Primary outcomes
Significant differences were noted in serum LDH levels of women
diagnosed with preeclampsia compared to controls (Figure 2 ).
Significant asymmetry was observed following assessment of the forest
plot with the majority of assessed studies falling into the level of
higher statistical significance (p<.01); thus, suggesting that
it might be attributed to other reasons than publication bias. The trim
and fill function included 11 simulated studies and the overall effect
following their addition was not significant (MD -334.42, 95%CI
-334.42, 58.81, p=.162 ). The meta-regression function did not
reveal a significant effect of the NOS score and publication year on the
outcomes of the meta-analysis. Trial sequential analysis revealed that
the required sample size was reached to ensure the adequacy of findings.
Prediction intervals did not reach statistical significance (-385.82,
744.45 IU/L).
Similarly, differences among mild preeclampsia cases and controls were
also noted. Significant asymmetry was observed that was also attributed
to other reasons than publication bias as the majority of studies fell
into the level of high statistical significance (p<.01). The
trim and fill function included 7 simulated studies and the overall
effect following their addition was not significant (MD 68.85, 95%CI
-22.75, 160.45). The meta-regression analysis revealed that the effect
was significantly skewed by the parameters of NOS and publication year;
thus, denoting that the asymmetry might be the result of these
parameters. The required information was exceeded with the first
fraction of studies as denoted by the trial sequential analysis results.
Prediction intervals fell into the non-significant level (-101.47,
453.74).
Severe preeclampsia cases had more pronounced differences compared to
control women (MD 371.86, 95% CI 258.76, 484.97 IU/L). Funnel plot
analysis revealed significant asymmetry that was not attributed to
publication bias. The trim and fill function included 7 simulated
studies and the overall effect following their addition maintained
statistical significance (MD 175.68, 95%CI 18.53, 332.83).
Meta-regression analysis revealed that both the NOS and publication year
influenced the results of the primary analysis; thus, denoting that the
asymmetry might be the result of these factors. Sample size was adequate
to support this finding.
Direct comparison of cases with mild with those that had severe
preeclampsia revealed that women with mild preeclampsia had
significantly lower levels of LDH (MD -162.04, 95% CI -217.18, -106.80
IU/L) (Figure 3 ). Ten simulated studies were added with the
trim and fill function which resulted in non-significant differences
among the two groups (MD -51.25, 95%CI -125.18, 22.67). As in the case
of the previous comparisons, funnel plot analysis revealed that
asymmetry was not attributed to publication bias, but meta-regression
analysis failed to reveal a significant effect of the quality of
included studies or of the year of publication to these findings. Sample
size was adequate to support the result of the meta-analysis.