Linear mixed effect models - am I right?

#1
Hey, at the moment I am trying to do an analysis in biostatistics with Panel-Data. There is also the biostatistic forum but as I am interested in regression analyisis, I hope this is the right forum.
It is a retrospectice analysis on patient data from a database, it's more or less routine information: laboratory and medication data over the past years. I would like to know if two different drugs have (=drug_a_vs_b) an impact on renal function. I have four timepoints. Renal function is changing over time, I need to put time as a factor into the analysis (time), Also, renal function is dependant on another drug where the blood concentration is measured regularly. I have this data for the above time-points f as well and need to adjust to that too (=concentration_drug_c). Also, I need to adjust to other factors as well.

My boss told me I should do a multivariable regression analysis with renal function at the last time-point as dependant variable with adjusting to various covariates and factors including drug blood concentrations of drug c at the other time-points as covariables. I am not sure if this is the correct analysis.

I have read that you can analyse this kind of data with a mixed effect model. I am using SPSS and the method is called mixed in the syntax. This is the code I would like to use and I am wondering if it is correct:

MIXED renal function BY sex drug_a_vs_b time WITH blood_concentration_drug_c daily_dose_d age
/CRITERIA=CIN(95) MXITER(100) MXSTEP(10) SCORING(1) SINGULAR(0.000000000001) HCONVERGE(0,ABSOLUTE) LCONVERGE(0, ABSOLUTE) PCONVERGE(0.000001, ABSOLUTE)
/FIXED=sex drug_a_vs_b blood_concentration_drug_c time b daily_dose_d age
/METHOD=REML
/REPEAT=time | SUBJECT(Patient) COVTYPE(AR1)
/COMPARE REFCAT(LAST) ADJ(BONFERRONI)

- The factor drug_a_vs_b (nominal data, two different drugs) is my major interest.
- Time (including 4 different visits; nominal) is different for each patient and changes over time
- Renal function is the dependant variable (interval data; is dependant on the above listed factors, covariable but especially on time)
- sex (male, female)
- age of the patients (interval data)
-concentration of drug c can influence renal function as well and may change over time (interval data)
-daily dose of drug d (also has an impact on renal function, actually it is ordinal data but there are a lot of different dosing variations, I would say it is interval data)

I would like to have the results for each time-point as well so I have included a post-hoc analysis with bonferroni correction

Do you think my repeat function with time (SUBJECT(PATIENT) with AR1 covariance type is correct?

Sorry for all the question...
 
Last edited: