A systematic review of complementary feeding practices in South Asian infants and young children: the Bangladesh perspective

Background Sub-optimal nutrition among children remains a problem across South Asia (SA). Appropriate complementary feeding practices (CFP) can greatly reduce this risk. The primary objective of this systematic review (SR) of CF studies was to assess timing, dietary diversity, meal frequency and influencing factors in children under two in Bangladesh. Methods Searches included English-language research published between January 2000 and June 2016 within MEDLINE, EMBASE, Global Health, Web of Science, OVID Maternity & Infant Care, BanglaJOL, Cochrane Library, CINAHL, POPLINE and WHO Global Health Library. Eligibility criteria: primary research concerning the adequacy of complementary feeding practices in South Asian children aged 0–2 years and/or their families. We excluded interventional papers and those focusing exclusively on breast-feeding. In total 45,712 titles and abstracts were screened against inclusion criteria, 860 of which received independent full text review by two reviewers. 36 papers relevant to Bangladesh were identified. The ‘EPPI-Centre Weight of Evidence Framework’ was used to objectively assess each study’s value in answering the review question. As per WHO Infant and Young Children Feeding Guidelines (IYCF), introduction of CF was assessed as the proportion of infants aged 6–8 months who received solid, semi-solid or soft foods. Search terms were: “children”, “feeding” and “Asians” with their derivatives. Two researchers undertook study selection, data extraction and quality appraisal. Results Three cohort, 30 cross-sectional and 3 mixed methods studies were included. Despite adopting the WHO IYCF Guidelines, sub-optimal CF practices were found in many studies. Timely initiation of CF practices ranged from 24 to 83%. Achieved minimum dietary diversity ranged from 25% to 44% and minimum meal frequency from 33% to 81%. Influencing factors included maternal education, poor knowledge of CF practices and socioeconomic variables. Conclusions This is the first systematic review to evaluate CF practices in Bangladesh. Despite adoption of the WHO IYCF guidelines, inadequate CFP remain in communities across Bangladesh. Trial registration PROSPERO Registration No: CRD42014014025. Electronic supplementary material The online version of this article (doi:10.1186/s40795-017-0176-9) contains supplementary material, which is available to authorized users.


Background
Undernutrition including stunting and suboptimal breast feeding accounts for 45% of all childhood deaths [1]. It is estimated that 70% of the world's stunted children live in Asia with Bangladesh having the second highest rate of child undernutrition in the world [1,2].
The WHO defines complementary feeding as: "The process starting when breast milk alone is no longer sufficient to meet the nutritional requirements of infants, and therefore other foods and liquids are needed, along with breast milk" [3].
Complementary feeding therefore focuses on bridging the gradual transition between 6 and 24 months from exclusive breastfeeding to solid foods eaten by the whole family alongside breastfeeding. Poor CFP have been linked to increased risks of respiratory and gastrointestinal infections alongside mortality [4,5]. Only 71% of Bangladeshi infants consume appropriate complementary foods by 6 to 8 months of age and often '[fall] off the growth curve' once CF begins [2,6]. Therefore, it is vital that WHO recommendations on CF are widely adopted [7]. Appropriate CF requires sufficient household food availability and adequate nutritional knowledge application by caregivers [8]. While vast numbers of studies exist on BF practices, CF remains less extensively studied.
The 2010 WHO Infant and Young Children Feeding (IYCF) guidelines, an internationally ratified framework adopted in Bangladesh, emphasises that as a global public health recommendation, infants should be exclusively breastfed for the first 6 months of life to achieve optimal growth, development and health [9]. Thereafter, infants should receive safe and nutritionally adequate complementary foods while breastfeeding continues for up to 2 years of age or beyond.
With no previously published systematic review identified, we aimed to assess the adequacy of CFPs based on IYCF recommended minimum dietary diversity and meal frequency, timing of introducing CF and barriers and promoters influencing CFPs amongst South Asian (SA) children.
To limit the scope of our review, we focused on SA families residing in Bangladesh, India, Pakistan and Developed Countries (DC). This would inform future work in developing and assessing the effectiveness of culturally appropriate interventions to improve CFPs across these communities.

