I am training to learn how to perform meta-analysis using R. I conducted a meta-analysis using the metaprop function in R with the provided dataset. The goal is to compute the pooled specificity using the code snippet below:
Data <- structure(
list(
detection = c(
"Fluorescence",
"Fluorescence",
"Fluorescence",
"Fluorescence",
"Fluorescence",
"Fluorescence",
"Fluorescence",
"Lateral flow",
"Lateral flow",
"Fluorescence",
"Fluorescence",
"Fluorescence"
),
TP = c(91, 139,
29, 59, 32, 32, 24, 18, 80, 77, 261, 51),
TN = c(62, 53, 20,
20, 40, 37, 22, 26, 40, 19, 127, 27),
FP = c(1, 0, 0, 1, 0, 2,
3, 0, 0, 0, 6, 0),
FN = c(25, 1, 0, 9, 3, 2, 1, 2, 27, 0, 7,
12)
),
row.names = c(NA, -12L),
class = c("tbl_df", "tbl", "data.frame")
)
I encounter multiple problems when I wanted to use metaprop to compute pooled specificity using this code:
library(meta)
metaprop(event = TN,
n = TN + FP,
studlab = rownames(Data2),
data = Data,
sm = "PLOGIT",
method.ci = "CP",
subgroup = Data$detection,
comb.fixed=FALSE,
comb.random=TRUE,
title = "Specificity")
This is the result:
Review: Specificity
Number of studies: k = 12
Number of observations: o = 506
Number of events: e = 493
proportion 95%-CI
Random effects model 0.9829 [0.9462; 0.9947]
Quantifying heterogeneity:
tau^2 = 0.7642; tau = 0.8742; I^2 = 0.0% [0.0%; 58.3%]; H = 1.00 [1.00; 1.55]
Test of heterogeneity:
Q d.f. p-value
Wald 3.80 11 0.9755
LRT 19.52 11 0.0523
Results for subgroups (random effects model):
k proportion 95%-CI tau^2 tau Q I^2
detection = Fluorescence 10 0.9761 [0.9374; 0.9911] 0.3949 0.6284 3.80 0.0%
detection = Lateral flow 2 1.0000 [0.0000; 1.0000] 0 0 0.00 0.0%
Test for subgroup differences (random effects model):
Q d.f. p-value
Between groups 0.00 1 0.9995
Details on meta-analytical method:
- Random intercept logistic regression model
- Maximum-likelihood estimator for tau^2
- Logit transformation
- Continuity correction of 0.5 in studies with zero cell frequencies
(only used to calculate individual study results)
Warning messages:
1: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
2: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Hessian is numerically singular: parameters are not uniquely determined
3: In vcov.merMod(res.ML) :
variance-covariance matrix computed from finite-difference Hessian is
not positive definite or contains NA values: falling back to var-cov estimated from RX
4: Ratio of largest to smallest sampling variance extremely large. May not be able to obtain stable results.
5: Ratio of largest to smallest sampling variance extremely large. May not be able to obtain stable results.
I have some questions:
Confidence Interval for Lateral Flow: The confidence interval for specificity in the "Lateral flow" subgroup is reported as [0.0000; 1.0000]. Given the absence of false positives in both studies using the "lateral flow" method, I expected a more precise interval. Why does it return (0-1)?
Interpretation of Warnings: The function generates several warning messages, and I'm unclear about their implications.
Any insights or suggestions on resolving these issues would be greatly appreciated. Thank you.
