Meta‐analysis of prevalence: I² statistic and how to deal with heterogeneity

Abstract

Over the last decade, there has been a 10‐fold increase in the number of published systematic reviews of prevalence. In meta‐analyses of prevalence, the summary estimate represents an average prevalence from included studies. This estimate is truly informative only if there is no substantial heterogeneity among the different contexts being pooled. In systematic reviews, heterogeneity is usually explored with I‐squared statistic (I2), but this statistic does not directly inform us about the distribution of effects and frequently systematic reviewers and readers misinterpret this result. In a sample of 134 meta‐analyses of prevalence, the median I2 was 96.9% (IQR 90.5–98.7). We observed larger I2 in meta‐analysis with higher number of studies and extreme pooled estimates (defined as <10% or >90%). Studies with high I2 values were more likely to have conducted a sensitivity analysis, including subgroup analysis but only three (2%) systematic reviews reported prediction intervals. We observed that meta‐analyses of prevalence often present high I2 values. However, the number of studies included in the meta‐analysis and the point estimate can be associated with the I2 value, and a high I2 value is not always synonymous with high heterogeneity. In meta‐analyses of prevalence, I2 statistics may not be discriminative and should be interpreted with caution, avoiding arbitrary thresholds. To discuss heterogeneity, reviewers should focus on the description of the expected range of estimates, which can be done using prediction intervals and planned sensitivity analysis.

The Impact of the COVID-19 Pandemic and Associated Control Measures on the Mental Health of the General Population

“…When interpreting the results, we distinguished between the average summary SMD from the meta-analysis and the expected SMDs shown in the range of the PrI. We estimated between-study heterogeneity by considering the SD of the random effects (τ) and the width of the PrI rather than the I 2 metric ( 18 , 19 ). To enhance interpretation, we transformed summary SMDs into mean differences on the Patient Health Questionnaire-9 scale for depression and on the General Anxiety Disorder-7 scale for anxiety, using the observed pooled SDs of a large representative study ( 20 ).…”

An epidemiological synthesis of emerging and re-emerging zoonotic disease threats in Cameroon, 2000–2022: a systematic review

“…For the pathogens with at least five studies, heterogeneity was greater than 75%. This is expected with a meta-analysis of prevalence studies because of the difference in the sampling frame, sampling technique, sampling size, diagnostic tool, and differing biomarkers reported as a result [26] .…”

Health facility delivery service utilization and its associated factors among women in the pastoralist regions of Ethiopia: A systematic review and meta‐analysis

“…In addition, studies indicated that a high I 2 value is not always synonymous with high heterogeneity, and, like this study, when only a small number of studies with no true heterogeneity are included in the meta‐analysis, I 2 will overestimate heterogeneity. 37 , 38 It could not be therefore difficult to utilize this paper for different purposes. The publication bias has also been overcome by conducting trim and fill analysis.…”