THE TSWAKA STUDY the Tswaka Stdy a Journey Into an Innovatve Public–Private–Private Researc Partnersip

newsnews24 THE TSWAKA STUDY The Tswaka Stdy A journey into an innovatve public–private–private researc partnersip

Saskia JM Osendarp, Dominic Schofeld Global Alliance for Improved Nutrition (GAIN), > This report summarizes the key lessons learned and Geneva, Switzerland discusses what it takes to make such partnerships work. The key lessons appear on pp. 28–29. Maaike J Bruins DSM Nutritional Products, Basel, Switzerland This is a report on the learnings gathered by a unique partner- Leon GJ Frenken ship, consisting of two private-sector partners, one NGO and an Unilever R&D, Vlaardingen, the Netherlands academic partner, established to perform a complex study: the Tswaka Nutrition Intervention study. Jane Badham The study was a community-based randomized controlled JB Consultancy, Johannesburg, South Afica trial carried out in North West province, South Afica, on the efcacy of two small-quantity lipid-based complementary food Klaus Kraemer supplements provided to children aged 6–12 months. The Tswa- Sight and Life Foundation, Basel, Switzerland ka study – which means ‘mixing’ in the local Setswana language, referring to the fact that the supplement should be mixed into Cornelius M Smuts home-cooked food – was completed in 2015, and its results have Centre of Excellence for Nutrition, recently been published.1 North-West University, Potchefstroom, South Afica The journey embarked on in 2010 presented many challenges before it was successflly completed following eight years of efort. This report summarizes the key lessons learned and dis- Key messages cusses what it takes to make such partnerships work. We hope that our experiences and insights will be of value to others en- > This is a report on the learnings gathered by a partnership tering into partnerships in the nutrition space. In the end, our established to perform a complex study: the Tswaka Nutri- joint success came down to the three Ps: Personal relationships tion Intervention study. and trust, Perseverance and determination, and the Passion to make this project a success. > The study was a community-based randomized controlled trial carried out in North West province, South Africa, on the Inception: aligning motivations efcacy of two small-quantity lipid-based complementary to create a single concept food supplements provided to children aged 6–12 months. In 2010, an NGO (GAIN) and a private-sector partner (Unilever) joined forces as a Funding Consortium to facilitate a research > The Tswaka study was completed in 2015 and its results project investigating a novel fortifed complementary food sup- have been recently published.1 plement. For the detailed design and execution of the intervention study, it was decided that an Academic Research Partner would be selected via a call for proposals. SIGHT AND LIFE

V OL.

33(1)

2019 One of the children participating intheTswakaOne ofthechildren participating Studyhasherweight taken as part of the bimonthly anthropometric measurement process ofthebimonthlypart anthropometric 25

