• Pneumonia

Dr Obiora et al1 have presented some interesting data. One could quibble about the validity of the nested case-control model, but this issue is important mainly in identifying the fact that the patients were not individually assessed, nor were their particular variables, apart from the ones of interest, benzodiazepines and pneumonia, taken into consideration. Part of the randomness was accounted for by the use of conditional logistic regression but, as …