Dietary risk factors, including inadequate intakes of fruits, vegetables, whole grains and excessive intakes of unhealthy foods (foods high in added sugar, Na and saturated fat), are the leading contributors to death and disability globally(Reference Afshin, Sur and Fay1). Dietary intake in early childhood has implications for child physical, social and mental well-being(2), placing children at an increased risk of developing a variety of non-communicable conditions later in life, including obesity and high blood pressure(Reference Rodrigues, Abreu and Resende3,Reference Deal, Huffman and Binns4) . As the dietary behaviours and food preferences learnt during early childhood frequently carry through into adulthood(Reference Mikkilä, Räsänen and Raitakari5–Reference Spence, Campbell and Lioret7), improving the diet of young children is paramount to reduce the burden of dietary risk factors in the population.
Early childhood education and care (ECEC) settings (inclusive of long day cares, preschools, nurseries, kindergartens and family day care) provide access to a large number of young children (United States (US) ∼ 60 % of children(8); Australia ∼50 % of children(9)), for prolonged and regular periods of time (on average 30 h per week), during a highly influential life stage(10,11) . As these settings are accessed by children and families across various socio-economic and demographic groups, they provide an opportunity to address health inequities in young children. Further, national regulations and quality assessment systems for the sector (e.g. National Quality Framework in Australia(12), Quality Rating and Improvement System in the US(13)) support the creation of environments that promote healthy eating behaviours. As such, ECEC is recommended by the WHO as an important setting for the implementation of public health nutrition interventions(14).
Over the past few decades, there has been considerable public health and research investment in the development and implementation of effective population-based interventions for improving child nutrition(Reference Yoong, Lum and Wolfenden15–Reference Matwiejczyk, Mehta and Scott17). Despite evidence of efficacy of these interventions, assessments of their ‘real-world’ effectiveness demonstrate substantially reduced effects on child nutrition(Reference Sutherland, Jackson and Lane18). A recent systematic review assessing the effectiveness of scaled-up public health nutrition interventions found effect sizes reported from scaled-up interventions were on average only 50 % of the effect size reported in preceding efficacy trials(Reference Sutherland, Jackson and Lane18). Unless such interventions can be successfully scaled-up whilst maintaining an effect that is meaningful to the population, they offer little benefit and represent significant research waste. While a range of factors, such as poor reach, lack of intervention adherence, fidelity and dose, may contribute to the reduced effects of these scaled-up interventions, the limited impact may also be due, in part, to selection and subsequently implementation, of interventions that are not well suited to the contexts in which they are to be delivered for population scale-up. Such interventions are therefore likely to encounter a range of barriers to implementation at scale.
To provide evidence to support policy makers and practitioners to more readily assess whether ECEC-based healthy eating interventions are amenable for scale, assessment of intervention scalability is recommended(Reference Milat, Newson and King19). Scalability is defined as ‘the ability of a health intervention shown to be efficacious on a small scale and or under controlled conditions to be expanded under real world conditions to reach a greater proportion of the eligible population, while retaining effectiveness’(Reference Milat, King and Bauman20). A range of tools have been designed to support scalability assessments(Reference Ben Charif, Zomahoun and Gogovor21). Such tools suggest that in addition to intervention efficacy/effectiveness, other factors are thought to influence decision making regarding the scalability of public health interventions. These factors include the expertise and resources required to deliver the intervention outside of the research environment, potential reach, cost, availability of delivery infrastructure, as well as fit within the local context(Reference Milat, Lee and Conte22,Reference Milat, Lee and Grunseit23) .
The reporting of data relevant to the factors of scalability as part of trials of nutrition interventions in ECEC would better inform scalability assessments to support public health decision making. Such information is crucial for end-users to increase the likelihood of selecting an intervention that can be successfully scaled-up to produce public health impact(Reference Yoong, Turon and Wong24,Reference Azevedo, Rodrigues and Londral25) . However, the extent to which such information is available within published reports of healthy eating interventions in this setting is unknown. A number of previous reviews in the ECEC setting have extracted some information relevant to intervention scalability(Reference Yoong, Lum and Wolfenden15,Reference Wolfenden, Barnes and Jones26,Reference Sanchez-Flack, Herman and Buscemi27) ; however, no previous reviews have sought to systematically examine the reporting of all scalability factors. As such, the aim of this study was to assess the extent to which the factors required to assess scalability are reported among healthy eating interventions conducted within the ECEC setting.