The motivation for the study was diferent for each of the research team. Therefore, two contracts had to be aligned and partners in the Funding Consortium. While GAIN was interested agreed upon: one bilateral grant agreement between GAIN and in working with private-sector partners on product formulation the research team, and one trilateral agreement between GAIN and demonstration of the feasibility of a market-based approach, and the two private-sector partners (the Funding Consortium), Unilever was interested in testing the efcacy of such a product who each contributed one-third of the research costs. before considering an innovation project on the feasibility of a Issues concerning intellectual property (IP) rights for data market-based approach. and for products, as well as publication and dissemination Unilever wanted to collaborate with GAIN on this research clearance procedures, had to be discussed at senior manage- because GAIN was considered a trusted intermediary between ment level in all the partner organizations. The entire contract Unilever and the Academic Research Partner. Initial discussions negotiations took more than a year and were eventually con- between Unilever and GAIN, however, identifed a potential con- cluded in 2011. fict of interest if GAIN were to work on a study assessing the efcacy of a product fom just one private-sector partner. Preparation phase: managing risks and building trust To address this issue, a second private partner, DSM, was The preparation phase turned out to be the most delicate phase invited to join the partnership. DSM wished to demonstrate ef- of the entire process. There were no products ready yet, and fcacy for a similar product formulation that contained certain negotiations started between the two private-sector partners extra ingredients with additional benefts for growth and devel- with the aim of creating a single joint development process for opment. Despite their diferent motivations, the three partners the study products. A joint development process was thought to recognized the benefts of sharing their complementary exper- be benefcial because this would allow for blinding of the two tise as well as sharing the research costs within the famework intervention arms by having two products comparable in taste of an NGO/private-partner Funding Consortium. and appearance, albeit with diferent compositions. As product To align the diferent motivations into one achievable goal development and product ingredient formulations are among a acceptable to all three partners took some time. There were company’s most valued IP, it took some time for the two pri- discussions about the requisite levels of transparency and vate partners to agree on a joint development agreement (JDA). openness and about the dynamics driving the choices and de- When an issue occurred during the actual production of one of cisions required for a successfl partnership. To avoid potential the study products, however, the private-sector partners decid- complications and to keep the study to a manageable size and ed to terminate the JDA and to continue frther scale-up and acceptable cost levels, it was designed in such a way that the production independently of one another. sample size would not allow direct comparison between the two In March 2013, the acceptability study among mothers and products. children with the frst product samples started in South Afica. The acceptability study yielded positive results but also revealed “The more organizatons you have to some mild side efects, assessed as being unrelated to the intervention products as such.3 Meanwhile, however, within the involve, the more important it is to private-sector partners, changes in leadership and company have regular, open communicaton”2 strategies had taken place, leading to a revised assessment of the potential reputational risk for all partners involved in this community-based trial among very young and vulnerable chil- While these initial discussions were still ongoing, and prior dren in a disadvantaged population. to DSM joining the partnership, GAIN and Unilever had already To better manage and control these risks, it was initially developed a call for proposals based on an initial concept note, decided to hire an independent monitor to remotely review and an independent expert committee was formed to select the the quality of the study execution. In addition, a Safety Moni- Academic Research Partner to perform the study. toring Board (SMB) was established to monitor adverse efects North-West University, South Afica, won the bid in partner- and identif any potential causal relationships to study par- ship with the South Afican Medical Research Council (‘the re- ticipation or test product. In addition, an independent study search team’) and started to work on a fll proposal, which had physician was hired to train and support the research team as to be adapted into a three-arm design when the second private and when needed. The study monitoring was frther upscaled partner, DSM, joined the Funding Consortium. by hiring an independent Contract Research Organization Simultaneously, contract negotiations had started. The idea (CRO) to ensure the quality and safety of the study by mon- was that GAIN would act as the ofcial study sponsor, and also itoring the implementation of the study on site and for data as an intermediary between the private-sector sponsors and the management on site. SIGHT AND LIFE | VOL. 33(1) | 2019 THE TSWAKA STUDY 27

