Hoppe et al. (27), graded B, examined in a double-blinded randomized trial (n=59) the effects of the two major milk fractions, whey and casein, and milk minerals, calcium (Ca) and phosphorus (P), on sIGF-I and glucose–insulin metabolism (see results in “Glucose–insulin metabolism”). Eight-year-old boys were randomized to receive 540 ml of milk-based drinks for 7 days, either: 1) whey with low mineral content (Ca and P); 2) whey with high mineral content; 3) casein with low mineral content; and 4) casein with high mineral content. The amount of other milk components aimed to be identical to the content in 1.5 L of skimmed milk.

Dietary intake was assessed using two repeated 3-day weighed food records (2 weekdays and 1 weekend day); the first kept before the intervention (days −3 to 0) and the second at the end of the intervention (days 5 to 7). Measurement errors in the dietary recordings were not considered, but the importance of maintaining usual dietary intake was emphasized to the families. Unintentionally, there were significant differences between the intervention groups with regard to several of the anthropometrical measurements, energy and milk intake. No interactions between milk mineral groups (high, low) and milk protein groups (whey, casein) were found, so the milk protein groups were combined.

Average daily protein intake was increased by 17% by the whey drink, from 58 g/day (2.23 g/kg/day, 13.0 PE%) to 68 g/day (2.56 g/kg/day, 15.4 PE%), and by 51% by the casein drink, from 68 g/day (2.30 g/kg/day, 14.3 PE%) to 103 g/day (3.44 g/kg/day, 23.4 PE%). In the whey group, there was no change in sIGF-I. No independent effects of a high milk mineral intake on sIGF-I were found. Increase in serum urea nitrogen (SUN), a marker for protein intake, and the molar ratio of sIGF-I/sIGFBP-3 was significantly higher in the casein group than in the whey group.

A limitation of the study, also mentioned by the authors, is that the subjects were allowed to eat their habitual diet, so there might be other factors in the diet contributing to the findings. However, the authors point out that this has been controlled for in the analyses. The results were not changed markedly after controlling for energy intake, protein intake, SUN, or milk intake. The authors conclude that casein stimulates circulating sIGF-I and that both milk protein fractions seem to be important, but different, in the growth-stimulating effect of milk.

Hoppe et al. (28), graded B, conducted an intervention study (n=24) to examine whether an increase in animal protein intake could increase concentrations of sIGF-I in serum(s) and the molar ratio of sIGF-I/sIGFBP-3 in pre-pubertal children. Eight-year-old boys with a habitual milk intake of at least 500 ml/day were asked to take an extra 53 g protein daily for a week – extra daily intake was 1.5 l of skimmed milk by 12 boys and 250 g of low fat meat by 12 other boys. In addition, they were asked to eat their normal diet ad libitum. Dietary intake was assessed using two repeated 3-day weighed food records (2 weekdays and 1 weekend day); the first kept before the intervention (days −3 to 0) and the second at the end of the intervention (days 5 to 7). Measurement errors in the dietary recordings were not considered, but the importance of maintaining usual dietary intake was emphasized to the families.

After 7 days, the average protein intake increased to 4.0 g/kg/day (+61%) in the milk group and to 3.7 g/kg/day (+54%) in the meat group. High intake of milk increased concentrations of sIGF-I (+19%) and sIGF-I/sIGFBP-3 (+13%), while no increases were seen in the meat group. The authors conclude that compounds in milk, and not a high protein intake as such, seem to stimulate sIGF-I, and that this might explain the effect of milk intake on growth seen in some studies.

Note: The total reported energy intake by the milk group increased by 13% and the group gained on average 550 g of weight during the intervention week compared with a 3% increase in energy intake and no change in weight in the meat group. SUN was used as a biomarker of protein intake, but SUN increased similarly in the two groups although reported protein intake per kg and day increased 20% more in the milk group (+1.7 g/kg/day vs. +1.4 g/kg/day in the meat group). There was no difference in intake at baseline (2.32 g/kg/day vs. 2.27 g/kg/day). No power calculation was reported.

Koletzko et al. (29), graded A, conducted a double-blind, randomized controlled trial in a European multicenter study in five countries to test the hypothesis that a higher early protein intake leads to more rapid growth in the first 2 years of life. Eight-week-old healthy formula-fed infants (n=1138) were randomly assigned to a low- or high-protein diet (cow-milk-based infant formula with 1.77 vs. 2.9 g protein/100 kcal and follow-on formula with 2.2 vs. 4.4 g protein/100 kcal). The protein content represented approximately the lowest and highest acceptable levels in the range given in the European Union (EU) directives from 1991. An observational group of infants exclusively breastfed for the first 3 months of life was also included in the study (n=619) (Note: Exclusive breastfeeding was defined as <10% of feedings or less than three bottles of formula/week).

Dietary intake was assessed by 3-day weighed records during three consecutive days (2 weekdays and 1 weekend day) at 3, 6, 12, and 24 months. Energy intake was not calculated for food records containing any breastfeeding as breast milk intake was only measured in a subgroup. Food records with energy intake greater than three SDs of the mean by months and those deemed incomplete or with reported concurrent illness were excluded. (Note: No details given about the number of excluded records.)

Infant formula with higher protein content was associated with higher weight in the first 2 years of life but had no effect on length. At 24 months, adjusted z-score for weight-for-length in the lower protein formula group was 0.20 (95% CI: 0.06, 0.34; P=0.005) lower than in the higher protein group and did not differ from that of the breastfed reference group. The effect of intervention was not different between countries. Compared with breastfed children, those fed high-protein formula had significantly higher z-scores for weight, length, weight-for-length, and BMI at 24 months. Analyses were also performed at 3, 6, and 12 months, and in general, the differences were greatest at 12 months for weight, weight-for-length, and BMI. The authors conclude that limiting dietary protein intake during infancy might be a way to decrease the risk of overweight and obesity in later life.