Methods
We undertook secondary data analysis(Reference Yoong, Turon and Grady28) of included studies identified by the Cochrane systematic review conducted by Yoong et al.(Reference Yoong, Lum and Wolfenden15), which aimed to assess the effectiveness of healthy eating interventions delivered in ECEC settings for improving child dietary intake in children aged 6 months to 6 years. The repurposing of data included within high-quality systematic reviews has been recommended as a way of reducing research waste, identification of research gaps and a way of addressing important public health policy and practice questions(Reference Yoong, Turon and Grady28).
Briefly, as per the inclusion criteria outlined by Yoong et al.(Reference Yoong, Lum and Wolfenden15), this included the following:
-
Randomised controlled trials (RCT), including cluster-RCT, stepped-wedge RCT, factorial RCT, multiple baseline RCT and randomised crossover trials;
-
Interventions conducted within the ECEC setting that offer care for children 6 months to 6 years, which includes formal paid care such as preschools, nurseries, long day cares, kindergartens and family day care services;
-
Interventions conducted with a range of participants, including (but not limited to) children attending the ECEC service; parents, guardians, or carers of children, and professionals responsible for the care provided to children attending an ECEC service (e.g. service directors, educators, volunteers, cooks or other employed staff) and
-
Healthy eating interventions containing a nutrition component that aims to influence child diet.
The current study was limited only to those studies included in the Cochrane review that reported on any child dietary intake outcomes which included consumption of food groups/specific foods; consumption of beverage types/specific beverages; intake of macronutrients and specific dietary components; overall diet quality and specific diet quality components. Studies not reporting such an outcome (including those that only reported on anthropometric outcomes) were excluded given the focus of this review. Included studies could be at any stage of scale-up (i.e. efficacy, effectiveness, implementation or dissemination) as long as they reported child dietary intake outcomes.
Identification of supporting evidence
As information regarding scalability factors may be reported in a range of publications beyond the primary trial outcome publication, we sought to comprehensively capture all peer-reviewed publications associated with an intervention to inform scalability assessments. This included forward and backward citation searches in Scopus of the included studies. The aim of this search was to identify any additional published data or information related to the included studies, reporting on, but not limited to, intervention development; effectiveness; implementation; dissemination; feasibility/acceptability; adaptations/fidelity; sustainability and cost-effectiveness outcomes.
Data extraction
Characteristics of included studies were extracted by pairs of independent reviewers, using Microsoft Excel, as per Yoong et al.(Reference Yoong, Lum and Wolfenden15). Data including first author, year, country, study design, delivery setting and participants and name and brief description of the intervention were extracted. Similar to previous reviews assessing the scale-up of nutrition, and obesity prevention interventions(Reference Sutherland, Jackson and Lane18,Reference McCrabb, Lane and Hall29) and based on proposed scale-up pathways for public health interventions(Reference Indig, Lee and Grunseit30), included studies were categorised as efficacy (primarily aiming to evaluate the effect of an intervention in ideal, controlled settings), effectiveness (primarily aiming to evaluate the effect of an intervention in real-world settings), implementation (primarily aiming to evaluate strategies to increase the uptake or adoption of an evidence-based intervention within real world settings) or dissemination (primarily aiming to evaluate the distribution of an intervention within real-world settings).
Scalability assessment
The extent to which data relating to the factors of scalability were reported by included studies were extracted according to the Intervention Scalability Assessment Tool (ISAT)(Reference Milat, Lee and Conte22,Reference Milat, Lee and Grunseit23) . Such an approach has been undertaken by two recent reviews(Reference Yoong, Turon and Wong24,Reference Azevedo, Rodrigues and Londral25) . The ISAT(Reference Milat, Lee and Grunseit23) was developed to support policy-makers and practitioners to make systematic assessments of the suitability of health interventions for scale-up within high-income country health and community settings. Briefly, the ISAT tool consists of three parts. Part A: considers the context in which the intervention is being deliberated for scale-up and consists of five domains: (1) the problem; (2) the intervention; (3) strategic/political context; (4) evidence of effectiveness and (5) intervention costs and benefits. Part B: explores the potential implementation and scale-up requirements and consists of five domains: (1) fidelity and adaptation; (2) reach and acceptability; (3) delivery setting and workforce; (4) implementation infrastructure and (5) sustainability. Part C: provides a brief summary of the information gathered in Parts A and B. All sections of included studies were reviewed for relevant data, including the Introductions, Methods, Results, Discussion, Conclusions, Acknowledgements, Funding, Conflicts of interest and Appendices/Supplementary material.
Review authors identified data related to key scalability domains as described in Table 1.
Table 1 Scalability domains, description and examples of relevant data for each domain
As we were interested in identifying whether such data were reported, we only systematically extracted data regarding availability and for each domain reported it as No: Data not reported; Partial: Data partially reported (i.e. one of the two items assessed for the domain was reported); Yes: Data fully reported. Only those domains assessing multiple factors within a single domain could be assessed as Partial (i.e. strategic and political context, fidelity and adaptation and reach and acceptability). Given the comprehensive nature of the ISAT domains, brief examples of the type and extent of data reported for each of the scalability domains have been described narratively. Scalability assessments were undertaken by one reviewer (AG) and checked by a second reviewer (JJ). In the case that one reviewer was an author on included studies (AG), the second (JJ) and third reviewer (ML), undertook and checked the scalability assessments, respectively. Discrepancies between reviewers were reconciled by consensus.
Analysis and synthesis
Review findings were synthesised narratively with descriptive statistics (frequencies and percentages) used to report the number of ISAT domains assessed as ‘Yes: Data fully reported’ for each study and the number of studies assessed as ‘Yes: Data fully reported’ ‘No: Data not reported’ and ‘Partial: Data partially’ reported for each of the ISAT domains.
Results
A total of thirty-eight studies (reported across forty-two articles) were included from Yoong et al.; a subgroup of the total studies included in the Cochrane review(Reference Yoong, Lum and Wolfenden15). Broadly, Yoong’s review found that healthy eating interventions in ECEC lead to small improvements in child diet quality and increased fruit consumption and vegetable consumption, however, did not have an effect on consumption of less healthy foods and sugar-sweetened drinks. A further 2246 titles were screened from the Scopus forward and backward citation search of included studies, identifying an additional thirty-three articles reporting relevant data, resulting in seventy-five included articles (see Fig. 1). Interventions were published between 2005 and 2022, and all were of a cluster RCT design. Interventions were most commonly conducted in the US (n 14), Australia (n 5), United Kingdom (n 2), Norway (n 2), Germany (n 2) and Belgium (n 2) (Table 2). All interventions were conducted within ECEC settings, fifteen of these included an additional component (beyond the intervention delivered in the ECEC setting) that was delivered in the child/family’s home(Reference De Bock, Breitenstein and Fischer31–Reference Vereecken, Huybrechts and Van Houte45) and two included the wider community(Reference De Coen, De Bourdeaudhuij and Vereecken46,Reference Iaia, Pasini and Burnazzi47) . Thirty-three studies were categorised as effectiveness studies, with five categorised as implementation. None of the studies were categorised as efficacy or dissemination (Table 2). Brief descriptions of the characteristics of the interventions can be found in Table 2, along with additional articles associated with an intervention (identified via citation searching).
Fig. 1 PRISMA diagram
Table 2 Intervention characteristics and relevant associated publications
NA, not applicable; RCT, randomised controlled trial; FCCH, Family Child Care Homes; PA, physical activity; ECEC, early childhood education and care; US, United States; UK, United Kingdom.
Scalability of healthy eating interventions
None of the studies reported on all ten domains of scalability (Table 3). Across studies, the reporting of domains ranged between four to nine domains. In total, twenty-three (61 %) studies reported on more than half (i.e. > 5) of the domains of scalability.
Table 3 Scalability assessments of included studies according to ISAT domains
All thirty-eight (100 %) studies described the problem, the intervention objective and key elements, the effectiveness of the intervention and the delivery workforce and setting. For all studies, ‘the problem’ was reported as the burden of disease and prevalence of poor dietary intake for the population of interest. While the intervention objectives were clearly described for all studies, the amount of detail describing the intervention elements was variable. For example, some studies provided only a brief description of the intervention(Reference Nekitsing, Blundell-Birtill and Cockroft48) whereas others provided detailed accounts of each intervention component(Reference Kristiansen, Bjelland and Himberg-Sundet39,Reference Pearson, Finch and Sutherland41) and reported according to TIDieR (template for intervention description and replication) guidelines(Reference Hoffmann, Glasziou and Boutron49). All studies reported on intervention effectiveness in improving child dietary intake (per inclusion criteria); however, few reported whether the intervention resulted in any adverse outcomes (e.g. negative impacts on child health or staff/parent attitudes)(Reference Fitzgibbon, Stolley and Schiffer32,Reference Fitzgibbon, Stolley and Schiffer33,Reference Kipping, Langford and Brockman37,Reference Pearson, Finch and Sutherland41,Reference Seward, Wolfenden and Finch50) . None described the relative advantage of the intervention being evaluated over existing interventions to address child dietary intake in the setting (e.g. comparison of any perceived differences in the strategic, political, economic or societal outcomes of the intervention over usual practice and/or alternate healthy eating interventions delivered in ECEC). The extent of information reported also varied for the delivery workforce and setting. While all studies reported the setting (ECEC, home and wider community) and the workforce delivering the intervention (most commonly ECEC staff, researchers and external nutrition experts), reporting of the number and description of formal qualifications of the individuals involved in delivering the interventions was variable(Reference Puder, Marques-Vidal and Schindler42,Reference De Coen, De Bourdeaudhuij and Vereecken46) .
