Subgroup, sensitivity analyses, and publication bias:
To test the robustness of the results and investigate the between-study
heterogeneity, we conducted subgroup analyses. Supplemental
Table 3 presents findings for the different subgroups. In terms of
total water intake and risk of all-cause mortality, a significant
inverse association was obtained in studies conducted in US countries.
However, a significant positive association was seen in studies
conducted in US countries and those that had applied food recall or
record for total water intake assessment. In terms of total water intake
and risk of CVD mortality a significant inverse association was seen in
studies that conducted in US and non-US countries, those that performed
in females and males, studies that had used FFQ for total water intake
assessment, studies with a >10 years follow-up duration and
those that controlled for BMI and energy intake in their analysis. In
terms of drinking water intake and all-cause mortality, a significant
inverse association was reached in studies that had applied FFQ for
drinking water assessment.
Findings from the sensitivity analysis using a fixed-effects model
revealed that the exclusion of any single study from the analysis did
not appreciably alter the pooled effect sizes. There were no missing
studies imputed in regions of the contour enhanced funnel plots. No
publication bias was found based on Egger’s linear regression test. The
application of the trim and fill method did not change the average
effect size, further suggesting that results were not affected by
publication.