I am analyzing a group of patients (n=1500) with arterial line insertions (n=2000), infection rate (n=50) and mortality (n=20) in a specific patient population.
Simple Chi-Square analysis and t-test (Fit y by x) reveals a significant association of line infections to increased mortality. - However a patient may have counted more than once, as I primarily look at relation of lines (n=2000) to mortality, but one patient (n=1500) can have multiple lines, or multiple admissions with lines, so this may bias the results.
How can I adjust the analysis for this bias to get the true effect of line related mortality per patient ?
I have a specific identifier (numeric) for every patient for every admission (n=1500). Do I use block, weight, by or frequency ?
Many thanks, Marc