Only five (13 %) studies reported on the cost domain. This reporting included the cost of delivering the intervention(Reference Kipping, Langford and Brockman37,Reference Iaia, Pasini and Burnazzi47,Reference Leis, Ward and Vatanparast51) and formal cost analyses (i.e. cost-effectiveness and cost-utility)(Reference Natale, Messiah and Asfour52,Reference Yoong, Grady and Wiggers53) . Regarding the sustainability domain, only six (16 %) studies reported on the sustainability of the intervention (i.e. assessed as reporting study outcomes ≥ 12 months post intervention(Reference Fitzgibbon, Stolley and Schiffer32,Reference Fitzgibbon, Stolley and Schiffer33,Reference Kristiansen, Bjelland and Himberg-Sundet39,Reference Iaia, Pasini and Burnazzi47,Reference Natale, Messiah and Asfour52,Reference Natale, Atem and Weerakoon54) . The sustainability of the required infrastructure (including funding, resources, processes and delivery workforce) for intervention delivery, however, was rarely reported.
Implementation infrastructure was reported for seventeen (45 %) of the studies(Reference Fitzgibbon, Stolley and Schiffer34,Reference Fitzgibbon, Stolley and Schiffer35,Reference Kipping, Langford and Brockman37–Reference Pearson, Finch and Sutherland41,Reference Iaia, Pasini and Burnazzi47,Reference Nekitsing, Blundell-Birtill and Cockroft48,Reference Leis, Ward and Vatanparast51,Reference Yoong, Grady and Wiggers53–Reference Vaughn, Hennink-Kaminski and Moore58) . The extent of information and content of this domain varied substantially. For example, some studies reported the intervention was already being scaled(Reference Kobel, Wartha and Lämmle38,Reference Puder, Marques-Vidal and Schindler42,Reference Pinket, Van Lippevelde and De Bourdeaudhuij59) , albeit little detail on the infrastructure and operational requirements for scale-up were provided, whereas others reported on the resource barriers to widespread implementation of the intervention(Reference Kipping, Langford and Brockman37,Reference Blomkvist, Wills and Helland55) .
Three ISAT domains (political and strategic context, fidelity and adaptation and reach and acceptability) contained two criteria, and therefore could receive a ‘Partial’ rating. Seventeen (45 %) studies fully reported on the political and strategic context domain of the ISAT(Reference De Bock, Breitenstein and Fischer31,Reference Fitzgibbon, Stolley and Schiffer35,Reference Grummon, Cabana and Hecht36,Reference Lumeng, Miller and Horodynski40,Reference Pearson, Finch and Sutherland41,Reference Vereecken, Huybrechts and Van Houte45,Reference Seward, Wolfenden and Finch50,Reference Leis, Ward and Vatanparast51,Reference Yoong, Grady and Wiggers53,Reference Natale, Atem and Weerakoon54,Reference Gans, Tovar and Kang56,Reference Pinket, Van Lippevelde and De Bourdeaudhuij59–Reference Ward, Vaughn and Burney64) , eighteen (47 %) studies reported partial data(Reference Fitzgibbon, Stolley and Schiffer32–Reference Fitzgibbon, Stolley and Schiffer34,Reference Kobel, Wartha and Lämmle38,Reference Kristiansen, Bjelland and Himberg-Sundet39,Reference Puder, Marques-Vidal and Schindler42–Reference Roberts-Gray, Ranjit and Sweitzer44,Reference De Coen, De Bourdeaudhuij and Vereecken46–Reference Nekitsing, Blundell-Birtill and Cockroft48,Reference Natale, Messiah and Asfour52,Reference Blomkvist, Wills and Helland55,Reference Vaughn, Hennink-Kaminski and Moore58,Reference Hu, Ye and Li65–Reference Zeinstra, Vrijhof and Kremer68) and three (8 %) studies did not report this domain at all(Reference Kornilaki, Skouteris and Morris57,Reference Namenek Brouwer and Benjamin Neelon69,Reference Witt and Dunn70) . For those studies partially reporting data, this included one study only reporting context and seventeen studies only reporting on funding sources. Overall, studies reporting on the context surrounding the intervention (n 18) provided varying accounts (e.g. alignment of the intervention into mandatory nutrition curriculum(Reference Başkale and Bahar60) policies and guidelines for the ECEC setting(Reference Yoong, Grady and Wiggers53), inclusion in state-sponsored nutrition programmes(Reference De Bock, Breitenstein and Fischer31), support from local, state and national governing organisations(Reference Vereecken, Huybrechts and Van Houte45,Reference Ward, Vaughn and Burney64) . The reporting of the source of funding received (or lack thereof) (n 34) was fairly consistent across studies.
