Age of introduction of complementary feeding and overweight in adolescence and adulthood: A systematic review

Abstract

Recent studies on early infant feeding suggest that the type of diet and age of starting complementary foods may have a positive correlation with overweight in adolescence and adulthood. This study aimed to systematically review the evidence on the relationship between the age of introduction of complementary feeding and excess weight in adolescence and adulthood (PROSPERO: CRD42017067764). The preferred reporting items for systematic reviews and meta‐analyses standards were used as a reference. Articles were researched on Medline via PubMed, Web of Science, Embase, Lilacs, Ovid, and Scopus, between June and December 2017. Descriptors were defined according to Medical Subject Heading. Of the 103 articles selected for complete reading, nine were included in the review. Of these, only three found an association between the age of introduction of complementary feeding and overweight. Despite these findings, the studies presented a high heterogeneity, mainly due to the difference in cut‐off points for the age of introduction of complementary feeding and the classification of overweight. We concluded that the evidence was not consistent enough to confirm the existence of the association.

Key messages.

• In this review, of the nine studies selected, only three found a significant association between the age of introduction of complementary feeding and excess weight in adolescence and adult life.

• Significant differences were observed between cut‐off points for the age of introduction of complementary feeding and different references for overweight classification.

• The evidences found were not considered consistent due to the high heterogeneity presented between the studies.

• Most of the studies presented more than 80% of adequacy of methodological quality and moderate risk of bias.

1. INTRODUCTION

Obesity is one of the most serious public health challenges of the 21st century. The problem is global in nature and has constantly been affecting many low‐ and middle‐income countries, particularly in urban areas (World Health Organization [WHO], 2018). Data from the WHO show that obesity rates worldwide tripled from 1975 to 2016. In 2016, more than 1.9 billion people over 18 years of age were overweight, 650 million of these being obese. The prevalence of excess weight in adults was 39% for overweight and 13% for obesity. Considering the young people, world data show overweight and obesity rates for children aged 5–19 years increased from 4% in 1975 to 18% in 2016 and 1% to 6% (boys) and 8% (girls), respectively (WHO, 2017).

The surge in the obesity epidemic has a particularly harmful effect on the cardiovascular health of children and adolescents (Friedemann et al., 2012; Funtikova, Navarro, Bawaked, Fíto, & Schröder, 2015) and the type of feeding adopted in the first years of life may be associated with overweight in adolescence and adult life (Matos, Barreto, Rodrigues, & Al, 2014; Stettler, 2007; Weng, Redsell, Swift, Yang, & Glazebrook, 2012). This is why the quality and quantity of foods consumed in the first years of life have been considered extremely important for child development and has lifelong repercussions (Saldiva, Escuder, Mondini, Levy, & Venancio, 2007; Silva, Venancio, & Marchioni, 2010).

According to the WHO, adequate infant feeding includes the practice of breastfeeding and the timely introduction of appropriate complementary foods. The child should preferably be breastfed for 2 years or more, exclusively for the first 6 months of life, with complementary feeding introduced from that age onwards (WHO, 2000). However, this is not a consensus. The American Academy of Pediatrics, the ESPGHAN Committee on Nutrition, and the European Food Safety Authority recommend the introduction of complementary feeding between 4 and 6 months of age without causing harm to the child's health, including overweight (American Academy of Pediatrics, 2005; ESPGHAN Committee on Nutrition, 2008).

Some researchers suggest that the introduction of complementary foods before 4 months of age constitutes an important risk factor for overweight in childhood, which may persist later in life (Fall et al., 2011; Schack‐nielsen, Sørensen, Mortensen, & Michaelsen, 2010; Seach, Dharmage, Lowe, & Dixon, 2010).

However, other authors do not support the evidence linking the age of introduction of complementary foods and later obesity (Garden, Marks, Simpson, & Webb, 2012; Lin, Leung, Lam, & Schooling, 2013; Neutzling et al., 2009; Symon, Crichton, & Muhlhausler, 2017).

Considering these divergences about the theme and the importance of the appropriate introduction of complementary feeding for growth and development in the childhood and their repercussions in later life, this study aimed to systematically review the evidence on the relationship between the age of introduction of complementary feeding and excess weight in adolescence and adulthood.

