What is the inverse relationship between smoking status and GLP-1 RA prescription rates, despite similar or higher indication rates for prescribing?