Methods
Due to the vast number of publications identified, this review summarises publications of CFP in SA families in Bangladesh only with concurrent reviews summarising publications of CFP in SA families in DC, Pakistan and India respectively. This was done in order to allow the reviews to go into sufficient depth on the details of CFP in each country. PRISMA guidelines/methodology were adhered to in this review.

Eligibility criteria
Studies were included if they met the following criteria: Participants: Children aged 0-2 years, parents, carers and/or their families Outcomes: Adequacy of complementary feeding (based on minimum dietary diversity and meal frequency), timing of introduction of CF and barriers/promoters to incorporating WHO recommended CFP Language: Studies published in English, or with translation available Year: Published from 2000 or later In the IYCF indicators, introduction of CF is assessed as the proportion of infants aged 6-8 months who receive solid, semi-solid or soft foods. In contrast, minimum dietary diversity is assessed by the proportion of 6-23 months of age who receive foods from 4 or more food groups. The 7 WHO IYCF recommended food groups consist of [7]; 1. Grains, roots and tubers (G) 2. Legumes and nuts (L) 3. Dairy products (e.g. milk, yoghurt, cheese) (D) 4. Flesh foods (e.g. meat, fish, poultry, and liver/organ meats) (F) 5. Eggs (E) 6. Vitamin A rich fruits and vegetables (A) 7. Other fruits and vegetables. (V) Whilst the consumption of iron rich or iron fortified foods is commonly assessed as a separate IYCF indicator, this was incorporated within dietary diversity for ease of interpretation.
Finally, minimum meal frequency (MMF) was assessed by the proportion of breastfed and non-breastfed children 6-23 months of age who receive solid, semi-solid, or soft foods (but also including milk feeds for nonbreastfed children) according to the minimum number of times or more per day; 2 for 6-8 months, 3 for 9-23 months and 4 for 6-23 months (if not BF) [8].
Due to the nature of the topic, all study types (qualitative, quantitative or mixed) were included to ensure the diversity of evidence was captured and summarised to be of relevance to both policy makers and health and social care professionals. We excluded studies focusing on exclusive breastfeeding and interventional studies, as such studies did not reflect the focus of this review, mainly CF practices.

Information sources
A search strategy was devised to search the following databases: MEDLINE, BanglaJOL, EMBASE, Global Health, Web of Science, OVID Maternity & Infant Care, The Cochrane Library, POPLINE, CINAHL and WHO Global Health Library. Searches were conducted in December 2014 and updated in June 2016.
Members of electronic networks on @jiscmail.ac.uk including minority-ethnic-health and networks (eg. South Asian Health Foundation) developed from the Specialist Electronic Library for Ethnicity and Health were contacted to request any additional or unpublished material from members of the networks. Bibliographies of included articles were also hand-searched for possible additional publications.

Search strategy
The search strategy included the terms "feeding", "South Asian" (including terms specifying all major subgroups) and "children". For example, the search strings used for MEDLINE were:

Study selection and data extraction
In total, 45,712 titles and abstracts were screened against inclusion criteria. Two reviewers assessed these papers independently with conflicts resolved by discussion with the team. In view of the large number of articles deemed eligible for full-text review, articles published before the year 2000 were excluded, ensuring that included papers were strongly focused on current practice, closely preceding and also following the 2010 WHO guidelines. In total, 44,852 titles and abstracts were excluded, as shown in Fig. 1.
This left 860 potentially eligible full text articles describing CF practices in SA children, which were independently reviewed by two reviewers. 136 full text Data was extracted by a single reviewer using a piloted modified worksheet including: country of study; study type; study year; study objectives; population studied, eligibility criteria and illness diagnosis; study design; sampling; data collection and analysis; feeding behaviours; adequacy of CF practices; timing of initiation of CF; comments to aid interventional development; weight of evidence. A second member of the research team checked each extraction. Extractions were then doublechecked to ensure accuracy.