Larnkjær et al. (30), graded B, performed a randomized controlled trial to study the effects of whole milk and infant formula on growth and IGF-l from 9 to 12 months of age in Denmark. Healthy infants (n=83) were randomized to receive either whole milk (WM) or infant formula (IF) and either a daily fish oil supplement or no supplement. Dietary intake was measured over seven consecutive days at 9 and 12 months using a pre-coded dietary record developed for children with portion sizes estimated from a portion size photo series. The infants in both groups consumed about 300 ml WM or IF per day. Measurement errors in the dietary records were not considered.

No effect of milk type on growth was found in this 3-month intervention study. [Note: Dropouts (17% of WM and 6% of IF) were 1.6 cm shorter at 9 months, which could be problematic as this was a study on growth in infancy. Also, the breastfeeding prevalence between 9 and 12 months was lower in the IF group, although including breastfeeding in the analysis made no difference.] Intake of WM significantly increased the PE%; PE% was 14.2 in WM and 11.4 in IF at 12 months, whereas no differences were found in weight and length between groups. The authors suggest that this could be due to the relatively short intervention period, and that there is a need for studies on the long-term effects of protein and milk protein.

Intake of WM increased sIGF-I in boys (P=0.034) but not in girls. Intake of fish oil had no effect on the outcomes. Including all infants, PE% was positively associated with sIGF-I at both 9 and 12 months after adjusting for sex and breastfeeding. Positive correlation was also found between sIGF-I and intake of WM and WM products adjusted for sex and duration of full breastfeeding. Boys had lower levels of sIGF-I and sIGFBP-3 at 9 and 12 months than girls. The authors conclude that the results suggest that high protein intake is positively associated with sIGF-I concentration in well-nourished infants.

Räihä et al. (5), graded B, studied breastfed and formula-fed healthy term infants. Infants who stopped breastfeeding before 28 days of age were randomly assigned to receiving one of the three study formulas up to 120 days of age. The three isocaloric formulas differed by their protein source: a conventional whey adapted starter formula with a whey/casein ratio of 60/40 and a protein content of 2.2 g/100 kcal was compared with two experimental formulas with a whey/casein ratio of 70/30 and a protein content of 1.8 g/100 kcal.

No differences were found between the four feeding groups for weight- and length-gains or for body mass indices (BMI). No differences in energy intakes between the formula-fed groups could be found, whereas protein intakes were lower in infants fed the 1.8 g/100 kcal formulas. Plasma urea levels of the infants fed the 1.8 g/100 kcal formulas were closer to those found in the breastfed infants. The authors conclude that a whey predominant formula with a protein/energy ratio of 1.8 g/100 kcal provides adequate intakes of protein from birth to 4 months of age.

Sandström et al. (31), graded B, did a (partly) randomized control study on the effects on growth of exchanging part of the protein in milk-based formulas with α-lactalbumin. All formulas contained 1.96 g protein/100 kcal, but with differences in the protein composition. Breastfed infants were controls. Weight gains in the α-lactalbumin-enriched formula groups were similar to that of the breastfed infants. The standard whey-predominant formula group gained significantly more weight than did the breastfed infants. All formula-fed infants had significantly higher plasma concentrations of most essential amino acids at 4 and 6 months than did the breastfed infants, and SUN was also higher in the formula-fed infants, which might indicate that the protein content of α-lactalbumin-enriched formula can be further reduced.

Dorosty et al. (32), graded B, studied the adiposity rebound (AR) among 5-year-olds in relation to protein intake at 18 months. Dietary intake was assessed through 3-day household measured records at 18 months (records were also collected at 8 months but were not used in the present study). Measurement errors in the dietary recordings were not considered. No evidence for an association between protein intake, or any other dietary variable, and timing of the AR were found. Children with very early (≤43 months) or early (from 49 but before 61 months) AR had parents with significantly higher BMI and were more likely to have at least one obese parent.

Gunnarsdottir et al. (3), graded B, studied BMI at 6 years in relation to size at birth, growth between 0 and 12 months, and food intake among a representative sample of Icelandic children. Dietary intake was assessed through monthly repeated 24-h records throughout the first 12 months and 3-day weighed records at 9 and 12 months of age. The children were weighed before and after breastfeeding. Measurement error in the dietary assessment was not considered.

Weight gain from birth to 12 months as a ratio of birth weight was positively related to BMI at the age of 6 years in both genders (β 2.9±1.0, P=0.008, and β 2.0±0.9, P=0.032 for boys and girls, respectively). Boys in the highest quartile of protein intake (E%) at the age of 9–12 months had significantly higher BMI at 6 years than the lowest and the second lowest quartiles (17.8±2.4 vs. 15.6±1.0 and 15.3±0.8, P=0.039 and P=0.01, respectively). Energy intake was not different between the groups. Together, weight gain at 0–12 months and protein intake at 9–12 months explained 50% of the variance in BMI among 6-year-old boys. The authors concluded that rapid growth during the first year of life is associated with increased BMI at the age of 6 years in both genders. In boys, high intake of protein in infancy could also contribute to childhood obesity. (Note: Bonferroni's correction for multiple tests was used when significant differences were found.)

Günther et al. (33), graded B, studied the AR among 7-year-olds in relation to protein intake between 12 and 24 months. Six repeated 3-day weighed records assessed dietary intake during the first 2 years, each covering 2 weekdays and 1 weekend day. In the present study, records from 12, 18 and 24 months were used. Measurement error in the dietary assessment was not considered, but all dietary variables were preadjusted for total energy intake using the residual method.

