RATIONALE. Soymilk and other processed soy products are at an important crossroads. Although the DGA, Health Canada, and various pediatric associations recognize fortified soymilk as the only non-dairy plant milk alternative equivalent to cow's milk and it can carry an approved health claim for coronary heart disease risk reduction based on the soy protein that it contains, soymilk is classified by the NOVA classification as an ultra-processed food (the opposite of the classification of cow's milk as an unprocessed or minimally processed food). To be an acceptable iso-sweet alternative to cow's milk, soymilk is also often sweetened with sucrose, which is designated as an added sugar, whereas the lactose that sweetens cow's milk is not (despite lactose in cow's milk being present in quantities that are double that of sucrose in soymilk products designed to be iso-sweet analogues of cow's milk). With near universal recommendations from major public health authorities to reduce the intake of both ultra-processed foods and added sugars and the FDA updating its "healthy" claim criteria to limit added sugars, the role of soymilk as a "healthy" non-dairy alternative to cow's milk is in serious question.
OBJECTIVES. The objective is to conduct a systematic review and meta-analysis of randomized controlled trials to assess the effect of soymilk in substitution for cow's milk and its modification by added sugars (sweetened versus unsweetened) on established cardiometabolic risk factors of clinical and public health importance and assess the certainty of the evidence using the Grading of Recommendations Assessment Development and Evaluation (GRADE) system.
DESIGN. The systematic review and meta-analyses will be conducted according to the Cochrane Handbook for Systematic Reviews of Interventions and reported according to the Preferred Reporting items for Systematic Reviews and Meta-Analyses (PRISMA).
DATA SOURCES. Medline, Embase, and The Cochrane Central Register of Controlled Trials (Clinical Trials; CENTRAL) will be searched using appropriate search terms supplemented by manual searches of references of included studies.
STUDY SELECTION. Randomized controlled trials of ≥3-weeks assessing the effect of soy milk in substitution for cow's milk on cardiometabolic risk factors will be included.
DATA EXTRACTION. Two or more investigators will independently extract relevant data. Authors will be contacted for additional information and any missing data will be computed/imputed using standard formulae.
RISK OF BIAS. Two or more investigators will independently assess risk of bias using the Cochrane Risk of Bias Tool 2. All disagreements will be resolved by consensus.
OUTCOMES. Outcomes will include changes in established markers of (1) adiposity (body weight, BMI, body fat, waist circumference, abdominal fat); (2) glycemic control (HbA1c, fasting plasma glucose (FPG), fasting plasma insulin (FPI), plasma glucose area under the curve (AUC), and 2h plasma glucose (2hPG)); (3) blood lipids (LDL-cholesterol, non-HDL-cholesterol, apolipoprotein B, HDL-cholesterol, triglycerides); (4) blood pressure (systolic blood pressure and diastolic blood pressure); (5) non-alcoholic fatty liver disease (NAFLD) (intrahepatocellular lipids (IHCL), alanine transaminase (ALT), aspartate aminotransferase (AST), and fatty liver index (FLI)); (6) inflammation (C-reactive protein (CRP)); (7) renal function and structure (creatinine, creatinine clearance (CrCl), glomerular filtration rate (GFR), estimated GFR (eGFR), albuminuria, albumin-to-creatinine ratio (ACR)); and (8) and uric acid.
DATA SYNTHESIS. Data will be pooled using the Generic Inverse Variance method for each outcome. As one of our primary research questions relates to the role of added sugars as a mediator in any observed differences between soy milk and cow's milk, we will stratify results by the presence of added sugars in soy milk (sweetened versus unsweetened) and assess effect modification by this variable on pooled estimates. Random effects models will be used even in the absence of statistically significant between-study heterogeneity, as they yield more conservative summary effect estimates in the presence of residual heterogeneity. Fixed-effects models will only be used where there is \<5 included studies. Paired analyses will be applied for crossover design trials. Heterogeneity will be assessed (Cochran Q statistic) and quantified (I2 statistic). Sources of heterogeneity will be explored (if there are \>=10 trial comparisons) by sensitivity analyses and a priori subgroup analyses (dose, comparator, follow-up, baseline levels, design, age, health status, funding and risk of bias). Meta-regression analyses will assess the significance of categorical and continuous subgroups analyses. Publication bias will be assessed (if there are \>=10 trial comparisons) by inspection of funnel plots and the Egger and Begg tests. Adjustment for evidence of funnel plot asymmetry or small study effects will be conducted by the Duval and Tweedie trim-and-fill method.
GRADE ASSESSMENT. To assess the certainty of the evidence, the investigators will use the GRADE system, an evidence-based grading system adopted by \>100 organizations (http://www.gradeworkinggroup.org/). It grades the evidence as high, moderate, low or very low quality based on the study design and a series of downgrades (risk of bias, imprecision, inconsistency, indirectness, publication bias) and upgrades (large magnitude of the effect, dose-response gradient, and attenuation by confounding). The investigators will follow the GRADE handbook (https://gdt.gradepro.org/app/handbook/ handbook.html) and use the GRADEpro GDT (gradepro.org) software.
ADD-ON SYNTHESIS (LACTOSE VERSUS ADDED SUGARS) To interrogate the question of the role of added sugars as a mediator of any observed effects of sweetened soymilk, we will conduct a sub-synthesis (a separate systematic review and meta-analysis of RCTs) of the effect of lactose (which is not considered an added sugar) versus added sugars (sucrose, HFCS, or fructose) irrespective of matrix (all comparisons with a matched matrix will be included). We will follow the same protocol and analysis plan as that for the main analysis.
KNOWLEDGE TRANSLATION PLAN. The investigators will follow the Ottawa model of Research for knowledge translation. The results will be disseminated through interactive presentations at local, national, and international scientific meetings and publication in high impact journals. Target audiences will include public health and clinical communities with an interest in nutrition and cardiometabolic diseases. Feedback will be incorporated and used to improve public health messages and key areas for future research will be defined. The PIs will network among opinion leaders to increase awareness and participate directly in the development of future guidelines.
SIGNIFICANCE. Both proposed projects will aid in knowledge translation related to the role of soy milk as a "healthy" non-dairy alternative to cow's milk, strengthening the evidence-base for guidelines development in the U.S., Canada, Europe, and beyond and improving health outcomes, by educating healthcare providers and patients, stimulating industry innovation, and guiding future research design.