2. METHODS

2.1. Research strategies

A systematic review was conducted and registered in the PROSPERO Protocol (York University) under the number CRD 42017067764, evaluating the relationship between the age of introduction of complementary feeding and overweight in adolescence and adulthood. The preferred reporting items for systematic reviews and meta‐analyses standards (Moher, Liberati, Tetzlaff, Altman, & Group, 2009) were used as reference. Articles were independently researched by two researchers (C. S. A. and E. P. M.) in the Medline databases via PubMed, Web of Science, Embase, Lilacs, Ovid, and Scopus between June and December 2017. The descriptors were defined based on the terms indexed by Medical Subject Heading: “Infant Nutritional Physiology” and 11 synonyms (Nutritional Physiology, Infant; Physiology, Infant Nutritional; Infant Nutritional Physiological Phenomenon; Infant Nutrition Physiology; Physiology, Infant Nutrition; Nutrition Physiology, Infant; Supplementary Feeding; Feeding, Supplementary; Supplementary Feedings; Complementary Feedings; Feedings, Complementary); “Obesity” “Body Mass Index,” and 5 synonyms (Index, Body Mass; Quetelet Index; Index, Quetelet; Quetelet's Index; Quetelets Index); “Abdominal Obesity” and 10 synonyms (Abdominal Obesities; Obesities, Abdominal; Central Obesity; Central Obesities; Obesities, Central; Obesity, Central; Obesity, Visceral; Obesities, Visceral; Visceral Obesities; Visceral Obesity); and “Overweight,” with the Boolean operators AND and OR being used in all their combinations.

Manually researched studies were also included in the reference list of selected articles. The search strategy considered the issue under investigation, structured by the acronym PECO (population: children under 2 years of age; exposure: age of introduction of complementary feeding; control: age of introduction of food after 6 months of age; outcome: presence of overweight in adolescence and adulthood). Only the terms for the E (exposure) and O (outcome) components were defined, in order to avoid undesirable specifics in the data search.

This review included published studies with a cohort‐type observational design, without restriction as to language and year of publication, which analysed the association between the age of introduction of complementary feeding (independent or exposure variable) and overweight (response variable or outcome) in adolescence and adulthood. The exposure variable (complementary feeding) was analysed according to the age (in months) at which complementary foods were introduced into the child's diet. The outcome variable was analysed according to anthropometric data (body mass index [BMI], waist circumference [WC], subscapular skinfolds [SSF], and body composition) and the classification criteria for overweight and obesity adopted in each study for adolescents and adults.

We did not include studies with children born preterm and those with pathologies (e.g., diabetes, mental disorders, kidney disease, hypertension, HIV, cancer, and Down syndrome), intervention studies, literature reviews, reports and editorials, samples composed of children older than 2 years of age at the time of exposure, and younger than 10 or over 60 years of age for outcome.

The articles identified in the databases were selected by two independent reviewers (C. S. A. and E. P. M.), who followed the selection criteria for the studies. First, the titles and abstracts were read in order to exclude certain studies. There were no differences of opinion between the two reviewers.

For five selected articles that did not provide all the necessary information for the meta‐analysis, the article's author was contacted by email in order to obtain more details and data. However, only one author replied. The reference list of articles found and thematic reviews were also searched, but no further studies were found. Figure 1 summarizes the selection process of the studies.

Figure 1.

Flowchart of the article research and selection process

2.2. Data extraction

The original articles included in the final list were read in full, and the information contained therein was recorded in an Excel® worksheet, created by a single researcher (C. S. A.), and standardized in order to evaluate quality and synthesize the evidence: author, year of publication, country of study, sample size, population, exposure assessment method, outcome assessment method, and description of results.

The results of the studies were not combined in meta‐analysis due to the absence of information for four studies, to the considerable heterogeneity among the studies in relation to the exposure measurement, as the age of introduction of complementary feeding, the concept of complementary feeding, and the classification of excess weight were not consistently categorized and/or standardized in all studies.