Girls in the highest tertile of habitual protein intake (E% and g/kg reference body weight [RBW]/day) had a significantly higher BMI SD score (BMI-SDS) at AR than those in the lower tertiles. In boys, there were no differences in BMI-SDS at AR between tertiles of habitual protein intake (E% or g/kg RBW/day). Boys in the lowest tertile of habitual protein intake (E%) tended to experience a later AR (P=0.07). But neither in girls nor in boys was age at AR significantly different between tertiles of habitual protein intake. The authors concluded that a higher habitual protein intake between the age of 12 and 24 months was associated with a higher BMI-SDS at AR in girls, but not in boys. There was no consistent relation between habitual protein intake in early childhood and timing of AR.

Günther et al. (34), graded A, analyzed the associations of different protein intakes during 6–24 months with BMI and percentage body fat (%BF) at 7 years, using data from the DONALD study in Germany. Dietary intake was assessed by 3-day weighed records during three consecutive days (2 weekdays and 1 weekend day) at 6, 12, 18, and 24 months. Breastfed infants were weighed before and after each feed. Goldberg's cut-off and Schofield equations were used to test the validity of the dietary assessment, and the residual model was used to adjust total protein intake for energy intake and sex. The median of these energy-adjusted protein intakes was used to distinguish different patterns of low and high protein intakes, and marked differences were observed between the low and high protein groups at 6 months (7–8 E% and 12 E%) and at 12 months (11–12 E% and 14–15 E%), respectively.

A consistently high protein intake at 12 and 18–24 months was independently related to a higher mean BMI SDS and %BF at 7 years and a higher risk of having a BMI or %BF above the 75th percentile. The analyses included adjustments for a large number of possible confounders, such as dietary factors (e.g. energy intake and breastfeeding) and parental characteristics (e.g. maternal overweight). Protein intake at 6 months was not associated with the outcomes. They conclude that their results suggest an association between high protein intakes during complementary feeding and the transition to the family diet with both a higher BMI and higher body fatness at 7 years of age.

Günther et al. (35), graded A, used data from the DONALD study to examine whether there may exist a critical period of protein intake for later obesity early in childhood and to analyze the relation between protein intake from different sources on BMI and %BF at 7 years of age. Dietary intake was assessed by 3-day weighed records during three consecutive days (2 weekdays and 1 weekend day) at 6, 12, 18, 24 months and yearly thereafter. Mean intake was calculated for the periods 18–24 months, 3–4 years, and 5–6 years. Breastfed infants were weighed before and after each feed, and 5% was added to the test weighing results to account for insensible water loss. Goldberg's cut-off and Schofield equations were used to test the validity of the dietary assessment, and the residual model was used to adjust total protein intake for energy intake and sex.

The ages of 12 months and 5–6 years were identified as critical periods at which higher total and animal, but not vegetable, protein intakes were positively related to body fatness at 7 years. Animal protein E% at 12 months was positively associated with BMI SDS at 7 years. The analyses were adjusted for several dietary and family characteristics, such as energy and fat intake and maternal overweight. When examining different sources of protein, dairy protein E% at 12 months, but not meat or cereal, showed a positive association with BMI SDS (p=0.02) and %BF at 7 years (P=0.07). The authors conclude that a higher intake of animal protein at 12 months, especially from dairy foods, might be associated with an unfavorable body composition at 7 years, and that the age of 5–6 years might represent another critical period of protein intake for later obesity risk.

Hoppe et al. (36), graded B, studied body size (BMI), body composition (percent body fat, %BF), and insulin-like growth factor I (sIGF-I) at 10 years of age among Danish children (n=142 at 9 months, 105 at 10 years) in relation to protein intake at 9 months and SUN at 9 months and 10 years. Dietary intake was assessed through 5-day weighed records at 9 months, including three weekdays and a weekend. (Note: It is not clear if measurement errors in the dietary recordings were considered. About one third of the children were breastfed at 9 months, but there is no information about whether breast milk intake was measured or if it is included in the nutrient calculations.) Adjustments were made for breastfeeding status at 9 months without any effect. Tanner stages were assessed at 10 years but not adjusted for in the analyses (96% of boys and 63% of girls had no sign of pubertal development).

In total, 7.8% of boys and 7.5% of girls were overweight, none were obese. SUN (mmol/l) at 9 months was a predictor for BMI and weight at 10 years. Protein intake (E%, g/day but not g/kg/day) at 9 months was a predictor for weight and height at 10 years. The associations remained when adjusting for parental body size, but were attenuated when adjusting for infant body size at 9 months.

Morgan et al. (37), graded B, examined body weight, length, and head circumference at 4, 8, 12, 16, 20, and 24 months in a cohort study of 144 infants. Dietary intake was measured through 7-day weighed food records at 4, 8, 12, 16, 20, and 24 months (meat intake calculated as average intake during 4–24 months). Meat intake from 4 to 12 months was positively and significantly related to weight gain. This association might be mediated via protein intake but was independent of energy, zinc, or iron intake. These findings remained after adjustment for potential confounding factors. There was no interaction between meat intake and breastfeeding on growth.

Öhlund et al. (4), graded B, did a follow-up at 4 years of age of healthy children (63 girls and 64 boys) previously followed prospectively from 6 to 18 months of age.. Monthly 5-day weighed records measured dietary intake at 6–18 months and at 4 years. BMI at 6–18 months was a strong predictor of BMI at 4 years. Intake of protein in particular, and also of total energy and carbohydrates at 17/18 months and at 4 years, was positively associated with BMI at 4 years. Although BMI at 6–18 months was the strongest predictor of BMI at 4 years, in the final multivariate models of the child's BMI, high protein intake at 17–18 months and at 4 years, energy intake at 4 years, and the father's, but not the mother's, BMI were also independent contributing factors. One limitation of the study was that physical activity was not included as a factor when estimating energy requirements.