There was a clear need to balance the risk exposure of all made in keeping the momentum fom the site visit with the parties involved. Whereas the private partners wished to imple- community and the feld workers. To keep the onsite research ment processes such as monitoring, a centralized database and team motivated, extra training opportunities were ofered by extensive medical supervision and reporting, the research team the fnding partners. wished to make sure that its credibility as an independent academic team would not be jeopardized. “The fac that there was a signed Having an intermediate party in the form of GAIN turned out to be critical, ensuring efective communication between the obligaton beteen the partners research team and the private-sector partners and coordinat- meant we had to move forward and ing negotiations about the changing needs of both the research team and the Funding Consortium. could not withdraw” In addition, a site visit by the Funding Consortium helped to provide a more realistic view of the circumstances at the study site and to improve understanding of the practical dilemmas The renewed confdence within the partnership allowed for the research team faced in day-to-day management of a com- more open discussions on the issues of sponsor involvement munity-driven study of this size in this setting. It also frther and authorship. The dilemma was whether the growing in- strengthened trust between the responsible individuals within volvement of the study sponsors in the various stages of study each of the partner organizations. Due to these personal rela- implementation needed to be made transparent in the publica- tionships and a shared commitment to succeed, it was possible tion of the study results at the level of coauthorship. For the re- to commence data collection in the fall of 2013. search team, sponsor authorship was considered to reduce the independence of the individual scientists involved, and there Data collection: trust and transparency was a realistic fear that it would also afect the credibility of While scaling-up of the production of the study product and the study results.4,5 negotiations on the involvement of the CRO were still ongoing, There did not seem to be a perfect solution. When industry research staf had already been hired, and mobilization in the sponsors are not included as coauthors, this can be construed community for recruitment had already taken place. By the time as an attempt to disguise their involvement. When they are in- the research team was in a position to start enrolling infants cluded as authors, however, this may raise questions as to their into the study, the children of mothers initially mobilized had infuence on the interpretation of fndings. In recent years, there aged over the 6 months (+ 2 weeks) enrollment criterion, with has been increased recognition that stricter rules may help safe- the result that they had to be excluded fom participation. This guard the independence and credibility of public–private re- caused distrust within the community, and the research team search partnerships.5 In the current partnership, this issue was had to step up its eforts to restore confdence in the study. At addressed by all partners agreeing on a joint author and sponsor the same time, the CRO, which was familiar with Phase III phar- contribution statement, according to the international criteria ma studies, had to get used to the complexity, uncertainty and for authorship. It was explicitly agreed that, while results would changing conditions in which community-based nutrition inter- be discussed with all partners, the Academic Principal Investi- ventions are carried out. gator had fll responsibility to decide on the fnal interpretation Anyone experienced in conducting trials with vulnerable and dissemination of the study’s results. populations in disadvantaged communities knows that unex- Data collection for the Tswaka study was completed in July pected challenges are the rule rather than the exception. Contin- 2015. Once the fnal statistical plan had been agreed and all de- uous communication with the community members is necessary scriptive statistics had been completed, a blind-review meeting to engage all individuals involved in the partnership. The Tswaka was organized at the study site in November 2015. In this meet- research team had to deal with rumors and distrust within the ing, with all fnding partners and the CRO present, data was pre- community, while at the same time trying to explain all this to sented and agreed prior to locking the database and de-blinding the representatives of the fnding partners, some of whom had the study. limited experience of working in these difcult settings and with this age group. GAIN played a central role in channeling initial Post-data collection phase bilateral discussions toward more central discussions through- Data management of a large study such as the Tswaka study is out the project, which helped to build trust at every level. complex and also prone to delays for various reasons. In this The study site visit by the sponsors was crucial in improv- case, one reason was the decision to outsource data manage- ing understanding between the partners. Investments were ment to a CRO with rigorous quality standards. 28 THE TSWAKA STUDY

The rigorous and extensive data quality and data transfer Issues to be managed steps for the collection of data on adverse events and mor- Setting up a new type of public–private nutrition research part- bidity required by the CRO’s data management system slowed nership can be challenging, as many issues need to be managed the research team’s speed of response. To overcome this, the along the way. research team had ended up keeping its own records. In con- Real and perceived risks in areas such as safety, confict of in- sequence, two independent databases on morbidity were kept. terests, project management, reputation and scientifc indepen- Given the size of the study, within a vulnerable age group, many dence need to be identifed fom the start of a partnership and mild adverse events occurred, making discrepancies in morbid- managed careflly – something that requires commitment and ity data almost inevitable. Having two datasets on morbidity fexibility fom all partners. On the other hand, such partnerships provided the opportunity to address and counter-check some of ofer unique opportunities for the diferent partners that cannot these discrepancies. However, this caused additional delays, as be realized individually. The private-sector partners, for exam- data analysis could only start afer the two databases had been ple, had the opportunity to carry out studies on a shared costs checked for inconsistencies and combined into a single, locked basis and to enjoy access to scientifc expertise, study teams, database. These delays put pressure on the availability of allo- study environments and key experts with in-depth knowledge cated statistical support staf. of the local context. The research team benefted fom access to In May 2016, almost a year afer completion of data col- quality study product development, as well as tolling capabili- lection, the databases were locked, and study codes were ties, access to quality monitoring services, and scientifc nutri- unblinded. The de-blinded dataset was only accessible to the ent knowledge, besides receiving appropriate fnding. For both research team and not to the fnding partners throughout data parties, learning fom each other’s expertise and developing analyses and the reporting process, to frther ensure the in- awareness of each other’s strengths and weaknesses increased tegrity of scientifc interpretation by the Academic Principal mutual understanding and will be a beneft in the case of fture Investigator. The preliminary results of the Tswaka study were collaborations. Last but not least, private–public partnerships presented at the Micronutrient Forum global conference in can contribute to the education of a new generation of scientists Cancún, Mexico, in October 2016. The discussions about the with experience in dealing with these types of partnerships. In interpretation of the study fndings – probably the most sen- this particular project, for instance, three PhD students were in- sitive part of any industry-sponsored research – took place volved. following the predefned disclosure statements. The Academic Principal Investigator’s fnal interpretation and wording of the Conclusions conclusion were flly endorsed. The infuence of industry fnding on the perceived credibility of nutrition research is currently under debate,4 and some guid- “Things cange ofen within ing principles for how to manage industry fnding of food and nutrition research have been proposed.5 The Tswaka study re- a business environment – search partnership showed that these public–private research what is a callenge and an interes partnerships can be successfl if managed careflly and trans- parently. Personal commitment, trust, perseverance, passion today may be diferent tomorrow. and patience were among the key success factors. Some of the Priorites and people cange.” main learnings fom this partnership, which may inform fture research partnerships, are summarized below.

