library(pwr)

Background information:

We are going to recruite diabetic patients in a clinical trial study with interventaional educate patients of knowlege with diabetes. Baseline information will be collected. After intervention, we are going to do 3 months and 6 months follow up. The results will be evaluated at the points of baseline, 3 month and 6 months. The major measurment will be focused on HBA1C values. Other measurements will be serume creatinine, weight, blood pressure, and FBG.

Mazzuca et al. reported a similar study before. In his study, 532 patients were recruted. Half of them in control group, the other half in intervational education group. After 11-14 months, experiment group patients experience significantly greater reduction in fasting blood glucose and HBA1C as compared with control subjects.

We only have one group, so around 250 patients to 300 patients will be sufficient according to Mazzuca’s paper. Baseline, 3 months and 6 months results of HBA1C results will be analysised using one-way ANAVO which is the same method used by Mazzuca.

Here is the power anlaysis use Cohen’s pwr power analysis package using R programing language

pwr.anova.test(k=3, f=0.1, sig.level=.05, power=.8)
## 
##      Balanced one-way analysis of variance power calculation 
## 
##               k = 3
##               n = 322.157
##               f = 0.1
##       sig.level = 0.05
##           power = 0.8
## 
## NOTE: n is number in each group

k= number of group, in our study that means “baseline”, " 3 months“,”6 months" three obvservation points. f=0.1 Cohen who designed the package of power analysis in r programming language suggested, 0.1, o.25, and 04 represent small,medium and large effect sizes respectively. After review literature and clinical experience, we choose 0.1 as effective size. sig.level we choose 0.05 which means p<0.05 as statistical significance. power = 0.8 means that a study (when conducted repeatedly over time) is likely to produce a statistically significant result 8 times out of 10. After we put all the parameter in the fomula and run in r enviorment, we have got n=322 as a study population number.

Therefore, I conclude to choose around 250 to 300 patients in this study.

Reference: 1. Mazzuca SA et al. The Diabetes Education Study: A Controlled Trial of the Effects of Diabetes Patient Education. Diabetes Care 9(1):1-10 · January 1986

  1. Quick-R Power Analysis http://statmethods.net/states/power.html