Zoonotic diseases are growing in prevalence, creating a need for integrated research, and we face an ever-growing replicability crisis due to publication, statistical, and literary biases. Consequently, there are significant literature gaps surrounding the ecology of zoonoses. This study provides insight to address these issues through meta-analytical evaluations of ecological, epidemiological, and interdisciplinary studies. We seek to ascertain whether these fields are reporting relevant, non-biased, results via post-hoc analyses (of heterogeneity). A random effects model was chosen to measure the normal distribution of effects,heterogeneity, and bias in 71 zoonotic studies. A literature search was conducted using 26 search terms across four databases. The number of search hits per term were recorded over a period of six months and averaged to investigate publication preferences and literary biases. The frequencies of reported statistics were calculated to determine data use preferences per field. The search yielded 7,291 effect sizes.
Results/Conclusions
The overall effect size was negative (-0.23) and crossed null, suggesting effects reported are no different than zero. The I2 and τ2 values calculated were greater than 98% and ranged from 1.6927 – 14.5967, suggesting heterogeneity is the result of bias. Kendall’s rank correlations were significant for epidemiological (-0.01772, p<0.0001, df=3000) and interdisciplinary studies (0.2854, p<0.0001, df=629), and significant for ecological studies (0.07, p=0.0009, df=1014) following bias correction. Egger’s regression tests (z) of standard error and variance were significant for ecological (p<0.0001) and epidemiological studies. Egger’s test (z) of sample size was significant for epidemiological studies. Egger’s tests (t) of sample size were significant for all studies. Egger’s regression tests (t) were all significant for interdisciplinary studies. Google Scholar held the highest average hits for 21 search terms. PubMed displayed preference towards public health with limited hits for ecological studies. PLoS ONE displayed preferences for ecology and infection-based terms. Elsevier yielded the most hits for field-based terminology but was otherwise limited. Proportions (n=67) were the most used statistic, followed by means (n= 9), regressions (n=45), effect sizes (n=42), frequencies (n=27), and correlations (n=15). Kendall’s tests suggest publication bias is a prominent issue across fields. The negative direction of this test indicates false-reporting of large effect sizes inhibits epidemiological research, while small sample size biases impact ecological and interidisciplinary studies. Egger’s tests provide supporting evidence for Kendall’s tests and suggest numerous forms of biases impact zoonotic studies.