Result synthesis
The eligible studies tended to address very broad research questions, were conducted using qualitative and/ or quantitative and/or descriptive methods, reported on diverse outcome measures and were not presented following standardised reporting guidelines (e.g. STROBE for observational studies or COREQ for qualitative research). Meta-analyses were therefore not undertaken.
To standardise study classifications, the formal definitions below were utilised and applied: 1. Cohort study; An observational study in which a group of patients are followed over time. These may be prospective or retrospective. 2. Cross sectional study; An observational study that examines the relationship between health-related characteristics and other variables of interest in a defined population at one particular time. 3. Mixed methods; A study that combines both quantitative and qualitative methodology.
In view of the considerable heterogeneity in studies identified in terms of methods, participants, interventions and outcomes, a narrative approach to synthesis was utilised using guidance developed from the University of York Centre for Reviews and Dissemination (CRD) and the Economic and Social Research Council (ESRC) [10,11].
The evidence reviewed is presented as a narrative report with results broadly categorised following IYFP indicators on; (1) adequacy of CFP comprising of dietary diversity, meal frequency, timing of introduction of CFP and consumption of iron-rich foods and (2) barriers/promoters influencing CF practices.
Barriers were defined as obstacles or impediments to achieving correct CFP whilst promoters were defined as supporters to achieving correct CFP. These were subcategorized into factors influencing at the family (e.g. family members), and organizational level (e.g. health care providers, hospitals, political bodies).

Quality assurance
The Centre for Reviews and Dissemination (CRD) guidance emphasises the importance of using a structured approach to quality assessment when assessing descriptive or qualitative studies for inclusion in reviews. However, it acknowledges the lack of consensus on the definition of poor quality with some arguing that using rigid quality criteria lead to the unnecessary exclusion of papers [10].
In our review, the EPPI-Centre Weight of Evidence Framework was used for objective judgements about each study's value in answering the review question. It examines three study aspects: Quality of Methodology, Relevance of Methodology and Relevance of Evidence to the Review Question and categorises them to Low, Medium or High. An average of these weightings is taken to establish the study's Overall Weight of Evidence. Studies with an Overall Weight of Evidence of Low are still included in the table of included studies but not discussed further within the results and discussion. This was performed by two independent reviewers, with additional arbitration by other team members where required [12].

Study and participant characteristics
The 36 studies consisted of 30 cross-sectional, 3 cohort and 3 mixed methods studies. Additional file 1: Table S1 denotes a summary of all the included studies. Figure 2 illustrates the study locations of 25 of the 36 included studies which provided mappable locations. Table 1 presents the Weight of Evidence awarded to each of the studies. The core narrative themes extracted from the papers are presented under the headings; adequacy and factors influencing CFP. The former is categorised further into dietary diversity, meal frequency, timing, iron fortification and food hygiene.

Adequacy of complementary feeding
As per the WHO IYCF indicators, adequacy of CFP is assessed according to minimum dietary diversity (MDD), meal frequency and timing of introducing CFP. These are detailed in the subheadings below.
Three studies use data from the nationwide Bangladesh Demographic Health Surveys (BDHS), which showed that in 2007 MDD was achieved in approximately 44% of children [6,8]. The 2007 BDHS combined groups M and E into a single "meat, fish, and eggs" group; this figure may therefore underestimate the true value. In the 2011 BDHS the reported rate had fallen to 25% [19]. A large cohort study conducted in rural subdistricts in 2011 provides an even lower number; Owais et al. (Weight of Evidence = High (WOE = H)) interviewed 2073 mother-child pairs over 9 months, and found that just 16% of children received the necessary four food groups per day to satisfy MDD [15]. conducted in 2010 on 1211 infants, found that 31.1% achieved MDD [16].
The food used for complementary feeding could be categorised into WHO IYCF food groups for 27 of the 36 studies. In general, foods were heterogeneously classified; many authors combined or added to the WHO food groups, and some did not specify the ingredients of categories -eg. "thick" foods, "gruel", "family foods", or "hotchpotch".
While maternal and child dietary diversity was associated, the proportions of each food group differed. Both Owais et al. and Nguyen et al. noted that mothers were less likely to provide flesh foods (F) and fruit and vegetables (A, V) to their children, even if these featured in the maternal diet [15,16]. Hackett et al. (WOE = M) described that while vegetables were perceived to be good complementary foods, just 14% of the women interviewed suggested feeding meat, fish, or other animalsource foods to young children [26].

Iron rich foods
As explored in the previous section, many studies mentioned meat being used for CF, but type of meat was not often specified. In Rawat et al. (WOE = M) 56% of infants were noted to be iron deficient in 20 rural sub districts. Fish was the most commonly consumed iron-rich food (i.e. by 19% of infants), whilst less than 14% of infants consumed plants as a source of iron [13]. Zongrone et al. (WOE-M) found that 51.46% of infants consumed iron-rich foods generally [8], and Rasheed et al. (WOE = H) found that 10% of infants 6-23 months consumed recommended levels of iron [14].

