Table 4.

GRADE assessment of the systematic review and meta-analysis of prospective cohort studies assessing the association between total nut consumption and cardiovascular disease outcomes

Outcome	No. of studies	Study design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerationsa	Relative (95%CI)	Quality
CVD incidence	3	Observational studies	Not serious	Not serious	Seriousb	Not serious	Dose-response gradientc	RR 0.85 (0.80 to 0.91)	⊕⊕⊕◯◯ LOW
CVD mortality	14	Observational studies	Not serious	Not serious	Not serious	Not serious	Dose-response gradientd	RR 0.77 (0.72 to 0.82)	⊕⊕⊕◯ MODERATE
CHD incidence	7	Observational studies	Not serious	Seriouse	Seriousf	Seriousg	Dose-response gradienth	RR 0.82 (0.69 to 0.96)	⊕◯◯◯ VERY LOW
CHD mortality	12	Observational studies	Not serious	Not serious	Not serious	Not serious	Dose-response gradienti	RR 0.76 (0.67 to 0.86)	⊕⊕⊕◯ MODERATE
Stroke incidence	7	Observational studies	Not serious	Not serious	Seriousj	Seriousk	None	RR 1.00 (0.92 to 1.09)	⊕◯◯◯ VERY LOW
Stroke mortality	11	Observational studies	Not serious	Not serious	Not serious	Seriousl	Dose-response gradientm	RR 0.87 (0.76 to 1.00)	⊕⊕◯◯ LOW
Hemorrhagic stroke	5	Observational studies	Seriousn	Not serious	Seriouso	Seriousp	Dose-response gradientq	RR 1.02 (0.77 to 1.34)	⊕◯◯◯ VERY LOW
Ischemic stroke	7	Observational studies	Not serious	Not serious	Seriousr	Seriouss	None	RR 0.99 (0.89 to 1.10)	⊕◯◯◯ VERY LOW
Atrial fibrillation	2	Observational studies	Not serious	Not serious	Serioust	Seriousu	None	RR 0.85 (0.73 to 0.99)	⊕◯◯◯ VERY LOW
Heart failure	2	Observational studies	Seriousv	Not serious	Seriousw	Seriousx	None	RR 1.00 (0.86 to 1.16)	⊕◯◯◯ VERY LOW

Abbreviations: CHD, coronary heart disease; CI, confidence interval; CVD, cardiovascular disease; GLST, generalized least squares trend; GRADE, Grading of Recommendations, Assessment, Development, and Evaluation; NOS, Newcastle-Ottawa scale, RR, risk ratio.

^aPublication bias could not be assessed in meta-analyses that included <10 trial comparisons. Therefore, for these outcomes, no downgrades were made for publication bias.

^bSerious indirectness for CVD incidence, as the included studies were conducted among health professionals and >50% of the weight (69.30%) was contributed by studies conducted among males.

^cUpgrade for a dose-response gradient, as the MKSPLINE dose-response analyses showed a significant nonlinear inverse relationship between total nut consumption and CVD incidence (P < 0.01); see Figure S27 in the Supporting Information online.

^dUpgrade for a dose-response gradient, as the MKSPLINE dose-response analyses showed a significant nonlinear inverse relationship between total nut consumption and CVD mortality (P < 0.01); see Figure S28 in the Supporting Information online.

^eSerious inconsistency for CHD incidence due to high degree of unexplained heterogeneity (I² = 74%, P = 0.001).

^fSerious indirectness for CHD incidence, as >50% of the weight (55.4%) was contributed by studies conducted among health professionals.

^gSerious imprecision for CHD incidence, as the 95%CI (0.69–0.96) overlapped with the minimally important difference for clinical benefit (RR 0.95).

^hUpgrade for a dose-response gradient, as the GLST dose-response analyses revealed a significant linear inverse relationship between total nut consumption and CHD incidence (P < 0.01); see Figure S29 in the Supporting Information online.

ⁱUpgrade for a dose-response gradient, as the MKSPLINE dose-response analyses showed a significant nonlinear inverse relationship between total nut consumption and CHD mortality (P < 0.01); see Figure S30 in the Supporting Information online.

^jSerious indirectness for stroke incidence, as >50% of the weight (72.7%) was contributed by studies conducted among health professionals.

^kSerious imprecision for stroke incidence as the 95%CI (0.92–1.09) overlapped with the minimally important difference for clinical benefit (RR 0.95) and harm (RR 1.05).

^lSerious imprecision for stroke mortality, as the 95%CI (0.76–1.00) overlapped with the minimally important difference for clinical benefit (RR 0.95).

^mUpgrade for a dose-response gradient, as the MKSPLINE dose-response analyses showed a significant nonlinear inverse relationship between total nut consumption and stroke mortality (P = 0.029); see Figure S32 in the Supporting Information online.

ⁿSerious risk of bias for hemorrhagic stroke, as >50% of the weight (68.7%) was contributed by studies considered to be at high risk of bias (NOS < 7).

^oSerious indirectness for hemorrhagic stroke, as >50% of the weight (68.7%) was contributed by studies conducted among health professionals and >50% of the weight (55.7%) was contributed by studies conducted among males.

^pSerious imprecision for hemorrhagic stroke as the 95%CI (0.77–1.34) overlapped with the minimally important difference for clinical benefit (RR 0.95) and harm (RR 1.05).

^qUpgrade for a dose-response gradient, as the MKSPLINE dose-response analyses showed a significant nonlinear inverse relationship between total nut consumption and hemorrhagic stroke (P = 0.01); see Figure S33 in the Supporting Information online.

^rSerious indirectness for ischemic stroke, as >50% of the weight (66.1%) was contributed by studies conducted among health professionals.

^sSerious imprecision for ischemic stroke, as the 95%CI (0.89–1.10) overlapped with the minimally important difference for clinical benefit (RR 0.95) and harm (RR 1.05).

^tSerious indirectness for atrial fibrillation, as only 2 available studies were conducted among males.

^uSerious imprecision for atrial fibrillation, as the 95%CI (0.73–0.99) overlapped with the minimally important difference for clinical benefit (RR 0.95).

^vSerious risk of bias for heart failure, as >50% of the weight (65.40%) was contributed by a study considered to be at high risk of bias (NOS < 7).

^wSerious indirectness for heart failure, as only 2 available studies were conducted among males.

^xSerious imprecision for heart failure, as the 95%CI (0.86–1.16) overlapped with the minimally important difference for clinical benefit (RR 0.95) and harm (RR 1.05).