Fidelity and adaptation were reported in full by 34 % of studies (n 13)(Reference De Bock, Breitenstein and Fischer31,Reference Fitzgibbon, Stolley and Schiffer32,Reference Fitzgibbon, Stolley and Schiffer34,Reference Kipping, Langford and Brockman37,Reference Pearson, Finch and Sutherland41,Reference Puder, Marques-Vidal and Schindler42,Reference Leis, Ward and Vatanparast51,Reference Yoong, Grady and Wiggers53–Reference Blomkvist, Wills and Helland55,Reference Vaughn, Hennink-Kaminski and Moore58) , partially reported by 32 % of (n 12) studies(Reference Fitzgibbon, Stolley and Schiffer33,Reference Fitzgibbon, Stolley and Schiffer35,Reference Grummon, Cabana and Hecht36,Reference Lumeng, Miller and Horodynski40,Reference Roberts-Gray, Ranjit and Sweitzer44,Reference De Coen, De Bourdeaudhuij and Vereecken46,Reference Nekitsing, Blundell-Birtill and Cockroft48,Reference Seward, Wolfenden and Finch50,Reference Gans, Tovar and Kang56,Reference Pinket, Van Lippevelde and De Bourdeaudhuij59,Reference Jones, Wyse and Finch61,Reference Ward, Vaughn and Burney64,Reference Zeinstra, Vrijhof and Kremer68,Reference Witt and Dunn70) and not at all by 34 % (n 13) of studies(Reference Kobel, Wartha and Lämmle38,Reference Kristiansen, Bjelland and Himberg-Sundet39,Reference Reyes-Morales, González-Unzaga and Jiménez-Aguilar43,Reference Vereecken, Huybrechts and Van Houte45,Reference Iaia, Pasini and Burnazzi47,Reference Natale, Messiah and Asfour52,Reference Kornilaki, Skouteris and Morris57,Reference Başkale and Bahar60,Reference Ray, Figuereido and Vepsäläinen63,Reference Hu, Ye and Li65–Reference Morris, Edwards and Cutter-Mackenzie67,Reference Namenek Brouwer and Benjamin Neelon69) . For those studies partially reporting data, this included eleven studies only reporting fidelity, and one study only reporting adaptations. Overall, the studies that reported on intervention fidelity (n 24), most often described compliance in delivery of the intervention components from the delivery workforce(Reference De Bock, Breitenstein and Fischer31,Reference Jones, Wyse and Finch61) , or implementation of the intervention among intervention recipients (i.e. staff and parents)(Reference Lumeng, Miller and Horodynski40,Reference Witt and Dunn70) . None of the studies reported how intervention fidelity would be monitored or maintained long term (e.g. any existing structures/processes or future plans for the monitoring or maintenance of intervention delivery). Overall, the reporting of adaptations of the interventions (n 13) covered planned modifications from pilot interventions(Reference Blomkvist, Wills and Helland55,Reference Fitzgibbon, Stolley and Dyer71,Reference Messiah, Lebron and Moise72) , in addition to unplanned adaptations during the intervention period(Reference Puder, Marques-Vidal and Schindler42,Reference Grady, Seward and Finch73) . The likely impact of these unplanned modifications on intervention effectiveness was rarely described(Reference Puder, Marques-Vidal and Schindler42).
