Association of Protein Intake during the Second Year of Life with Weight Gain-Related Outcomes in Childhood: A Systematic Review

There is accumulating evidence that early protein intake is related with weight gain in childhood. However, the evidence is mostly limited to the first year of life, whereas the high-weight-gain-velocity period extends up to about 2 years of age. We aimed to investigate whether protein intake during the second year of life is associated with higher weight gain and obesity risk later in childhood. We conducted a systematic review with searches in both PubMed®/MEDLINE® and the Cochrane Central Register of Controlled Trials. Ten studies that assessed a total of 46,170 children were identified. We found moderate-quality evidence of an association of protein intake during the second year of life with fat mass at 2 years and at 7 years. Effects on other outcomes such as body mass index (BMI), obesity risk, or adiposity rebound onset were inconclusive due to both heterogeneity and low evidence. We conclude that higher protein intakes during the second year of life are likely to increase fatness in childhood, but there is limited evidence regarding the association with other outcomes such as body mass index or change in adiposity rebound onset. Further well-designed and adequately powered clinical trials are needed since this issue has considerable public health relevance.


Introduction
The World Health Organization (WHO) recommends exclusive breastfeeding during the first six months of life followed by continued breastfeeding along with nutritious complementary foods up to the age of 2 years or beyond [1]. Human milk has a lower protein content compared with typical infant formulas [2,3]. The amount of protein and the amino acid profile in breast milk is specific and genetically regulated for each mammal species, adapted to its requirements [4,5]. Breastfed infants exhibit lower weight gain velocity and a subsequent lower obesity risk later in life compared with those fed a cow's milk-based formula [6][7][8].
There is compelling evidence-from observational studies, as well as some randomized clinical trials (RTC)-demonstrating that a high protein intake during the first year of life is associated with higher body mass index (BMI), higher fat mass, as well as increased risk of overweight or obesity later in life [9][10][11][12]. An RCT designed with this purpose showed an effect of the lower-protein formula in reducing BMI at 2 years and later in

Materials and Methods
We used the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) method to systematically review the articles that assessed the effects of protein intake during the second year of life on childhood health later in life. We conducted the review using the electronic database PubMed ® /MEDLINE ® (https://www.ncbi.nlm.nih. gov/pubmed/) and the Cochrane Central Register of Controlled Trials (https://www. cochranelibrary.com/central) and by searching clinical trials (including RCTs or prospective cohort studies) or reviews published up to 30 May 2020. The search terms conducted in each database are presented in Appendix A. We selected all the studies in humans that fulfilled the aim (effects of protein intake during the second year of life on all outcomes related to weight gain and the implicated mechanisms) without outcome restriction but published in English or Spanish. Study population was restricted to healthy term infants that did not present special needs or altered growth (i.e., we discarded all the studies performed in preterm or low-birth-weight infants as well as all the interventions performed in ill patients). Studies performed in low-income areas (in which there might be a high undernutrition risk or interventions aimed to improve growth in those areas) were also excluded from the analysis. In addition to the studies obtained by database searches, reference lists from reviews, meta-analyses, and the retrieved articles were checked over to identify further relevant studies.
As described previously, our main objective was to assess the effects of the protein intake during the second year of life on excessive weight gain and obesity risk development later in childhood. Primary outcomes were as follows: • All the fatness measurement approaches (like subcutaneous skinfolds (millimeters), % body fat (%BF), fat mass (FM) (kilograms), or fat mass index (FMI) (kilograms/meter 2 ), • Body mass index (BMI) (kilograms/meter 2 or z-score), and • Risk of excessive weight (overweight and obesity risks) (RR, OR, or frequency).
In addition, we also investigated other secondary outcomes: • Weight gain velocity (grams/month, z-scores or those that classified as rapid or normal growth), and • Adiposity rebound (age in months or BMI at the adiposity rebound onset).
When possible, the principal summary measures were reported as differences in means, RR, or.
Literature search was conducted by one author (NF). Questions or uncertainties were solved by discussion with other authors (JE, VL, and RC). The initial search consisted of screening titles and abstracts, and the second step consisted of reviewing full-text articles to confirm or discard the study selection.
The information extracted from each individual study was as follows: first author and year of publication, study design, population and sample size, protein intake assessment (methods and measures), assessed health outcomes, effects and comments, and quality of the evidence.
For each included study, the quality of the evidence was assessed following the SING method that classifies the evidence in eight grades (ranged from 1++ (corresponding to high-quality metanalyses, systematic reviews of RCTs, or RCTs with a very low risk of bias) to 4 (corresponding to expert opinions)) and according to the classification, establishes the study as one of four grades of recommendation from A (when there is at least one metanalysis, systematic review, or RCT rated as 1++ and directly applicable to the target population or a systematic review of RCTs or a body of evidence consisting principally of studies rated as 1+ directly applicable to the target population and demonstrating overall consistency of results) to D (when there is only evidence level 3 or 4 or extrapolated evidence from studies rated as 2+) [29]. This method was developed by the Scottish Intercollegiate Guidelines Network for the NHS in Scotland to improve the current system for grading guideline recommendations.

