Analytic Studies: Observational Study Designs

Cohort Studies Analytic Epidemiology Prospective Cohort Studies Retrospective (historical) Cohort Studies Part 2 Case-Control Studies Nested case-control Case-cohort studies

Dr. H. Stockwell Case-crossover studies

Determinants of disease: Analytic Epidemiology Epidemiology: Risk factors

„ Identifying the causes of disease „ A behavior, environmental exposure, or inherent human characteristic that is associated with an important health „ Testing hypotheses using epidemiologic related condition* studies „ Risk factors are associated with an increased probability of disease but may „ Goal is to prevent disease (deterrents) not always cause the diseases

*Last, J. Dictionary of Epidemiology

Analytic Studies: Cohort Studies Panel Studies

„ Healthy subjects are defined by their „ Combination of cross-sectional and exposure status and followed over time to cohort determine the incidence of disease, symptoms or death „ Same individuals surveyed at several poiiiints in time „ Subjects grouped by exposure level – exposed and unexposed (f(reference group, „ Can measure changes in individuals comparison group) Analytic Studies: Case-Control Studies Nested case-control studies

„ A group of individuals with a disease „ Case-control study conducted within a (cases) are compared with a group of cohort study individuals without the disease (controls) „ Advantages of both study designs „ Also called a retrospective study because it starts with people with disease and looks backward for ppprevious exposures which might be relevant to the development of the disease

Other case-control designs Intervention Studies

„ Case-cohort studies – controls sampled from entire source population – those at „ Also called Experimental studies risk at the start of the study „ Randomized Controlled Clinical Trials

„ Case-crossover – cases serve as own „ Community or field trials controls- good for events that have acute onset times

Randomized Controlled Trials Community Trials

„ Preventive „ Random allocation is at community level

„ Intervention or other group such as a school

„ Therapeutic Randllfldom allocation of volunteers to experimental or control procedure to dfldetermine impact of experimental exposure on outcome Why do we do an analytic study? Measures of Association

„ Most often to look for a relationship Goal: between an exposure and disease „ A single measure that estimates the association between an exposure and the risk of developing disease „ Does exposure to factor X „ Calculate ratios of the measures of disease freqqyuency i/dikfdiincrease/decrease your risk of disease (usually incidence) This ratio is called a relative risk Y? or „ Calculate the difference between the two measures „ Need a measure of this association of disease frequency called risk difference or attributable risk

Measures of association Relative Risk Estimates

„ Relative risk and risk difference/ „ Cohort studies- calculate relative risk attributable risk are the two most „ Case-controldl studies – calllculate odds frequently used measures used in ratio to estimate relative risk epidemiology

„ Cross-sectional studies – calculate the „ Together they provide a more complete prevalence rate ratio pihihicture than either measure al one

Cohort Studies: Calculating the Relative Risk Two kinds of Incidence

• It indicates the likelihood of developing the „ Cumulative Incidence disease in the exposed group relative to those not exposed. Number of new cases of disease over a specified time period Total population at risk • It is the ratio of the incidence of disease among exposed group divided by the incidence of disease among non-exposed „ Incidence Density Number of new cases of disease over a specified time period • The RR estimates the strength of an Total person-time of observation association between exposure and disease.

• RR= IE / I0 Relative Risk Risk Ratio

• If the incidence measure being used is Disease/Outcome cumulative incidence , the relative risk is called Risk ratio = a / a+b D D the risk ratio c / c+d E ab

Exposed E cdd • If the incidence measure being used is CI incidence density, the relative risk is called the exposed = rate ratio CI unexposed

Risk Ratio Meaning of RR

„ 1.0 = no association: There is no relation between the Exposure and Disease Rate ratio = a / PYe

c / PYu „ > 1.0 = Exposure is a risk factor for the Disease: Persons with the exposure are more likely to get Disease ID exposed = „ < 1.0 = E is inversely related to D: Persons with ID unexposed the exposure are less likely to get Disease

The relative risk is RELATIVE Relative Risk to the baseline incidence

„ The relative risk estimate is relative to CIE = .0026 Risk ratio (RR) = 1.0 the baseline incidence CI0 = .0026

CI = .49 E Risk ratio (RR) = 1.0 CI0 = .49

5 IDE = .062 per 100 PY 5 Rate ratio (RR) = 1.0 ID0 = .062 per 100 PY Cohort Study: Risk Ratio Risk Ratio

Cumulative incidence= OC Use Infection OC Use: Infection Yes No Total CI in Exposed/CI in Unexposed Yes No Total Yes 27 455 482 Yes 27 455 482 No 77 1831 1908 „ 27/482= .056 (CI in Exposed) Total 104 2286 2390

No 77 1831 1908 „ 77/1908=.040 (CI in Unexposed

Total 104 2286 2390 „ Risk Ratio= .056/.040 = 1.4

Rate Ratio Calculate the rate ratio

Question: HRT use and CHD Incidence density among E CHD RR = Yes No Incidence density among E + 54308.7 Study sample: 30 HRT + HRT -- 60 d 51477.5 30 women taking HRT developed CHD after Rate ratio = ID = 30 / 54308.7 = .0005524 54,308.7py of follow-up E ID0= 60 / 51477. 5 = .0011655 RR = 0.0005 / 0.00116 = .47 or .5 60 women not taking HRT developed CHD after 51,477 .5py of follow -up Interpretation: women using HRT had only half (.5) the risk of developing CHD as did non users.

Cross-sectional studies: Calculating the prevalence rate ratio Prevalence Rate Ratio

„ PRe=27/61=.443 „ Prevalence rate among exposed Exposure Outcome smoking Depression No Total „ PRu=26/88=.295 Yes Prevalence rate among unexposed „ PRR= . 443/. 295 Yes 27 (a) 34 (b) 61 =1.5 „ The prevalence of the outcome (depression) is No 26 (c) 62 (d) 88 1.5x more common among those exposed Total 149 (smokers) than those unexposed (non- smokers) NO AUDIO ON THIS SLIDE

Case-control Studies Review Quiz 2: Rationale

Question 1 „ Cannot calculate incidence in case- control studies , so cannot calculate 450 /(450 + 20) Prevalence Rate Ratio = = 1.27 relative risk 1150 /(1150 + 380)

„ Estimate relative risk in case-control Question 2 studies byyg calculating the odds ratio This cohort study measured cumulative incidence so (OR) you would calculate a risk ratio. 160 /(160 + 80) 0.667 RR = = = 6.009 40 /(40 + 320) 0.111