Scaglioni et al. (38), graded B, studied growth of healthy children (singleton birth, healthy parents) from birth to age 5 years. Dietary intake was assessed at 1 and 5 years-of-age through age-adjusted food frequency questionnaires (FFQs) complemented by a 24-h recall to standardize the usual serving size. Measurement error in the dietary assessment was not considered. The prevalence of overweight at the age of 5 years was strongly associated with parental overweight (P<0.0001), and overweight children had a higher intake of protein at the age of 1 year than non-overweight children (22E% vs. 20E%, P=0.024). In children born from overweight mothers, the prevalence of overweight at the age of 5 years tended to be higher in bottle-fed than in breastfed ones (62.5% vs. 23.3%, P=0.08). The authors concluded that parental overweight is a major risk factor for childhood overweight in the first years of life, but an early high protein intake may also influence the development of adiposity.

Skinner et al. (39), graded B, found in a prospective study of 70 children that their BMI at 8 years was negatively predicted by age of AR and positively predicted by their BMI at 2 years. Dietary intake was assessed through nine sets of non-consecutive 3-day data collections (2 days’ recordings and one 24-h recall). Measurement error in the dietary assessment was not considered. Longitudinal intake (between 2 and 8 years) of protein and fat (grams/day and E%) were positively related to BMI at 8 years, but mean carbohydrate intake over the same time period was negatively related to BMI at 8 years. A weakness, also mentioned by the authors, was that there were no observations to verify normal food intake nor was total energy expenditure measured by doubly-labelled water (DLW) methods.

Van Vught et al. (40), graded B, studied associations between intakes of total protein as well as the amino acids arginine (ARG) and lysine (LYS) in the habitual diets of normal weight and overweight 6-year-olds and subsequent linear growth, body fat, and fat free mass (FFM) at 9 years. Dietary intake was assessed through a 7-day estimated food record. Reporting of dietary intake was evaluated by comparing reported energy intake with estimated energy requirements. Children reporting an energy intake below BMR × 1.3 or above BMR × 2.0 were excluded.

High ARG intake was associated with linear growth among girls. Furthermore, in girls, fat mass index (FMI) had a stronger inverse association with high ARG intake, if it was combined with high LYS intake, instead of low LYS intake (P=0.03). No associations were found in boys although the change was in the same direction as for the girls. The authors conclude that in pre-pubertal girls, linear growth may be influenced by habitual ARG intake, and body fat gain may be relatively prevented over time by the intake of the amino acids ARG and LYS.

In another study, Van Vught et al. (41), graded B, studied associations between intakes of protein, ARG, and LYS among 8–10-year-olds and subsequent 6-year change in body composition (fat-free mass index [FFMI] and FMI). Dietary intake was assessed through one 24-h recall.

Note: Measurement errors were discussed but not measured. The authors suggest that among lean girls, high protein intakes at 8–10 years may decrease subsequent body fat gain and increase fat free mass gain depending on the available amounts and combinations of ARG and LYS.

Budek et al. (14), graded B, studied concentrations of serum insulin-like growth factors (sIGF-I) in relation to intake of total dairy and meat protein in pre-pubertal 8-year-old boys. Dietary intake was assessed using a 3-day weighed food record (2 weekdays and 1 weekend day). Measurement errors in the dietary recordings were not considered, but all dietary variables were preadjusted for total energy intake using the residual method. Free sIGF-I was positively associated with total protein (p<0.01) and dairy protein (p=0.06) but not with meat protein. (Note: Meat protein included red meat, poultry, and fish, but not pork. Plant protein was estimated from the difference between total protein intake and intake from dairy, meat, and egg.) Measurement errors in the dietary recordings were not considered, except for mentioning that dietary assessment in children is difficult.

Hoppe et al. (13), graded B, examined associations between protein intake, sIGF-I concentrations, and height in a cross-sectional study of 2.5-year-old healthy children (n = 90) in Denmark. Dietary intake was reported through a 7-day estimated food record with pre-coded response categories. Amounts were given as household measures or as standard portion sizes estimated from a picture booklet. Measurement errors in the dietary recordings were not considered. The 10th, 50th, and 90th percentiles of protein intake were 2.4, 2.9, and 4.0 g/kg/day, respectively; 63% was animal protein.

In multiple linear regressions with adjustment for sex and weight, height and sIGF-I concentration were positively associated with each other as well as with intakes of animal protein (g/day) and milk but not with those of vegetable protein or meat. The authors concluded that milk intake was positively associated with sIGF-I concentrations and height. An increase in milk intake from 200 to 600 ml/day corresponded to a 30% increase in circulating sIGF-I. The authors suggest that milk compounds have a stimulating effect on sIGF-I concentrations and, thereby, on growth.

Kourlaba et al. (42), graded B, studied interaction effects between energy and macronutrient intakes and angiotensin-converting enzyme 1 (ACE) I/D polymorphism on adiposity-related phenotypes among 1–5-year-olds in the Greek Genesis study. Dietary intake was assessed through 3-day records (2 weekdays and 1 weekend day) combining weighed food records (by staff during nursery hours) and 24-h recall. Measurement errors in the dietary recordings were discussed but not measured.

Significant interactions were found between the ACE I/D polymorphism and protein intake on BMI and being overweight. Stratified analyses revealed that protein intake was associated with BMI and being overweight only among carriers of the D-allele (i.e. DD or ID genotypes). Protein intake was higher among those ‘at risk of being overweight’ or ‘overweight’ compared with their normal-weight counterparts. The authors conclude that the results suggest that the ACE I/D polymorphism may act as a modifying factor in the response of adiposity-related phenotypes to diet and that further research is required to confirm their findings.

