Can severity of symptoms be used as an outcome measure in trials of non-ulcer dyspepsia and helicobacter pylori associated gastritis?