2.3. Evaluation of methodological quality

The evaluation of methodological quality was performed by two independent researchers (C. S. A. and J. C. D. P.), using the Checklist Downs and Black (appendix 2) for randomized and nonrandomized studies, composed of four scales: report, external validity, internal validity (bias and confusion), and power of the study. The report assesses whether the information provided will allow the reviewer to evaluate studies without bias. External validity addresses the extent to which the findings from the study can be generalized to the population from which the study subjects were derived. Internal validity refers to the applicability of study results to the subject under investigation. The power of the study analyses whether the negative results for the study could be due to chance. In order to evaluate the methodological quality of the studies included in this review, only the questions applicable to cohort studies were used (Downs & Black, 1998). The quality score of each article corresponds to the sum total of items analysed as positive, with a minimum score of 14 and maximum score of 18 points. The articles were considered as having adequate methodological quality if they obtained a score equal to or greater than 80% of the maximum score (Rodrigues, 2013).

2.4. Risk of bias assessment

The risk of bias assessment of the included studies was performed by two independent researchers (C. S. A. and J. C. D. P.). The Research Triangle Institute Item Bank (appendix 3), proposed by Viswanathan and Berkman (2012), was used as a reference. The Research Triangle Institute Item Bank covers a variety of different study designs, which include 13 questions for observational studies (Viswanathan & Berkman, 2012). High risk of bias was considered when the study had three or more responses classified as unclear or negative, moderate risk when up to two responses were rated as unclear or negative, and low risk of bias when no responses were considered unclear or negative (Sedlar et al., 2017).

3. RESULTS

3.1. Search results

According to the article search strategy, 18,222 articles were found by reading titles and abstracts in the databases accessed. Of these, 785 duplicate articles were excluded and 17,403 excluded after reading the title and abstract. The full text of 137 articles was read, 34 of which were selected from the references of the articles, and only 9 were included in the systematic review. The reasons for excluding the articles were mainly unsuitability in terms of the study's objectives, as they did not evaluate the age of introduction of complementary feeding and overweight in adolescence and adulthood or had an outcome set at under 10 years of age (Figure 1).

3.2. Description of included studies

The studies included in this review were developed in the following countries: South Africa, Australia, Brazil, China, France, Philippines, Guatemala, Greece, India, England, Portugal, and Turkey. Table 1 shows the summary of characteristics and main results of the studies. In total, 33,329 individuals between 10 and 15 years of age (for adolescence) and between 32 and 42 years of age (for adulthood) participated in the studies; eight studies were of birth cohorts using a representative sample of the population, and one study was of samples from schools and paediatric hospitals. The information was collected between 1959 and 2013 and published between 2009 and 2017. Most of the studies referenced the introduction of complementary feeding as the age at which solid foods were introduced into the child's diet. Only one study (Fall et al., 2011) referenced WHO recommendations (6 months; WHO, 2000) and one (Lin et al., 2013) those of the American Academy of Paediatrics (4 months; American Academy of Pediatrics, 2005).

Table 1.