Maillard et al. (43), graded B, studied growth in relation to dietary intake among 5–11-year-old non-obese pre-pubertal children. Dietary intake was assessed through 1-day records, that is, one out-of-school-weekday. Intake data were validated with Schofield's equation and Goldberg's & Black's cut-off. They found that associations between adiposity and protein intake as E% or energy intake were non-significant. All adiposity indices (except waist girth and BMI) were significantly and inversely associated in girls with energy intake from carbohydrates (%EIC, all P-values <0.02), and positively with energy intake from fat (%EIF, all P-values <0.05). Similar but non-significant trends were observed in boys. The relationships were not linear, and thresholds close to current dietary recommendations were highlighted.

Manios et al. (44), graded B, studied anthropometrical indices in relation to nutrient intake (based on 3-day food records; 2 weekdays, 1 weekend day) among children aged 1–5 years. Intake data were validated with Schofield's equation and Goldberg's & Black's cut-off. Bonferroni correction was used. For both fat and carbohydrate, a substantial percentage of toddlers and preschoolers had usual intakes outside the acceptable macronutrient distribution range, whereas protein was less than this range.

No statistically significant differences were seen in protein E% between children with normal weight (17.1±1.6), at risk of overweight (17.1±1.5), or overweight (17.1±1.5). However, children ‘at risk of overweight’ and ‘overweight’ had higher intake of total energy, protein (g/day), and fat (g/day) compared with their normal-weight counterparts (p<0.001 for all), whereas no differences were found for micronutrient intakes.

Closa-Monasterolo et al. (45), graded B, aimed to investigate whether sex modulates the responses of relevant biochemical parameters and growth to different protein intakes early in life. In randomized controlled trials (RCT) in five European countries [the same participants as in (29)], formula-fed infants were assigned to receive formula with lower or higher protein content (cow-milk-based infant formula with 1.77 or 2.9 g protein/100 kcal) from a median age of 14 days. The protein content represented approximately the lowest and highest acceptable levels in the range given in the EU directives from 1991.

Protein (g/day) and energy intake (kcal/day) was assessed by 3-day weighed records during three consecutive days (2 weekdays and 1 weekend day) at 3 and 6 months. Measurement errors in the dietary recordings were not considered. Outcomes (e.g. sIGF-I axis parameters, weight, length, BMI, and leptin) were measured at 6 months. The authors conclude that their findings indicate that the endocrine response to a high protein diet early in life may be modulated by sex. The sIGF-I axis of female infants showed a stronger response to the intervention, but there was no enhanced effect on growth.

Budek et al. (46), graded B, compared the short-term (7 days) effect of high milk and high meat intake on bone turnover during pre-puberty among 24 Danish boys aged 8 years assigned to either 1.5 l milk per day or 250 g meat per day (both containing 53 g protein) given together with the habitual diet for 7 days. The participants all had a habitual milk intake of at least 500 ml/day before the start of the study. Dietary intake was assessed using two repeated 3-day weighed food records (2 weekdays and 1 weekend day); the first kept before the intervention (days −3 to 0) and the second at the end of the intervention (days 5 to 7). Measurement errors in the dietary recordings were not considered, but the importance of maintaining usual dietary intake was emphasized to the families.

There were no significant differences between the groups’ intake of total energy (MJ/day) or protein (g/day, g/kg/day) neither before nor after the intervention. In the milk group, the proportion of energy from carbohydrates increased slightly and the proportion from fat decreased with an average 8%-units (from 34.6 E% to 26.7 E%), while the meat group decreased the proportion of energy from carbohydrates with on average 10%-units (from 56.6 to 46.8 E%) and increased the proportion from fat with about 2%-units.

At equal protein intake, milk, but not meat, decreased markers for bone formation and resorption after 7 days. The authors suggest that this effect was more likely due to some milk-derived compounds, rather than to the total protein intake. The milk group also had significantly higher calcium intake compared with the meat group and this could affect the decline observed in bone turnover in the milk group. Whether the decline in bone turnover markers promotes higher bone mineral accretion during growth needs to be further studied according to the authors.

Alexy et al. (47), graded B, examined the association of long-term protein intake and dietary potential renal acid load (PRAL) with bone variables in a subgroup of 229 children and adolescents (6–18 years) in the DONALD study. Proximal forearm was measured once (cross-sectionally) at 6–18 years of age, and dietary intake was measured through yearly 3-day weighed records over the 4-year period prior to bone measurements. Goldberg's cut-off and Schofield equations were used to test the validity of the dietary assessment (38% had non-valid records and were excluded). Collinearity diagnostics detected no evidence for collinearity between the dietary factors.

Reported protein intake measured as g/day was significantly lower in prepubescent boys and girls compared with pubescent boys and girls, while the intake measured as g/kg/day was higher in the younger children. Long-term protein intake was positively associated with all bone variables (periosteal circumference, cortical area (CA), bone mineral content (BMC), and polar strength strain index at the proximal diaphyseal radius) in these healthy children and adolescents, also after adjustment for age, sex, and energy intake, and control for forearm muscularity, BMI, growth velocity, and pubertal development. Children with a higher dietary PRAL (i.e. indicating an inadequate intake of alkalizing minerals from vegetables and fruit) had significantly less CA and BMC. Calcium intake had no significant effect on any bone variable. The authors emphasize the importance of evaluating the whole diet when studying dietary influence on bone health.