Food preparation
Six of 36 studies discussed food preparation. Two studies noted that infants' foods were rarely prepared separately from the rest of the family's foods [14,36] H) found that 59.0% of 6-11 month olds, and 59.8% of 12-24 month olds, were fed at the recommended frequency for breastfed children [14]. This was similar to Paul (WOE = M) who found that frequency was appropriate in 53.5% [35]. In Saleh et al. (WOE = M), frequency rates varied by age and exclusive breastfeeding status; they found that 33% of non-exclusively breastfed 6-8 month olds were fed 3 times per day, compared to 81% of exclusively breastfed 12-23 month olds [40].
Three foods; they found that Dal was the most fed daily food, being consumed >1 per day by 19.7% of included children [39].
Timely initiation at 6 months was often reported, but proportions ranged widely. Moore et al. (WOE = M) reported that the mean age for initiation was 6.8 months [33]. Zongrone (WOE = M) and Sultana (WOE = M) reported that CF was timely in around 80%, and 83.1%, respectively [8,39]. This contrasted with Paul (WOE = M), who reported that timely initiation at 6 months occurred in 48.4% [35]. found that 58% of foster children were fed CF before 4 months, compared to 14% in the non-fostered comparison group [42]. Burhan Uddin et al. (WOE = M) did not find a difference in appropriate initiation when comparing tribal and nontribal mothers, with 87% of Garo and 84% Non Garo mothers starting CF before 6 months of age, with 13% and 16% respectively starting CF after 6 months of age [20].
Regarding the extremes of very early or late CF, Kamruzzaman et al. (WOE = M) noted that CF was not initiated until after 12 months by 1 of 54 infants in their sample, and Moore et al. also found initiation took place between 12 and 14 months for a small number of infants [28,33]. Regarding very early CF, Akhtar et al. described 2.4% of their sample commencing CF at 2-3 months, and Saha et al. found that 0.4% of their sample were given CF before 2 months of age [22,37]. Rasheed et al. (WOE = H) described women who fed CF at 2-3 months "when the baby cried due to hunger" and Helen Keller International (WOE = M) described gruel being given before 3 months [14,18].

Factors associated with IYCF practices
We identified numerous factors that influenced CFP. These are summarised in Table 2 as either a Barrier or Promoter and subcategorized as either acting at family or organizational level. Due to conflicting study findings, factors may appear as both a Barrier and Promoter. 16 promoters and 22 barriers influencing CFP were identified.

Promoters
Eight studies identified promoters at the organizational level. Promoters were: Residence in urban area, Interventions (e.g. health and nutrition educators) advocating and stimulating families to practice WHO recommendations on CFP, Mother having higher number of antenatal check-ups, Higher household food security, Higher household wealth index, Presence of local officials to promote food already in household to be used as CF.
The most common promoter of CF at the organisational level was higher household wealth index, identi-  [6,37,45,46].
Fifteen studies identified promoters at a family level. Promoters were: Father's occupation, Mother with secondary or primary education, Knowledge on appropriate CFP recommendations, The practice of responsive feeding (e.g. strategies to overcome poor child appetite), Higher number of food groups consumed in maternal diet, Education on health nutrition by health workers, Mothers produced sufficient milk to feed baby for over 6 months, Mother engages with media sources, Child in question is male, Child's taste and behavioural response to appropriate CF given (e.g. perceived preference).
The most common promoter of good CF at the family level was levels of maternal education, identified in seven studies by Kabir [6,16,23,31,38,39,45]. It is notable that two studies found that children were likely to receive improved CF if they were male [26,37].