Results
A total of 3982 references were revealed after duplicate removal and screened. In addition, we obtained 42 manuscripts from the reference lists of reviews and studies, which were also assessed for inclusion. After reading the title and the abstract, a total of 3971 studies were excluded and 53 full-text items were assessed for inclusion. From those, 39 were finally excluded (Appendix A, Table A1) and 14 were included in the systematic review ( Figure 1). Reasons for exclusion were mainly being a narrative review, not assessing the protein intake during the second year of life, or associating the protein intake with outcomes measured at the same time period instead of at later ones. Studies included in quantitative synthesis (meta-analysis) (n =0)  Table 1 summarizes the included studies, grouped by assessed outcomes. Although 14 references were included, those manuscripts provide data from nine different studies with several reported in more than one included publication. We only found published data from one RCT [30], while all other studies were observational cohorts or secondary analyses of an RCT designed with a different purpose. In addition, we detected one registered RCT that is still ongoing (TOMI trial, NTC02907502). • Anthropometry (BMI z-score, %BF, skinfolds, and overweight risk) at 5 years, rapid growth (0-2 y) (defined as weight z-score increases >0.67 (this change was is equivalent to a quartile increase)) Children with rapid growth (0-2 years) exhibited higher BMI z-score (−0.016 ± 0.99 vs. 0.41 ± 0.90, p < 0.001) at 5 years. However, the distribution of children with rapid growth was similar between groups of sustained or not high protein intakes during the first 2 years (H-H vs. H-L). Sustained higher protein intake during the first 2 years of life was modulating BMI z-score at 2 years (β = 0.36 ± 0.13, p = 0.005 for the H-H group compared to the H-L) but had no effect on the longitudinal change of BMI z-score between 2 and 5 years.

2+
Low biases except for a moderate risk of follow-up bias Higher protein intake at 12-18 months was associated with higher BMI z-score at AR onset only in females. BMI z-score (internal z-score) was −0.11 and -0.66 in the upper and lowest terciles of protein intake at 12-18 months of age, respectively. Age at AR onset was independent of protein intake at 12-18 months in both genders.

