Lifelink (depending upon data source) will tend to exaggerate dynamic RXs (by dynamic I mean those where a change has occurred, new, switch, add etc.). This occurs because to get ideal data you need a 100% pharmacy panel with no gaps or omissions and perfect matching of patients between pharmacies/data sources.
In reality you will have gaps and inconsistencies which means you see individual RXs in locations with no history or following RXs (i.e. a single Rx). The system will label these as 'New' and hence put it in the 'New to Brand' basket. However this could be a patient who has been on the same drug every month, but normally takes there prescriptions to a different (non-panel) pharmacy and on this occasion happened to do something different. Hence what is really a repeat RX appears dynamic.
As a general rule, metrics like NTB are useful to track, but best used as a relative and directional measure (i.e. is it going up or staying the same) rather than as an absolute figure.
Similar caution should be exercised with any compliance/persistence/adherence studies measured from this data source, since it will suffer from the same problem - patients appearing to be missing RXs when they go elsewhere, and lots of short durations of treatment.