`metacont` and `forest`: studies with available central tendency but not spread are not counted in `metacont$K`

[guido-s]

`metacont` and `forest`: studies with available central tendency but not spread are not counted in `metacont$K`

[guido-s]

I cannot reproduce the problem with the following R code:

m1 <- metacont(10 + 1:4, 1:4, c(1:2, NA, 4),
               10 + 1:4, 1:4, c(1:2, NA, 4))
m1
forest(m1)

m2 <- metacont(10 + c(1:4, NA), c(1:4, NA), c(1:2, NA, 4, NA),
               10 + c(1:4, NA), c(1:4, NA), c(1:2, NA, 4, NA))
m2
forest(m2)

Which R code are you using to generate the forest plot?

BTW, k is the number of studies contributing data to the meta-analysis which is 10 in both your examples.

[Rscho314]

The behaviour seems to appear by setting allstudies = F in forest in my code. Sorry, I should absolutely have mentioned that!

BTW, k is the number of studies contributing data to the meta-analysis which is 10 in both your examples.

Indeed, that was what I meant. In metacont, k does not seem to account for the study with only central tendency and forest seems to always display k studies when using allstudies = F, therefore missing the last one.

So this exhibits the behaviour I described:

m1 <- metacont(10 + 1:4, 1:4, c(1:2, NA, 4),
10 + 1:4, 1:4, c(1:2, NA, 4))
forest(m1, allstudies = F)  # yields Warning Message: In forest.meta(m1, allstudies = F) : n.stud != sum(sel)
m2 <- metacont(10 + 1:4, c(1:2, NA, 4), c(1:2, NA, 4),
10 + 1:4, c(1:2, NA, 4), c(1:2, NA, 4))
forest(m2, allstudies = F)

forest(m1, allstudies = F) does not display study 4.

Perhaps this is intended and I am misunderstanding?

[guido-s]

Thank you for pointing out this bug (and the example R code).