Survival Analysis

Question 1

what is survival analysis and advantage

Answer 1

= how long it took an event to occur
- account for loss to follow up
~ account for animals left study before event of interest occured = censoring

Question 2

examples of events that could study using survival analysis

Answer 2

• motion sickness in pigs
  ~ exposure whether or not ate before transport
  ~ survial time until sickness occured
• mange in dogs
• task learning in chimpanzees
  ~ the event of interest not always death (even though survival)

Question 3

Types of censoring

- right censoring

Answer 3

• when an animal is lost to a study, before the outcome of interest occurs.
• Or study ends and the animal has not experienced the outcome of interest.

Question 4

Types of censoring

- interval censoring

Answer 4

• when an animal is only observed periodically (e.g., every 6 months), all that is known is if the event occurred in the preceding time window, but the precise time is not known

Question 5

Types of censoring

- left censoring

Answer 5

• event occurred before the animal was observed
  (and consequently the animal may not be put in the study, unless depending on if multiple events are permitted e.g. mastitis)

Question 6

evaluating the effect of factors on survival times

Answer 6

• Lots of different types of survival models
• non-parametric models have no assumptions
• Using our data, we can create Kaplan Meier “survival curves”
• compare the curves using the log-rank test to see if they are statistically different from one another
• 1.00 = 100% animals dont have effect
• p value represents results of log rank test = at least one of the curves is significantly different from others but dont know where that difference lies
  ~ suspect one line is significantly diff due to distance between lines (two could be closer together and one further out)
• step down = the event occured
• vertical line = an animal was censored

Question 7

Cumulative hazard

Answer 7

• instead of survival on y axis = cumulative hazard
• go up from 0
• hazard function directly related to survival function
• may logically make more sence presenting response