Table 7.

GRADE evidence profile table^a.

Quality assessment							Strength of Evidenceb	Summary & Justification
No of studies	Design (Ref.)	Limitations	Inconsistency	Indirectness	Imprecision	Dose-response
Cognition-related outcomes: Cognitive function, Mood, Stroke
5	RCTs [35–39]	Serious limitations: Overall ROB was high for 60% of trials and SC for 40% of trials. Studies were limited mostly due to concerns arising from the randomization process and deviations from intended interventions.	No serious inconsistency: No trials reported significant differences in mood and total cognitive scores between walnut and control groups. Significant differences were observed in population subgroups and/or subdomains only.	No serious indirectness: Clinical and/or validated cognition-related tests used.	Imprecise: All studies reported wide CIs or other measures of variance.	Not applicable: no within study comparisons of different walnut intake amounts.	⊕⊕OO LOW	None of the 5 RCTs found a significant effect of walnut on mood or cognitive function in complete study populations, though subanalyses and/or subdomains demonstrated a walnut effect. Due to concerns regarding risk of bias and imprecision in measures, we conclude that the SOE for the effects of walnut intake on cognition-related outcomes is low.
7	Observational [20,29–34]	Some limitations: Overall ROB was SC for 86% of trials and low for 14% of trials. Studies were limited mostly due to participant selection, self-reported walnut intake, and incomplete or selective reporting of results.	No serious inconsistency: Significant differences were reported between walnut and control groups for mood (100%) and total cognitive function scores (80%). The remaining study reported a significant effect of walnut on a subdomain of cognitive function.	No serious indirectness: Clinical and/or validated cognition-related tests used.	Some imprecision: 43% of studies reported wide CIs and/or small sample sizes.	Dose-response is present for all cognition- related outcomes.	⊕⊕OO LOW	There are insufficient data to support a hypothesis on the associations between walnut intake and mood or stroke, as ≤ 2 studies reported on these outcomes. However, 3 cross-sectional studies and 1 prospective cohort study found significant associations between walnut intake and cognitive function. Despite moderate imprecision and concerns of bias across these studies, the demonstration of a dose response effect upgraded the SOE for the association between walnut intake and cognition-related outcomes to a low rating.
Inflammation
17	RCTs [41,44, 50–58,61–66]	Some limitations: Overall ROB was low for 44%, SC for 44%, and high for 12% of trials.	No serious inconsistency: Across studies reporting inflammation markers, 60–100% did not find a significant effect of walnut compared to control.	No serious indirectness: Biomarkers of inflammation.	Some imprecision: Results were imprecise across studies, as indicated by moderate or high statistical heterogeneity of meta-analyses.	Not applicable; no within study comparisons of different walnut intake amounts.	⊕⊕⊕O MODERATE	The majority of studies (81%) reporting inflammation outcomes observed no effects of walnut. Additionally, meta-analyses found no significant effects of walnut on inflammation (hsCRP, VCAM, ICAM, TNFa, E-selection, IL-6). Due to concerns of bias and imprecision, the SOE for the effects of walnut intake on inflammation outcomes was rated as moderate.
Glucose homeostasis
13	RCTs [40–49,59, 60,62]	Some limitations: Overall ROB was low for 31%, SC for 54%, and high for 15% of the trials reporting glucose outcomes. Studies were limited mostly due to concerns arising from the randomization process, deviations from the intended intervention, missing outcome data, and selection of reported results.	No serious inconsistency: 70% of studies did not find a significant effect of walnut on HbA1c and none of the included studies found an effect of walnut on HOMA-IR.	No serious indirectness: HbA1c and HOMA-IR are validated measures of glucose homeostasis.	Imprecise: Results were imprecise across studies, as indicated by moderate to large statistical heterogeneity in meta-analyses. The 2 studies that could not be meta-analysed also raised concerns of imprecision, due to large measures of variance and lack of quantitative data.	Not applicable; no within study comparisons of different walnut intake amounts.	⊕⊕⊕O MODERATE	70% of RCTs found no significant effect of walnut on HbA1c and no effects were observed on HOMA-IR. The meta-analyses did not find an overall significant effect of walnut on HbA1c or HOMA-IR. Due to concerns of bias and imprecision, SOE for the effects of walnut intake on glucose outcomes was rated as moderate.

^aCI: confidence interval; GRADE: Grades of Recommendation; RCT: randomized-controlled trials; ROB: risk of bias; SC: some concerns; SOE: strength of evidence.

^bSymbols indicate the following strength of evidence: ⊕⊕⊕⊕, HIGH (further research is very unlikely to change our confidence in the estimate of association); ⊕⊕⊕O, MODERATE (further research is likely to have an important impact on our confidence in the estimate of association and may change the estimate); ⊕⊕OO, LOW (further research is very likely to have an important impact on our confidence in the estimate of association and is likely to change the estimate); and ⊕OOO, VERY LOW (any estimate of association is very uncertain).