Reach and acceptability were reported in full for sixteen (42 %) studies(Reference Fitzgibbon, Stolley and Schiffer34,Reference Kipping, Langford and Brockman37,Reference Lumeng, Miller and Horodynski40–Reference Puder, Marques-Vidal and Schindler42,Reference Roberts-Gray, Ranjit and Sweitzer44,Reference Vereecken, Huybrechts and Van Houte45,Reference Iaia, Pasini and Burnazzi47,Reference Nekitsing, Blundell-Birtill and Cockroft48,Reference Leis, Ward and Vatanparast51,Reference Yoong, Grady and Wiggers53,Reference Gans, Tovar and Kang56,Reference Pinket, Van Lippevelde and De Bourdeaudhuij59,Reference Jones, Wyse and Finch61,Reference Ward, Vaughn and Burney64,Reference Hu, Ye and Li65) , partially for eighteen (47 %) studies(Reference De Bock, Breitenstein and Fischer31–Reference Fitzgibbon, Stolley and Schiffer33,Reference Fitzgibbon, Stolley and Schiffer35,Reference Grummon, Cabana and Hecht36,Reference Kobel, Wartha and Lämmle38,Reference Kristiansen, Bjelland and Himberg-Sundet39,Reference De Coen, De Bourdeaudhuij and Vereecken46,Reference Seward, Wolfenden and Finch50,Reference Natale, Messiah and Asfour52,Reference Natale, Atem and Weerakoon54,Reference Blomkvist, Wills and Helland55,Reference Kornilaki, Skouteris and Morris57,Reference Vaughn, Hennink-Kaminski and Moore58,Reference Başkale and Bahar60,Reference Ray, Figuereido and Vepsäläinen63,Reference Namenek Brouwer and Benjamin Neelon69,Reference Witt and Dunn70) and was not reported for four (11 %) studies(Reference Reyes-Morales, González-Unzaga and Jiménez-Aguilar43,Reference Lerner-Geva, Bar-Zvi and Levitan66–Reference Zeinstra, Vrijhof and Kremer68) . For those studies partially reporting data, this included fifteen studies only reporting reach and three studies only reporting acceptability. Overall, of those studies reporting reach (i.e. the number and representativeness of participants, relative to the target population) (n 31), this was often reported in the context of the trial evaluation (e.g. consent and attrition rates)(Reference Ray, Figuereido and Vepsäläinen63,Reference Blomkvist, Helland and Hillesund74) , rather than reach of the intervention to ECEC services, staff, children and parents (if applicable)(Reference Ward, Chow and Humbert75). Overall, reporting on the acceptability (n 19) of the intervention (or components of) among any end-user or stakeholder was most commonly from the perspective of ECEC staff and parents. Intervention acceptability included formal measurement via questionnaires(Reference Pinket, Van Lippevelde and De Bourdeaudhuij59,Reference De Craemer, Verbestel and Verloigne76) or qualitative interviews(Reference Kipping, Langford and Brockman37,Reference Langford, Jago and White77) with participants with findings reported in study results and brief statements of acceptability reported in discussions(Reference Iaia, Pasini and Burnazzi47,Reference Hu, Ye and Li65) .
Discussion
This is the first study to assess the extent to which the factors required to assess scalability have been reported among healthy eating interventions in the ECEC setting. We found that despite a substantial number of RCTs evaluating the impact of healthy eating interventions on child dietary intake, the reporting of factors important to assess scalability within these interventions is scarce, with no studies reporting on all ten factors assessed. Across studies, the reporting of domains ranged between four and nine domains. In total, twenty-three (61 %) studies reported on more than half (i.e. > 5) of the domains of scalability. The studies reporting the highest number of scalability factors were Yoong et al.’s feedAustralia(Reference Yoong, Grady and Wiggers53), Leis et al.’s Healthy Start Départ Santé(Reference Leis, Ward and Vatanparast51) and Kipping et al.’s NAPSACC UK(Reference Kipping, Langford and Brockman37). These three studies fully reported on all factors, with the exception of sustainability, and were published between 2019 and 2020 – more recently than other studies included in this review. Further two of these studies were classified to be at the ‘implementation’ stage of scale-up. These findings may be a result of the growing prominence of implementation research, guidance on developing implementation strategies(Reference Fernandez, ten Hoor and van Lieshout78) and measuring implementation outcomes in this setting(Reference Ben Charif, Zomahoun and Gogovor21), in addition to the benefits of employing hybrid designs to simultaneously evaluate intervention effectiveness and implementation(Reference Curran, Bauer and Mittman79), which are aligned to some of the factors recommended to assess intervention scalability. This finding also highlights that the opportunity and appropriateness of reporting domains of scalability may differ based on the type of study and stage of scale-up. As trials move through the translational pipeline from efficacy through to dissemination the focus becomes less about the internal validity of an intervention, with greater emphasis on external validity, and therefore broad assessments of intervention impact in the real world (with greater consideration to scalability domains such as acceptability, reach for example).
