Tests for secular trend...am I on the right track?

#1
Hi-

This is not homework, but a project rather for both my work and graduate program. I collected data from a large outpatient clinic in my area on positive cultures for Staphylococcus aureus to evaluate the changes in proportion of antibiotic resistant cultures from the year 2000 through 2005 (6 years of data).

Basically, I have a datafile with each row representing a "case" (i.e. a positive S. aureus culture) and the following columns: patient age, patient sex, home zipcode, year cultured, and the culture's susceptibility to antibiotics (coded as 1=resistant, 0=susceptible).

Based on research I have done thus far, it appears that chi square can be used to assess whether there is a statistically significant association between time (year cultured) and the proportion of resistant cultures. Is this appropriate for looking at this type of data? I have performed the chi-square test already and the value was 140 (df=5; p<0.05). For the rows, the "year" values are 2000, 2001, 2002, 2003, 2004, 2005 and for the columns "0" (susceptible cultures) and "1" (antibiotic resistant cultures). Can the age groups be analyzed in a similar manner, in terms of looking for an association between age group and proportion of antibiotic resistant cultures within each age group)?

I have also heard that you can do logistic regression for something like this to evaluate trends...would there be an advantage to doing this? I am trying to keep analysis fairly simple because I am hoping to use the results as an education tool targeting local physicians who may not have much time to delve into complex results.

I hope my questions are not too confusing...I do have some background in biostatistics, but I am a little unsure of myself with this and I want to do it correctly. Regards,

Jonah
 
#2
What you have written about the chi-square test you are using sounds right, and yes it could be applied to age groups. Alternatively age and year could both be variables used simultaneously in a logistic regression.