Consensus from an expert panel on how to identify and support food insecurity during pregnancy: A modified Delphi study

There is little agreement on the number of participants required to achieve consensus [23]. Linstone and Turoff [19] recommend a minimum of ten participants, acknowledging that when increased beyond this number, the Delphi can become labour intensive, with a large amount of data being gathered. While Okoli and Pawlowski [38] suggest a sample of between 10 and 18 sufficient to achieve consensus. Furthermore, it has been reported that improvements in reliability once the number of experts in the panel rises above 15 are negligible [39]. As a result of the level of time commitment required from the panel members, Hanafin and Brooks [40] suggest attrition rates of 16–28% should be expected per round. Allowing for this level of attrition and aiming to have at least ten contributors in the second round, we aimed to recruit a sample of 15–20 experts. While there is little agreement on the size of the sample [39], critical in the membership of the panel is that it is balanced in terms of the composition of members from different areas of expertise and experience [41, 42].

Participants were recruited through professional networks of the researchers via email, direct contact through publicly available contact information, and via social media (Twitter and LinkedIn) to reach a broad audience. Participants were also invited to share the invitation to the first round with people in their network who might also have expertise in nutrition, pregnancy, and food insecurity. The recruitment material included a link where potential participants could read the plain language and informed consent statement and provide their contact details.

Those who expressed their interest in being involved in the study through the recruitment procedure described above were provided with a link to an online survey via email. This modified Delphi collected data via two rounds of consensus.

Round 2: Participants who completed round one were sent a summary via email of the current research that seeks to address food insecurity in pregnancy. This summary was based on a systematic review [15] completed by the authors and describes the current situation of food insecure pregnant women and current evidence-based interventions. The summary of previous evidence was provided in round two, rather than earlier, to discourage the bandwagon effect [43], a common limitation of Delphi, allowing for free flowing ideas to be generated in round one. This material was designed to orientate the expert to the focus of the study, a process that has been found to be a useful way to build the research relationship and provide the experts with an easy summary of the current evidence [44].

A list of all suggested priorities and ways to respond to food insecurity during pregnancy, based on the outcomes of round one and the current literature, were compiled into a new survey and emailed to each expert who was involved in round one. Experts were asked to rate each item in terms of importance on a 5-point Likert scale (1 = not important at all, 2 = not very important, 3 = moderately important, 4 = very important, 5 = extremely important). A free‐text space was provided for feedback on the items and to comment on their decision-making process. The experts were given two weeks to complete their responses and were sent a reminder via email after the two weeks had lapsed.

Participants were asked to rate the importance of a range of considerations when supporting or identifying a pregnant woman who is (or who they suspect to be) food insecure or hungry (Table 2). Of the nine statements posed, eight achieved consensus, with five achieving 100% agreement. Mean scores for eight of the nine statements were over 4 out of a possible 5. Agreement and rank prioritisation were consistent. Participants identified linking women with appropriate social care services such as emergency community food assistance as both the most important consideration and as the highest priority. The item that was ranked as the lowest priority and where only 64% of participants indicated that this was extremely or very important was providing women with food literacy or nutrition education.

Participants were asked to rate the importance of a range of actions when addressing food insecurity during pregnancy (Table 3). Of the 14 statements posed, ten achieved consensus, with one achieving 100% agreement and seven achieving 91% agreement. All the actions that achieved consensus were system level actions, that could be achieved through policy or institution level cooperation. Agreement and rank prioritisation were consistent. Mean scores for 10 of the 14 statements were above 4 out of a possible 5. Participants identified creating a social care arrangement specifically for food insecure pregnant women, one that might include access to nutrition supplements, and care and support for pregnant women that focuses on reducing stigma and blame, as the key priority and the most important action. Items that were ranked as the lowest priorities included providing women with food literacy or nutrition education, linking women with other health care services, and directly providing food (either through food parcels or via commercial meal kits).

Participants were asked to rate the importance of a range of activities that could be implemented to address food insecurity during pregnancy (Table 4). Of the nine statements posed, four actions achieved consensus, two achieved 100% agreement, one was determined to be approaching consensus (73% agreement). Mean scores for five of the nine statements were above 4 out of a possible 5. Actions that were highly ranked were those which can be influenced at the clinic level, including routine food security screening, the introduction of clinical practice guidelines, and referral to emergency and community food assistance; these items are consistent with other questions in round two and results of round one. Items that were ranked as the lowest priorities included providing women with food literacy or nutrition education, and directly providing food (either through food parcels or via commercial meal kits).

Items that achieved consensus that could be implemented at the institution level, for example at individual clinics or hospitals or indeed in all hospitals, include routine screening and linking pregnant women with a range of services, both internal and external to the clinical setting. Including routine screening and then providing programs that link food insecure households or individuals to the services via health care settings is a solution that has grown out of the suggestion that supporting food security can lead to improvements in population health [46, 47]. There is increasing interest in routine screening and the role of healthcare systems in addressing food insecurity in non-pregnancy healthcare settings [48, 49]. The International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) Z codes (Z55–Z65) allow for the classification and documentation of the social determinants of health in electronic medical health records. The official guidelines for coding and reporting of the ICD-10-CM suggest that all clinicians, not just physicians involved in the care of a patient, document the Z codes to report on the social determinants of health of the patient [50]. Screening tools that are based on or match these codes could be incorporated into existing screening mechanisms or could be used as the basis a more comprehensive referral system.

