Consistent bad MOS scores...repeated 1000 in CDR latency values

I'm doing some reporting on my CDR data and have WAY too many low MOS scores ~2.25.   Looking at the stats on a LOT of calls I see latency of 1000 (both max and avg....CSV fields 37 and 39).  

Is this some max value that we report on...and I'm seeing more than that...but it's rounding off?    

I was on one of the calls that showed this behavior and it wasn't a 1s latency call. 

 

Any ideas?