[R-meta] Combining percentages, odds, categorical values

[Mika Manninen]

Hello all,

I am conducting a meta-analysis on the effect of an intervention on
decision making. I have pre and post scores (m, sd, n) for both the control
and intervention groups.

Most of the studies measure decision making in 1) percentages (appropriate
decisions/appropriate + inappropriate decisions). Some studies report the
same in 2) odds (appropriate decisions/inappropriate decisions). Lastly,
some studies use a 3) categorical or continuous variable for decision
making (ranging from 1-3 or 1-5 - e.g. not appropriate at all, somewhat
inappropriate ... very appropriate).

What would be the best way to aggregate the effects from individual studies
in this case? Would it be as straightforward as excluding the
categorical/continuous measures and simply deal with probabilities
(transform odds to probabilities)? In this case, I would be meta-analysing
percentages.

Also, as many studies have several effects coming from them (between
correlations are unknown), would I best be advised to use RVE (robumeta) in
this instance?

Below is a script to reproduce some of my dataset:

structure(list(study = c(1, 2, 3, 4, 5, 5, 6, 7, 7, 8, 8, 9, 9, 10), effect
= c(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13,14),
               eprem = c(1.05, 1.46, 4.16, 1.68, 2.8, 2.32, 0.71,
0.699157641395908, 0.522983521248916, 0.18, 0.65, 0.78, 0.65, 69.52),
               epresd = c(0.91, 0.99, 2.5, 1.05, 0.25, 0.3, 0.2, 0.33,
0.11, 0.16, 0.16, 0.18, 0.29, 15.14),
               epostm = c(1.63, 2.37, 3.43, 2.52, 2.68, 2.43, 0.85,
0.838264299802761, 0.667349027635619, 0.53, 0.73, 0.67, 0.63, 83.38),
               epostsd = c(1.23, 1.81, 2.01, 1.72, 0.58, 0.38, 0.16, 0.35,
0.09, 0.34, 0.15, 0.22, 0.38, 19),
               en = c(35, 49, 14, 36, 17, 17, 41, 8, 8, 18, 25, 22, 22, 16),
               er = c(0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5,
0.5, 0.5, 0.5, 0.5),
               cprem = c(0.93, 1.47, 4.7, 1.42, 2.85, 2.22, 0.69,
0.579690652320108, 0.508771929824561, 0.57, 0.57, 0.68, 0.68, 72.14),
               cpresd = c(0.99, 1.23, 4.08, 1.18, 0.13, 0.19, 0.22, 0.24,
0.16, 0.29, 0.29,  0.3, 0.3, 13.41),
               cpostm = c(1.06, 1.75, 2.32, 1.8, 2.99, 2.23, 0.73,
0.711538461538462, 0.467311715481172, 0.59, 0.59, 0.74, 0.59, 74.55),
               cpostsd = c(0.74, 1.17, 2.07, 1.09, 0.03, 0.45, 0.1, 0.52,
0.13, 0.31, 0.31, 0.29, 0.34, 17),
               cn = c(37, 52, 21, 39, 18, 18, 21, 8, 8, 42, 42, 24, 24, 16),
               cr = c(0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5,
0.5, 0.5, 0.5, 0.5),
               measure = c("odds", "odds", "odds", "odds", "categ",
"categ", "perc", "perc", "perc", "perc", "perc", "perc", "perc", "perc")),
          row.names = c(NA, -14L), class = c("tbl_df", "tbl", "data.frame"))



Thank you very much in advance,
Mika

	[[alternative HTML version deleted]]