Cue-based versus scheduled feeding for preterm infants transitioning from tube to oral feeding: the Cubs mixed-methods feasibility study

Background There is a lack of evidence of the effect of cue-based feeding compared with scheduled feeding on important outcomes for preterm infants. Objectives The objectives were as follows: (1) to describe the characteristics, components, theoretical basis and outcomes of approaches to feeding preterm infants transitioning from tube to oral feeding; (2) to identify operational policies, barriers and facilitators, and staff and parents’ educational needs in neonatal units implementing cue-based feeding; (3) to co-produce an intervention for feeding preterm infants in response to feeding cues; (4) to appraise the willingness of parents and staff to implement and sustain the intervention; (5) to assess associated costs of implementing cue-based feeding; (6) to determine the feasibility and acceptability of a future trial; (7) to scope existing data-recording systems and potential outcome measures; and (8) to determine stakeholders’ views of whether or not a randomised controlled trial of this approach is feasible. Design This was a mixed-methods intervention development and feasibility study comprising (1) a systematic review, case studies, qualitative research and stakeholder consensus; (2) the co-production of the intervention; (3) a mixed-methods feasibility study; and (4) an assessment of stakeholder preferences for a future evaluation. Setting Three neonatal units in the UK (two level 3 units and one level 2 unit). Participants Developmentally normal, clinically stable preterm infants receiving enteral feeds (n = 50), parents (n = 15 pre intervention development; n = 14 in the feasibility study) and health-care practitioners (n = 54 pre intervention development; n = 16 in the feasibility study). Intervention An evidence-informed multicomponent intervention comprising training, a feeding protocol, feeding assessment tools, supplementary training materials [including posters, a film and a narrated PowerPoint (Microsoft Corporation, Redmond, WA, USA) presentation] and the ‘Our Feeding Journey’ document. Main outcome measures The main outcome measures were recruitment and screening rates, infant weight gain, duration of the intervention, feeding outcomes, implementation outcomes (contextual facilitators and barriers, acceptability, adoption, appropriateness and fidelity) and stakeholder preferences for a future evaluation. Results The systematic review of 25 studies concluded that evidence in favour of cue-based feeding should be treated cautiously. The case studies and qualitative research highlighted contextual barriers to and facilitators of the implementation of cue-based feeding. The telephone survey found that many neonatal units are considering implementing cue-based feeding. We recruited 37% of eligible infants, and there was good retention in the study until discharge but a high loss to follow-up at 2 weeks post discharge. The mean number of days from intervention to transition to full oral feeding was 10.8, and the mean daily change in weight gain was 25 g. The intervention was acceptable to parents and staff, although there was dissatisfaction with the study documentation. Intervention training did not reach all staff. A cluster-randomised design with a composite outcome was suggested by stakeholders for a future study. Limitations The intervention was available only in English. Intervention training did not reach all staff. There was low recruitment to qualitative interviews and observations. Only a small number of medical staff engaged in either the training or the interviews. Conclusions It is feasible to implement a cue-based feeding intervention with improved training and documentation. Further work is needed to assess the feasibility of a future trial, noting evidence of existing lack of equipoise. Future work The next steps are to digitalise the intervention and conduct a survey of all neonatal units in the UK. Study registration This study is registered as PROSPERO CRD42018097317 and ISRCTN13414304. Funding This project was funded by the National Institute for Health Research (NIHR) Health Technology Assessment programme and will be published in full in Health Technology Assessment; Vol. 25, No. 74. See the NIHR Journals Library website for further project information.

Can MCDA guide transdisciplinary endeavors? A framework applied to co-developing a flood forecasting system in West Africa

Climate change is projected to increase flood risks in West Africa. The EU Horizon 2020 project FANFAR co-designed a pre-operational flood forecasting and alert system for West Africa in four workshops with 50–60 stakeholders from 17 countries, adopting a Multi-Criteria Decision Analysis (MCDA) process. Firstly, we aimed to find a robust configuration of the FANFAR system. We document empirical evidence of MCDA, including stakeholder analysis, jointly creating 10 objectives, and 11 FANFAR system configurations. Stakeholders found it most important that the system produces accurate, clear, and accessible flood risk information, well before floods. Monte Carlo simulation and sensitivity analyses helped identifying three configurations that were robust despite uncertainty of expert predictions and different stakeholder preferences, elicited in group sessions. Secondly, we investigated if problem structuring helps focus early technical system development. Although partly achieved, full MCDA was necessary to provide convincingly robust configurations. Thirdly, we critically analyzed MCDA based on literature from sustainability science and transdisciplinary research. Our proposed framework consists of three steps: co-design (joint problem framing), co-production (doing research), and co-dissemination and evaluation of integrated knowledge. MCDA met many requirements, but not all. In step 1, participatory MCDA with problem structuring provides a consistent methodology, and can identify stakeholders and shared objectives to foster joint understanding. MCDA successfully contributes to step 2 by combining interdisciplinary expert knowledge, integrating conflicting stakeholder preferences, handling uncertainty, and providing unambiguous, shared results. Many elements of step 3 are not met by MCDA. We discuss this framework and using MCDA for transdisciplinary hydrology research that engages with stakeholders and society.

Human-Centric Design Requirements and Challenges for Enabling Human-AI Interaction in Engineering Design: An Interview Study

AI technologies are enabling the development of not only active tools that provide decision-support, but also interactive tools that seek human input and feedback. As interactive tools facilitate human-AI interaction, their design needs to be informed by human-centric requirements, that is, the needs of the users of such tools. In the context of engineering design, there is a gap in our understanding of designing intelligent tools that facilitate human-AI interaction. To fill this gap, the research question of this study is, What are the human-centric design requirements for the design of AI agents to enable human-AI interaction in engineering design contexts? To answer this question, we conducted an interview study with faculty members in engineering design. The faculty predominantly discussed engineers, designers, and engineering design students as the potential stakeholders who would directly benefit from human-AI interaction. For such stakeholders, we identify several human-centric design requirements and challenges in designing AI tools that facilitate human-AI interaction in engineering design. We find that the requirements focused on the need to understand the stakeholders’ cognition and the engineering design contexts. The results of our study point to the need for the theory of mind in AI agents to enable them to infer stakeholder preferences while engaging in engineering design activities.