On 2016 Apr 20, Jennifer S Walsh commented:

It wasn't possible to get perfect matching for all the factors, so it was done in order of: gender, age, height, postcode and smoking. Because smoking was the lowest priority, there were a few pairs who were matched on the other factors but not on smoking. The number of smokers was low overall, and we don't think the imperfect matches for smoking had any influence on the study results.

On 2016 Apr 15, Jose M. Moran commented:

There is an issue that needs to be addressed. Controls were recruited to be individually matched to an obese participant by sex, age (±3 years), height (±5 cm), postcode and smoking status (current smoker or nonsmoker). If controls were matched 1:1 by the smoking status, how it is possible that the percentage of current smokers differs between normal and obese subjects across all the age groups (except in men aged 25-45 years n=6)?

On 2016 Apr 15, Jose M. Moran commented:

There is an issue that needs to be addressed. Controls were recruited to be individually matched to an obese participant by sex, age (±3 years), height (±5 cm), postcode and smoking status (current smoker or nonsmoker). If controls were matched 1:1 by the smoking status, how it is possible that the percentage of current smokers differs between normal and obese subjects across all the age groups (except in men aged 25-45 years n=6)?

On 2016 Apr 20, Jennifer S Walsh commented:

It wasn't possible to get perfect matching for all the factors, so it was done in order of: gender, age, height, postcode and smoking. Because smoking was the lowest priority, there were a few pairs who were matched on the other factors but not on smoking. The number of smokers was low overall, and we don't think the imperfect matches for smoking had any influence on the study results.