All discussions, mostly academic in nature, were docu- Lessons learned mented for fll transparency, while the fnal manuscript was 1. Establishing alignment between multiple partners shared with the industry partners for clearance without fr- involves managing tensions between objectives and agen- ther editing rights. das. It takes time, it should be transparent and it should With the Tswaka study project and fnding coming to an of- clearly identif how the respective partners’ priorities will fcial end, the remainder of the work to be done (fnal statistical be managed. analyses, fnal report writing and manuscript preparation) be- came dependent on the personal time and commitment of the 2. Having a clearly defned, achievable and shared public Funding Consortium team involved. This was not a problem, as health goal is important for a successfl public–private the team was determined to bring this project to a successfl partnership.5 completion. And so it happened. SIGHT AND LIFE | VOL. 33(1) | 2019 THE TSWAKA STUDY 29

3. When a consortium is being established, it is essential to have all required knowledge and expertise represented Six voices say what they are most proud in the team and all roles and responsibilities clearly defned of having achieved through the Tswaka Study before commencing the study. Research Partnership

4. Full transparency is required fom the start regarding the VOICE 1 status of negotiations between fnding partners, as well as “ We developed a very difcult product specially designed for regarding product development, to allow realistic deadline older infants, and we achieved it with so many partners.” setting by the research team. VOICE 2 5. For research projects that involve a high reputational “ We completed an extraordinarily complex process to risk for the sponsor and/or the research team, having an implement what we thought was a straightforward idea. independent third partner (a ‘coordinator’) can be helpfl. We have done great research and established the ground- The coordinator can align the academic and private sector, work for a more systematic approach for organizations and can communicate regularly and directly with all consor- to work together.” tium partners, in an efort to make best use of industry involvement (e.g., optimizing study and product quality) VOICE 3 and to minimize direct industry involvement in the execu- “ We have learned a great deal for our organizations regarding tion of the study and direct interactions with researchers. partnerships, and also for ourselves personally. We can do things better in future, but we achieved our goal in the end.” 6. Accepting industry fnding involves dealing with diferent priorities and comes at a cost for researchers: VOICE 4 perceived and real confict of interests, and perceived and “ I estimate that I spent more than 9 weeks in conversations real reputational risks, will need to be managed; lines of to keep this study going – but we have done it, and many communication will need to be kept open; and all roles and from the community and research team have had opportuni- responsibilities will need to be clarifed and agreed upon ties as a result of it.” up font in well-documented author and sponsorship involvement statements, preferably before contracts are VOICE 5 signed. Throughout the project, excellent communication “ We have shown that these products are benefcial – that is is required between all partners, facilitated by the important for children.” coordinator. VOICE 6 7. Within organizations, strategies and leadership change all “ We managed to complete a unique PPP, and we overcame the time, making it difcult to rely on continuous leadership enormous hurdles through everyone’s commitment and and project ownership for a multiple-year research project. without jeopardizing our relationships.” Champions within the various organizations are essential to push the project forward and bridge the diferent internal needs with academic needs. Disclosure statement Authors’ roles 8. Unexpected issues will arise and cause delays and additional expenses, as in every complex project – something Saskia JM Osendarp: Consultant, Young Child Nutrition all partners need to deal with. Having an independent coor- and Partnership at GAIN dinator is again helpfl for aligning partners and managing expectations. Cornelius M Smuts: Academic Principal Investigator of the study

