Table 3. Summary of transfer function models explaining the monthly %MRSA by use of each antimicrobial drug class^a.

Antimicrobial classb	Average delay (months)	Direction of effectc	p value	R2 d
Combinations of penicillins with β-lactamase inhibitors	2 4	Positive Positive	0.04 0.01	0.92
β-lactamase–resistant penicillins	0 6	Negative Positive	0.02 0.002	0.90
Macrolides	1	Positive	0.0001	0.93
Penicillins with extended spectrum	1	Positive	0.03	0.91
Third-generation cephalosporins	1	Positive	0.04	0.90
β-lactamase–sensitive penicillins	6	Positive	0.04	0.89
Combinations of sulfonamides and trimethoprim, including derivatives	4	Positive	0.02	0.90
Fluoroquinolones	4	Positive	0.0004	0.92
Second-generation cephalosporins		No relationship
Other antibacterialse	0	Positive	0.002	0.91
Tetracyclines	4 7	Positive Negative	0.03 0.0007	0.91
Aminoglycosides		No relationship
Lincosamides	7	Positive	0.02	0.89
First-generation cephalosporins		No relationship
Carbapenems	3	Positive	0.03	0.90

^aMRSA, methicillin-resistant Staphylococcus aureus.
^bGlycopeptide use is not presented in this table because it showed an inverse relationship with %MRSA. In other words, %MRSA explained the monthly variations of glycopeptide use and not the reverse (Discussion).
^cPositive direction of effect: increase in antimicrobial use results in increase in %MRSA and inversely. Negative direction of effect: increase in antimicrobial use results in decrease in %MRSA and inversely.
^dAll models include the variable %MRSA with a 1-month delay and a p value < 0.0001.
^eAmphenicols, monobactams, other quinolones, imidazoles, fusidic acid, and nitrofurantoin derivatives.