Several observations of a distant solar object must be taken before determining its orbit, but in the case of objects discovered beyond Neptune's orbit, just how many are required over what period of time (or what portion of arc) before a reasonably accurate orbit can be determined, sufficient, say, to be added to the Minor Planet Database or a JPL ephemeris?

I'd be interested to know to what degree of certainty orbital elements for objects beyond 30 AU can be determined when we have only observed them for (so far) short portions of their orbits. What about 60 AU, or 90 AU?