Bounds et al. (48), graded B, studied total BMC (g) and bone mineral density (BMD, g/cm²) measured by dual energy X-ray absorptiometry (DEXA) scan at 8 years among 52 children (25 boys, 27 girls) in relation to an averaged longitudinal nutrient intake between 2 and 8 years (measured as average of 3 days [2 days with estimated food records and one 24-h recall] times nine occasions). Measurement errors in the dietary recordings were not considered, but it is stated that the parents received training from registered dieticians. Mothers’ BMC was also assessed by DEXA and the time children spent in sedentary activities was assessed through a questionnaire at 6 and 7 years of age. Multivariate models predicting children's total BMC and BMD at 8 years of age were developed using stepwise regression procedures adding all potentially independent variables.

Factors positively related to children's BMC at age 8 years in the final models included longitudinal intakes (ages 2 to 8 years) of protein, phosphorus, vitamin K, magnesium, zinc, energy, and iron; height; weight; and age. Factors positively related to children's BMD at age 8 years included longitudinal intakes of protein and magnesium. Female sex was negatively associated with both BMC and BMD at age 8 years in contrast with previous studies, that have been unable to find sex differences in pre-pubertal children. Children's bone mineral indexes at ages 6 and 8 years were strongly correlated (r=0.86, P<0.0001 for BMC; r=0.92, P<0.0001 for BMD).

Budek et al. (49), graded B, studied BMC and bone area (BA) determined by DEXA, in relation to milk and meat protein intake at 17 years. Dietary intake was reported through a 7-day estimated food record with pre-coded response categories, supplemented with open-ended alternatives. Physical activity level was recorded using a 24-h recall questionnaire, assessing total hours per day spent at four different activity levels. Time spent on high activity level was used in the statistical analyses.

Total protein intake (∼1.2 g/kg) and milk protein intake (∼0.3 g/kg), but not meat protein intake (∼0.4 g/kg), was positively associated with size-adjusted BMC. The positive association between milk protein intake and size-adjusted BMC remained significant after correction for energy, calcium, and physical activity and did not seem to be mediated via current serum sIGF-I. None of the analyzed protein sources were significantly associated with size-adjusted BA. The authors suggest that some components of milk protein may promote bone mineralization.

Note: Measurement errors in the dietary recordings were not considered, except for mentioning in the discussion that the dietary assessment might include bias due to inadequate recording (however, no details are given). Pubertal stages were not considered.

Budek et al. (14), graded B, also studied concentrations of sIGF-I and markers for bone-turnover (serum osteocalcin [sOC], bone-specific alkaline phosphatase [sBAP], and C-terminal telopeptides of type l collagen [sCTX]) in relation to intake of total dairy and meat protein in pre-pubertal 8-year-old boys. Dietary intake was assessed using a 3-day weighed food record (2 weekdays and 1 weekend day). Measurement errors in the dietary recordings were not considered, except for mentioning that dietary assessment in children is difficult. All dietary variables were preadjusted for total energy intake using the residual method. Note: Meat protein included red meat, poultry, and fish, but not pork. Plant protein was estimated from the difference between total protein intake and intake from dairy, meat, and egg.

Dairy protein was negatively associated with sOC but not significantly associated with sBAP and sCTX. Further analyses showed that dairy protein decreased sOC at a high meat protein intake (>0.8 g/kg), whereas meat protein increased sOC at a low dairy protein intake (<0.4 g/kg). Total and meat protein intake was positively associated with sBAP but not significantly associated with sOC and sCTX. Free sIGF-I was positively associated with total and dairy (P=0.06) protein but not with meat protein. The authors suggest that their results indicate that dairy and meat protein may exhibit a distinct regulatory effect on different markers for bone turnover.

Hoppe et al. (50), graded B, studied associations between dietary factors and whole body BMC and BA determined by DEXA in a cross-sectional analysis of 105 10-year-old Danish children. Dietary intake was reported through a 7-day estimated food record with pre-coded response categories, supplemented with open-ended alternatives. Goldberg's cutoff was used to test the validity of the dietary assessment, and no implausibly low energy intakes were reported.

The mean intakes of calcium, protein, phosphorus, and sodium were 1226 mg, 78 g, 1523 mg, and 3.3 g, respectively, all considerably higher than the Nordic recommendations. In bivariate analyses, BMC and BA were positively correlated with height and weight and with intakes of energy and several nutrients. In multivariate analyses, size-adjusted BMC was positively associated with calcium intake, and size-adjusted BA was positively associated with dietary protein, and negatively associated with intakes of sodium and phosphorus. Inclusion of pubertal stages in the multiple linear regressions did not alter the outcome.

Libuda et al. (51), graded B, examined as part of the DONALD cohort study, the effects of nutrients (e.g. protein, calcium, and vitamin D) on bone parameters and compared their effects sizes with those of two known predictors of bone development: bone-related muscle mass and androgen levels. Long-term nutrient intakes were assessed by 3-day weighed dietary records and calculated as the mean of each dietary record in the 4 years before measurement of bone and muscle variables, for example, diaphyseal BMC, CA, and muscle area.

Of all considered nutrients, only protein showed a trend for an association with BMC (β= +0.11; P=0.073) and CA (β = + 0.11; P=0.056). None of the other dietary variables was associated with the bone parameters. The size of the bone anabolic effect of protein was partly comparable with that of androstenediol. The authors conclude that their results suggest that bone-related muscle area has the strongest effect on bone status of healthy prepubertal children followed by the sex steroid androstenediol and protein intake, which was found to be the strongest dietary predictor of diaphyseal bone.

