Despite the strengths of using a large resource such as UK Biobank, this study has some limitations. In particular, there is a trade-off between obtaining the large sample sizes generally needed for genetic studies and the depth of phenotyping that is practicable in such large populations. Our spirometric definition of COPD was not based on bronchodilator reversibility testing, although we have shown previously that by limiting inclusion to individuals with GOLD stage 2+ spirometry, most of these individuals are likely to have COPD according to more rigorous criteria.13 Similarly, our definition of asthma was based on self-reporting of doctor-diagnosed disease. These limitations might have reduced our ability to identify some novel disease associations, although we were able to replicate many known associations using this approach.