Summary of studies included in the systematic review and risk of bias assessment

Author	Year	Country	Age	N	Population	Introduction of CFa	Measure of outcomes	Results	RTIb
Garden et al.	2012	Australia	11 years	370	Childhood Asthma Prevention Study (CAPS) study	Age <3 m or >3 m Early introduction <3 m	BMI ≥ P85—overweight/obesity (CDC, 2000)	No association between age of CF and BMI	Moderate
Fall et al.	2011	Brazil India South Africa Philippines	32 years	8.322	Pelotas cohort New Delhi birth cohort Birth of 20 cohort Cebul Longitudinal Health and Nutrition Survey (CLHNS) cohort	0.01–3.0 m; 3.01–6.00; 6.01–9.00; 9.01–12.00; 12.01–18.00; >18 m Late introduction >6 m	BMI ≥ 25 kg m−2—overweight BMI ≥ 30 kg m−2—obesity WHO, 1995 WCc, SSFd	Inverse association between age of CF introduction and overweight, WC, and SSF	Moderate
Leary et al.	2015	England	15 years	4.750	Avon Longitudinal Study of Parents and Children (ALSPAC) cohort	≤2 m, 3 e ≥ 4 m early introduction <2 m	Tanita TBF 305 body fat analyse—percentage analysis	No association between age of CF introduction and body composition	Moderate
Lin et al.	2013	China	14 years	7.184	Hong Kong children cohort	<3 m; 3–4 m; 5–6 m; 7–8 m; >8 m Early introduction <3 m	BMI ≥ 25 kg m−2—overweight IOTF, 2000	No significant association between age of CF introduction and overweight	Moderate
Moschonis et al.	2017	England	13 years	6.124	ALSPAC cohort	<5 m; ≥5 e < 6 m; ≥ 6 m Late introduction ≥4 m	BMI ≥ 25 kg m−2—overweight IOTF, 2000 Bioimpedance, DEXA	No significant association between age of CF introduction and BMI and body composition	Moderate
Neutzling et al.	2009	Brazil	11 years	1.204	Pelotas cohort	<4 m ou > 4 m Early introduction <4 m	BMI > P85 overweight BMI > P97 obesity WHO, 1995	No significant association between age of CF introduction and overweight and obesity	Low
Schack‐nielsen et al.	2010	Denmark	42 years	5.068	Copenhagen perinatal cohort	<4 m e ≥ 4 m Early introduction <4 m	BMI ≥ 25 kg m−2—overweight BMI ≥ 30 kg m−2—obesity WC (WHO, 2007)	Late introduction of CF is associated with smaller BMI and WC values in adulthood	High
Seach et al.	2010	Australia	10 years	307	Melbourn Atopy Cohort (MAC) cohort	Age of introduction in weeks (late ≥20)	BMI z‐score BMI ≥ 25 kg m−2—overweight IOTF, 2000	Late introduction of CF is associated with a smaller chance of overweight and obesity	Moderate
Vehapoglu et al.	2014	Turkey	14 years	4.990	Paediatric Clinic of the Paediatric Hospital of Istambul	<4 m; 4–5 m; ≥6 m Early introduction <4 m	BMI/age/gender BMI > P85 < P94—overweight BMI ≥ 94—obesity Neyzi et al., 2006	No significant association between age of CF introduction and overweight and obesity	Moderate

^aCF: complementary feeding.

^bRTI: Research Triangle Institute Item Bank.

^cWC: waist circumference.

^dSSF: subscapular skinfolds.

3.3. Age of introduction of complementary feeding

From the studies included in this review, seven studies investigated the association between the age at which solid foods were introduced and overweight in adolescence, and two studies evaluated this relationship in adulthood (Fall et al., 2011; Schack‐nielsen et al., 2010). The age of introduction of complementary feeding was recorded as a continuous variable in months, and the definition as early or late varied among the studies: For Garden et al. (2012), Leary, Lawlor, Smith, Brion, and Ness (2015), and Lin et al. (2013), early introduction was defined as the child is being less than 3 months of age; for Neutzling et al. (2009), Schack‐nielsen et al. (2010), and Vehapoglu, Turkmen, Nursoy, and Ozkaya (2014), less than 4 months; and for Fall et al. (2011), Moschonis et al. (2017), and Seach et al. (2010), at an age greater than 4 and 6 months.

3.4. Overweight in adolescence and adulthood

In order to evaluate the overweight, Schack‐nielsen et al. (2010) used the WC, and Fall et al. (2011) used the WC and SSF as indicators of excess weight in addition to BMI. The study by Leary et al. (2015) used the body fat percentage indicator, evaluated by means of the Tanita TBF 305® device, and Moschonis et al. (2017) used the Bioimpedance and dual X‐ray absorptiometry to evaluate body composition. The BMI classifications used were WHO, 1995 and 2007; CDC, 2000; IOTF, 2000; and Neyzi's reference for Turks (Neyzi et al., 2006). In four studies (Fall et al., 2011; Lin et al., 2013; Moschonis et al., 2017; Seach et al., 2010), the BMI classification of IOTF (Cole, Bellizzi, Flegal, & Dietz, 2000) was used.

3.5. Quality of methodology and risk of bias

According to the methodological quality assessment proposed by Downs and Black (1998), it was observed that the majority of the studies presented more than 80% of adequacy as far as quality was concerned, with percentages varying between 83% and 94%; post hoc exploratory analysis was not done in any of the studies. Two studies were classified as being of inadequate quality: Moschonis et al. (2017) with 72% and Vehapoglu et al. (2014) with 44%. The questions that factored into this evaluation the most were losses and follow‐up characteristics that were not considered and described in the study; the sample used not being representative of the population; and the results not being clearly described.

