pubhlth 244: discussion 8

if strata-specific are diff…

effect modifier, report strata specific RR

if strata-specific are similar…

calculate thhe adjust RR & find magnitude & report if it is important based on the 10% rule

bias that occurs in case-control & cohort/experimental

• participation differs on exposure & disease

• more likely

information bias in case-control & cohort

• bias will always occur whrn info differes on exposure & disease

• more likely

information bias branches

• recall bias

• interviewer bias

• measurement error

recall bias & solution

• try to remember something

• solution: look at existing records or closed ended standardized questions

interview bias & solution

• inteviewer’s conscious or unconscious personal beleifs that influence assessment

• solution: asking for closed ended questions in an objevtice (not subjective)

measurement error & solution

• solution: use mor accurate source of info avail

• validarion from several data sources

ex: in a cohort study of sodium & hypertension, dietary socisum is assessed using a single 24 hr recall, which doesn’t accurately reflect

measurement bias

ex: in a case control studye of congenital anomalies

recall bias

ex: in a study examining whether-longer term air pollution

selection bias

ex: in a study of maternal alcohol use and preterm birth, interviewrs

interviewer bias

effect modification

variable that changes the strength or direction of the association b/w an exposure & outcome

effect of the exposure on the outcome is diff depending on the level of the effect modifier multiplicative scale vs additive scale