Mark et al. (52), graded B, is a continued analysis of the data from the intervention study by Hoppe et al. (27), studying the effects of milk-derived compounds on bone formation and resorption during growth in a 7-day randomized double-blind study among 8-year-old Danish boys with a habitual milk intake not exceeding 500 ml/day. Markers for bone-turnover (serum osteocalcin [s-OC], bone-specific alkaline phosphatase [s-BAP], and C-terminal telopeptides of type l collagen [s-CTX]) were measured. The boys received one of four milk drinks, that is, whey protein with low or high content of minerals or casein protein with low or high content of minerals. The amount of whey and casein was identical to the content in 1.5 l of milk. Dietary intake was assessed using two repeated 3-day weighed food records (2 weekdays and 1 weekend day); the first kept before the intervention (days −3 to 0) and the second at the end of the intervention (days 5 to 7). Measurement errors in the dietary recordings were not considered, but the importance of maintaining usual dietary intake was emphasized to the families.

The intake of milk drinks containing whey protein increased sOC at the low content of milk minerals, whereas it decreased sOC at the high content of milk minerals (P<0.05). In contrast, the intake of milk drinks with casein protein increased sOC both at low and high content of milk minerals. They concluded that whey and casein differently affected sOC in 8-year-old boys depending on the content of milk minerals, but that it did not seem to affect other markers for bone turnover.

Five studies, (14, 47–50), report a positive association between total protein intake and BMC and/or other bone variables in childhood and adolescence. The short-term intervention study by Budek et al. (46) conclude that at equal protein intake, milk, but not meat, decreased markers for bone formation and resorption after 7 days. Three of the six studies were performed in 8-year-olds, one in 10-year-olds, and one in 6–18-year-olds, both adjusting for pubertal stage, and one in 17-year-olds not adjusting for pubertal stage. Only two of the six studies considered the validity of the dietary assessment.

Based on the above, we conclude that evidence is limited-suggestive (grade 3) for a positive association between total protein intake and BMC and/or other bone variables in childhood and adolescence. The two papers found in the complementary search support this conclusion, although we do not consider it enough to change the evidence grading.

Berkey et al. (53), graded B, studied the relation of childhood diet and body size to menarche and adolescent growth in a longitudinal study of 67 Caucasian girls in Boston, United States. Data were collected prospectively from birth to 18 years during the 1930s–40s. Dietary intake was assessed through dietary history interviews every 6 months up to 11 years of age, and annually thereafter. The reliability of total protein intake was estimated to be 71%, and the validity was tested by correlation between reported daily protein intake and the child's rate of growth of muscle in the lower leg.

They found that age at menarche, age at peak height growth velocity and peak height growth velocity were all associated with diet and body size earlier in childhood. For peak growth velocity, the same three factors emerged in all age periods, that is, more calories, more animal protein, and lower BMI were consistently associated with higher peak growth velocity (factors closer to puberty more important). Timing of puberty (age at menarche and age at peak growth velocity) was predicted by protein intake and height. Girls with higher animal protein (energy adjusted) intake and less vegetable protein at 3–5 years had earlier menarche. Higher dietary fat intake at 1–2 years was associated with earlier peak growth, and higher calorie intake at 1–2 years gave higher peak height velocity. Girls with higher animal protein intake at 6–8 years had earlier peak growth. Later age at menarche was associated with lower age at peak growth (r = 0.81, p<0.05) and lower peak growth velocity (r=-0.41, p<0.05).

Günther et al. (54), graded A, used data from the DONALD study in Germany to examine the association between dietary protein intake in early- and mid-childhood and timing of puberty; the ages of take-off of the pubertal growth spurt (ATO), age at peak height velocity (APHV), and menarche in girls/voice break in boys. Dietary assessment method is described in Günther et al. (35) BMI, growth, body composition, and sIGF-I above.

Higher animal protein intake (E%) at 5–6 years was associated with earlier puberty. Similar tendency was seen at 3–4 years. Vegetable protein intake was associated with a later ATO. Children with higher animal protein intake at 3–4 and 5–6 years had earlier APHV, while those with high vegetable protein intake had later APHV. Higher animal protein intake (especially milk) at 3–4 years tended (p=0.06) to be, and at 5–6 years was (p=0.02), associated with earlier menarche/voice break and later for high vegetable protein intake. Adjustment for confounders did not change all of these associations.

The authors conclude that whereas higher animal protein intake at 5–6 years might be related to an earlier ATO, APHV and menarche/voice break, higher intakes of vegetable protein at 3–4 and 5–6 years were associated with delayed puberty.

Remer et al. (55), graded B, investigated as part of the DONALD-study the associations between adrenal androgen (AA) secretion with early and late pubertal markers in a prospective study of 109 Caucasian children. Urine samples were used to measure urinary steroid profile, which was used to determine total AA secretion. Energy and animal protein intake was estimated from 3-day weighed dietary records 1 and 2 years before puberty onset.

AA predicted earlier ages at Tanner stage 2 for pubic hair and breast and genital development, but this was independent of protein intake. Intake of animal protein was independently negatively associated with the ATO and with the APHV and negative association of borderline significance was seen between animal protein intake and age at menarche in girls and voice break in boys. Therefore, the authors concluded that animal protein intake might be involved in earlier pubertal growth.

A paper by Shi et al. (56), graded B, was also based on the DONALD-study and included 137 healthy pre-pubertal 3–12-year-old children. The aim was to investigate if adrenal androgen (AA) production in childhood is associated with body composition and dietary intakes. Diet was, as in the prior study, tested by 3-days’ dietary registration and urine samples were used to measure urinary steroid profile, which was used to determine total AA secretion. Fat mass (FM) predicted most of the variation in AA (5%, p<0.0001) and total protein intake to some extent (1%, p<0.05). The authors concluded that body fat may relevantly influence the prepubertal adrenarchal androgen status and animal protein may also make a contribution.

