A CASE STUDY ON �TYPE-II DIABETES MELLITUS

DONE BY :-

JEROME INFANT SALETH N A

​

STUDY OBJECTIVE

• To confirm effect of Drug T, once daily versus Drug C, once daily in con-trolling glycemia after 26 weeks in subjects with type II diabetes mellitus with or without oral anti-diabetic treatment.
• This is done by comparing the difference in change from baseline in HbA1c after 26 weeks to a non-inferiority margin of 0.4% 
• Demonstrating non inferiority shows that the new treatment has an acceptable level of efficacy to be adopted into clinical practice as an alternative to the current standard therapy

​

​

​

• Primary Endpoint :-
• Change from baseline in glycosylated hemoglobin (HbA1c) (%) at 26weeks.
• Secondary Endpoint:-
• Proportion of subjects with HbA1c < 7% at 26 weeks.

ENDPOINTS

SAMPLE SIZE DETERMINATION

• Estimate sample size is based on confirming non-inferiority for Drug T by margin of 0.4% under the assumptions of a one-sided test of size 2.5%, power of 90% , a mean treatment difference of 0.0%, standard deviation of 1.0%.

​

Sample Size determination formula for a non inferiority parallel group trial

This is also verified in R

Using the package

“SampleSize4ClinicalTrials”

function- ssc_meancomp

• Subject
• Treatment
• Age (18-65)
• Region (Africa, Asia, Europe, North America)
• Sex (Male, Female)
• Previous oral-antidiabetic treatment (>=2 OAD, <2 OAD)
• HbA1c

​

​

DATA SIMULATION

Simulated Data

10% Missing Data

Descriptive Statisitcs

Required Source Variables

• In the first stage created 5 copies of the dataset, with the missing values replaced by imputed values (used predictive mean matching method). 
• In the second stage fitted the ANCOVA model to each of the imputed datasets. Analysis of covariance is used to test the main and interaction effects of categorical variables on a continuous dependent variable, controlling for the effects of selected other variables
• Each of the model were then averaged together to get the overall estimate (Using Least Square Means).
• Estimated associations in each of the imputed datasets will differ because of the variation introduced in the imputation of the missing values, and they are only useful when averaged together to give overall estimated associations.

MULTIPLE IMPUATION FOR HANDLING MISSING VALUES

PRIMARY END POINT

Observed Mean and Standard deviation Of HbA1c for baseline and week 26 (simulated data)

Results of ANOCOVA

Independent Samples Confidence Interval

μ₁ - μ₂ = (M₁ - M₂) ± ts_(M1 - M2)

where:

M₁ & M₂ = sample means�t = t statistic determined by confidence level�s_{(M1 - M2)} = standard error = √((s²_p/n₁) + (s²_p/n₂))

​

HbA1c(%) Change from baseline

0.002 (-0.172, 0.176)

Absolute difference in reduction of HbA1c(%) by Drug T and Drug C

• Forest Plot
• To Check for if the drug is Non inferior or not

Since the upper confidence interval is less than 0.4%,

we conclude that Drug T is non-inferior to Drug C

Observed Mean Of HbA1c at different time points

Change from baseline in HbA1c at different time points

Proportion of subjects with HbA1c<7% at �26 week

Drug T

Drug C

SECONDARY END POINT

The odds ratio for Drug T taking Drug C as the reference category is 1.034 which implies that there is only 3.4% increase in the proportion of subjects with HbA1c less than 7% at week 26 for Drug T compared to drug C.

Which again indicates that both the drug are acting almost similarly in reducing HbA1c.

Logistic Regression Results

• The Drug T and Drug C contribute almost equally (Absolute difference: 0.002%) in reducing HbA1c from baseline at week 26 considering a non-inferiority margin of 0.4%.
• The Drug T and Drug C also contribute almost equally (OR: 1.034) in increasing the proportion of subjects having HbA1c <7% at week 26 considering a non-inferiority margin of 0.4%.
• So, the Drug T is non inferior to Drug C in controlling glycaemia after 26 weeks in subjects with type 2 diabetes mellitus.

​

CONCLUSION