Humanitarian emergencies affect millions of people each year. In 2014, an estimated 141 million people were affected by natural disasters and 59.5 million were forcibly displaced by violence and conflict [1].

On the other hand, breastfeeding is the single most effective intervention to save children’s lives and could prevent 13 % of all deaths among children less than five years if practiced optimally, while good complementary feeding could prevent another 6 % of deaths [2]. The recommendations for optimal infant and young child feeding practices include starting breastfeeding within an hour of birth; exclusive breastfeeding for the first 6 months of life; continued breastfeeding for at least 2 years; and introduction of nutritionally-adequate and safe complementary (solid) foods at 6 months together with continued breastfeeding up to 2 years of age or beyond [3]. However the latest Countdown to 2015 report highlights that, according to surveys in around 50 countries, only 50 % and 39 % of mothers reported early initiation of breastfeeding and exclusive breastfeeding for 6 months, respectively, and 67 % recounted adequate timing of introduction of complementary feeding [4].

In addition to the already poor infant and young child feeding practices in normal circumstances, there are further challenges during emergencies to ensure that optimal infant and young child feeding is protected, promoted and supported. These include environmental factors, misconceptions about breastfeeding, inappropriate feeding prior to and during disasters, food insecurity hampering the procurement of nutritious food, competing household needs, and psychological trauma that affects child care practices [5]. Moreover, donations of breast-milk substitutes (BMS) and non-targeted distributions of BMS to non-breastfed children might also interfere with and undermine breastfeeding.

Not breastfeeding exposes a child to a greater risk of infectious diseases such as diarrhoea. For example, after the earthquake in Yogyakarta, Indonesia in 2006, a study found that rates of diarrhoea doubled among infants who had been given donated infant formula compared with those who had not [6]. During the floods in Botswana in 2006, infants hospitalised with diarrhoea were 30 times more likely not to be breastfed compared with infants without diarrhoea [7].

Most of the evidence about the effectiveness of interventions aimed at improving breastfeeding and complementary feeding has been obtained in non-emergency settings and there is a dearth of data during emergencies. A review of research on health interventions in humanitarian crises concluded that more evidence is required on the impact of infant and young child feeding interventions in emergencies (IYCF-E) [8]. Another paper recently pointed out that “Some elements of emergency responses promoted by the international community have so far received too little study and require empirical study, including the effectiveness of various recommendations on infant and young child feeding approaches in emergencies” [9]. Moreover a review of IYCF-E involving several humanitarian stakeholders highlighted the needs for evidence-based programmes [5].

As a first step towards filling these gaps, we aimed to develop and prioritise a list of research questions on IYCF-E.

We used the Child Health and Nutrition Research Initiative (CHRNI) systematic method for setting priorities in child health research [10] and adapted it to the specific context. The CHRNI framework not only aims at creating novel knowledge, but also emphasises the importance of implementation and the better application of current knowledge.

To develop the research questions we interviewed 46 people: 27 people representing 14 NGOs, 5 people representing 4 UN agencies, 4 donors representing 4 agencies and 10 people representing 9 research institutions. Saturation was attained since no new information emerged from the last people to be interviewed. Interviews could not be arranged with two people initially contacted. Forty-eight IYCF-E research questions were developed from the interviews. Most of them (56 %) were classified in the research category of “existing and new interventions”, especially for the research components “evaluating interventions” (9 questions) and “programme design and implementation” (13 questions). Eleven (23 %) and 10 (21 %) questions concerned “policy and systems” and “basic epidemiology” (Additional file 1).

The on-line survey to prioritise IYCF-E research questions was answered by 27 of the 34 people who received the survey. The overall response rate was 79 %, including 100 % response rate from Save the Children staff (13 people), 77 % response rate from the IFE Core Group (10 people) and 50 % response rate from academics who were not members of the IFE Core Group (4 people). The responders represented 8 NGOs, 2 UN organisations and 6 research institutions.

The overall scores were generally high and ranged from 86.2 for the question ranked 1, to 65.5 for the question ranked 48th (Additional file 1). The number of responses “not sufficiently informed” was generally low with median and rank of 1 (0–10) for answerability, 0 (0–5) for operational relevance, 1 (0–6) for disease burden and 1 (0–7) for originality. Agreement was generally good: median agreement score was 69.9, ranging from 51.7 to 80.1.

This analysis indicates that the main gap in evidence on IYCF-E relates to programme implementation, including programme design and evaluation of interventions, as indicated by both the prominence of this topic in the research questions derived from interviews with key informants, and its prominence in the highest ranked research questions. However this could be due to the large number of operational staff included in the survey, compared with the smaller number of academics.

