Graphics
Forest Plot for Odds Ratio
(FSTG01)
1. Forest Plot for Odds Ratio (with subgroup analysis)
- The
fstg01template produces the standard forest plot for odds ratio. - Users are expected to subset the parameter of interest
(e.g.
PARAMCD == "BESRSPI") in pre-processing. - Users are expected to subset the arm variable to keep only the two
arms to compare
(e.g.
ARM %in% c("A: Drug X", "B: Placebo")). - By default, the plots displays a subgroup analysis for
"SEX","AGEGR1"and"RACE". - Unstratified analysis is provided by default.
- The plots displays by default the Total number of subjects, the odd ratio and the 95% confidence interval, and, for each arm, the number of subject, the number of responders and the proportion of responders.
proc_data <- log_filter(
syn_data,
PARAMCD == "BESRSPI" & ARM %in% c("A: Drug X", "B: Placebo"), "adrs"
)
run(fstg01, proc_data)
2. Forest Plot for Odds Ratio (with a user-defined confidence level)
The confidence level of the confidence interval can be adjusted by
the conf_level argument.
3. Forest Plot for Odds Ratio (with p-values and/or different statistics)
The interaction p-values and a different set of statistics can be
displayed using the stat_var argument. Note that the users
are expected to select a method for p-value computation. see
[tern::prop_diff_test].
4. Forest Plot for Odds Ratio (with user-defined subgroup analysis)
The subgroups arguments controls which variables are
used for subgroup analysis. If NULLthe subgroup analysis is
removed.
5. Forest Plot for Odds Ratio (with stratified analysis)
The strata_var argument is used to pass the columns used
for stratified analysis.
run(fstg01, proc_data, strata_var = "STRATA1")
#> Warning in coxexact.fit(X, Y, istrat, offset, init, control, weights = weights,
#> : Ran out of iterations and did not converge
#> Warning in s_odds_ratio(df = l_df[[2]], .var = "rsp", .ref_group = l_df[[1]], :
#> Unable to compute the odds ratio estimate. Please try re-running the function
#> with parameter `method` set to "approximate".
#> Warning in coxexact.fit(X, Y, istrat, offset, init, control, weights = weights,
#> : Ran out of iterations and did not converge
6. Forest Plot for Odds Ratio (without proportional sizing of the odds ratio symbol)
The col_symbol_size argument controls the size of the
odds ratio symbols which are by default proportional in size to the
sample size of the subgroup. If NULL the same symbol size
is used for all subgroups.
Forest Plot for Hazard Ratio
(FSTG02)
1. Forest Plot for Hazard Ratio (with subgroup analysis)
- The
fstg02template produces the standard forest plot for hazard ratio. - Users are expected to subset the parameter of interest
(e.g.
PARAMCD == "OS") in pre-processing. - Users are expected to subset the arm variable to keep only the two
arms to compare
(e.g.
ARM %in% c("A: Drug X", "B: Placebo")). - By default, the plots displays a subgroup analysis for
"SEX","AGEGR1"and"RACE". - Unstratified analysis is provided by default.
- The plots displays by default the Total number of events, the hazard ratio and the 95% confidence interval, and, for each arm, the number of events and the median time to event in month.
proc_data <- log_filter(
syn_data,
PARAMCD == "OS" & ARM %in% c("A: Drug X", "B: Placebo"), "adtte"
)
run(fstg02, proc_data)
2. Forest Plot for Hazard Ratio (with p-values and/or different statistics)
The interaction p-values and a different set of statistics can be
displayed using the control argument. More details about
the control options are available in
[tern::extract_survival_subgroups]
run(
fstg02,
proc_data,
stat_var = c("n_tot", "n", "ci", "hr", "pval"),
control = list(conf_level = 0.9, pval_method = "likelihood")
)
3. Forest Plot for Hazard Ratio (with user-defined subgroup analysis)
The subgroups arguments controls which variables are
used for subgroup analysis. If NULLthe subgroup analysis is
removed.
4. Forest Plot for Hazard Ratio (with stratified analysis)
The strata_var argument is used to pass the columns used
for stratified analysis.
run(fstg02, proc_data, strata_var = "STRATA1")
#> Warning in coxph.fit(X, Y, istrat, offset, init, control, weights = weights, :
#> Loglik converged before variable 1 ; coefficient may be infinite.
5. Forest Plot for Hazard Ratio (without proportional sizing of the hazard ratio symbol)
The col_symbol_size argument controls the size of the
hazard ratio symbols which are by default proportional in size to the
number of events in the subgroup. If NULL the same symbol
size is used for all subgroups.
Kaplan-Meier Plot (KMG01)
1. Kaplan-Meier Plot (without comparative statistics)
- The
kmg01template produces the standard Kaplan-Meier Plot. - Users are expected to select a particular parameter for analysis.
- Users are expected to select the treatment groups to compare, otherwise, all treatment groups available in the input datasets will be plotted.
- The comparative statistics are not included by default.
- The estimation of median survival time per treatment group by default.