In terms of individual factors to assess scalability, we found the problem, the intervention, effectiveness and the delivery workforce and setting were the most frequently reported, with relatively low reporting of the domains of fidelity and adaptation and reach and acceptability among included studies. These findings are broadly similar to reviews assessing the scalability of home telemonitoring-based interventions(Reference Azevedo, Rodrigues and Londral25) and infant obesity prevention interventions(Reference Yoong, Turon and Wong24), also based on the ISAT. Data relating to the cost and sustainability domains, however, were the least reported factors, with only 13 % and 16 % of included studies reporting these, respectively. These findings are similar to reviews assessing implementation interventions within the ECEC setting(Reference Wolfenden, Barnes and Jones26,Reference Sanchez-Flack, Herman and Buscemi27) , however, are in contrast to reviews outside of the ECEC setting which found the cost domain to be reported in 43–77 % of studies and sustainability reported in 50–77 % of studies(Reference Yoong, Turon and Wong24,Reference Azevedo, Rodrigues and Londral25,Reference Gyamfi, Vieira and Iwelunmor80) . The practical implications of these findings are substantial as this lack of information means that decisions whether to scale-up ECEC-based healthy eating interventions (or not) are being made in the absence of critical evidence regarding budgets and infrastructure (resources, including processes and delivery workforce) required to implement these interventions and the longer-term impact (or lack thereof) of such interventions. Consideration of the long-term availability of the required infrastructure for intervention delivery, alongside the use of guides and frameworks to support the development and selection of implementation strategies likely to facilitate intervention sustainability, may represent examples of how researchers can plan for sustainability(Reference Fernandez, ten Hoor and van Lieshout78).
It is important to recognise that the variability in reporting of scalability factors within the current, and across other reviews, may be due to the type of information conventionally reported within journal articles, with some domains (particularly those related to implementation) only receiving more attention in recent years. There are also substantial challenges for researchers in terms of being able to measure and report on every factor of scalability while considering participant burden and funding constraints. Often the limited and competitive funding for research is insufficient to cover the costs for collection of data relating to all domains of scalability, in particular long-term follow-up (sustainability) or for formal economic evaluations. Further, design requirements of included studies (e.g. presence of a control arm) and challenges in conducting comparative effectiveness and factorial trials likely contribute to the lack of reporting regarding the relative advantage of the intervention over existing interventions (within the effectiveness domain). Previous research suggests there are differing levels of perceived importance of scalability domains across different health conditions, settings, contexts and individuals (researchers, policy-makers and practitioners)(Reference Milat, King and Bauman20). While differing levels of importance have been identified for public health(Reference Milat, Bauman and Redman81) and nutrition and physical activity interventions broadly(Reference McKay, Naylor and Lau82), this is yet to be explored within the ECEC setting specifically.
As the weighting of scalability domains is likely to impact recommendations on whether an intervention should be scaled-up or not(Reference Azevedo, Rodrigues and Londral25,Reference Lee, Milat and Grunseit83) , investigation into the relative importance of some factors of scalability to decision makers and how these should be defined and measured(Reference Ben Charif, Zomahoun and Gogovor21,Reference Azevedo, Rodrigues and Londral25) , in the ECEC setting is warranted. This should be conducted from multiple perspectives, including researchers, policy makers, practitioners, funding bodies, ECEC staff, ECEC governing and advocacy bodies (e.g. the Australian Children’s Education and Care Quality Authority, Canada’s Federal Secretariat on Early Learning and Child Care, the Child and Adult Care Food Program in the US), and the community(Reference Ben Charif, Zomahoun and Gogovor21). Such information could guide future reporting of public health nutrition interventions.
Given the failings of public health nutrition interventions to retain their effectiveness when implemented at scale(Reference Sutherland, Jackson and Lane18,Reference Lane, McCrabb and Nathan84) , it is recommended that interventions be designed, evaluated and reported with scalability in mind(85). The process of designing for scale and evaluating scalability, in addition to the outcomes of intervention scalability assessments, should be reported and published(Reference Gyamfi, Vieira and Iwelunmor80) to facilitate transparency and support decision making by policy makers and practitioners looking to implement and scale public health interventions(Reference Lee, Milat and Grunseit83,Reference Barnes, Grady and Yoong86,Reference Zamboni, Schellenberg and Hanson87) . A recent brief by Barnes et al.(Reference Barnes, Grady and Yoong86) provides an example of this, detailing how scalability was prioritised within the evaluation of a web-based program to improve child nutrition in ECEC, with a scalability assessment guided by the ISAT. A number of other avenues may also facilitate improvements in the reporting of factors of scalability. For example, policy and practice decision-makers should advocate and/or require such processes and data be reported or collected, prior to selecting public health interventions to be scaled-up. In addition, funding bodies and journals could employ guidelines which prioritise the evaluation and reporting of such data. For example, the SUCCEED project (standards for reporting studies assessing the impact of scaling strategies) aims to develop reporting guidelines for scaling studies and could be recommended for studies that have a public health application(Reference Gogovor, Zomahoun and Ben Charif88).