2+
Unknown risk of follow-up bias because dropouts are not reported Most of the included studies addressed the effect of protein intake on different outcomes related with the risk of overweight, such as BMI, fat mass index (FMI), percentage of body fat (%BF), subcutaneous skinfolds (SbSK), rapid weight gain, adiposity rebound (AR), or risk of excessive weigh [30][31][32][33][34][35][36][37][38][39]41,42,44,45]. We did not find studies assessing the effect of protein intake during the second year of life on blood pressure or neurodevelopment.
Most of the studies assessed the protein intake through dietary diaries over 3 to 7 days, but these were performed at different timepoints during childhood between 12 and 24 months of life. Additionally, the studied health outcomes were assessed at different ages. Different authors analyzed protein intake data with different approaches: some studies addressed total protein intake whereas others differentiated between protein sources, and some studies used protein intake as continuous variable while others categorized subjects in groups of low vs. high intake or compared intervention groups according to the protein content of the formula in an RCT. All these differences made the comparisons between studies difficult and prevented us from performing a pooled quantitative study of the effect using a meta-analysis. Results are presented below, grouped per health outcome.
Wall et al. reported an RCT with low degree of bias, which corresponds to a quality of evidence of 1+. All the other included references were longitudinal prospective cohort studies, so the quality of evidence was considered as 2+ for most of them. The most relevant biases were found for attrition, as 5 out of the 13 cohort study references showed a moderate risk of bias, and another 5 of them a high risk of bias (loss to follow-ups over 30% or 55%, respectively). Most of the studies showed a low confusion bias, but two of them (from the same cohort) showed a high risk [35,44], and another study [37] a moderate risk, depending on the confounders that were missing in their analyses. Selection bias was low in almost all the studies. Some of the cohorts had specific characteristics that differed from the healthy term infant population, which was our main focus (i.e., TEDDY study in newborns with genetic diabetes risk [42] and Gemini study with relatively high proportion of preterms [39]). These differences hampered the extrapolation to the general population. For one study [31], neither selection criteria nor the study sample characteristics were reported in the publication. Therefore, we categorized the study as unknown risk of selection bias. Finally, all the studies showed a low risk of information bias except for one [41], in which risk of bias was classified as high. A summary of the biases assessed in all the included references is shown in the Appendix B, Tables A2 and A3.

Results
Wall et al. [30] performed an RCT in 160 healthy one-year-old infants that were randomly assigned to consume cow's milk (3.1 g protein/100 mL, control) or a Growing up milk Lite (1.7 g protein/100 mL, intervention) until the age of 2 years (Wall 2019). Apart from the differences in protein content, the products included other differences as the control group was provided with whole pasteurized and homogenized cow milk (supplied as powder) and the intervention group consumed a cow's milk-based Growing up milk fortified with micronutrients (including vitamin D and iron), probiotics, and prebiotics (i.e., a symbiotic). Results showed that the intervention significantly reduced FMI (0.045 kg/m 2 , p = 0.026) and %BF (2.09%, p = 0.047) vs. control at 2 years.
Percentage of body fat was also assessed by Gunther et al. in the DONALD Study cohort that followed 203 healthy term infants from 6 months to the age of 7 years [32,33]. In one of the references [32], the authors analyzed the protein intake from 6 months to 2 years and categorized infants as high (H) or low (L) protein intake, if they were above or below the median at each age, respectively. They added the effect of continuing in the H group during the second year (intakes either at 18 or 24 months or the mean of the two values). They analyzed the association between remaining or not in the H group of protein intake during the first 2 years (H-H vs. H-L). They observed that the H-H group had a significant higher (16%) fat mass compared with the H-L group at 7 years. In addition, the H-H group showed a twofold odds ratio for having overfatness at 7 years (2.28 (95% CI: 1.06, 4.88; p = 0.03)). In the same study, considering the protein intake as a continuous variable (% of energy) and splitting by type of protein (total, animal, plant, and dairy), Gunther et al. found no association between these intakes at 18 or 24 months with %BF at 7 years [33]. Additionally, children in the H-H group at 2 years had a slightly higher %BF (β = 0.67 ± 0.31, p = 0.03) at 5 years compared with children in the H-L group [34].
In a French Cohort (ELANCE Study), Rolland-Cachera et al. showed that protein intake at 2 years was directly correlated with increased subcutaneous skinfolds at 8 years in 112 healthy children (r = 0.20, p = 0.04) [31].

Conclusions
In summary, results from one RCT and two cohort studies that gathered a total of 521 children point to an association between a high protein intake during the second year of life and higher fatness in childhood. However, the timing of the outcome assessment was very variable (from 2 to 7 years), and there were discordant results when the protein intake was analyzed separately according to the food source or evaluated as a continuous variable. Therefore, we conclude that there is moderate evidence (B, 1+/2+) for a direct association between total protein intake during the second year of life and body fat mass.

