Ambrosini 2016.

Methods	Study design: prospective cohort study. Analyses for cohorts: cohort analysis: mean nutrient intakes across increasing quintiles of DP1a, DP1b and DP2 z‐scores estimated by using linear regression. Then, GEEs applied to investigate longitudinal associations between DP z‐scores and fat mass index (FMI) z‐scores. These models regressed FMI on DP z‐score at the previous time point by using DP z‐scores at 7, 10, and 13 years of age and FMI z‐scores at 11, 13 and 15 years of age. Models adjusted for time‐varying covariates (i.e. age, dietary misreporting, physical activity, Tanner stage) and fixed covariates (sex, maternal social class). CIF subsample analysis: linear regression used to model DP1a and DP2 z‐score at ages 5 and 7 years on FM (kg) at age 9 years. How were missing data handled? Cohort: lost to follow‐up at 7 years (6404/14,536, 44%); at 11 years (7542/14,536, 52%); at 13 years (8554/14,536, 59%) and at 15 years (9192/14,536, 63%). Study website contained details of all participants; reasons for attrition not provided by authors. Data analysis included all available data for the different time points. CIF subsample: complete data on diet and BC available for 521 (36%) children at ages 5 and 9 years and 682 (48%) children at ages 7 years and 9 years. Effect of missing data assessed (no data reported). Number of study contacts: 7 (at age 5, 7, 9, 10, 11, 13 and 15 years). Period of follow‐up (total period of observation): 4 years (CIF subsample from 5 to 9 years); 8 years (whole cohort from 7 to 15 years). Periods of recruitment: 1 April 1991 and 31 December 1992. Sample size justification adequately described? Yes. For a normally distributed quantitative trait (e.g. weight), a sample of 10,000 would be 80% certain to be able to show a difference of 0.19 SD as statistically significant if just 2% of the population had relevant exposure, whereas for a population of 1000, there would be sufficient power to demonstrate a difference of 0.62 SD (Golding 2001) Sampling method: convenience sample. Birth cohort that recruited pregnant women in Avon, UK. Of the 14,472 birth outcomes, 14,062 were live births and 13,988 were alive at 1 year. An additional 713 children whose mothers were initially invited but had not enrolled were recruited later. Total baseline cohort therefore included 14,701 children who were alive at 1 year. Of these, 8297 children attended clinics at age of 7 years. CIF sample: random subsample of 1432 children selected from births in the cohort that occurred in last 6 months of recruitment. Study objective: objective 1 (CIF subsample): to identify a DP that explained DED, FD and % energy from fat and analyse its association with fatness in children aged 5‐9 years. Objective 2 (whole cohort): to examine longitudinal relationships between a DP characterised by DED, % energy from fat and FD and FM in children aged 7‐15 years. Objective 3: to identify DPs characterised by high‐sugar content, HF content, or both, and their longitudinal associations with adiposity in children aged 7‐15 years. Study population: children and adolescents aged 5‐15 years in Avon, UK.
Participants	Baseline characteristics (reported for 2 groups: overall cohort and subsample of cohort) Overall cohort Age (mean in years): overall (n = 8224) 7.5 (SD 0.3); boys (n = 4174) 7.5 (SD 0.3); girls (n = 4050) 7.5 (SD 0.3). Sex: 49.25% girls. Ethnicity: majority of mothers white. Education: NR. Income: "...slightly more affluent than the national average." Pubertal stage: NR. Parental BMI: NR. Child total energy (kJ): overall (n = 7285); 7200 (SD 1300). Child total fat (%TE): overall (n = 7285); 35.5 (SD 4.4). Child total protein: NR. Child total CHO (%TE): overall (n = 7285); 54.5 (SD 5.1). Child physical activity: NR. Child physical inactivity or screen time or both: NR. Child CVD risk (excluding fatness): NR. Child body fatness, weight (kg): overall (n = 8211) 25.95 (SD 4.75); boys (n = 4164) 25.9 (SD 4.5); girls (n = 4047) 26.0 (SD 5.0); BMI at 7 years: overall (n = 8210) 16.25 (SD 2.06); boys (n = 4163) 16.1 (SD 1.9); girls (n = 4047) 16.4 (SD 2.2). CIF subsample (n = 521) Age (mean in years): 5.2 (SD 0.1). Sex: 45.68% girls. Ethnicity: majority white. Education: maternal education (n = 514 n, %): CSE: 38 (7.29), vocational: 45 (8.64); O level: 182 (34.93); A level: 162 (31.09); degree: 87 (16.69). Income: NR. Pubertal stage: NR. Maternal BMI (kg/m2): prepregnancy maternal BMI (n = 521) by adiposity status at age 9 years (median, IQR): normal adiposity 22.1 (20.6‐24.2); excess adiposity 23.8 (22.2 to 26.8). Child total energy (kJ): 6217 (SD 1395). Child total fat (%TE): 36.19 (SD 4.11). Child total protein (%TE): 13 (SD 2.04). Child total CHO (%TE): 53.81 (SD 5.98). Child physical activity: NR. Child physical inactivity or screen time or both: TV watching at 5 years (n = 498): 1 hour/day 27.5%; 1‐2 hours/day 44.4%; > 2 hours/day 28.1%. Child CVD risk (excluding fatness): NR. Child fatness:weight at 5 years by adiposity status at age 9 years (median, IQR): normal adiposity 19 (17.6‐20.4), excess adiposity 21.1 (19.0‐23.0); BMI at 5 years by adiposity status at 9 years (median, IQR): normal adiposity 15.6 (15.0‐16.3); excess adiposity 17.2 (16.3‐18.2). Included criteria: for cohort analysis, participants of ALSPAC cohort with follow‐up data at ages 7‐15 years were included. For analysis of CIF sample, eligible participants had available data on diet and BC at ages 5, 7 and 9 years. Excluded criteria: NR. Brief description of participants: aged 5‐15 years in ALSPAC cohort, Avon, UK. Total number completed in cohort study: 4729 (at 15 years). Total number enrolled in cohort study: 7285 at age 7 years (CIF subsample: 790 at age 3.6 years).