Strengths and limitations
A number of limitations in the design of the current study need to be acknowledged. First, included studies were restricted to those identified in a previous review. Some relevant studies reporting on healthy eating interventions in ECEC (not meeting the Cochrane systematic review criteria) may therefore not be captured here; however, it is likely this review provides a comprehensive list of all ECEC-based healthy eating RCTs. Second, the appropriateness of assessing the domains of scalability solely within published journal articles should also be considered, as the content of journal publications are impacted by journal requirements. The ISAT identifies a variety of information sources that can be drawn upon for completing scalability assessments in addition to published literature, including any available evaluation reports, grey literature, practice-based information and expert option(Reference Milat, Lee and Grunseit23). While it would be helpful for all studies to report on the domains of scalability, we recognise journal articles are not the sole source of information for policy makers and practitioners, who will likely use such evidence reported here, in addition to other sources of data (e.g. local data on workforce capacity, local policies) when making judgements to inform selection of interventions for scale. Additionally, while the current study provides an overview of which domains of scalability are reported within included journal articles, we did not systematically extract data relating to the content of each domain. Third, we employed a crude approach to categorising study stage of scale-up, based on study primary aims and outcomes, the delivery environment and delivery personnel. As the transition from efficacy to effectiveness exists on a continuum(Reference Gartlehner, Hansen and Nissman89), future reviews may benefit from employing a more comprehensive approach to classifying study stage of scale-up (e.g. PRagmatic Explanatory Continuum Indicator Summary-2)(Reference Loudon, Treweek and Sullivan90), including assessment of a range of study design features (e.g. participant eligibility, intervention flexibility and analysis approach). Finally, although the issue of equity may be captured within the adaptation and reach domains, it is not explicitly assessed within the ISAT, and therefore within this review. Future scalability assessments should include consideration of the ability of an intervention to address health inequities (or ensure they do not contribute to the maintenance and/or exacerbation of health disparities at a minimum) as recommended by the WHO’s ExpandNET framework(91).
Despite limitations, a strength of the review should be noted. While the psychometric properties of tools to assess the scalability of interventions are yet to be established(Reference Ben Charif, Zomahoun and Gogovor21), we utilised the ISAT, one of the more comprehensive and methodologically sound tools for assessing scalability. The ISAT is considered to yield content validity as there was a well defined and rigorous process for developing tool content (including an explicit theoretical, conceptual and practical basis for the tool items and systematic item review by experts)(Reference Milat, Lee and Conte22); and to only have minor methodological flaws, compared with the majority of scalability measures which have important methodological flaws(Reference Ben Charif, Zomahoun and Gogovor21). Given increasing use of the ISAT(Reference Yoong, Turon and Wong24,Reference Azevedo, Rodrigues and Londral25,Reference Lee, Milat and Grunseit83,Reference Barnes, Grady and Yoong86,Reference Calnan, Lee and McHugh92) , the findings of this study yield relevant information for policy makers, practitioners, program managers and researchers and identifies gaps for researchers seeking to undertake research in the field.
Conclusion
This review found that while a substantial number of RCTs have evaluated the impact of ECEC-based healthy eating interventions on child diet, the reporting of key scalability domains particularly cost/cost-effectiveness and sustainability remain scarce. At present, there is insufficient information for policy makers and practitioners to select ECEC-based public health nutrition interventions that are able to be delivered at scale, while maintaining meaningful effects on health outcomes. Reporting on all factors required for assessing scalability should be considered to support policymakers and practitioners selecting ECEC-based public health nutrition interventions for scale-up.
Acknowledgements
We would like to thank Nancy Garter and Dylan Price for their contribution to the manuscript.
Financial support
This project was funded by internal research funds provided by the University of Newcastle, School of Medicine and Public Health and the National Health and Medical Research Council (NHMRC) funded Centre for Research Excellence (National Centre of Implementation Science - APP1153479). The funding bodies had no role in the design, analysis or writing of this article. A.G. is supported by a Heart Foundation Postdoctoral Fellowship (102 518). S.L.Y. is funded by a Heart Foundation Future Leader Fellowship (106 654). L.W. is supported by an NHMRC Investigator Grant (APP1197022). Infrastructure support was provided by Hunter New England Population Health and University of Newcastle.
Conflict of interest
There are no conflicts of interest.
Authorship
A.G. and S.L.Y. conceived the idea. A.G., S.L.Y., J.J. and L.W. contributed to the methods. A.G., J.J. and M.L. undertook data extraction. A.G. and J.J. drafted the manuscript. All authors provided critical comments and final approval for the manuscript.
Ethics of human subject participation
Not applicable.
Open access