While screening for food insecurity does not exist in the Australian health care setting, screening exists in health care settings in other countries, where it has been found to be part of a successful approach when seeking to address food insecurity [51‐53]. Both the American Academy of Family Physicians, and the American Academy of Paediatrics recommend that food insecurity is routinely screened [54]. A systematic review of routine screening in the health care setting found that screening is generally conducted via a brief screening tool comprising one or two items [54]. As the results of this current study demonstrate, there is an appetite for routine screening from clinicians to identify and assist food insecure pregnant women. In Australia, there is work underway that seeks to highlight the link between health care settings and actions that address food insecurity. For example Kerz, Bell [48], have validated a brief tool to be used in a Australian paediatric health care setting, with findings suggesting that food insecurity is prevalent among families of children attending paediatric outpatient hospital appointments. While McKay and colleagues [14] have demonstrated that a brief tool measuring food insecurity could be used in a clinical setting as a component of a referral pathway for pregnant women who are identified as food insecure. It is possible that this two-item tool could be included in the electronical medical health record and incorporated into standard practice. While there is evidence for the importance and acceptability of screening for food insecurity, practical considerations need to be made. For this reason, researchers suggest that short screening measures be employed [14, 48]. Short screening measures are limited in that they do not allow for the assessment of the severity of food insecurity, however, they are more appropriate for busy clinical settings and take into consideration barriers including time constraints and increased workloads experienced by clinicians, while still being able to determine food secure status [55]. While it is clear that clinicians consider food insecurity screening as an important component of antenatal healthcare [16, 56], there are a variety of barriers that may prevent them from asking their patients about their household food security. This includes a lack of guidelines, uncertainty surrounding responsibility for screening, inadequate clinical knowledge and training, and time constraints [16, 56].

While approaches to screening are gaining traction, there remain gaps in how to best link food insecure pregnant women to the services they need. A recent systematic review exploring interventions that have sought to address social needs during pregnancy care after standard or routine screening, found that while there are evidence based interventions for family and domestic violence, there are few interventions for other social needs, including for people who have been identified as food insecure [57]. A different review found that despite an increase in the number of care settings that screen for food insecurity, those who are referring food insecure or hungry women and families, are largely referring them to external services (for example, in the USA supplemental nutrition assistance program, SNAP, and women, infants, and children supplemental nutrition assistance program, WIC, are common places of referral). However, referral and assistance can be less formal and include providing information about emergency and community food assistance, as well as providing assistance via community health workers or social workers [58]. Research suggests that these linkage programs have positive health outcomes [59].

Experts rated community level solutions, such as linking women to appropriate income support and social care services, and referrals to intimate partner violence services. The consideration of these community levels services reflects an acknowledgement that while food insecurity has clinical implications, the solutions may lie outside the hospital setting, with structural and systemic level changes known to take considerable time, highlighting the need for immediate solutions to solve short term need. While the definition of food insecurity is a lack of appropriate food, there are many reasons that individuals and households are unable to access food. Estimates suggest that 13% of Australian’s live below the poverty line [60] due to rising living costs, stagnant wage growth, and unemployment and underemployment. Poverty predisposes low-income individuals towards a suboptimal diet [61]. Many people who live below the poverty line are also in receipt of government welfare payments, however, these payments may be insufficient to cover the basic costs of living, increasing stress and pressure on individuals and households. There is a body of research that highlights the role of poverty and income on chronic food insecurity [62], with many people who live on low incomes forgoing food for other basic living expenses [63]. Low-income neighbourhoods are more likely to have limited options for fresh produce and whole grains [64], and are more likely to have access to fast-food outlets and convenience stores [65]. Low income households are often forced to make decisions about the food they can purchase, sometimes in the absence of health considerations [66]. For many individuals and families, the experience of food insecurity and its impact on diet is greater than limited funds and lack of access to healthy food, and while many households experience short term or acute food insecurity, for other families, the experience of food insecurity can be long term, often intergenerational [67, 68]. While there is evidence to suggest that at current levels, government-provided income support is below the poverty line for most families, it can provide some mitigation from the more serious impacts of food insecurity [69, 70] and as highlighted by experts in the current study, referral to appropriate income support should be included in any approach to address household food insecurity in a health care setting.

