11-11-2016 10:22 AM
I am analyzing data from a two treatment, two period crossover design. I have bioavailability parameters of area under the curve, peak time, peak value etc. The protocol evaluable subset is based on different parameters, therefore it is possible 1 or more subjects may have a missing value on one of the treatments for one or more of the bioavailability parameters. Can they still be used in the proc mixed model for that parameter or must I subset my data for each parameter to only complete cases.
11-29-2016 01:05 PM
Provided the data are "missing at random", then the estimates from PROC MIXED are valid. However, if the missing data are not at random (say due to cumulative toxicity and subject dropout), then you should consider other options.