What scenario would cause an NA (missing data) entry rather than a zero?

NA specifically means that the endpoint cannot apply to a scenario and leaving a 0 would skew association results - for example, for prostate cancer females are NA, or for open-angle glaucoma cases the less-specific "other glaucomas" entries are NA.

In some endpoints we have done stricter delimitation of controls, meaning that we remove controls based on certain criteria. For example, in _EXALLC endpoints, all cancers have been removed from the controls, and in _EXMORE endpoints a stricter control cut off has been used. These individuals will appear NA in the data.

NA entries are perfectly acceptable and will be skipped in analyses without adding any spurious controls, especially if you're working with more strictly delimited data. However, if you're confident that your data should not have any NA entries and a substantial number are present, we suggest contacting [e-mail].