How can I design a cascaded IIR bandpass butterworth filter for my data x?

I have data with the intention to filter them offline. I would like to apply a cascaded IIR bandpass butterworth filter. I have a Fs of 5000 and would like to filter low pass (12 Hz) and high pass (10 Hz), order is 3. If I am correct, I have to use the sosfilt function? My current code is:

[c,b,a] = butter(3,[0.004 0.0048]/(5000/2), 'bandpass');

sosbp = zp2sos(c,b,a)

x_filter = sosfilt(sosbp, x);

However, something is wrong since data are not filtered to the desired cut-off frequencies? Can anyone provide help?