2.9 Outcomes and prioritization
One important outcome of interest was to gather and summarize data on
the frequency of detection from cross-sectional studies, and measures of
association (95% confidence intervals) in case-control and case reports
studies. For case-control studies the measure of association reported
was the odds ratio. In addition to that clinical presentations, type of
samples and diagnostic tools used for the detection of porcine
astroviruses was reported.
For case-control studies, when the OR was not reported it was calculated
if the necessary data were provided. Odds ratios were calculated using a
two-by-two frequency table and the 95% confidence intervals were
calculated using a formula described by Suzmilas et al., 2010. When
there were zero values in the two-by-two table (which make an OR
incalculable), 1 was added to each cell when calculating OR for
detection of PoAstVs. The statistical software RStudio Version 1.2.1335
was used to create a forest plot using R Package meta (R Core Team R et
al., 2013) which summarizes the quantitative findings. A decision was
made a priori that if the heterogeneity from the forest plot was
significant i.e., P-value of heterogeneity ≤ 0.05, the effects should be
summarized using a random effects model (Fletcher et al., 2007).
Otherwise, a high P value (P > 0.05) suggests that
the heterogeneity was insignificant, and results would be summarized
using a fixed effects model.
The subjective reasoning based on the information provided by each paper
was used to determine whether each of the nine Bradford Hills criteria
was fulfilled. The evidence collected for each Bradford Hill’s criteria
presented in results with final judgment as to whether the viewpoint was
fulfilled or not.