Results
In the DONALD Study cohort, results showed that infants classified in the higherprotein group until 2 years (H-H group) presented significant higher BMI z-scores at 2 years (β = 0.36 ± 0.13, p = 0.005) [34] and BMI z-scores at 7 years (from 0.4 to 0.8 standard deviations (SD) depending on confounders adjustment) [32] compared with the H-L group, but was not associated with the BMI z-score trajectory between 2 and 5 years [34]. Assessing this association as a continuous variable (energy % from protein), protein intakes at 12 months modulated BMI z-score at 7 years, but protein intakes at 18 or 24 months did not [33].
The effect of protein intake during the second year of life on the BMI later in childhood was analyzed in a further six cohorts [31,[35][36][37][38][39]41]. In five of them [31,36,38,39,41], the results showed a significant association of the protein intake on the BMI z-score measured at different ages (ranging from 4 to 11 years). On the other hand, Cowin et al., in a subsample of the ALSPAC Cohort study in UK (n = 389) [35], found no association between protein intake at 18 months on the BMI z-score at 2.5 years.
In the CAPs cohort [37] (in which the authors observed an association between protein intake at 18 months and BMI z-score at 8 years), BMI z-score growth trajectories from birth to 11.5 years were constructed with the growth mixture model to classify children in one of the three more representative classes (normal growth, early persistent BMI increase, and late increase). Protein intake at 18 months did not differ between children in the different growth trajectory groups.

Conclusions
We found three studies including a total of 962 children that discarded or did not demonstrate an association between protein intake during the second year of life and the later BMI z-score or BMI z-score trajectory [33,35,37]. The other seven references including 3285 children (including some manuscripts on the same cohort studies that showed no association in a different analysis) found higher protein intake during the second year of life was associated with a higher BMI z-score in childhood [31,32,34,36,38,39]. All of the studies included here were longitudinal cohorts. Since results were highly heterogeneous and some studies showed considerable risk of bias, we conclude that there was a weak evidence of this association (D, 2+/2−).

Included Studies
The association between protein intake and the risk to develop overweight or obesity was investigated in three cohorts (The DONALD cohort [32,34], the Gemini cohort [39], and the TEDDY Study [42]).

Results
In the DONALD cohort, the risk of overweight was more than twofold higher at 7 years if protein intake was persistently high during the first 2 years of life [32]. In the same cohort, Karaolis-Danckert showed an increased risk of being overweight at 5 years (27% vs. 15%, p = 0.003) among children who had a rapid weight gain pattern from birth to 2 years [34]. Pimpin et al. showed, in a twins cohort (n = 2435), that protein intake (measured at a mean age of 21 months) was associated with a trend of increased OR of being overweight or obese at 3 years (OR: 1.10, 95%CI 0.99-1.22, p = 0.075), while this was not maintained at 5 years of age [39]. In the TEDDY study (a multicenter cohort of 5563 infants at risk of having diabetes), protein intake at 1-2 years was not associated with the overweight risk at 5.5 years [42].

Conclusions
In three different well-conducted cohort studies with a total of 8247 children, no consistent effect of protein intake during the second year of life on the overweight risk in childhood was demonstrated. Discrepancies between results could be due to the different ages at the outcome assessment in different studies. So, we conclude that the evidence for this association is low (C, 2+).

Included Studies
Two of the included studies assessed the effect of protein intake on weight gain [34,39].

Results
Pimpin et al., in the Gemini cohort (n = 2435), showed that those children consuming, at a mean age of 21 months, more than 16.3% of total energy from proteins (which corresponded to the top of the two quintiles) had higher weight gain until the age of 5 years (0.330 kg (95%CI 0.182-0.478) compared with children with the lowest intakes) without different length growth [39].
Results from the DONALD cohort (n = 249) showed that having a sustained consumption of higher protein intake during the first 2 years of life (H-H group) was not significantly associated with the probability to be classified as performing rapid weight gain in the same period (defined as weight increases >0.67 SD) compared to the H-L group [34]. Children with rapid early growth (0-2 years) showed higher %BF (16.7% vs. 18.0%, p = 0.02) and a higher risk of increased fatness (17% vs. 7%, p = 0.02) at 5 years, but protein intake during the first 2 years was not modulating changes in %BF trajectories beyond the 2 years.