- More arguments in the
g_kmandcontrol_coxphfunctions can be passed through, please use the Help to find out more information.
proc_data <- log_filter(syn_data, PARAMCD == "OS", "adtte")
run(kmg01, proc_data, dataset = "adtte")
2. Kaplan-Meier Plot (with comparative statistics)
To enable the comparative statistics (hazard ratio and p-value), the
argument annot_coxph needs to be set to TRUE. The compare
group is determined by the levels in the factorized variable of
treatment group and the first level is used as reference group in the
statistics.
proc_data <- log_filter(syn_data, PARAMCD == "OS", "adtte")
run(
kmg01,
proc_data,
dataset = "adtte",
annot_coxph = TRUE,
control_annot_coxph = tern::control_coxph_annot(x = 0.33, y = 0.42)
)
3. Kaplan-Meier Plot (without censoring marks)
To suppress the censoring marks, set the argument
cencor_show to FALSE.
proc_data <- log_filter(syn_data, PARAMCD == "OS", "adtte")
run(kmg01, proc_data, dataset = "adtte", censor_show = FALSE)
4. Kaplan-Meier Plot (without estimation of median survival time)
proc_data <- log_filter(syn_data, PARAMCD == "OS", "adtte")
run(kmg01, proc_data, dataset = "adtte", annot_surv_med = FALSE)
5. Kaplan-Meier Plot (with statistical annotation of either median or min of survival time)
To add the statistics annotation, use the function
annot_stats. Options are min or
median.
proc_data <- log_filter(syn_data, PARAMCD == "OS", "adtte")
run(kmg01, proc_data, dataset = "adtte", annot_stats = "median")
Mean Plot (MNG01)
1. Plot of Mean and Confidence Interval (with Table Section)
- The
mng01template produces the standard mean plot. - Note that the template
mng01is quite general. The users are expected to specify the analysis dataset and the visit variable in therunfunction, and select the parameters prior to therunfunction. - The table of summary statistics is included by default.
- The variable Analysis Value
AVALis used for plotting by default. - If the input dataset contains results of the same analyses in
multiple units,(e.g. SI/CV units in
ADLB), please make sure that the parameters in appropriate units are selected in advance.
proc_data <- log_filter(syn_data, PARAMCD == "DIABP", "advs")
run(mng01, proc_data, dataset = "advs", x_var = c("AVISIT", "AVISITN"))
#> $`Diastolic Blood Pressure`
2. Plot of Mean and Confidence Interval of Change from Baseline of Vital Signs
proc_data <- log_filter(syn_data, PARAMCD == "DIABP", "advs")
run(mng01, proc_data, dataset = "advs", x_var = c("AVISIT", "AVISITN"), y_var = "CHG")
#> `geom_line()`: Each group consists of only one observation.
#> ℹ Do you need to adjust the group aesthetic?
#> $`Diastolic Blood Pressure`
3. Plot of Mean (+/-SD) (Changing the Statistics)
To change the statistics, use the argument interval_fun.
Options are mean_ci, mean_sei,
mean_sdi, median_ci,
quantiles,range.
proc_data <- log_filter(syn_data, PARAMCD == "DIABP", "advs")
run(mng01, proc_data, dataset = "advs", x_var = c("AVISIT", "AVISITN"), interval_fun = "mean_sdi")
#> $`Diastolic Blood Pressure`
4. Plot of Mean and Confidence Interval (Modify Alpha Level)
To change the alpha level of the confidence interval, use the
argument
control = control_analyze_vars(conf_level = <0.xx>).
Note that this is only in effect when interval_fun is set
to mean_ci.
proc_data <- log_filter(syn_data, PARAMCD == "DIABP", "advs")
run(
mng01, proc_data,
dataset = "advs", x_var = c("AVISIT", "AVISITN"),
interval_fun = "mean_ci", control = tern::control_analyze_vars(conf_level = 0.80)
)
#> $`Diastolic Blood Pressure`
5. Plot of Mean and Confidence Interval （With Number of Patients Only)
proc_data <- log_filter(syn_data, PARAMCD == "DIABP", "advs")
run(mng01, proc_data, dataset = "advs", x_var = c("AVISIT", "AVISITN"), table = "n")
#> $`Diastolic Blood Pressure`
6. Plot of Mean and Confidence Interval (without Table Section)
proc_data <- log_filter(syn_data, PARAMCD == "DIABP", "advs")
run(mng01, proc_data, dataset = "advs", x_var = c("AVISIT", "AVISITN"), table = NULL)
#> $`Diastolic Blood Pressure`
A new argument has been added to control the theme (e.g. setting the angle of the axis); see an example below:
ggtheme <- ggplot2::theme(
panel.grid = ggplot2::element_line(colour = "black", linetype = 3),
panel.background = ggplot2::element_rect(fill = "white"),
legend.position = "top",
axis.text.x = ggplot2::element_text(angle = 22, hjust = 1, vjust = 1)
)
run(mng01, syn_data, dataset = "adlb", ggtheme = ggtheme)
#> $`Alanine Aminotransferase Measurement`
#>
#> $`C-Reactive Protein Measurement`
#>
#> $`Immunoglobulin A Measurement`