Hi All
I'm trying to run a rather simple fractional polynomial regression model for the first time. We know for sure that there is a non-linear relationship between our outcome of interest (HbA1c or glycated haemoglobin) with the time metric of interest (duration of disease). We are interested in the relationship (and naturally the visualization) of HbA1c since diagnosis in our study cohort, given a set of covariates. The simplest model (without the confounders) is:
where duration is a continuous variable in months ranging from 0 to 48 months. I get the following error message:
''the variable on which fractional polynomial terms are to be calculated contains nonpostive values. you must rescale the variable by specifying option scale, or specify option zero or catzero''
I understand that this might be a problem with the duration variable as it contains values = 0 for some subjects (those with HbA1c data right from diagnosis onward). How does one exactly specify the zero option or rescale the variable in the syntax above? I did go through the help file for 'fp' but wasn't able to solve my problem i.e. I haven't really understood the issue with having some values = 0 and how this effect the regression model.
Thanks!
/Amal
I'm trying to run a rather simple fractional polynomial regression model for the first time. We know for sure that there is a non-linear relationship between our outcome of interest (HbA1c or glycated haemoglobin) with the time metric of interest (duration of disease). We are interested in the relationship (and naturally the visualization) of HbA1c since diagnosis in our study cohort, given a set of covariates. The simplest model (without the confounders) is:
Code:
fp<duration>,dim(3): mixed hba1cmol <duration> ||, mle nostderr, if diagdate>=2011 & duration<43
''the variable on which fractional polynomial terms are to be calculated contains nonpostive values. you must rescale the variable by specifying option scale, or specify option zero or catzero''
I understand that this might be a problem with the duration variable as it contains values = 0 for some subjects (those with HbA1c data right from diagnosis onward). How does one exactly specify the zero option or rescale the variable in the syntax above? I did go through the help file for 'fp' but wasn't able to solve my problem i.e. I haven't really understood the issue with having some values = 0 and how this effect the regression model.
Thanks!
/Amal
