortium (U-TIRC). The target population included household and non-household contacts of people newly diagnosed with pulmonary TB in these clinics.

As envisioned by the design partner, the HCD process consists of three phases: inspiration, ideation and implementation (figure 1). For this intervention, the inspiration phase involved understanding community beliefs and priorities through interviews and focus group discussions. The ideation phase involved generating new ideas through brainstorming sessions, rough prototyping and the testing and refining of these prototypes through participant feedback. The inspiration and ideation phases of the HCD process resulted in development of the ‘Tuli Wamu Nawe’ (Luganda for ‘We are with you’) strategy. This strategy emphasises a team-based approach, in which community health workers are supported by ‘health riders’, who transport health workers to patient homes and sputum specimens to clinics. This strategy also includes printed informational materials to help community health workers to educate patients diagnosed with TB and to help patients educate their household and close contacts. The Tuli Wamu Nawe strategy is now in the implementation phase and undergoing evaluation in a stepped wedge cluster randomised trial. Both implementation and effectiveness outcomes are under evaluation, including the incremental proportion of contacts completing screening and testing, and the incremental number of contacts diagnosed with microbiologically confirmed TB.

Figure 1

Overview of HCD-informed design process for TB contact investigation. The process of creating the HCD-informed solution for TB contact investigation in Kampala, Uganda, took place over 20 weeks. During these 20 weeks, the designers and healthcare professionals completed all three phases of HCD from inspiration through ideation and implementation, with the aim to build empathy with the target community and then to propose, create, test and refine solutions rooted in people’s actual needs (8). Activities within each phase are listed at the bottom of the figure. After 20 weeks of creating, testing, and iterating, the HCD-informed intervention was ready for implementation. Figure courtesy of HCD, human-centred design; TB, tuberculosis.

Cost-effectiveness of HCD-informed TB contact investigation

In deciding whether to pursue HCD, expenditure data on the (future) intervention are rarely available. Thus, to better replicate the decision-making process, we assume availability of budget estimates (rather than expenditure data) and take a health systems perspective. To estimate costs in our motivating example, we sourced itemised budgets from both and U-TIRC. In doing so, we assumed that the incremental cost of the HCD-informed intervention (relative to the standard of care) includes both an up-front cost associated with the inspiration and ideation phases (ie, costs that must be incurred before any clients benefit from the potential intervention) and an ongoing cost associated with the implementation phase (ie, personnel, travel, and equipment costs). For purposes of providing a transparent decision-making framework, we assumed that the inspiration/ideation phase cost (ie, cost of design) is fixed, whereas the implementation phase cost scales linearly with the number of clients reached.

We compared the budgets from this project to actual expenditures for other HCD-informed interventions and for prior contact investigation activities in Uganda to confirm that cost estimates were reasonable. All costs were converted and inflated to 2020 US dollars using the World Bank gross domestic product (GDP) deflator for the USA at 3600 Ugandan shillings to 1 US dollar.15 Future costs were discounted at 3% per year, as recommended by the US Panel on Cost-Effectiveness in Health and Medicine.16

To estimate corresponding effectiveness (ie, health benefit), we first estimated the number of TB cases that might be identified through contact investigation, using data from a meta-analysis of household contact investigation.17 18 Given that contact investigation may still occur without the HCD process, we considered two scenarios: one in which the HCD-informed intervention was ‘incremental’ (ie, without HCD, contact investigation would still be performed, but less efficiently—defined here as diagnosing 40% of all household cases), and one in which this intervention was ‘disruptive’ (ie, without HCD, no contact investigation would be performed). Based on projections from the ongoing trial, we assumed that this ‘disruptive’ intervention could reach 720 household contacts—regardless of underlying TB status—if implemented across six clinics for 2 years. We then used a published transmission model calibrated to South Africa to estimate the corresponding cost per disability-adjusted life year (DALY) averted based on the cost per case detected, over a 2-year time horizon.19 We also explored potential economies of scale from reaching larger client volumes.

