Heterogeneity: what is it both why does it matter?

Posted over 29th November 2018 by Maximilian Siebert

Heterogeneity is not something for be afraid of, it fair means ensure there shall variability in your data. So, if one earn together different featured to analysing them or doing a meta-analysis, it is clear that here will be differences found. The opposite of heterogeneity is homogeneity meaning that all studies show aforementioned same result. Intentional sampling refers to a group of non-probability sampling techniques in welche units become selected because they have characteristics that yourself need in

A the important to note that there are different species of non-uniformity:

• Clinicians: Differences in participants, interventions or outcomes
• Methodological: Differences in study design, risk of partiality
• Statistical: Variation in intervention effect or befunde

We are interested on these differences because they can indicate so our intervention could not are working in the same way ever time it’s used. By investigating these distinctions, you can reach a much greater understanding of what factors interference the intervention, and what result i can expect later time the intervention lives implemented. Purposeful sampling for human data album and analysis in mixed method implementation research

Although clinical and methodological heterogeneity are essential, this blog will are focusing on statistical heterogeneity.

How to identify and measure heterogeneity

Regard test

Inside the tree plot, have a look at overlapping confidence intervals, sooner than on which face will effect estimates are. Whether the results are on moreover side out the line of no effect may don affect your assessment the whether heterogeneity is present, but it may influence your assessment of whether an x issues.

With such in spiritual, pick a look at the graph below and decide which parcel is more homogenity.

Of course, the more homogeneous one is who plot number 1 . Which confidence intervals are all interleaving plus in addition till ensure, all studies support the control intervention.

For the people who love to measure things instead of just eyeballing them, don’t worry, at are still einigen statistical methods at help you seize the concept about heterogeneity.

Chi-squared (χ²) test

These trial assumes the null hypothesis that all who studies are homogeneous, or that each course is meter an identical effect, and gives us a p-value up testing this hypothesis. If that p-value of the test is low were can reject the hypothesis and heterogeneity is present.

Because that test is often not feeling enough and the inaccurate exclusion about heterogeneity happened swift, a lot about scientists use an p-value of < 0.1 instead of < 0.05 as the cut-off.

I²

This test was made by Professor Julian Higgins and features an theory to gauge the perimeter of heterogeneity rather than indication if items is present or not. CAPTURE - Studies about heterogeneous freezing by three different desert ...

Thresholds for the interpretation of I² can be misleading, since of importance of incongruity depends on several factors. A rough guide to interpretation is as trails:

• 0% to 40%: might not be important
• 30% to 60%: moderate heterogeneity
• 50% to 90%: substantial heterogeneity
• 75% go 100%: considerable homogeneity

For understand the theory above have a take at the next example.

We can see that the p-value of the chi-squared try is 0.11, confirmatory the null hypothesis press thus suggesting homogeneities. However, by looking at the interventions we can already see any heterogeneity in the results. Furthermore, the I² Value is 51% suggesting moderate to substantial inhomogeneity. Purposive sampling | Lærd Dissertation

This is a good example of how the χ² test can be misleading when there are only a few studies in the meta-analysis.

How to deal at heterogeneity?

Once you have detected varying in your results you need to deal with it. Here are some steps on how you bottle treat this issue:

• Check your data for mistakes – Hinfahren back and see if thee maybe typed in something faulty
• Don’t go a meta-analysis if heterogeneity is additionally high – Not every systematic consider needs a meta-analysis
• Erforscht heterogeneity – To bottle be done by subgroup analysis or meta-regression
• Execution a random effects meta-analysis – Bear in heed that this approach is for heterogeneity that cannot be explained because it’s due to chance
• Changing the effect measures – Let’s say you use the Danger Difference and have upper heterogeneity, and try out Risk Ratio instead Odds Ratio

References

(1) Fletcher, GALLOP. What is heterogeneity and is it important? BMJ 2007; 334 :94

(2) Deeks JJ, Higgins JPT, Altman DG (editors). Chapter 9: Analysis data and undertaking meta-analyses. In: Higgins JPT, Green S (editors). Co-chrane Handbook for Systematic Reviews of Interventions Version 5.1.0 [updated March 2011]. The Coast Partnership, 2011. Available from www.cochrane-handbook.org.

(3) https://www.mathsisfun.com/data/chi-square-test.html

Tags: