Observational / Non-experimentalTo determine Both Exposure & Disease have already occurredwhether or notan associationexists between a measure exposuredisease and aparticular riskfactor. Uses Comparison Group Retrospective Effect to Cause Back in time
Why case-control design for study of rarediseases? Consider some rare disease say some cancer (leukemia)• Crude Annual Incidence = 3.4/100000 (< 15 years)• Cohort Study: A year of observation on a million children to identify 34 cases• Sample of 34 cases: Sub-divided in 2 or more exposure categories• What about conducting case-control design?
Research questions• Is OC use associated with MI in women?• Is current IUD use associated with PID?• Is OC use associated with the risk of breast cancer?• Is age at first coitus associated with cervical cancer?
Selection of cases Cases may be located from Select cases after the hospitals, clinics, disease diagnostic criteria and registries, screenings, etc. definition of the disease is clearly established Incident cases are preferable to prevalent cases for reducing • Conceptual definition • Recall bias • Obesity defined as • Over-representation of cases of body fat Case percentage > 33% long durationdefinition • Operational definition The most desirable way to obtain cases is to include all incident cases • Body Mass Index > 30 in a defined population over a specified period of time
Why controls?• Controls estimate the exposure rate to be expected in cases if there were no association between exposure and disease
Selecting Controls Controls shouldCome from thesame populationRepresentative
Sources of cases and controlsCases ControlsAll cases diagnosed in the Sample of general populationcommunityAll cases diagnosed in a Non-cases in a sample of thesample of the population populationAll cases diagnosed in all Sample of patients in allhospitals hospitals who do not have the diseaseAll cases diagnosed in a Sample of patients in the samesingle hospital hospital who do not have the diseaseAny of the above methods Spouses, siblings or associates of cases
Ascertainment of Disease and exposure status • Death • Questionnaires and interviews, certificates, information from • Disease a surrogateExternal Internal (spouses orsources: registries, sources: mother of • Hospital and children), physicians biological sampling( e.g. records etc. antibody)
Bias in Case-Control studies Selection biasAny trend in the • Unequal chance of getting into studycollection, analysis, Berkson’s biasinterpretation, • Variable rate of hospitalizationpublication or review affecting case selectionof data that can lead Neyman fallacyto conclusions that • Incident case Vs prevalent caseare systematically Detection biasdifferent from the • Due to closer medical attention,truth (Last, 2001) detection of endometrial cancer was more in a group using estrogen
Matching Matching is Matching may be defined as the of two types: process of selecting the controls so that they are similar tothe cases in certain Individual Group matching characteristics, matching. such as age, race, sex, Case selection Select one case socioeconomic status, and Select next hospital occupation. Calculate proportions admission that matches age and Control selection sex as control
Assessing ExposureExposure is usuallyan estimate unless pastmeasurements are available
Calculating the Odds Ratio Disease Status CHD cases No CHD (Cases) (Controls) Exposure Smoker 112 176 Status Non- 88 224 smoker Total 200 400 AD 112 x 224Odds Ratio = = = 1.62 BC 176 x 88
Calculating the Odds Ratio Disease status CHD (Cases) No CHDExposure (Controls)status Smoker 112 (a) 176 (b) Non smoker 88 (c) 224 (d) 200 400
Interpreting the Odds RatioThose with CHD are 1.62 times more likely to be smokers than thosewithout CHD orThose with CHD are 62% more likely to be smokers thanthose without CHD
Inference OR<1 OR=1 OR>1 Odds of exposure Odds of Odds of exposureOdds exposure are for cases are for cases are lesscomparison equal among greater than the than the odds ofbetween cases cases and odds of exposure exposure forand controls controls for controls controls Exposure ExposureExposure as a reduces Particular increasesrisk factor for disease risk exposure is not a disease riskthe disease? (Protective risk factor (Risk factor) factor)
Advantages of Case-Control Studies• Quick and easy to complete, cost effective• Most efficient design for rare diseases• Usually requires a smaller study population than a cohort study
Disadvantages of Case-Control Studies Uncertainty of exposure-disease time relationship Inability to provide a direct estimate of risk Not efficient for studying rare exposures Subject to biases (recall & selection bias)
Cohort StudiesAt baseline (1st observation point)Subjects are all disease free Exposure is used to classify subjects into exposed or unexposed groups Subjects are followed to document incidence (2nd observation point)
Types of Cohort StudyProspective cohort studyRetrospective (historical) cohortstudyCombination of Retrospective andProspective cohort study.