Generalisable framework for estimating cost-effectiveness of HCD

Using our estimates above as a benchmark, we used sensitivity analyses to create a more general framework to estimate the incremental cost-effectiveness of a generic HCD-informed intervention. We considered three scenarios for the up-front cost of the HCD process: equal to the projected cost of HCD in the TB contact investigation intervention above, 50% lower and 50% higher. We also considered three scenarios for the incremental per-client cost of implementing the HCD-informed component of the intervention (US$0.10, US$1 and US$100), to illustrate the wide range of potential costs that might be anticipated across different types of global health interventions. Using these estimates of incremental HCD cost, we then plotted the projected incremental cost-effectiveness of HCD (cost per DALY averted) according to the number of clients projected to be reached and the incremental number of DALYs averted per client (ie, DALYs averted with HCD minus DALYs averted without HCD). These estimates are then compared with different willingness-to-pay thresholds to provide decision-makers with an initial estimate of whether undertaking an HCD process would likely be considered cost-effective under prevailing cost-effectiveness thresholds. In making this comparison, we assumed that cost-effectiveness could be evaluated based on society’s willingness-to-pay for health utility (ie, DALYs averted), and that incremental cost-effectiveness thresholds could be used as a proxy for willingness-to-pay thresholds.


Based on detailed budgetary review, we estimated the up-front cost of the HCD process for TB contact investigation as US$356 000, including design, creation, and testing of contact investigation over a 20-week period from inspiration to implementation. We estimated the corresponding incremental cost of implementing the HCD-informed component of the intervention (eg, printing materials, supporting ‘health riders’) as US$0.41 per client (household contact) reached. Assuming a baseline prevalence of 3.1% of active TB in household contacts,17 we estimated that HCD-informed household contact investigation could detect 22.3 cases of active TB across six clinics over 2 years (table 1).

Table 1

Estimated number of active TB cases detected and cost per DALY averted for HCD-informed TB contact investigation in Uganda

Assuming a low (2.2%) and high (4.4%) prevalence of active TB in household contacts resulted in detecting 15.8 and 32.7 cases, respectively. Assuming that, without HCD, contact investigation would not occur (ie, HCD as ‘disruptive’), we estimated the incremental cost-effectiveness of the HCD-informed intervention as US$17 700 per case detected, or US$51 200 per DALY averted with the baseline prevalence of active TB. The incremental cost-effectiveness in low and high prevalence settings was estimated as US$24 900 and US$12 400 per case detected or US$72 100 and US$36 100 per DALY averted, respectively. Cost-effectiveness estimates were less favourable if we assume that contact investigation would still occur in the absence of HCD, with similar costs but lower efficiency (ie, HCD as ‘incremental’).

The incremental cost-effectiveness of an HCD-informed intervention improves as the number of people reached by the intervention increases and the cost of the HCD process decreases (figure 2). For example, in the case of HCD-informed TB contact investigation in Uganda, the incremental cost-effectiveness of the intervention was estimated at US$306 900 per DALY averted if only 120 contacts could be reached, but it improved to US$8400 per DALY averted if 2400 contacts were reached and the up-front costs of the HCD process were cut in half. Similarly, if more DALYs could be averted per client reached, the estimated cost-effectiveness of the HCD process was projected to improve (figure 3). For example, in the primary scenario for HCD-informed TB contact investigation, we estimated that 0.011 DALYs would be averted per contact reached. For an intervention with similar cost that was instead able to avert 0.1 DALYs per client reached, incremental cost-effectiveness would fall from US$51 200 to US$5630 per DALY averted (figure 3, red dot to white dot).

Figure 2

Incremental cost-effectiveness of a human-centred design (HCD)-informed tuberculosis contact investigation intervention, according to number of clients reached. The x-axis shows variation in the incremental number of clients reached as a result of implementing the strategy that emerged from the HCD process. The y-axis shows the estimated incremental cost-effectiveness ratio (ICER, measured in cost per disability-adjusted life year (DALY) averted), assuming an up-front HCD cost of US$356 000 (orange line, as estimated in the project itself) or US$178 000 (green line, 50% HCD cost), plus US$0.41 per contact reached. Labels indicate ICER estimates for 120, 720 (point estimate), and 2400 incremental contacts reached. As the number of clients reached increases, cost-effectiveness estimates become more favourable.

Figure 3

Incremental cost-effectiveness of a human-centred design (HCD)-informed tuberculosis contact investigation intervention, according to number of clients reached and disability-adjusted life years (DALYs) averted per contact reached. Contours show thresholds of the incremental cost-effectiveness ratio (ICER, measured as cost per (DALY) averted) of an HCD-informed intervention costing US$358 000 up-front and US$0.41 per client reached, as a function of number of clients reached (x-axis) and DALYs averted per client reached (y-axis). The primary estimates corresponding to estimates in table 1 and figure 2 (720 contacts reached, 0.011 DALYs averted per contact reached) are shown by the red dot in the lower left. The white dot illustrates the incremental cost-effectiveness (US$5630/DALY) of a similar intervention that could avert 0.1 DALYs per client reached.