The top 10 research questions embraced both interventions to protect and promote breastfeeding as well as to improve complementary feeding. The former included the design of effective re-lactation interventions and the long-term effect of IYCF-E interventions on children. Although there is good evidence that re-lactation works for women in both developed and developing countries [15–17], there is only anecdotal evidence of success of programmes aimed at re-lactation during emergencies and there are no publicly available guidelines. Some guidance gives general principles but no practical application for actual implementation of programmes [12, 18]. The global long-term positive effects of breastfeeding, including increase in cognitive performance, and possible reduction in overweight and diabetes have been well documented [19]. However, long-term effect of breastfeeding might be different in emergency contexts where children are exposed to a series of hazards.

Questions on the effectiveness of complementary feeding strategies reflect the fact that the effectiveness of different complementary feeding interventions to prevent wasting and stunting, such as the distribution of food supplements and/or transfers of cash, is globally high on the agenda of the humanitarian nutrition community at the moment but with no clearly definitive results yet on their nutritional impact and benefit [20]. Recent research projects have been initiated to further explore this matter [21].

The ability of mothers to care for children in highly challenging humanitarian situations in which caretakers are exposed to stress and trauma was also a main concern. The line of interrogation relied especially on how to provide effective psychological support to caretakers in absence of skilled staff. WHO has initiated a series on low-intensity psychological interventions [22]. The integration of some of those interventions with IYCF-E programmes and the evaluation of their effectiveness on care-taker’s and child’s well-being would permit an advance in this area.

The design of adequate responses in emergency contexts in which the use of BMS was common also stood out as one of the major areas needing research, with three out of ten questions related to this issue: the design of IYCF-E programmes in these contexts; the extent of the use of cash transfers to buy breast-milk substitutes; and the pros and cons of distributing ready-to-use infant formula compared with powdered infant formula plus kit for safer use of BMS if use of BMS is necessary. This recognises the current challenges faced by humanitarian organisations in emergency contexts such as the Syria and Ukraine crises. There would be a good opportunity to further evaluate effectiveness of IYCF-E programmes in these crises. Along these lines, new interim operational guidance has been released for the feeding support of infants and young children in refugee and migrant transit settings in Europe [23] and is currently under evaluation.

Finally, questions focused on evaluating the impact of interventions, such as estimating the coverage of IYCF-E programmes and the impact of specific IYCF-E programmes on nutritional status, morbidity and mortality, were also judged highly. Although effectiveness of breastfeeding interventions has been largely demonstrated in non-emergency contexts [24], there is a dearth of evidence in emergency contexts where populations and programme implementation are faced with specific challenges. Only 4 % of published papers and grey-literature investigating nutrition interventions in emergency between 1980 and 2013 assessed IYCF-E interventions [8]. Impact evaluation is not only central to evaluating programme impact and for accountability, but also to convince donors of the crucial needs for IYCF-E programmes. Indeed, nutrition programmes in emergency contexts tend to focus on the management of acute malnutrition rather than on protecting and improving IYCF practices.

To our knowledge, this paper is the first to identify, rank and extensively describe major research topics on IYCF in emergencies, a neglected and difficult theme to work in. Some of the research questions described had a narrow scope and could be answered with limited resources, such as a literature review or secondary data analysis, while others would require substantial primary data collection at considerable cost. Our analysis complements a similar exercise conducted recently focusing on the management of acute malnutrition in infants less than 6 months [13] but had a wider focus to include both non-malnourished children and children from birth to 2 years.

The study had some limitations. First, some reporting bias might have occurred when the questions were written from the interviews with miss-interpretation of the notions expressed by the interviewees. However, to minimise this risk, discussions were held to clarify any imprecise points during the interviews. Interviews were also recorded.

Second, there might also have been some selection bias arising from the choice of survey respondents. We decided to share the survey with a limited number of known experts in the field to maximise the response rate rather than to open the survey to a wider group, who might have been less motivated to respond. The results and findings therefore only represent the views of a restricted number of people chosen for their expertise but the response rate was good so the results are representative of the group itself. In addition, those people are the key experts in the domain and represent prominent NGOs, the major UN agencies and academic institutions, giving credibility to the results of the study. However, no representatives of governments or affected communities, who might have had a different point of view, were involved in the study. Before undertaking a research project derived from the research questions, it would be advisable to seek government and community representatives’ point of view to complement this study.

Third, the number of respondents from Save the Children was high and this might have biased the results towards the views of this organisation. Agreement scores were, however, high. Moreover there was concordance between the responses of Save the Children staff and the staff of those other organisations (data not shown).

Fourth, no weighting was applied to the four judging criteria, which is different to the CHRNI method. No participants suggested weighting, and the authors of the paper deemed all judgement criteria to be of equal worth. The data can still be re-analysed using a weighting system at a future stage.

Finally, some research questions have a wide scope and more precise questions will need to be defined before a research project can be undertaken.

This study provides an insight into the scope and perceived importance of research questions on IYCF-E by experts in the subject. These questions could form the basis for future research on IYCF-E and could be integrated into the research agenda of stakeholders. The results of studies based on these questions will be important to fill the evidence gap in IYCF-E, to improve IYCF-E programming and, ultimately, to contribute to a reduction in morbidity and mortality among infants and young children.