In the sensitivity analyses on seroprotection, heterogeneity disappeared when analyses were stratified by antigen content (≤9 μg) for the A/H3N2 and B strains, by other causes of immunosuppression excluding transplantation for the same strains, and by injection site and by micro injection system for B strain.
Many of the findings in the unadjusted analyses disappeared when the analyses were controlled for baseline mental health at T1.
We detected a dispersal-aggregation trade-off consistent with basic theory in analyses including within-strain variation, but this relationship disappeared in analyses focused only on among strain variation suggesting the relationship is weak or due to non-genetic effects.
The statistical significant associations found in the non-calibrated analyses disappeared in the calibrated analyses and bootstrap analyses.
Those authors evoked the same kind of LBA artefact to explain what they considered as an artefactual result, as support for it disappeared in analyses including fewer sites but with fewer missing data [ 32].
Second, we found a significant correlation between SAI and memory for faces (Faces I; r = −0.31, p = 0.02), although, again, it disappeared when analyses were performed separately within each age group (r ≤ |0.17|).
There is some evidence of such confounding in relation to ethnic groups, i.e., relationships between genotype and phenotype that were found in multiethnic populations disappeared when analysed separately in each ethnic group [41], [42].
On the other hand, differences in diffusion anisotropy of the anterior corpus callosum fibres between FTD and Alzheimer's disease disappeared when analysed in terms of Dra.
The oldest transfused RBCs were strongly associated with all of the in-hospital postoperative adverse outcomes except for MACCEs in the univariate analyses but the predictive effects disappeared in multivariable analyses adjusting for confounders, except for postoperative wound complications (Table 3, see also Additional file 2).
In univariate analyses, estimates of nonsusceptibility from MYSTIC were consistently higher than those from SENTRY, but these differences disappeared in multivariate analyses, which supports the epidemiologic relevance of findings from the 2 programs, despite differences in surveillance strategies.
In the adjusted analyses, the association between unknown and probable diabetes mellitus with poor SRH, found in the crude analyses, disappeared when age was included in the model in women and men, but also by inclusion of working status alone in women.
