|
Abstract:
|
The Diabetes Prevention Program is a multi-center, completely randomized clinical trial targeting at lowering diabetes occurrences. 3000 people with high risk were recruited and randomized to placebo, life-style training or Metformin group. They were followed quarterly with lab measures and quality of life surveys. One of the primary outcome of interest is microangiopathy, which is defined as developing one or more of the three binary symptoms: retinopathy, nephropathy and neuropathy. The three outcomes are regulated by the same mechanism of diabetes development, thus correlated. They also have its unique characteristics. Four different statistical models are explained and applied to simulated data, corresponding powers calculated. The differences are illustrated through a compliance study with data also from DPP. Finally, take-home message is given for studying correlated binary outcomes.
|
