Biostatistics for pain study, NP t-test or GLM ?

#1
This might be a very basic question, but I keep getting different answers, so I'll ask it in this forum of experts :)
Material: Back pain prospective cohort study of n=200 patients all suffering from LBP at the initial consultation. A questionnaire is filled at time=0 years and time=10 years. The outcome is back pain (VAS score 0-10), disability (RMDQ 0-23). Patients are divided into two groups based on MRI-finding with +/- disc protrusion at time=0 years, group sizes: +disc protrusion n=80, -disc protrusion n=120. At 10-year follow-up groups, based on original MRI findings, are: +disc protrusion n=70, -disc protrusion n=100.
QUESTION: Is disc protrusion associated with outcome (back pain score and disability)?
METHOD: 1) Non-paired t-test at time= 0 and 10 years to find a difference in outcome between the groups?
OR
2) General linear model with repeated measures
OR
3) Mixed Anova
Thank you for answering! Peter
 
#3
Yes only those two measurements, it's a long term follow up study. Within the questionnaire there are questions regarding: spine surgery, other treatments etc over the 10 year period. No difference in number of spine surgeries, other treatments, BMI, alcohol consumption, smoking etc exist between the groups when analysed by NP t-test at 10y follow up