Correction: Performance of several types of beta-binomial models in comparison to standard approaches for meta-analyses with very few studies

Correction: BMC Medical Research Methodology 22, 319 (2022)

https://doi.org/10.1186/s12874-022-01779-3

Following publication of the original article [1], the authors requested to update the following errors.

Location	Incorrect	Correct
Abstract: Background	Delete the last sentence in the Background part of the Abstract: “Therefore, we extended this model to a version that respects randomisation.”
Abstract: Results	Delete the last sentence in the Results part of the Abstract: “The beta-binomial model respecting randomisation did not improve performance.”
page 3, right column, 1st paragraph	“The same formulas hold true for the number of events in the treatment arm with , , , , and .”, has to be changed to .	“The same formulas hold true for the number of events in the treatment arm with , , , , and .”
page 3, right column, 3rd paragraph
page 3, right column, 8th paragraph	“In the BBST, the event probability in the control arm is random but the treatment effect is considered to be fixed across all studies. Thus, although the BBST is a true random effects model, from a meta-analytic point of view, it is a model with a fixed treatment effect.”	“In the BBST, the event probability in the control arm is random and the same holds for the event probability in the treatment arm , so the treatment effect is random as well (and with probability 0). As the study-specific treatment effect is random, the BBST is a true random effects model, also from the standard inverse-variance meta-analytic point of view.”
page 4, left column, 2nd paragraph	“To deal with the aforementioned properties of BBST as a fixed effect model that disrespects randomisation, we implemented another BBM (BBFR) where the treatment effect …”	“We implemented another BBM (BBFR) where the treatment effect…”
page 4, left column, between 2nd and 3rd paragraph
page 4, left column, 3rd paragraph	Delete the sentence “By adding a random effect to the treatment effect, this model respects the randomisation to a study arm of a specific study.”
page 4, right column, 8th paragraph	Delete the sentences “Like the BBST, the GLMM is a random effects model. But as the treatment effect is random it is more comparable to meta-analytic REMs than the BBST.”
page 15, left column, 3rd paragraph	“The attempt to extend the BBST by a random treatment effect term for the study, due to concerns about disrespecting the randomisation, failed.”	“The attempt to extend the BBST by a random treatment effect term for the study failed.”
	“As BBFR uses a random effect attached to the treatment effect in addition to the random intercept, this behaviour is to be expected, because only then “enough” additional heterogeneity remains to be estimated, i.e. not all heterogeneity goes into the random intercept.”	“As BBFR uses a random effect attached to the treatment effect in addition to the random intercept, this behaviour is to be expected, because only then “enough” additional heterogeneity remains to be estimated.”
page 15, right column, 2nd paragraph	Delete the sentence “One could try a BBFR with more quadrature points than 1, but this might not work because of convergence problems.”

The original article [1] has been corrected.