Conclusions
Data from two cohorts showed inconsistent results and a low evidence (C, 2+) for an association of protein intake during the second year of life with the child weight gain velocity until the age of 5 years.
3.5. Effects of Total Protein Intake on Adiposity Rebound (Secondary Outcome) 3.5.1. Included Studies One of the proposed mechanisms derived from the early protein hypothesis is related with an increased growth velocity that could lead to an earlier adiposity rebound (AR). Three of the included studies investigated how the protein intake in the second year of life modulated the AR process. This outcome was evaluated as age at the AR onset, as well as BMI z-score at the AR onset [31,44,45].

Results
The effects of protein intake on AR were first investigated by Rolland-Cachera et al. in 1995 [31]. In the ELANCE cohort, protein intake (% energy from protein) at 2 years was associated with an AR at younger age (r = −0.2, p = 0.02) and with a higher BMI increase after the AR, which led to a subsequent association between the protein intake at 2 years and BMI at 8 years (r = 0.22, p = 0.03)). Moreover, those children with an early AR (before 4 years) had consumed higher protein intakes at 2 years compared with those showing a late AR (after 8 years) (16.6 ± 2.1% vs. 14.9 ± 2.1%, p < 0.01).
Dorosty et al., in a subsample of the ALSPAC Study cohort (n = 889, approximately 10% of the total sample), classified the children in three groups according to their age at AR (very early (<43 months), early (between 44 and 61 months), and late (>61 months), and found these not related to different protein intakes at 18 months [44].
The DONALD Study also explored the effect of protein intake on the BMI at the adiposity rebound. In girls, protein intake at 12-18 months was associated with an increase in the BMI z-score at the adiposity rebound onset (BMI z-score −0.61 vs. −0.08, p = 0.01; at the lower and higher tertile of protein intake, respectively), whereas there was no association in boys. Additionally, there were no age differences at adiposity rebound [45].

Conclusions
Results from three cohorts including 1314 children showed an uncertain effect of protein intake during the second year of life on the AR advancement, with considerable heterogeneity. The heterogeneity was probably increased due to differences in the timepoints of protein intake assessment (12-18 months, 18 months, or 2 years), and the fact that one study only considered intake at one timepoint that cannot be considered representative of the protein consumption during the whole year. Considering the risk of bias in the included studies, we conclude that the evidence of this uncertain association was low (C, 2+) but cannot be discarded.

Discussion
There is high-quality evidence in the scientific literature indicating that higher protein intake during the first year of life has a lasting programming effect increasing later obesity risk [5,8,14,[46][47][48]. For protein intake during the second year of life, our review shows some indications for possible lasting effects, but the available evidence is based primarily on observational studies and does not allow firm conclusions.

Effect of Protein Intake during the Second Year of Life on Body Composition
There was high-quality evidence for a higher body fatness induced by increased protein intake during the second year of life as a short-term effect (at age 2 years) provided by an RCT (Gumli trial) [30]. The results that support the increased body fatness in the short term and later in life (at 2 and 7 years, in the DONALD cohort study) were provided by moderate evidence and hold some heterogeneity [32,34]. An extended follow-up of the Gumli trial sample would be valuable to improve the evidence of a long-term effect, if any.
All in all, it seems that the adipogenic effect exerted by higher protein intake during the first 12 months [49,50] could take place during a longer period, until the age of 2 years.

Effect of Protein Intake during the Second Year of Life on Body Mass Index
Apart from body composition and fatness, an indirect assessment of obesity could be addressed by the BMI measurement. Even though BMI is not able to differentiate if the excessive weight comes from fat or fat-free mass, this index is a convenient low-cost rule of thumb used to categorize a person according to body weight.
Some cohorts were able to demonstrate a significant association between protein intake during the second year of life and BMI [31,32,36,38,39,41], however, the biases of most of the studies were substantial, and this reduced the certainty of the effect observed.
The DONALD study found an association of protein intake during the second year of life with BMI z-score at 2 and 7 years [32,34]. However, they could not confirm an association between protein intake at 2 years and BMI trajectories between 2 and 5 years. These results could suggest that changes exerted during the first two years of life remain permanent rather than programming a later change in weight gain velocity.
The study by Garden et al. reported an association between protein intake during the second year and BMI z-score at 8 years that was not maintained later at age 11.5 years [36,37]. This could be because many other lifestyle patterns influencing BMI could take place; and also because puberty changes occurring during this period actually take place at different rhythms. In fact, the lack of information about the puberty stage is a weakness of the study performed by Garden et al. that did not control for this confounding factor. At 8 years, most children are still prepuberal, and we do not foresee an interference with the observed results. However, at 11.5 years, differences in maturity are considerable, and this could partially hide a possible long-term effect of early protein intake, if any, that could reappear later.
In summary, available evidence from study cohorts suggests a possible association between protein intake during the second year of life and an increased BMI z-score later on. There was low evidence of this association due to heterogeneity and the considerable biases observed in the cohort studies.

Effect of Protein Intake during the Second Year of Life on Later Obesity Risk
Another important primary outcome of interest was the effect on excessive weight (either overweight or obesity) risk. Analyzing this outcome is an attempt to assess the clinical relevance of the possible induced effects, as it reflects a well-recognized pathologic condition rather than an association with a risk factor.
The DONALD cohort supported the hypothesis of an increased overweight risk by having a persistently higher protein intake during the first two years of life (compared to only the first year of life) [32]. The results from the Gemini cohort did not confirm this hypothesis. The apparently contradictory results could be due to several factors: the Gemini cohort was performed in twins and included also preterms, thus, possibly lowbirth-weight infants with different growth patterns compared to our target population, which could hinder extrapolating conclusions. On the other hand, in the Gemini cohort, protein intake was assessed at a mean age of 21 months, within a range of 17-34 months. Thus, some of the intake data were obtained during the third year of life instead of during the second. Furthermore, these results were obtained at a unique timepoint, which could not be representative of what happened during the second year. However, the results from the DONALD study, obtained from two different timepoints and classifying children according to an apparently persistent high protein intake, could be more representative of the high-protein diet during a longer period.

Effect of Protein Intake during the Second Year of Life: Outcomes Potentially Related with the Mechanism Inducing Increased Obesity Risk
Finally, we reviewed the evidence on potential mechanisms through which protein intake could increase later BMI or obesity risk. Rolland-Cachera et al. suggested that protein intake during the first 2 years would accelerate weight gain velocity, resulting in an anticipated adiposity rebound (AR), which could lead to a greater BMI from then onwards [31], and a possible earlier onset of puberty [51]. Regarding the association between protein intake during the second year and the accelerated weight gain between 2 and 5 years, the results from the DONALD and Gemini cohorts provided contradictory results. This could be due to several reasons: on one hand, children from the Gemini cohort had a higher proportion of preterms and low-birth-weight infants that could have a different growth pattern and/or a different maturation rhythm. On the other hand, it is also possible that a higher protein intake during the second year could accelerate weight gain later on, but not enough to achieve the 0.67 SD threshold, which was the one used in the DONALD cohort study. A possible hypothesis could be that increased protein intake during the first two years of life may accelerate weight gain and the development of the adipose tissue during this period; the exerted changes remain during childhood, but do not modify changes in trajectories of weight gain and fat mass gain beyond that age. In any case, the available evidences were not able to confirm this hypothesis.
The anticipated AR as a consequence of increased protein intake during the second year of life could not be confirmed by Dorosty et al. [44] in the ALSPAC cohort. Gunther et al., in the DONALD cohort, found a higher BMI z-score at the AR onset in girls with higher protein intake compared with girls with lower protein intake (association not found in boys) [45]. The discrepancies in the effects of early protein intake between genders have been previously observed in other studies, and a possible explanation could be that the IGF-1 axis response to proteins in female infants could be stronger than that of male infants [52].
In summary, the available evidence could not consistently support the hypothesis that the protein intake during the second year of life could lead to earlier AR onset.

