Statistical Analysis Of Hypoglycemia Events In A Diabetes Study

Diabetes is a group of metabolic diseases in which there are high blood glucose levels over a prolonged period.The key consideration of treatment is reducing the blood glucose level,but too low blood glucose level also causes problems.Thus,monitoring blood glucose level after taking study drug is one important safety endpoint.Especially when some subjects overtake study drugs,will it lead to more hypoglycemia events,or even severe hypoglycemia events?To answer this question,one global pharmaceutical factory conducted a crossover study with 2 periods.About 64 subjects were randomized to 2 treatment groups,and investigators monitored hypoglycemia events within 84 hours after subject took the double dose drug in each period.In this study,the analysis endpoints include incidence,rate,time to event and time to event of recurrent events.By evaluating these endpoints and performing statistical analysis,we want to get enough evidence that new drug benefits more than control drug for safety consideration.We started the analysis from the test of independence,and chi-squire test and odds ratio methods were used.Then we introduced logistics regression,and we could also get to know the effect of other covariates.To analyze event rate,we used negative binomial regression.Moreover,we also used survival analysis methods and Anderson-Gill model to analyze the survival related endpoints,such as time to event and time to event of recurrent events.In general,we got the results as we expected.That is,new drug has better safety results than control drug based on our statistical analysis.Considering we used differ-ent analysis endpoints and implemented different statistical methods,and we got the consistent results,we have high level confidence that our results are reliable and robust.