Bias in Epidemiologic Study Designs – Flashcards

Accuracy

The study provides an UNBIASED estimate of what it claims to estimate

The results from the study can be GENERALIZED to some o ther population.

Precision

A systematic departure from the truth (A SYSTEMATIC ERROR) Any systematic error in the design, conduct or analysis of a study that results in a mistaken estimate of an exposure's association with disease. i.e., produces a biased estimate of OR,RR,PRR etc.

A systematic error = bias ---Selection Bias ---Information Bias Random Error=chance, sampling variability, sample size ---p values ---confidence intervals

how the study group is chosen Errors due to systematic differences in relevant characteristics b/w those who are included in study and those who are not ---Can be introduced by: * The investigator (how you choose subjects) * Personal Choices of Individuals (e.g. self-selection bias) * How persons with disease are diagnosed in the community (e. g. exposure suspicion bias) NO SELECTION BIAS if study population looks like target population...as in if prevalence of study = prevalence of target YES SELECTION BIAS if study population DOES NOT look like target population...as in prevalence of study /= prevalence of ta

Prevalence/Incidence Bias Self-Selection or Membership Bias Non-Response Bias in Case-Control Studies Loss to Follow-up/Withdrawa Bias (Cohort Studies) Berkson's Bias

Case-control / cross-sectional studies, results when prevalent cases, rather than incident cases, are used Prevalent cases over-represent cases of long duration bc those who die or are rapidly cured are less likely to be included Selective mortality-OR towards the null Selective survival-OR away from the null Avoid by using incident cases, if prevalent must be used: -use cases with short interval b/w onset and diagnosis, -include cases from entire spectrum of disease -count and obtain exposure information from deceased cases ex. colorectal cancer-women have higher life expectancy; so when using prevalent cases, you will have more proportion of women cause more women live longer

Characteristics of an individual may consciously or unconsciously affect membership in a certain group e.g. have to have a certain degree of health to jog---affects association of jogging and hearth dz Healthy Worker Effect-morbidity and mortality is lower in workers than in the general population (which includes non workers) difficult bias to avoid; choose comparison groups w/ same selection pressures, try to equalize from beginning CAN ONLY BE AVOIDED BY RANDOMIZATION-bc individuals can not choose their exposure status, it is chosen for them e.g. joggers and regular people----exercise effect on heart disease ---joggers also have other good habits such as good food to eat so this is selection bias

People who agree to participate in study may be different in terms of exposure, or other important characteristics than those who don't participate in studies---may increase or decrease risk estimate depending on factors related to non-response Case and Controls e.g. women who work tend to have more pap smears, so if controls in a case control study choose females who stay at home...then there is bias

Cannot determine whether bias is present based on follow-up proportions in exposed and unexposed alone Bias can be present if follow-up proportions are equal or absent if they are unequal Incidence outcome is different among exposed who are followed, compared to those who are lost, (or for unexposed and lost) then a biased risk estimate may result Can be reduced by minimizing to follow-up Assess by comparing lost group to followed group e.g. in alcohol study subjects are lost; if the incidence of diabetes is lower in drinkers who remain in your study than in drinkers who are lost to follow-up you wound under-estimate the true incidence; if it's higher, you would over estimate

Joseph Berkson Refers to selective factors that lead hospital cases or controls to be systematically different from all cases or controls in the population they represent occurs when a combination of exposure and disease lead to increase in hospitalization e.g. children with asthma and emotional disorders arrive at hospital more than just asthma; so a study with hospitalization of asthma would be biased if these children were chosen as cases or controls

inaccuracy in measurement or classification of exposure, outcome, or covariates; results in measurement error / misclassification Recall Bias Interviewer Bias Family Information bias

An error in the measurement of categorical variables Can be Non-differential or Differential

Classification error DOES NOT depend on disease status or exposure status If error of exposure classification is same for cases and controls, bias will usually be TOWARDS NULL; UNLESS: -mis-measured confounding factors -if >2 categories of the variable

Classification error DOES depend on disease status or exposure status If error of exposure classification is NOT the same for cases and controls, the bias can go EITHER TOWARD OR AWAY from the NULL depending on the nature of Misclassification

An error in the measurement of continuous variables

Differences among cases and controls in accuracy or completeness of recall of exposure Results in OR away from null when cases more accurately remember exposure than controls or when cases over-report exposure history Avoid by validating exposure history from independent source or objective measures Use of "sick" controls may help equalize this bias

Systematic Error due to interviewer's subconscious or conscious gathering of selective info in different groups results in an over-estimate of association when interviewers probe for more information among cases avoid by blinding interviewers to case/controls status; train interviewers and use standardized forms Assess by re-interviewing a sample of study subjects e.g interviewer in a case control study subconcsiouly selects a specific subject more than any other leading to bias.

Occurs when cases are more aware of their family history because the occurrence of their disease has stimulated discussion or investigation of past disease history in the family Results in an increased (odds) estimate, since cases may be more aware of a positive family history than are contorls Avoid by validating disease status of family members e.g. those with a disease will remember more about that disease in their family history---introduces biases