Limitations for the Present Systematic Review
The most relevant limitation of this systematic review is the lack of specific and welldesigned RCTs investigating the effect of protein intake during the second year of life on later obesity risk. We only found one published RCT [30] in which differences between the tested study products were not only restricted to the protein content but also differed in other nutrients such as vitamin D and iron, probiotics, and prebiotics. Moreover, risk of bias from the observational studies we found was moderate or high in most of cases.

"Importance for Public Health: A Window for Prevention"
The novelty of our systematic review is that we focus on the effect of protein intake during the second year of life, whereas previous reviews did not differentiate this period from the first year [14,53,54]. Thus, if protein intake during this period was associated to later obesity risk, there would be an additional reason to promote breastfeeding or formulas with a lower protein content beyond the first year of life. We found moderate evidence for the association between protein intake during the second year of life and increased body fatness at 2 years. Evidence supporting either an increased risk of overweight or overfatness later in life was inconclusive. Overall, the quality of the evidence on effects on overweight and obesity is limited and based on observational cohort studies. Firm conclusions may be derived from randomized controlled intervention trials. In the Cochrane database of registered clinical trials, an ongoing RCT is reported, with more than 1600 children already recruited as infants (TOMI trial, Clinical trials.gov, NTC02907502). Funding: This research received no external funding.

Conflicts of Interest:
The authors declare no conflict of interest.

Appendix A
The search terms conducted in each database were the following. In PubMed ® /MEDLINE ® (https://www.ncbi.nlm.nih.gov/pubmed/), a database search was performed including items published until end of May 2020 (publication date limit up to 31 May 2020). We searched using the following strategy without limiting search fields in order to maximize the number of items found: Combination of both searches retrieved a total of 4417 items that, after removing duplicates, dropped to 3755.
In the Cochrane Central Register of Controlled Trials (https://www.cochranelibrary. com/), searches were performed at the end of May 2020 without any further publication date limit. The following search terms were introduced at the Title, Abstract, or Keyword search fields: The combination of both searches retrieved a total of 1967 items that, after removing duplicates, reduced to 1747. Table A1. Excluded studies and their reasons for exclusion.

Reference
Reasons for the Exclusion Table A1. Cont.

Reasons for the Exclusion
Hörnell 2013 [14] Not focused on protein intake during the second year of life Kourlaba 2008 [69] Health outcomes associated to protein intake at the same timepoint (beyond 2 years) Larnkjaer 2012 [15] Narrative review Lind 2016 [70] Narrative review Manios 2008 [71] Health outcomes associated to protein intake at the same timepoint (beyond 2 years) Michaelsen 2012 [72] Narrative review Michaelsen 2014 [73] Narrative review Morgan 2004 [74] They did not evaluate the effect of protein intake beyond the 12 months of life on the assessed outcomes O'Sullivan 2016 [75] No data about protein intake (only specific food items) Patro-Golab 2016A [53] Not focused on protein intake during the second year of life Patro-Golab 2016B [76] Not focused on protein intake during the second year of life Pearce 2013 [55] Not focused on protein intake during the second year of life Redsell 2015 [77] Not focused on protein intake during the second year of life Rolland-Cachera 2013 [78] Outcomes measured only in adulthood (20 years of age) Rolland-Cachera 2016 [79] Narrative review Ruel 1995 [80] No data about protein intake reported Santos 2015 [81] No data about protein intake reported Switkowski 2019 [82] Protein intake assessed at a mean age of 3 years. Data specifically from the second year of life was not reported. Tang 2018 [9] Narrative review Van Vught 2009 [83] Health outcomes associated to protein intake at the same timepoint (beyond 2 years) Voortman 2015 [84] Not focused on protein intake during the second year of life Weker 2019 [85] They did not report any association between protein intake and the health outcomes Yang 2013 [86] Narrative review

Appendix B
A summary of the biases assessed in all the included references is shown in the following Tables A2 and A3.