With regard to the risk of bias, as far as the nine studies included in the review are concerned, only one study (Neutzling et al., 2009) was classified as having a low risk of bias. Seven studies were included in the classification of moderate risk of bias; it was not made clear in the studies whether they took into account important variations in the execution of the study based on the proposed protocol, and whether evaluations of the impact of loss to follow‐up were carried out. In Schack‐nielsen et al. (2010) study, the selection of the control group was considered inadequate; the confounding variables were not taken into account in the analyses, and they did not show reliable results; therefore, it was classified as having a high risk of bias.

3.6. Association between age of introduction of complementary feeding and overweight in adolescence and adulthood

In this review, nine studies investigated the relationship between the age of introduction of complementary feeding and BMI in adolescence, and one of these studies study evaluated only the percentage of fat in adolescents. Of the studies that did not find an association between the age of introduction of complementary feeding and overweight in adolescence and adulthood, one study (Vehapoglu et al., 2014) used an unadjusted analysis model; one study (Garden et al., 2012) showed adjusted analyses, and four studies used adjusted and unadjusted models (Leary et al., 2015; Lin et al., 2013; Moschonis et al., 2017; Neutzling et al., 2009).

Garden et al. (2012) used parental BMI, ethnicity, smoking in gestation, and parental occupation after child birth as confounding variables; Leary et al. (2015) used age, gender, parental factors, social factors, birth, and gestational weight; Lin et al. (2013) used gender, gestational age, birth weight, weight from 0 to 3 months, breastfeeding, parity, maternal age at birth, socio‐economic status, and passive smoking; Moschonis et al. (2017) used gender, birth weight, maternal age, maternal educational level, and prepregnancy BMI; Neutzling et al. (2009) used gender, skin colour, birth weight, family income, maternal schooling, smoking in pregnancy, and prepregnancy BMI.

With respect to the age of introduction of complementary feeding, whether early or late, Garden et al. (2012), Leary et al. (2015), and Lin et al. (2013) found no association between the early introduction of complementary feeding (under 3 months) and overweight and changes in body composition in adolescence. For early introduction of supplementary feeding at less than 4 months, Neutzling et al. (2009) and Vehapoglu et al. (2014) did not show a relationship between this age and overweight in adolescence. Moschonis et al. (2017) found that the late introduction of complementary feeding, at an age greater than 6 months, did not influence BMI in adolescence.

In contrast, three studies did find an association between the time of introduction of complementary feeding and overweight in adolescence and adulthood. After adjusting for the confounding variables (age, gender, maternal socio‐economic status, type of housing, ethnicity, and birth weight), Fall et al. (2011) found that the late introduction of complementary foods was inversely associated with lower BMI, WC, and SSF values in adults, 0.21 kg m⁻² (95% CI [−0.03, 0.45]), 0.45 cm (95% CI [−0.11, 1.01]), and 2.6% (95% CI [−0.9, 6.10]), respectively, in categories established according to an increase in the age at which complementary feeding was introduced. When the age of introduction of supplementary feeding was greater than 6 months, the analyses were attenuated and the BMI, WC, and SSF went down to 0.21 kg m⁻² (95% CI [−0.03, 0.45]), 0.45 cm (95% CI [−0.11, 1.01]), and 2.6% (95% CI [0.6, 5.4]), respectively. Schack‐nielsen et al. (2010) found that the introduction of complementary feeding before 4 months was associated with an increased risk of overweight and obesity (β −0.94, CI [−0.86, 1.02]; β 0.94, CI [0.83, 1.08]) and of the WC (β −0.25, IC [−0.49, −0.010]) in adulthood (42 years), even when adjusted for the following variables: maternal age, gender, prepregnancy BMI, gestational weight gain, birth weight, social status, maternal marital status, smoking during pregnancy, prematurity, and duration of gestation. Using unadjusted models and ones adjusted for maternal education, type of housing, parental occupation, parental smoking, child care, and maternal BMI, Seach et al. (2010) found that late introduction of complementary feeding reduces the chances of overweight and obesity in adolescence (OR 0.903 per week, 95% CI [0.841, 0.970], P < 0.005).