Selection of study subjectsGeneral population• Whole population in an area• A representative sampleSpecial group of population• Select group • occupation group / professional group (Dolls study )• Exposure groups • Person having exposure to some physical, chemical or biological agent • e.g. X-ray exposure to radiologists
Obtaining data on exposurePersonal interviews / mailed questionnaireReviews of records • Dose of drug, radiation, type of surgery etcMedical examination or special test • Blood pressure, serum cholesterolEnvironmental surveyBy obtaining the data of exposure we can classify cohorts as • Exposed and non exposed and • By degree exposure we can sub classify cohorts
Selection of comparison groupInternal comparison• An internal comparison group consists of unexposed members of the same cohort.• An internal comparison group should be used whenever possible, because its characteristics will be most similar to the exposedExternal comparison• More than one cohort in the study for the purpose of comparison• e.g. Cohort of radiologist compared with ophthalmologistsComparison with general population rates• If no comparison group is available we can compare the rates of study cohort with general population.• Cancer rate of uranium miners with cancer in general population
Follow-upTo obtain data about outcome to be determined (morbidityor death)• Mailed questionnaire, telephone calls, personal interviews• Periodic medical examination• Reviewing records• Surveillance of death records• Follow up is the most critical part of the studySome loss to follow up is inevitable due to death change ofaddress, migration, change of occupation.Loss to follow-up is one of the draw-back of the cohort study.
ANALYSISCalculation of incidence ratesamong exposed and nonexposed groupsEstimation of risk
Incidence rates of outcome Disease status TotalExposurestatus Yes No Yes a b a+b (Study cohort) No c d c+d (comparis on cohort) a+c b+d
Relative Risk (RR)• A ratio that measures the risk of disease among the exposed to the risk among the unexposed• RR Numerator: Incidence rate in the exposed• RR Denominator: Incidence rate in the unexposed
Example: Calculating the Relative Risk Disease Status CHD cases No CHD TOTAL (Cases) (Controls)Exposure Smoker 112 176 288 Status Non- 88 224 312 smoker A/(A+B) 112 / 288 Relative Risk = = = 1.38 C/(C+D) 88 / 312
Example: Interpreting the Relative Risk Relative Risk = 1.38The risk of developing CHD is 1.38 times higher for a smokerthan for a nonsmoker. orThe risk of developing CHD is 38% higher for a smoker thanfor a nonsmoker.
Inference RR<1 RR=1 RR>1 Risk for disease isRisk comparison Risk of disease is equal Risk for disease is lower in the exposedbetween exposed for exposed and higher in the exposed than in theand unexposed unexposed than in the unexposed unexposed Exposure reducesExposure as a risk disease risk Exposure increases Particular exposure isfactor for the disease risk (Protective not a risk factordisease? (Risk factor) factor)
Attributable Risk Incidence of disease among exposed – incidence of disease among non exposed Incidence of disease among exposed a/a+b – c/c+dAR = a/a+b
Smoking Lung cancer Total YES NOYES 70 6930 7000NO 3 2997 3000 73 9927 10000 Find out RR and AR for above data
• Incidence of lung cancer among smokers 70/7000 = 10 per 1000• Incidence of lung cancer among non-smokers 3/3000 = 1 per thousandRR = 10 / 1 = 10 (lung cancer is 10 times more common among smokers than non smokers)AR = 10 – 1 / 10 X 100 = 90 % (90% of the cases of lung cancer among smokers are attributed to their habit of smoking)
Advantages of Cohort StudiesTemporality: Exposure precedes outcome Efficient for studying May be used to study because the cohort is rare exposures multiple outcomesdisease free at baseline Allows for calculation of incidence of diseases in Minimizes recall bias exposed and unexposed individuals
Disadvantages of Cohort Studies Loss to follow-up Tend to be expensive • When multiple outcomes or(large sample size) and specific disease incidence is time consuming (long the outcome of interest, bias follow-up period) can be a serious problem Inefficient to study rare diseases
Disadvantages of Cohort Studies (cont.)• Nonparticipation (selection bias) – it cannot be assumed that those who chose to participate had the same prevalence of exposures nor incidence of disease as those who did not participate – A difference in prevalence of exposure in nonparticipants will not bias the results – A difference in rate of disease among nonparticipants will bias the results