How to Determine typical rate of progression with haphazardly collected data?

We are studying visual acuity (VA) in extant patient records. Some patients were tested once. Some patients many times. All at different ages. All have the same disease.

We would like to use these age vs. VA data to determine the typical course of progressive loss of VA in the population. What is the right approach to doing this, mindful that individuals may progress at different rates, but also that there must be some "typical" course.

SPSS preferred (could do MATLAB or something else in a pinch).

Thanks!

Random effects mixed linear modeling seems right, but how to frame it?