mpcr: Estimation of treatment effects in matched-pair cluster...

In zhenkewu/mpcr: Estimation of treatment effects in matched-pair cluster randomized trials (MPCR)

Description

Estimation of treatment effects in matched-pair cluster randomized trials by calibrating covarite imbalance between clusters.

Usage

mpcr(datinput, arm = "tx", cluster = "team", pair = "sitenew",
  outcome = "sf36pcs32", X_nm_all = c("race", "ageatint", "hcc",
  "livesalone", "education", "s10", "sf36mcs", "sf36pcs", "gender", "h1"),
  X_nm_binary = c("livesalone", "education", "gender"), X_nm_cat = c("race",
  "s10", "h1"), X_nm_cont = c("ageatint", "hcc", "sf36mcs", "sf36pcs"))

Arguments

datinput	The data set should have columns corresponding to the primary outcome, treatment assignment, pair IDs, cluster IDs, covariates used for covariate-calibration;
arm	The name of treatment assignment indicator. For two-arm trials, this variable takes value in 0,1: 0 for control, 1 for treatment;
cluster	The variable name for cluster IDs;
pair	The variable name for pair IDs;
outcome	The variable name for the primary outcome;
X_nm_all	The vector of covariate names that enter the covariate-calibrated analysis
X_nm_binary	The vector of binary covariate names;
X_nm_cat	The vector of categorical (>2 categories) covariate names;
X_nm_cont	The vector of continuous covariate names.

Value

• Tables:

  • Table 1 - cluster sample sizes; calibrated and uncalibrated outcome comparisons;

  • Table 2 - check covariate imbalances within each pair;

  • Table 3 -

    • 1st level analysis: maximum likelihood estimate (MLE); permutation tests;

    • 1st and 2nd level analysis:
      MLE; profile MLE;
      Bayes estimate with uniform shrinkage prior [link to paper];
      permutation tests.

• Figures:
  

  Check second level dependence for crude analysis
  

  √{v^{crude}_p} vs δ^{crude}_p,

  and covariate-calibrated analysis

  √{v^{calibr}_p} vs δ^{calibr}_p.

References

Wu, Z., Frangakis, C. E., Louis, T. A. and Scharfstein, D. O. (2014), Estimation of treatment effects in matched-pair cluster randomized trials by calibrating covariate imbalance between clusters. Biometrics. doi: 10.1111/biom.12214

zhenkewu/mpcr documentation built on May 4, 2019, 10:19 p.m.