9. Building personal relationships early in the project (e.g., by Klaus Kraemer: Managing Director of Sight and Life Foundation, visits to the study sites and face-to-face meetings between a Swiss foundation fnded partially by DSM partners) is important for creating trust, mutual understanding and commitment. Maaike J Bruins: Study Coordinator and Senior Scientist Nutrition at DSM 30 THE TSWAKA STUDY

Leon GJ Frenken: Project Leader at Unilever ity of data collection throughout the study via an independent study-monitoring agency and via regular phone and face-to-face Dominic Schofeld: President of GAIN meetings throughout the data collection process.

Jane Badham: Managing Director at JB Consultancy, 3) During data analysis and interpretation of results consultant in end-of-project evaluation The API was responsible for data analysis. The fnal interpretation of data was jointly discussed among API, sponsor and co-fnders. However, fnal decisions on the in- Cornelius M Smuts received traveling support fom Unilever, terpretation and dissemination of results rested with the API. DSM and Sight and Life. Correspondence: Saskia Osendarp, Saskia JM Osendarp is a consultant to GAIN and was working 10th Floor, 1201 Eye Street, NW Washington, on this partnership for Unilever R&D fom 2010 to 2011 before DC 20005-3915, USA starting as a consultant with GAIN. Email: [email protected]

GAIN was the sponsor of the study and is a non-proft organization working in micronutrient nutrition. References 01. Smuts CM, Matsungo TM, Malan L, Kruger HS, Rothman M, Kvalsvig Maaike J Bruins is employed by DSM Nutritional Products, JD, et al. Efect of small-quantity lipid-based nutrient supplements a supplier of vitamins, carotenoids, and omega-3 and -6 on growth, psychomotor development, iron status and morbidity nutritional lipids. among 6- to 12-month-old infants in South Africa: a randomized controlled trial. Am J Clin Nutr. 2019 Jan;109(1):55–68. Leon GJ Frenken is employed by Unilever R&D, Vlaardingen, 02. This paper is a summary of the proceedings of the Tswaka study the Netherlands. end-of-project meeting in December 2017, attended by all partners of the Tswaka study research. The quotes in this paper are quotes DSM and Unilever were co-funders of the study and from the participants of this meeting. provided the test products free of charge. None of the other 03. Rothman M, Berti C, Smuts CM, Faber M, Covic N. Acceptability of researchers has any conﬂict of interest. novel small-quantity lipid-based nutrient supplements for complementary feeding in a peri-urban South African community. Food This research was supported by GAIN, DSM and Unilever. Inter- Nutr Bull. 2015 Dec;36(4):455–66. doi: 10.1177/0379572115616057. pretation of data was jointly discussed among the Principal In- Epub 2015 Nov 8. vestigator of the study (Cornelius M Smuts), sponsor (Saskia JM 04. N estle M. Corporate funding of food and nutrition research, science Osendarp) and co-fnders (Maaike J Bruins and Leon GJ Frenken); or marketing. JAMA Intern Med. 2016 Jan;176(1):13–4. doi:10.1001/ however, fnal decisions on interpretation and dissemination of jamainternmed.2015.666. results rested with the Academic Principal Investigator of the 05. Alexander N, Rowe S, Brackett RE, Burton-Freeman B, Hentges EJ, study (Cornelius M Smuts). Kretser A, et al. Achieving a transparent, actionable framework for The responsibilities of the sponsor and co-fnders in this public-private partnerships for food and nutrition research. Am J project are as follows: Clin Nutr. 2015 Jun;101(6):1359–63.

1) During preparation GAIN and Unilever developed the initial study idea and followed this with a Request for Proposals. GAIN, DSM and Unilever reviewed, commented and approved the study proposal. DSM and Unilever developed, produced, shipped and ensured the quality of the study products.

2) During data collection The Academic Principal Investigator (API) was responsible for data collection. GAIN, DSM and Unilever monitored the qual-