Introduction to Statistical Modelling
Training session with Dr Helen Brown, Senior Statistician, at The Roslin Institute, December 2015.
************************************************
These training sessions were given to staff and research students at the Roslin Institute. The material is also used for the Animal Biosciences MSc course taught at the Institute.
************************************************
*Recommended Youtube playback settings for the best viewing experience: 1080p HD
************************************************
Content:
Logistic Regression: Example
-Data recorded on 549 patients undergoing renal transplant
-Outcome: Death within 5 years of transplant
-Independent variables: Many to consider :
--- age, gender, smoking, illnesses, etc
Model to assess effect of smoking
-First assess effect of smoking only
-Logistic regression fitting: smoker (1=yes, 0=no)
-Smoking has significant effect on survival at 5 years post transplant (p=0.005)
Expressing results from a logistic regression
-Exponential of estimate gives odds ratios
-Odds ratio comparing smokers and non-smokers:
Now adjust model for age
-Check that smoking effect is independent of fact that smokers could be older
-Logistic regression fitting: age and smoker
-Smoking still has a significant effect on survival at 5 years post transplant (p=0.01) after adjusting for age
-Odds of death is 2.22 times higher in smokers (compare to 2.46 in last model before adjustment)
-Odds of death is 1.06 times higher with each additional year of life
Does the effect of smoking vary with age?
-Fit interaction between age and smoking
-No significant interaction between age and smoking
--- Risk caused by smoking does not vary with age
Note: ‘Main effects’ for age and smoking cannot be interpreted directly :
--- ‘Age’ effect = effect of age in non-smokers
--- ‘Smoke’ effect = effect of smoking at age=0 (ignore - not in range of data analysed!)
- Tags
-