Figure 4 presents estimates of incremental cost-effectiveness of HCD at three different estimates of up-front HCD costs (US$178,000; US$356,000 and US$534,000) and incremental HCD costs per client reached (US$0.10, US$1, US$20, and US$100), under different assumptions regarding the number of clients reached and the incremental number of DALYs averted per client reached. For example, assuming an HCD-informed intervention for which an incremental 2000 clients were estimated to be reached and an incremental 0.05 DALYs were estimated to be averted per client reached, estimated cost-effectiveness of the HCD process was US$1800 per DALY averted assuming US$178 000 up-front and US$1 per client reached, versus US$3562 per DALY averted assuming US$356 000 up-front and US$0.10 per client reached (table 2). For this specific scenario, table 2 presents estimates of the size of the programme and/or DALYs averted per client that would need to be achieved to achieve cost-effectiveness at thresholds of US$500 or US$1000 per DALY averted.

Table 2

Estimated incremental cost-effectiveness of a representative HCD-informed intervention

Figure 4

Estimated cost-effectiveness of human-centred design (HCD)-informed global health interventions. Contours represent the estimated incremental cost-effectiveness of an HCD-informed health intervention, according to the number of clients reached (x-axis, in thousands) and disability-adjusted life years (DALYs) averted per client reached (y-axis). The first row (‘Low HCD Cost’) assumes an up-front HCD cost of US$178 000 (in 2020 US dollars); the second (‘Mid HCD Cost’) assumes US$356,000; and the third (‘High HCD Cost’) assumes US$534 000. The first column (‘Low Intervention Cost’) assumes an incremental HCD cost of US$0.10 per client reached; the second (‘Mid Intervention Cost’) assumes US$1; and the third (‘High Intervention Cost’) assumes US$100.


In this study, we developed a framework to assess the cost-effectiveness of HCD-informed health interventions in high-burden countries. This framework illustrates the importance of considering the up-front cost of the HCD process in the context of the overall intended scope of the programme (number of clients reached), magnitude of health benefits (DALYs averted per client), ongoing incremental cost of the HCD-informed intervention per client reached, and willingness-to-pay. We illustrate the application of this framework to an HCD-informed intervention for TB contact investigation in Uganda, demonstrating that increasing the volume of clients reached and decreasing the up-front cost of HCD can improve cost-effectiveness estimates by a factor of more than 10. This framework can be useful to decision-makers in LMICs who must consider whether the anticipated benefits of HCD can justify the often large up-front costs.

To date, most cost-effectiveness analyses of global health interventions have tended to ignore the substantial up-front costs that are often required to design interventions in a manner that is responsive to local priorities. When such design is not required (eg, a standardised intervention already exists), cost-effectiveness thresholds may be easier to meet. For example, a trial of universal HIV testing and treatment in Zambia estimated that the intervention would need to avert 0.019 DALY per person reached to be cost-effective.20 By contrast, our framework suggests that, if such an intervention would require a design process costing US$178 000 up-front plus US$0.10 per client, nearly 8000 clients would need to be reached in order for an intervention of similar effectiveness to be cost-effective at a threshold of US$1000 per DALY averted (figure 4, upper left). Thus, ignoring the importance of the design process can result in interventions that do not achieve the intended health impacts, but ignoring the cost of the design process can result in overly optimistic estimates of cost-effectiveness. Future economic evaluations of global health interventions should consider the degree to which HCD (or similar design processes) would be required to make the intervention effective in the context of local priorities—and if so, should formally consider the cost of the design process as part of the intervention’s cost-effectiveness.

An important finding from this analysis was that, in estimating the cost-effectiveness of the HCD process, the up-front cost of HCD tended to be a more important consideration than the incremental per-client cost. For example, in table 2, there was minimal difference between interventions for which the incremental cost of HCD per client was estimated to be US$0.10 versus US$1.00, but substantial differences existed when varying the up-front cost by±50%. In general, as seen in figure 4, an HCD process costing US$178 000 up-front could be justified at a cost-effectiveness threshold of US$1000 per DALY averted if a sufficient number of clients could be reached—whereas a process costing US$534 000 could rarely, if ever, be justified at this threshold. These findings argue for the importance of innovations to reduce the up-front cost of HCD for global health interventions, while retaining their quality.