Barriers
Eleven studies identified barriers at the organizational level. Barriers were: Residence in rural areas, Bureaucratic policies on IYCF practices and outreach of information on appropriate CFP to public, Mothers having no antenatal check-ups, Living in areas that flood, Low Household wealth index, Lower household food security, Lack of provision of maternity leave.
The most commonly identified barrier at the organisational level was a poor household wealth index, by seven studies - Kabir [6,22,24,25,42,45,46]. Two studies identified that flooding poses a barrier to good CF practices [24,25].
Eighteen studies identified barriers at a family level. Barriers were: Father's occupation, Mother with poor education, Father with poor education, Lack of knowledge on CFP, Maternal death leading to fostering of child, Psychosocial care during feeding, Mothers' strategies to overcome poor child appetite(e.g. force feed child refusing CF), Limited engagement of mother with media, Cultural factors and taboos, Lack of support by family members influence CFP, Higher maternal parity, Perceived /actual inadequacy of Mother's breast milk supply to breastfeed for 6 months, Time allocation of mother to household chores, Child's taste and behavioural response to CF given, and Illness of parent.

Discussion
To our knowledge, this is the first systematic review to assess CFP in Bangladesh. We identified that in many SA families in Bangladesh, WHO IYCF standards on minimum dietary diversity, meal frequency, and timing of introducing CF were not being met.
1 study [6] Mother engages with media sources; newspapers, radio, TV Use of local officials to promote food already in households to be used as CF 1 study [14] Lower household food security 4 studies [15,24,37,44] Lack of provision of maternity leave 1 study [44] eggs, fruit and vegetables [15][16][17]. Whilst raising awareness on MDD by improving knowledge is likely to be beneficial, the delivery mode of such an intervention needs to be established [6]. With evidence of an increasing prevalence of iron deficiency anaemia within 3 months of CF initiation, supporting mothers' use of iron rich or fortified foods is essential with some advocating the use of universal distribution of iron fortified foods for low households [13,18,19,47]. Levels of maternal education have been identified as an influence on CFP. Some have argued that as most rural girls have become wives prior to the age of 18 despite laws preventing this, impeding schooling, tougher legal enforcement is necessary [41]. In contrast, some have argued that social and financial empowerment of future mothers is a more effective upstream intervention alongside better education and caregiving support [13]. This may be particularly important for working women, given that two studies in this review identified maternal employment as a barrier to appropriate CF [24,36].
Household wealth indexes were identified by several studies as associated with CF practices. Long term interventions such as improving secondary education and financial aid may be beneficial in improving household wealth indices [24].
Improved education on appropriate CFP and household wealth is unlikely to lead to practice changes in isolation; behavioural change interventions addressing local cultural beliefs may be equally necessary to improve CFP [45]. As this study identified that family members can influence CF, interventions should be aimed at wider communities and relatives, not just parents [14,26,35]. Alongside this, improved basic amenities in flood prone, rural and slum areas is necessary to reduce the morbidity associated with gastrointestinal infections secondary to poor food hygiene [24,37].

Strengths and limitations
The strengths of our systematic review are derived by searching 10 databases utilising broad search strings, performing an updated search in June 2016 and having two reviewers undertake study selection, data extraction and quality assessment. By identifying factors that promote and are barriers to CFP, this study could assist with the development of interventions to improve CFP in future.
A key limitation of this paper is the exclusion of a large number of papers published before the year 2000 at full text review. Although this limited the range of the search, it allowed the final paper to be focused on current practice especially in the context of WHO IYCF recommendations. Other limitations include excluding papers which solely focused on children over 2 years, where CFP described in their younger years may have been missed, and papers not published in English, which would have both added to the diversity of CFP described.
A further limitation is that in several studies where there was overlap between children under and over 2 years and/or SAs by Indian, Pakistani and Bangladeshi origin, CFPs described and attributed to the whole study population maybe incorrect. Furthermore, in the Diversity section, we did not develop a composite measure of the quantities of the foods used, only the frequency with which they appeared in the studies.
Whilst we excluded interventional studies which may have described CFP in their study population, this is unlikely to be the primary focus of such studies and therefore unlikely to significantly affect our systematic review.
Finally, whilst we attempted to contact numerous authors to identify relevant grey literature for our review, due to the breadth and depth of the field of nutritional research this study is unlikely to be exhaustive with publication bias likely to be present.

Conclusion
This systematic review has highlighted CFPs covering dietary diversity, frequency, timing, hygiene, preparation, and the factors that influence them, which may provide guidance for development of context-tailored interventions. Despite adoption of the WHO IYCF guidelines, inadequate CFP remain in SA communities across Bangladesh. Increased awareness of this issue can pave the way for improvements in the future.

Additional file
Additional file 1: Table S1. Description of data: A summary of all included studies. (DOCX 57 kb)