Interventions	Description of exposure for cohorts Overall cohort Time span: 8 years. Dietary assessment method: estimated food records/diaries. Frequency: single 3‐day DR (non‐consecutive days) at 7 (baseline),11 and 13 years. Components of dietary exposure: energy‐dense, HF, low‐fibre DP (DP1a); energy‐dense, high‐sugar, HF, low‐fibre DP (DP1b) versus non‐energy‐dense, high‐sugar, LF DP (DP2a; DP2b). CIF subsample Time span: 5 years. Dietary assessment method: estimated food records/diaries. Frequency: single 3‐day DR at 5 (baseline) and 7 years; components of dietary exposure: energy‐dense, HF, low‐fibre DP (DP1a). See Table 9; Table 10; Table 11; Table 12; Table 13; Table 14; Table 15; Table 16; Table 17; Table 18 for details of total fat intake exposure per outcome.
Outcomes	Body fat Body fat (kg). FMI z‐score. FMI z‐score > 80th percentile, odds. Height Height (cm).
Identification	Sponsorship source: UK Medical Research Council, Wellcome Trust and the University of Bristol. Country: UK. Setting: community. Comments: ALSPAC. Author's name: Gina L Ambrosini. Institution: School of Population Health, The University of Western Australia, Perth, Australia; Medical Research Council Human Nutrition Research, Cambridge, UK. Email: gina.ambrosini@uwa.edu.au. Declaration of interests: Yes. "no conflicts of interest." Study ID: Ambrosini 2016. Type of record: journal article.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Were adequate outcome data for cohorts available? All outcomes	Unclear risk	Attrition relevant to eligible analyses for FMI was 35% (2556/7285) over 8 years. For eligible analyses for BMI and height in CIF subsample, attrition over 1.5 years was 11% (84/790), and over 4 years for body fat was 7.3% (38/521). Authors reported that children who attended clinics for follow‐up were more likely to come from more affluent or better‐educated families than were children who did not attend clinics (data NR), and that there were no significant differences in dietary and anthropometric variables between children with complete data compared to those who did not (data NR).
Was there matching of less‐exposed and more‐exposed participants for prognostic factors associated with outcome or were relevant statistical adjustments done? All outcomes	Low risk	Most prognostic variables adjusted for. Parental BMI not assessed during study period. Data analysis of CIF subsample adjusted for prepregnancy maternal BMI and overweight status.
Did the exposures between groups differ in components other than only total fat? All outcomes	Unclear risk	NR.
Can we be confident in the assessment of outcomes? All outcomes	Low risk	Standard methods used for measurement of weight, height and body fatness (DEXA).
Can we be confident in the assessment of exposure? All outcomes	Low risk	Repeated 3‐day food diaries (non‐consecutive days) completed by parent or child, with parental assistance. Authors assessed dietary misreporting of energy intake.
Can we be confident in the assessment of presence or absence of prognostic factors? All outcomes	Low risk	Repeated measurements of total physical activity performed using accelerometer. Mean time spent by children watching TV reported by parents at 4.5 years. Pubertal status self‐reported at 11 and 13 years (using validated diagrams). Parental socioeconomic information and prepregnancy heights and weights were self‐reported.
Was selection of less‐exposed and more‐exposed groups from the same population? All outcomes	Low risk	All participants of the ALSPAC.