This has been an unresolved issue ever since the total knee replacement (TKR) was first done in 1968. Today 365,000 TKRs are done each year in the United States. At first the kneecap (patella) was replaced. Early problems with fracture, loosening, and rupture of the tendon around the patella caused surgeons to rethink this decision.
Some studies were done of patients who all had a resurfaced (replaced) patella. Others reported findings on studies with patients keeping their own patellae (nonresurfaced). A single answer to which was better couldn't be reached.
It seems there are too many factors to compare. There are many different kinds of implants to choose from. Different surgeons come to the (operating) table, so-to-speak with a wide range of experience and surgical methods.
Even the patient's diagnosis can make a difference. Some patients have one knee replaced while others have both knees replaced. It's not always possible to tell if the results vary because of one of these factors or because the patella was or wasn't replaced.
There remains a need for high quality studies to be done in this area of orthopedics.