No it doesn't. First test showed NPV of 97% with lower cut-off. Data variability can be caused by many factors. If ALL hospitals showed crappy data, there wouldn't be variability. Depends on hospitals and protocols. There were only 8 hospitals used so even if only 2 messed it up, throws the whole data off.
If I had to guess, I would imagine that the variability is due to the fact that MRP 8 and MPR 14 are non-specific inflammatory markers associated with macrophages. MRP 8/14 can be elevated in several different inflammatory conditions, such as rheumatoid arthritis, lupus, type 1 diabetes, cardiovascular diseases, and gastrointestinal diseases.
Given these multiple sources of false elevation, the "variability" in the data might be explained by too many people at one or more sites coming to the ER with belly pain caused by these other inflammatory conditions.