Hi, I'm reading a paper which looks at the effects of empagliflozin on long-term kidney disease in type 2 diabetes. The data in this paper is extracted from the EMPA-REG outcome trial. But the trial uses half the number of patients in the placebo group as they do in the test empagliflozin group. Now can statistical analysis correct for this or would this have a major impact on the credibility of the data?
The paper is "Empagliflozin and Progression of Kidney Disease in Type 2 Diabetes" in the New England Journal of Medicine, 2016. If anyone has read this paper and has further opinions on the trial then I'd be very pleased to hear them as I intend to use some of this data for a university project.
Thank you.
To answer my own question, I've just found out that the paper mentions 2 doses in the empagliflozin group so bearing this in mind, the patient numbers do balance out.
But this paper does still present a few questions such as;
1. How empagliflozin exerts its reno-protective effects? There are some papers leaning towards its cellular action on mesangial cells but further opinions would be welcome please?
2. Also the FDA have issued warnings about the use of SGLT2 inhibitors such as empagliflozin due to their link with diabetic ketoacidosis, but in this trial empagliflozin doesn't seem to have increased this risk. Now the stats of diabetic ketoacidosis events was established in this trial by inputting relevant endpoint search-terms into MedDRA. What are your opinions of this technique of capturing end-points in clinical trials? Also if anyone has any opinions on whether attitudes towards SGLT2 inhibitors are changing then those responses would be very helpful, thank you.
It is not necessarily a problem, because statistical significance does not only depend on this.
As long as the 'power' for the outcome is preserved, then it is okay.
You can double check the quality of the study by looking at the table of baseline characteristics. If there are no statistical differences in baseline characteristics between the 2 groups, then the process of randomization is most likely valid.
And there are stat tactics like 2:1 randomisation, which one uses when there is difficulty getting the numbers in a particular group.
Best is always to have a great statistician on the team.....
the recent DECLARE study answers your DKA question.
There is a statistical increase, but the absolute increase in DKA is probably not clinically significant- if you increase your surveillance in the clinical domain.
The question of reno-protection: the nephrologists can come in here.
My 2c worth:
Reno protection is probably multi-factorial: better diabetes and BP control, esp due to weight loss. The drug needs to reduce intra-glomerular pressure to counteract some of the early Diabetic renal damage.
Single nephron studies may help answer this.
I will follow further discussion, with interest
- Badges
- Science topic
- Similar topics
- Social Psychology
- Attitude