Using Epidemiology to Identify the Cause of Disease: Cohort study Sakda Arj-ong Vallipakorn, M.D., MSICT., MSIS., PhD.(Clinical Epidemiology) Pediatrics, Pediatric Cardiology Emergency Medicine, Pediatric Emergency Medicine Ramathibodi Hospital, Mahidol University
What is the risk factor of disease Smoking CA Lung C-reactive protein Heart diseases
Assign exposure / intervention? yes no Experimental Observational Comparison group? yes no Analytic Descriptive Start with Outcome or Exposure E O E O E/O Case-control Cohort Cross sectional
Assign exposure / intervention? yes no Experimental Observational Comparison group? yes no Analytic Descriptive Start with Outcome or Exposure E O E O E/O Case-control Cohort Cross sectional
Cohort study cohort Group whose members share a significant experience at a certain period of time or have one or more similar characteristics. People born in the same year, for example, are the birth cohorts (generation) for that year. Similarly, married men or those who smoke are cohorts of the other married men or other smokers.
Bad Training Environment Burn out Residents Good Training Environment
Criteria for cohort study 1. Do not have the outcome at the time that study started 2. F/U time should be sufficient for the outcome to be expressed 3. Members of the cohort should be observed over the full period of F/U
Framingham study Begun in 1949 Identified factors associated with CHD 5,209 men and women, aged 30-59 Study ran for 30 years Continues with Framingham Offspring study
Types of cohort study Prospective cohort study Retrospective cohort or historical cohort study
Past Present Future NO Dz Diseases Cohort assembled Prospective cohort Follow-up
Past Present Future Cohort assembled Rama Personal Retrospective cohort Follow-up Obesity Cohort assembled Prospective cohort CVD, Stroke Follow-up
Retrospective or historical cohort study
1991-1998 Present MMR MMR vaccine Mump Measles Rubella Non MMR vaccine
Advantage Temporal sequence Concurrent Exposure Outcome Rare exposure; occupational setting
Advantage Can assess the relationship between exposure and many diseases ( one : many) n 1 n Breast cancer melanoma n Ovarian Cancer
Disadvantages Cannot be used for rare disease
Disadvantage Long latency period Exposure Outcome Time consuming Expensive/Budget loss Loss F/U
Selection of exposed population Common or Rare exposure?
Selection of exposed population Common exposure & common disease ability to complete and accurate information
Source of information 1. Preexisting record 2. Interview, questionnaire 3. Direct physical exam 4. Laboratory
Source of outcome data 1. Death certificate 2. Periodic health examination 3. Physician record, hospital discharge
Loss to F/U How long? Exposure Outcome Time
Outpatient record Town s resident list Military records Employment records Small amount of Loss F/U Loss F/U equally between expose and non expose
Way to express and compare risk
Incidence rate or risk of developing disease
Develop disease Not develop disease Incidence Rate Exposed a b a a+b Non exposed c d c c+d
Developed lung CA Not developed Lung CA Incidence Rate Smoked 100 900 0.10 Nonsmoked 10 990 0.01
Measures of effects in Cohort Study Risk difference Relative risk
Risk difference Incidence rate Expose incidnce rate Non-expose Incidence rate smoking incidnce rate Non-smoking
Developed lung CA Not developed Lung CA Incidence Rate Smoked 100 900 0.10 Non-smoked 10 990 0.01 Risk difference = 0.1-0.01 = 0.09 = 9%
Relative risk The ratio of the incidence rate between exposed and non-exposed RR = incidence in exposed incidence in non-exposed
Developed lung CA Not developed Lung CA Incidence Rate Smoked 100 900 0.10 Non-smoked 10 990 0.01 Relative risk = 0.1/0.01= 10
Interpreting Relative Risk IR Ex = IR non RR = 1 (no association) IR Ex > IR non RR > 1 (risk effect) IR Ex < IR non RR < 1 (protective effect