Definition A systematic error in collecting or interpreting observations found in the study design Types of Bias Types of BiasBiasDescriptionMitigationAccumulation Effectpatients sometimes must be exposed to a risk factor for aprolonged period of time before they develop a clinically detectable resulte.g., patients must smoke for many pack-years before bronchogenic carcinoma developstry to follow study participants for as long as is feasibleConfoundinga third factor is either positively or negatively associated with both the exposure and outcomeconfounders arenot in the causal pathwayif not adjusted for, can distort true associationeither towards or away from the null hypothesisrandomizationensures similar baseline characteristics between control and exposure/experimental groupsuse intention-to-treat analysis to preserve randomization even if participants change study treatmentsmatchinggroup similar participants into study pairsstratificationanalyze in separate subgroups determined by a potential confounderrestrictiononly include groups with specific features in the sampleadjustmentcan only adjust for confounders that areknown and measureablecrossover studiessubject acts as own controlSelection Biassampled population is not representative of the population researchers are trying to studydue to non-random selection of study participantssampling (ascertainment) biascertain individuals are more or less likely to be selected for a study group, leading to incorrect conclusionsnon-response biase.g., participants who pick up the phone may be less sick than participants who don'thealthy worker effectsamples with employed subjects only may be healthiervolunteer biaspeople who volunteer for a study may be different in some fundamental way from those who do not volunteerlate-look biaspatients with severe disease are less likely to be studied, because they die or are otherwise unavailable, making a disease look less severee.g., a group of HIV+ individuals are all asymptomaticalso can have opposite effecte.g.,people with more mild disease are cured before the study takes place and only persistently sick folks are included in the study, making a disease seem more severeBerkson biashospitalized study subjects are more likely to have a greater burden of illness than other possible subjectsattrition biasthose lost to follow-up may be different from those who remain in the studyrandomizationinclude patients in multiple settings (outpatient, hospitalized)study designs that are longitudinal in nature rather than cross-sectionalgather maximal information on participantsMeasurement Biasinformation is gathered in a way that distorts the information or misclassifies study participantsinterviewer biassubjects in one group are interviewed in a different way than anotherdifferences due to interviewing style disrepencies are falsely attributed to group differencesstandardize data collectionRecall Biassubjects with the disease are more likely to recall the exposure of intereste.g., parents of children with cancer recall exposure to a chemical reducing follow-up time in retrospective studiesPerformance Biasresearchers treat groups differently or subjects alter their behavior/responses due to study group awarenessHawthorne effectsubjects alter their behavior when they know they are being studiedprocedure biasresearcher decides assigment of treatment versus control and assigns particular patients to one group or the other nonrandomlypatient decides assigment of treatment versus controlblindingLead-Time Biassubjects appear to survive longer when in reality their disease was detected earliercommon with improved screeninge.g.,a cancer screening test is deemed to increase survival when in reality the disease was picked up earlier, increasing the time from detection to deathuse mortality rate instead of survival time in screening studiesestimate lead time and add that to survival in unscreened groupDesign Biasthe control group is inappropriately non-comparable to the intervention groupallocation biasdifference in the way participants are placed in control versus experimental groupse.g.,all zebras in control group and all lions in exposure grouprandomizationmatchingCognitive Biasobserver bias (pygmalion effect)investigator inadvertently conveys her high expectations to subjects, who then produce the expected resulta "self-fulfilling prophecy"golem effect is the opposite: study subjects decrease their performance to meet low expectations of investigatorconfirmation biasresearcher ignores results that do not support their hypothesisresponse biasparticipants do not respond accurately because they are concerned about the social desirability of their responses or misinterpret the question double blindinginclude positive and negative resultsSurveillance Biasoutcomes are more likely to be detected in certain groups because of increased monitoringe.g.,a certain skin disease being detected more often in hypertensive patients because they have more physician visits than non-hypertensive patientsresearchers may falsely attribute hypertension to causing the skin diseasematch participants on similar likelihood of surveillance Examples of Effects that are NOT Bias