Another important consideration when evaluating the cost-effectiveness of HCD-informed global health interventions is the local willingness-to-pay for health. This threshold may vary widely across contexts and perspectives, whereas GDP per capita was traditionally used as a standard threshold for willingness-to-pay for one DALY averted,21 it has been more recently argued that decision-makers should consider either higher thresholds, based on individuals’ valuation of a year of statistical life,22 or lower thresholds based on revealed willingness-to-pay.21 The sensitivity of the willingness-to-pay threshold to the assessment of HCD cost-effectiveness speaks to the importance of defining this threshold in each local decision-making context.

Fuge et al highlighted the diversity in HCD methods in a pattern analysis that spanned 809 HCD case studies. These authors found that 87% of interventions used ‘hear methods’ (inspiration phase) compared with ‘create and deliver methods’ (ideation and implementation phases).23 These authors also compared projects involving industrial designers or engineers within and projects using designers.23 They found that designers tend to have much higher usage in the initial ‘hear stage’ (inspiration phase), with greater preference for methods that involve end users at an early stage of the design process. Future cost-effectiveness analysis could consider comparing HCD-informed interventions applied by laypeople versus professional design teams.

Our streamlined analytical framework makes a number of simplifying assumptions. First, we assume that the up-front and per-client costs of HCD (as well as the number of clients reached and DALYs averted per client) can be estimated separately and also in incremental fashion relative to a non-HCD standard of care. This challenge was also noted in another HCD intervention carried out in Nigeria, Ethiopia and Tanzania, where researchers and designers faced challenges in measuring the total design cost and isolating the cost of HCD.24 In such cases, this framework may still be useful to decision-makers in highlighting the costs for which assumptions must be made in order to estimate the incremental cost-effectiveness of HCD. Our primary analysis is also anchored on a single emblematic intervention and thus cannot account for the full spectrum of on-the-ground realities of HCD implementation. We intend to validate this framework using data from an ongoing trial of the intervention in our motivating example; future refinement could also explore alternative HCD methodologies as well as applications of HCD to other interventions, comparing those interventions to the framework presented here. Finally, this framework simplifies a wide array of ongoing costs. Quality control, refresher training, monitoring and evaluation are considered as ‘per-client’ implementation costs when, in reality, many of these costs scale non-linearly with the number of clients served. As such, this simplified framework should only be used for initial planning and should not be interpreted as a full cost-effectiveness analysis of any specific HCD-informed intervention.

In summary, even though global health is a common focus area in HCD research,25 a framework for estimating the cost-effectiveness of HCD-informed interventions in global health has been lacking. This research provides a transparent, quantitative framework, whereby global health decision-makers can obtain ‘first-pass’ estimates as to whether an HCD process is likely to be cost-effective, based on a small number of inputs. This framework illustrates the importance of considering the up-front cost of HCD in any economic evaluation of an HCD-informed intervention, and also of future work to make HCD available at lower cost (yet high fidelity) in the global health context.

Data availability statement

Data are available in a public, open access repository. The full dataset and statistical code are available in a public Github repository. DOI: 10.5281/zenodo.5854794.

Ethics statements

Patient consent for publication

Ethics approval

This study does not involve human participants.



  • CL and JHL are joint first authors.

  • Handling editor Seye Abimbola

  • Twitter @trishkamu

  • CL, JHL, PT, JLD and DD contributed equally.

  • Contributors CLi and DD conceived and developed the framework. CLi and JHL carried out the analysis. CLi, JHL, and DD took the lead in writing the manuscript. JLD, PT, AG, AT, and AK implemented the motivating example research project this evaluation is based on. CLa was in charge of the HCD process and provided insight on HCD for the manuscript. DD takes responsibility for the overall content as the guarantor. All authors provided critical feedback and helped shape the research, analysis and manuscript.

  • Funding NIH 5R01AI104824, PI J. Lucian Davis

  • Competing interests Chris Larkin is an employee of, a human-centered design firm. Neither nor the National Institutes of Health had any input as to the analysis or the decision to publish.

  • Patient and public involvement Patients and/or the public were involved in the design, or conduct, or reporting, or dissemination plans of this research. Refer to the Methods section for further details.

  • Provenance and peer review Not commissioned; externally peer reviewed.