4. DISCUSSION

This review aimed to systematically review the evidence on the relationship between the age of introduction of complementary feeding and overweight in adolescence and adulthood. In order to accomplish this, rigorous systematic methods were used to guarantee the selection of the best quality studies currently available. However, no consistent evidence was found between the age of introduction of supplementary feeding and overweight in adolescence and adulthood.

Most of the studies did not find a statistically significant association, mainly after analyses adjusted for potential confounders.

Bearing in mind the WHO's definition of complementary food (WHO, 2000) as the “process of introducing solid or liquid foods when breast milk or other milk is no longer sufficient to meet the nutritional needs of the child”, no study has used this concept as a definition for the exposure variable. All of them referred to complementary feeding as being only the introduction of solid foods, without mentioning liquids, such as juices and others, which could be sweetened and interfered with the evaluated outcomes, this being an important divergence found among studies.

A large number of studies reported the age of introduction of complementary feeding in months, and only one study did so in weeks (Seach et al., 2010). In addition, the cut‐off point for evaluation of the introduction of complementary feeding differed among the studies and did not follow WHO guidelines, that is, start at 6 months of age. Some used early introduction as <3 months (Garden et al., 2012; Leary et al., 2015; Lin et al., 2013) and <4 months (Neutzling et al., 2009; Schack‐nielsen et al., 2010; Vehapoglu et al., 2014) to assess exposure. This lack of standardization of the age of introduction, whether late or early, contributed to increase the heterogeneity of this information among the studies included in this review, so that it was difficult to systematize and analyse the data, making it impossible to consolidate consistent and reliable evidence on the relationship between early introduction of supplementary food and overweight in adolescence and adulthood.

According to the WHO, the age‐appropriate recommendation for the introduction of complementary feeding is based on the maintenance of exclusive breastfeeding until the sixth month of life, taking into account, among other factors, the benefits of exclusive breast milk supply, the maturation of the child's digestive system at that age, when she/he is able to take in foods other than breast milk or infant formula, and the increased nutritional requirements due to the intense growth and development at this stage (Cole et al., 2000; WHO, 2000). However, these recommendations have been questioned by more recent studies suggesting that the introduction of supplementary feeding can be done between 4 and 6 months, age at which the gastrointestinal and renal functions are sufficiently mature allowing complementary feeding to be processed without causing harm to the child's health and taking into account that there is no convincing scientific evidence that avoidance or delayed introduction of potentially allergenic foods reduces allergies, even in infants considered at risk for the development of allergy (American Academy of Pediatrics, 2005; ESPGHAN Committee on Nutrition, 2008; European Food Safety Authority, 2009). In European countries, for example, there is no consensus regarding the age of introduction of complementary feeding. In Belgium, Greece, Ireland, and Spain, the recommendation is between 4 and 6 months; in Denmark, Lithuania, Sweden, and the United Kingdom, the recommendation is 6 months of age, and in Germany, the introduction of complementary feeding should begin no earlier than 5 months and no later than 6 months (European Food Safety Authority, 2009). The ESPGHAN Committee on Nutrition (2008) considers that the gastrointestinal and renal functions are matured around 4 months.

An important datum observed in the studies was the manner of collecting information on the age of introduction of complementary feeding. Most of the studies were performed when children were aged between 3 and 24 months in the cohort records. In one study, a questionnaire asking about the age of introduction of complementary feeding was sent to the mothers of adolescents (Lin et al., 2013), and in another, this information was collected retroactively when the adults were 42 years old (Schack‐nielsen et al., 2010), counting on the mothers' ability to remember such an event, which could affect the internal validity of the evidence, influencing the results.

In addition, another factor that may have contributed to the inconclusive findings of the studies was the lack of standardization of important potentially confounding variables. In this review, it was observed that most of the studies used regression analyses to control the effect of these variables on the relationship between exposure and outcome, the most used being birth weight, educational level, socio‐economic status, smoking, and parental BMI.

