Confidence Intervals - proportions for stratified designs

Introduction

As documented here, there is requirement to use methods other than the wald method, for CI’s for proportions. In certain cases, for example, when you have a baseline imbalance of stratification factors, the stratified Miettinen-Nurminen 1985 method is reccommended.

Packages

There are two identified packages that can do this in R:

• {ratesci}
• {cicalc}

This page is a placeholder whilst further content is being developed. Running code as shown below was found to give the same results in the examples tested.

{ratesci}

NOTE that: skew = FALSE option, indicates no skewness correction which gives us the miettinen-nurminen method. 

NOTE: The data contain are aggregated with 1 row per strata of results.
Eg.
strata  x1  n1  x2  n2
grp1    5   10  3   12
grp2    1   3   7   15
etc.

ratesci_out <- scoreci(
    x1 = strat_counts$x1,  #n events in trt 1 (Active)
    n1 = strat_counts$n1,  #n subjects in trt 1 (Active)
    x2 = strat_counts$x2,  #n events in trt 2 (comparator/placebo)
    n2 = strat_counts$n2,  #n subjects in trt2 (comparator/placebo)
    distrib = "bin",  #binomial distribution
    contrast = "RD",  #risk difference
    level = 0.95, # 95% ci
    skew = FALSE, # disable skewness to get MN method
    stratified = TRUE, #combine strata
    weighting = "MH", #MH method as reccomended by MN method
  )

{cicalc}

NOTE: set resp as: 1 = responder (event of interest), 0 = non-responders
      set trt as: 0= Numerator Active, 1= Denominator Comparator/Placebo
      set strat as: 
NOTE: The data are not aggregated, there is 1 row per subject identifying their trt, resp and strat categorization
Eg.
subj  strat   resp  trt
1     grp1    1     1
2     grp1    0     0
3     grp1    1     0
etc

NOTE: The following implements stratified mittinen nurminen
cicalc_out <- ci_prop_diff_mn_strata(
  x = resp, 
  by = trt, 
  strata = strat, 
  data = dataqc2
  )
  
  
NOTE: Other options are available - see package for more detail:
ci_prop_diff_mn_strata(
  x,
  by,
  strata,
  method = c("score", "summary score"),
  conf.level = 0.95,
  delta = NULL,
  data = NULL
)