06-04-2018 02:19 AM
I would like to learn how to use PROC LOGISTIC to determine how age relates to survey results.
More specifically, let's use this example:
A group of patients are surveyed every year and they choose Yes or No for some questions and use freeform text to answer other questions.
My SAS program flags patients who have changed their answer from this year compared to last year and which is also in a detrimental direction, meaning a decline in their condition, in which case health_decline_flag = 1.
I would like to ascertain using logistic regression how the patient's age may be affecting the result of the dichotomous variable health_decline_flag (1 = decline, 0 = no decline) and which questions (which have a question code in the table) are being affected most by age.
Additionally, some questions (containing freeform text) which are not comparing this year's result to last year's result would like to be analysed to determine if a patient uses a certain word in the freeform text then how much health_decline_flag is affected. For example, let's say a patient is living with a person who is caring for them but also carries a health risk (for example they are an elderly smoker) and that person is caring for the patient informally and the patient is say over a certain age which would indicate their carer may not live over the long term. In other words, the older these people become the greater the risk that this care will be lost and therefore the patient is at greater risk of losing support which they depend upon.
Any advice on how to conduct a stepwise analysis on this will be greatly appreciated and Liked promptly.