Rogers et al. (57) graded B, investigated as part of the ALSPAC study, associations between diet at three time points in childhood (3, 7 and 10 years) and age at menarche (AAM). AAM was categorized as before or after 12 years and 8 months, a point close to the median age in this cohort of British girls. Diet was assessed by FFQ at 3 and 7 years and by a 3-days’ un-weighed food diary at 10 years.

Total and animal protein intakes at 3 and 7 years were positively associated with AAM ≤12 years and 8 months. Meat intake at 3 and 7 years was also strongly positively associated with reaching menarche by 12 years and 8 months. The authors concluded that their data suggest that higher intakes of protein and meat in early to mid-childhood may lead to earlier menarche. Note: energy intake at 3 and 7 years is not stated and thus is not possible to deem as credible or not.

Three studies, one American study from the 1930–40s and two from the German DONALD-study found an association between increased intake of animal protein in early childhood (around 5–6 years) and earlier puberty. The third paper from the DONALD-study looked at prepubertal hormone levels and concluded that body fat and animal protein intake may increase the hormone levels.

Based on the fact that three of the papers of the original search came from the same group, we first concluded that evidence was slightly too weak to grade it as probable that increased intake of animal protein in childhood is related to earlier puberty, but as the paper found in the complementary search support this conclusion, we judge that the evidence grade increases to probable (grade 2).

Hoppe et al. (27) graded B, examined in a double-blinded randomized trial the effects of the two major milk fractions, whey and casein, and milk minerals (Ca and P) on sIGF-I and glucose–insulin metabolism (details about the methodology BMI, growth, body composition, and sIGF-I). Average daily protein intake was increased by 17% by the whey drink and by 51% by the casein drink. In the whey group, fasting insulin increased by 21%. No independent effects of a high milk mineral intake on insulin were found.

Increase in SUN was significantly higher in the casein-group than in the whey group. Conversely, whey increased fasting insulin more than did casein. A limitation of the study, also mentioned by the authors, is that the subjects were allowed to eat their habitual diet, so there might be other factors in the diet contributing to the findings. However, the authors point out that this has been controlled for in the analyses. The results were not changed markedly after controlling for energy intake, protein intake, SUN or milk intake. The authors concluded that whey protein stimulates fasting insulin.

Sandström et al. (31) graded B, did a randomized control study on the effects of exchanging part of the protein in milk-based formulas with α-lactalbumin on insulin levels (and other outcomes). All formulas contained 1.96 g protein/100 kcal, but with differences in the protein composition.

Three different formulas were used; 1) whey-predominant standard infant formula (11% α-lactalbumin, 14% glycomacropeptide [GMP]); 2) α-lactalbumin -enriched formula (25% α-lactalbumin), with GMP accounting for 15% of the protein content and 3) α-lactalbumin-enriched formula (25% α-lactalbumin), with GMP accounting for 10% of the protein content. Breastfed infants were the controls. Measurement errors in the dietary recordings were not considered.

All formula-fed infants had significantly higher plasma concentrations of most essential amino acids at 4 and 6 months than did the breastfed infants, and SUN was also higher in the formula-fed infants. Insulin and leptin concentrations did not differ between groups.

Based on the low number of studies, we conclude that evidence is limited-inconclusive (grade 4) that milk protein fractions are related to glucose–insulin metabolism in infancy and childhood.

Ulbak et al. (58) graded B, investigated in a cross sectional part of a randomized trial if macronutrients affected blood pressure in 2.5-year-old Danish children. The randomized trial concerned intake of n-3 long chain fatty acids of the mothers while lactating. Mothers kept 7-days food registration when the children were 2.5 years, the age when blood pressure was measured. Measurement errors in the dietary recordings were not considered. E% protein was significantly associated with both systolic and diastolic blood pressure, and g protein/day also significantly associated negatively with systolic blood pressure. The authors concluded that higher protein intake at the age of 2.5 years is associated with lower blood pressure.

Based on the low number of studies, we conclude that evidence is limited-inconclusive (grade 4) that higher protein intake is associated with decreased systolic blood pressure in young children.

Morgan et al. (37) graded B, studied health outcomes from complementary foods, especially growth and neurocognitive outcome from meat consumption and breastfeeding. Neurocognitive outcome was determined from the mental and psychomotor scales of the Bayley Scales of Infant Development II at the age of 22 months. Red and white meat consumption was calculated from 7-days’ weighed food intake diaries kept regularly at 4-month intervals from 4 months of age to 24 months. Measurement errors in the dietary recordings were not considered.

Total meat intake from all registrations at 4–12 months, that is, three times the 7-days’ registration (p<0.02) as well as 4–16 months (four registrations) (p<0.013) were positively and significantly related to psychomotor developmental indices. The authors concluded that meat intake from 4 to 16 months was associated with psychomotor development (and increased weight gain) possibly via its protein content. (Note: can the effect of ‘larger child eating more’ be excluded? Larger children, in this population from birth weight 2500 g, are likely to be more developed.)

Rask-Nissilä et al. (59) graded B, tested in 496 children, participating in the prospective STRIP project (the Special Turku Coronary Risk Factor Intervention Study), if dietary and other factors were associated with neurodevelopment. Neurodevelopment was tested at the age of 5 years by using a collection of developmental screening tests. Dietary intake was tested at 8, 13, and 18 months of age and thereafter twice a year. Measurement errors in the dietary recordings were not considered. In boys, stepwise logistic regression showed that protein intake predicted by positive association the outcome in speech and language skill tests at the age of 5 years.

Based on the low number of studies, we conclude that evidence is limited-inconclusive (grade 4) that higher protein intake is associated with improved neurodevelopment in children.