The age at which the outcome (overweight) was assessed did not vary much between the studies: 10–15 years for studies on adolescents and 32 and 42 years for adults. The most commonly used outcome measure was BMI. Some studies used BMI, WC, and SSF; others used BMI and body composition, and only one study used body composition alone. Seach et al. (2010) used the BMI on z‐score and the BMI classification by IOTF, from which a healthy and unhealthy BMI classification was created to evaluate outcome. In four studies, the classification used was that of the IOTF, which uses the adult cut‐off points to evaluate adolescents, and the same for the reference for Turks (Neyzi et al., 2006). Other parameters for classification of overweight by BMI were CDC (2000), WHO (1995), and WHO (2007). This fact makes it difficult to compare the results of these studies, because the cut‐off points differ, significantly influencing the classification of the outcome and the systematization of the results, which also contributes to increasing the heterogeneity among the studies.

The included studies varied in terms of sample size. Two studies had small samples: Garden et al. (2012) with n = 307 and Seach et al. (2010) with n = 307, which could decrease their power to detect a positive association between exposure and outcome. The other studies ranged from an n sample of 1,204 (Neutzling et al., 2009) to 8,322 (Fall et al., 2011), and of these, only two found a positive but inconsistent association.

For the studies that showed a positive association between the age of introduction of complementary feeding and overweight in adolescence and adult life, the methodological quality evaluation was superior to 80% (Fall et al., 2011; Schack‐nielsen et al., 2010; Seach et al., 2010). Through the external validity analysis, only in the study of Fall et al. (2011), the results can be generalized for the population from which the participants were selected. In a large number of studies, the characteristics of the patients with loss to follow‐up were not described, which may interfere in the reliability of the results. In the studies that found no positive association, only Vehapoglu et al. (2014) presented a low percentage of methodological quality (44.4%), mainly because the confounding variables, follow‐up time and losses, and participant recruitment period were not determined in the study.

Regarding the assessment of risk of bias, Schack‐nielsen et al. (2010) did not implement reliable and valid measures to evaluate the inclusion and exclusion criteria, and their results were not considered reliable when the limitations of the study were taken into account. Thus, the study was classified as having a high risk of bias. Moderate risk of bias was found in most studies included in this review (seven studies). This result was mainly due to the uncertainties in the evaluations of some cohort questions. Performance bias was one of those most found in all the studies, seen in the nonelucidation of important variations in the studies' execution based on the proposed protocol. Attrition bias was found in some studies that did not refer to the assessment of the impact of loss to follow‐up. And finally, there was an evaluation bias: The results of some studies were also not reliable when their limitations were taken into consideration. Only the study by Neutzling et al. (2009) presented a low risk of bias.

This study was limited by the small number of studies included in the systematic review, which may have been due to predefined methodological criteria, such as the type of study design and the outcome age (adolescents and adults). Unpublished articles were not included in this review, because the search was restricted to electronic database and the manual search of reference lists, which may have omitted some eligible studies.

5. CONCLUSION

In this systematic review, no consistent evidence was found between the age of introduction of supplementary feeding and overweight in adolescence and adulthood. The factors that most contributed to this result were the high heterogeneity among the studies, mainly due to the lack of standardization regarding the cut‐off point of the age of introduction of complementary feeding, each study's concept of complementary feeding, and the classification of overweight. This lack of standardization may be due to a number of factors, including the failure to adopt the recommendations of the WHO regarding the ideal age of introduction of complementary feeding, or to the lack of guidance for mothers regarding this practice. It is also worth mentioning the importance of conducting more consistent studies that investigate not only the relationship to the age of introduction but also the types of food introduced and the way in which these foods are offered to children and their relationship with the increase of overweight in later phases of life.

CONFLICTS OF INTEREST

The authors declare that they have no conflicts of interest.

CONTRIBUTIONS

CSA and PRFC proposed the idea and concept of this review. CSA and EPM carried out the searches and selection of articles. CSA made the complete reading of the selected articles and the extraction of the data. CSA and JCDP carried out the evaluation of methodological quality and risk of bias. CSA wrote the manuscript. PRFC, VAOQ, MLPS, and AMOA participated in the article review process. CSA, PRFC, VAOQ, EPM, JCDP, MLPS, and AMOA edited and approved the final manuscript.

Araújo CS, Costa PR d F, Queiroz VA d O, et al. Age of introduction of complementary feeding and overweight in adolescence and adulthood: A systematic review. Matern Child Nutr. 2019;15:e12796 10.1111/mcn.12796