dat.pignon2000.RdResults from studies examining mortality risk in patients with nonmetastatic head and neck squamous-cell carcinoma receiving either locoregional treatment plus chemotherapy versus locoregional treatment alone.
dat.pignon2000The data frame contains the following columns:
| id | numeric | study id number |
| trial | character | trial abbreviation |
| OmE | numeric | observed minus expected number of deaths in the locoregional treatment plus chemotherapy group |
| V | numeric | corresponding variance |
| grp | numeric | timing of chemotherapy: 1 = adjuvant, 2 = neoadjuvant, 3 = concomitant |
The purpose of this meta-analysis was to examine the mortality risk in patients with nonmetastatic head and neck squamous-cell carcinoma receiving either locoregional treatment plus chemotherapy versus locoregional treatment alone. For 65 trials, the dataset provides the observed minus expected number of deaths and corresponding variances in the locoregional treatment plus chemotherapy group. Based on these values, we can estimate the log hazard ratios with OmE/V and the corresponding sampling variance with 1/V.
The trials were also divided according to the timing of the chomotherapy: (1) adjuvant, after the locoregional treatment, (2) neoadjuvant, before the locoregional treatment, and (3) concomitant, chemotherapy given concomitantly or alternating with radiotherapy.
Pignon, J. P., Bourhis, J., Domenge, C., & Designe, L. (2000). Chemotherapy added to locoregional treatment for head and neck squamous-cell carcinoma: Three meta-analyses of updated individual data. Lancet, 355(9208), 949–955. https://doi.org/10.1016/S0140-6736(00)90011-4
medicine, oncology, hazard ratios
### copy data into 'dat' and examine data
dat <- dat.pignon2000
head(dat)
#> id trial OmE V grp
#> 1 44 Pitie-74 -1.8 14.5 1
#> 2 45 GETTECadj 11.8 57.1 1
#> 3 46 Int 0034 -7.4 80.6 1
#> 4 102 JHCFUS -5.6 8.5 1
#> 5 97 TMH R-4 0.4 6.4 1
#> 6 103 KKD-86 1.8 8.7 1
### load metafor package
library(metafor)
### calculate log hazard ratios and sampling variances
dat$yi <- with(dat, OmE/V)
dat$vi <- with(dat, 1/V)
head(dat)
#> id trial OmE V grp yi vi
#> 1 44 Pitie-74 -1.8 14.5 1 -0.12413793 0.06896552
#> 2 45 GETTECadj 11.8 57.1 1 0.20665499 0.01751313
#> 3 46 Int 0034 -7.4 80.6 1 -0.09181141 0.01240695
#> 4 102 JHCFUS -5.6 8.5 1 -0.65882353 0.11764706
#> 5 97 TMH R-4 0.4 6.4 1 0.06250000 0.15625000
#> 6 103 KKD-86 1.8 8.7 1 0.20689655 0.11494253
### meta-analysis based on all 65 trials
res <- rma(yi, vi, data=dat, method="EE", digits=2)
res
#>
#> Equal-Effects Model (k = 65)
#>
#> I^2 (total heterogeneity / total variability): 48.25%
#> H^2 (total variability / sampling variability): 1.93
#>
#> Test for Heterogeneity:
#> Q(df = 64) = 123.67, p-val < .01
#>
#> Model Results:
#>
#> estimate se zval pval ci.lb ci.ub
#> -0.11 0.02 -4.67 <.01 -0.16 -0.06 ***
#>
#> ---
#> Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
#>
predict(res, transf=exp)
#>
#> pred ci.lb ci.ub
#> 0.90 0.85 0.94
#>
### only adjuvant trials
res <- rma(yi, vi, data=dat, method="EE", subset=grp==1, digits=2)
res
#>
#> Equal-Effects Model (k = 8)
#>
#> I^2 (total heterogeneity / total variability): 10.56%
#> H^2 (total variability / sampling variability): 1.12
#>
#> Test for Heterogeneity:
#> Q(df = 7) = 7.83, p-val = 0.35
#>
#> Model Results:
#>
#> estimate se zval pval ci.lb ci.ub
#> -0.02 0.07 -0.34 0.74 -0.16 0.11
#>
#> ---
#> Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
#>
predict(res, transf=exp)
#>
#> pred ci.lb ci.ub
#> 0.98 0.85 1.12
#>
### only neoadjuvant trials
res <- rma(yi, vi, data=dat, method="EE", subset=grp==2, digits=2)
res
#>
#> Equal-Effects Model (k = 31)
#>
#> I^2 (total heterogeneity / total variability): 4.96%
#> H^2 (total variability / sampling variability): 1.05
#>
#> Test for Heterogeneity:
#> Q(df = 30) = 31.57, p-val = 0.39
#>
#> Model Results:
#>
#> estimate se zval pval ci.lb ci.ub
#> -0.06 0.03 -1.64 0.10 -0.12 0.01
#>
#> ---
#> Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
#>
predict(res, transf=exp)
#>
#> pred ci.lb ci.ub
#> 0.95 0.88 1.01
#>
### only concomitant trials
res <- rma(yi, vi, data=dat, method="EE", subset=grp==3, digits=2)
res
#>
#> Equal-Effects Model (k = 26)
#>
#> I^2 (total heterogeneity / total variability): 66.13%
#> H^2 (total variability / sampling variability): 2.95
#>
#> Test for Heterogeneity:
#> Q(df = 25) = 73.81, p-val < .01
#>
#> Model Results:
#>
#> estimate se zval pval ci.lb ci.ub
#> -0.21 0.04 -5.43 <.01 -0.28 -0.13 ***
#>
#> ---
#> Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
#>
predict(res, transf=exp)
#>
#> pred ci.lb ci.ub
#> 0.81 0.76 0.88
#>