Neuroepidemiology Teaching Course: Non-Experimental Neuroepidemiology Chisinau Moldavia September 26 2012 Design and Methods of Cohort Studies Giancarlo Logroscino MD PhD Department of Neurology and Psychiatry University of Bari. Italy Veterans By SEP default http://www The h 1 378 nytimes.co health Health The%20N By%20DE 20080924 11948178 The%20fe Veterans% ttp%3a% DENIS New 24 2Y G Veterans Gain U.S. Benefits for Lou Gehrig’s Disease •E-Mail •Print •Reprints •Save •Share •Linkedin •Digg •Facebook •Mixx •YAHOO! BUZZ •Permalink By DENISE GRADY September September 28, 2008 23, 2008 Published: The federal government will provide disability pay, lifetime health care and death benefits for all veterans with Lou Gehrig’s disease, the Department of Veterans Affairs said Tuesday, saying the disease was linked to military service. Skip to next paragraph RSS Feed •Get Health News From The New York Times » All veterans with the illness will be eligible, regardless of when or where Some general statements about epidemiology Study Design: Two Steps • We determine whether there is an association between a factor or a characteristics and the development of a disease • We establish appropriate inferences regarding a possible causal relationship based on the patterns of the association that has been found Goal of Epidemiology • Quantification of causal relationship between exposure and disease (K. Rothman 1986) • Measurement of the association focuses on quantitative index that characterizes the strength of the association • Quantification of an association is a measure of state of nature Null Effect The Natural History of a Disease Gordis L, Epidemiology 2000 W. B. Saunders Company Philadelphia, PA Preclinical Phase Clinical Phase Outcome (A) Biological Onset (P) (S) Pathological Evidence Signs and Symptoms of Disease time (M) Medical Care sought (D) Diagnosis (T) Treatment Time and the Induction Period of Disease Preclinical Phase Clinical Phase Exposure Biological Onset time Signs and Symptoms of Disease Diagnosis time Classification of Studies Descriptive Studies Populations Frequency Distribution by time Place Person Population Survey Analytic Studies Experimental Studies •Counting exposure and outcomes in individuals •Test causal hypotheses •Uncontrolled assignment •Counting exposure and outcomes in individuals •Test causal hypotheses •Controlled assignment CaseControl Cohort Randomized Clinical Trial Anatomy of the Cohort Study • Define the hypothesis • Define the independent variable (risk factor) -Exposed cohort -Unexposed cohort • Define the dependent variable (disease) • Perform measures -Initial study/baseline assessment -Follow-up and detection of incident cases • Statistical analysis -Incidence and relative-risk -Person-years, survival curves • Interpretation of results (bias) COHORT STUDIES • Study design Then follow to see whether First Select Totals Disease Develop Disease Not develop Exposed a b a+b Not exposed c d c+d Incidence Rates of disease a a+b c c+d COHORT STUDIES • Results of a Hypothetical Cohort Study of Smoking and Coronary Heart Disease(CHD) Then follow to see whether Develop CHD First Select Smoke cigarettes Do not smoke cigarettes Totals Do Not develop CHD Incidence per 1000 per year 84 2916 3000 28.0 87 4913 5000 17.4 RR= Ie/ Ine= Do we need observational studies? Individuals and Populations Observational epidemiolgy and Cohort Studies Experimental Study (RCT) Observational Study POPULATION OTHER-THAN RANDOM ALLOCATION (e.g., Self Selection) RANDOM ALLOCATION GROUP A GROUP B GROUP B Animal Models •Complete Control of Exposure •Random Allocation Individuals and Populations Lost in translation: treatment trials in the SOD1 mouse and in human ALS. Benatar M Neurobiology of Disease 2007 Apr;26(1):1-13 Why did the US Government take that decision about Veterans and ALS? Veterans By SEP default http://www The h 1 378 nytimes.co health Health The%20N By%20DE 20080924 11948178 The%20fe Veterans% ttp%3a% DENIS New 24 2Y G Veterans Gain U.S. Benefits for Lou Gehrig’s Disease •E-Mail •Print •Reprints •Save •Share •Linkedin •Digg •Facebook •Mixx •YAHOO! BUZZ •Permalink By DENISE GRADY Published: September 23, 2008 The federal government will provide disability pay, lifetime health care and death benefits for all veterans with Lou Gehrig’s disease, the Department of Veterans Affairs said Tuesday, September 28, 2008was linked to military service. saying the disease Skip to next paragraph RSS Feed •Get Health News From The New York Times » All veterans with the illness will be eligible, regardless of when or where they served. The 10-year cost for death and disability benefits is projected at $505,839,000, said Tom Pamperin, the deputy director of the compensation and pension service at the Veterans Affairs Department. That figure does not include health care costs. Inference from Epidemiological Evidence Savitz DA Oxford University Press 2003: 8 • Association between exposure and disease among study participants • Causal effect of exposure on disease in study population • Causal effect of exposure on disease in external population • Prevention of disease through elimination or reduction of the exposure • Public Health Impact from elimination or reduction of the exposure Gulf War Veterans: the ALS occurrence • 2482333 individuals who were on active duty or activate reserve • 696118 deployed to the Gulf region • 516 ALS cases from passive and active ascertainment • 96.2% successfully contacted Occurrence of ALS among Gulf War Veterans Horner et al, Neurology, 2003; 61:742-749 DMDC-reported Deployment Status: Population EEC Onset July 1991 (n=86) Men only (n=105) Total 2.1(1.3-3.2) 1.8(1.2-2.8) 1.7(1.2-2.6) Unit Active Reserves-National Guard 2.2(1.4-3.6) 3.4(1.0-11.3) 2.0(1.2-3.2) 2.7(0.9-8.4) 2.0(1.3-3.1) 2.2(0.7-6.3) Service branch Air Force Army Marine corps Navy 3.1(1.4-6.7) 2.1(1.1-4.3) 0.9(1.2-4.9) 1.9(0.7-5.1) 2.1(0.9-4.9) 2.7(1.3-5.3) 1.1(0.3-4.8) 1.0(0.3-3.0) 2.6(1.2-5.6) 1.8(0.9-3.3) 1.1(0.3-4.6) 1.4(0.6-3.4) Anatomy of the Cohort Study 1.Define the hypothesis 1.War RF for ALS 2.Define the independent variable (risk factor) 2. Deployment in GW -Exposed cohort Lack of deployment in GW -Unexposed cohort 3. Define the dependent variable (disease) 3.ALS 4.Perform measures -Initial study/baseline assessment -Follow-up and detection of incident cases 4.Deployment Status Ascertainment Detection of ALS cases 5.Statistical analysis -Incidence and relative-risk -Person-years, survival curves 5 Measurement of RR 6. Interpretation of results (bias) Population Concept of a Cohort Study Bhopal R. Concept of Epidemiology. An Integrated Introduction to the Ideas, Theories, Principles and Methods of Epidemiology. 2002 Oxford University Press, Oxford Time 0/Now NE NE E NE NE NE E E NE ALS E Induction Period now NE NE E E ALS NE ALS E NE NE E NE NE NE ALS ALS E E E E Time 1/Future E E NE ALS CHD future Excess of Incidence of ALS in Young Gulf War Veterans Haley, Neurology, 2003; 61:750-756 3.50 3.19 3.00 2.50 2.27 2.16 SMR (Ratio O:E) 2.03 2.00 1.67 1.55 1.50 1.07 0.94 0.91 1.00 0.50 0.00 0.00 1991 1992 1993 1994 1995 1996 1997 1998 1991-1994 1995-1998 ALS and The Gulf War Veterans: a Possible Cluster • Onset of disease in young age groups that have a very low rate in the general population • There is an increasing slope of the epidemic curve: the excess of cases is still to come • The excess in incidence probably triggered by some environmental exposure ALS mortality by military service and birth cohort Weisskopf et al Neurology: 2005;64:32-37 Relative Risk of ALS by Service During War Periods (1989-1998) Weisskopf et al Neurology: 2005;64:32-37 10 1 1.0 1.7 1.9 1.97 1 1 No N 2w > ars 2 wa Mi o Wa war l i ta rs ry rtime Se rvi ce p for trend <0.004 What is the Cause of ALS excess of cases among Gulf War veterans? ALS and Military Service: Hints for Causality Exposure precedes the outcome The risk peaks after a time period (induction period) The risk increase with increasing dose of exposure Advantage and Disadvantages Disdvantages Advantages • Measure relative risk • Less misclassification • Less measurement bias • Changes in exposure and gradation of exposure • Multiple disease for the same risk factor • Non-respondent/volunteer bias (selection) • Diagnostic suspicion bias (measurement) • Attrition •Time and expense (big sample) •Difficult to study uncommon diseases • Study may alter outcome • Ethical problems • Limited and fixed number of hypotheses (risk factors) Potential Contribution of Study Design Bhopal R, Concepts of Epidemiology, Oxford 2002 Criteria Cross-sectional Case-control Cohort Trial Temporality Sometimes Sometimes Often Usually Strength or Dose-response Sometimes Often Always Always Experimental confirmation Sometimes (repeated studies) Seldom Sometimes (natural changes) Always Specificity Sometimes Yes (disease) Yes (risk factor(s)) Biological Plausibility Not Directly Not directly Not directly Consistency Yes Yes Yes Yes (risk or preventive factor(s)) Not directly Yes