In addition to the physical experience of hunger and the physiological impacts of poor nutrition, there are also psychological implications for food insecurity. Work from the USA has identified a relationship between receipt of welfare and intimate partner violence, finding that intimate partner violence was associated with negative health outcomes and greater material hardship [71]. There is a significant amount of evidence highlighting the risk of domestic violence during pregnancy [72], and there is emerging evidence suggesting a relationship between food insecurity during pregnancy and intimate partner violence. According to Ricks, Cochran [73], food insecurity can be linked to violence in three main ways. First, economic abuse can produce food insecurity, as one partner in the relationship controls or restricting access to finances by the other partner [74]. Second, many individuals who escape an abusive relationship rely on financial assistance and low-wage jobs for survival, therefore lacking the financial ability to secure food. Third, some evidence to suggest that a food insecure environment may increase the rate of violence [75, 76]. The relationship between food insecurity, pregnancy, and intimate partner violence is unidirectional. Pregnant women who are food insecure are more likely to experience violence from an intimate partner [76, 77], and there is a predictive effect of intimate partner violence on food insecurity in longitudinal studies [78]. There are a range of factors for intimate partner violence, including drug and alcohol use, prior violence, traditional attitudes to gender and a range of socio-demographic characteristics [79]. As highlighted by experts in this study, there is a need to consider intimate partner violence when considering food insecurity. Screening for intimate partner violence already exists in most antenatal care in Australia, and as such most practitioners working in this space will already be able to screen for intimate partner violence, posing an opportunity to include food insecurity screening at the same time [80]. Promisingly, a recent systematic review suggests that pregnant women who are experiencing intimate partner violence with or without other mitigating factors are likely to benefit from screening, referral, and supportive counselling [81].

The main community level responses to food insecurity and hunger in Australia are through emergency and community food assistance; foodbanks and pantries, soup kitchens, and school lunch and breakfast programs [68]. While these community solutions play an important role in the charitable response to food insecurity and hunger, many people who use these services experience shame and stigma [82], and various restrictions on how and when they can be used means they are generally not able to meet all the needs of those who are experiencing hunger and food insecurity [83]. While many Australian emergency and community food assistance providers refer clients on to other services including family violence and income support services [84], to date, there has been limited evidence to suggest that clinicians refer patients to emergency and food assistance providers, nor that this is something that they would consider doing [16]. Unlike the more formal system in the USA, in Australia, these services are typically informal and are not a part of a systemic approach to food insecurity through partnership with government, health care, and charity. Encouraging research from the USA highlights a growing body of evidence demonstrating positive partnerships between healthcare systems and local food assistance programs as a way to reduce food insecurity and hunger and assist people to access the services they need [58, 85]. This research suggests that such a partnership may result in increased food intake, including increased fruit and vegetable intake [86], and better health outcomes [87, 88].

While most high-income countries have some form of government assistance for those who are experiencing hunger or food insecurity, these programs can vary widely. As described above, the main response of governments to food insecurity and hunger in Australia is through emergency and community food assistance. This sector has grown since the 1990s, most rapidly in the past decade. The sector in Australia operates through both formal and informal networks and is comprised of food banks, food pantries, soup kitchens, and meals programs operated by charities [89]. Foodbank Australia, Australia’s largest re-distributor of community food, has reported an increase in the number of people accessing their services, with almost one million people receiving food from Foodbank each month in 2021 [90]. This system is complemented by a range of government-provided or -run welfare programs that provide income support for aged, disabled, parents of young children, people seeking employment, and family tax benefits that help low-income families with the cost of raising children. Unlike the system in the USA, there is no large government program that is specifically designed to provide food assistance. The USA in comparison has multiple federal programs that are specific to food aid. The main federal food assistance programs include the Supplemental Nutrition Assistance Program (SNAP); the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC); and the National School Lunch Program (breakfast, lunch, and summer meals) [91].

Experts identified structural solutions to food security as possible ways to address food insecurity during pregnancy and as interventions that could be tried. While Australia does not currently have a comprehensive food security program like that of the USA, we may have reached a time when key stakeholders need to advocate for one. Enrolment in federal food assistance programs in the USA is associated with improved outcomes across multiple dimensions, including food security, nutrition, health, development, and health care costs [92, 93]. Importantly, the WIC program has been found to significantly reduce food and nutritional security among both pregnant women and children [94, 95]. Such a program allows for a clear referral pathway for women who are identified as food insecure when pregnant and ensures a systematic approach to ensuring adequate food and nutrition [52, 96].

While it is clear that solutions to targeting food insecurity in pregnancy have been acknowledged by experts in this area, there are a number of limitations that need to be taken into consideration. The aim of the study was to include 15–20 experts in round one in order to maintain the 10 needed for round two. Despite a very board recruitment campaign, this was not achieved, and more participants may have provided a diversity of opinions. However, most participants (11 of 12) who completed round one responded to round two, thereby eliminating the problem of attrition common in other Delphi studies. Secondly, COVID-19, and time and financial constraints meant that a face-to-face Delphi was unachievable. We chose a modified Delphi to attract a broad range of experts from all over Australia. Having an in-person meeting may have altered the results or provided more solutions. This study does not include the voices of consumers, and as such, solutions that may have been identified by those experiencing food insecurity may have been missed. However, a recent Australian study with pregnant women and their experience of hunger and food insecurity suggested that that while over half (57.1%) were comfortable that their health care provider might asked them about their household food security, most (61.6%) did not expect to be asked [97]. Finally, as described all participants were clinicians and were not experienced in policy or in working at a system level, this could be a limitation in both the suggestions provided here and the possible outcomes of this work if taken up by clinicians. Despite these limitations, the consensus in this study is a strength and provides support for the results presented here.