Dr Sridevi NH
MBBS (MD)
Community medicine
chikmagalur
Contents
 Introduction
 Hierarchy of scientific evidence
 Indications for cohort study
 Elements of a cohort study
 Types
 Advantages and disadvantages
 Summary
 References
Introduction
 Cohort study is a type of analytical study.
 Cohort is defined as a group of people who share a
common characteristic or experience within a
defined time period.
what is cohort study???
 Compare - occurrence of events /outcomes among
groups of people whose exposure status differs
‘naturally’.
Indications for cohort study
 There is good evidence of association between
exposure and disease.
 Exposure is rare, but the incidence of disease
among exposed is high.
 When attrition of study population can be minimized.
 When ample of funds are available.
Elements of cohort study
 Defining the study question
 Selecting the study population
 Measure exposure
 Follow up of participants
 Ascertaining outcome
 Analysis and results
Defining the study question
 A study question is framed.
 Important terms should be defined.
 The diagnostic and eligibility criteria of the disease
must be defined beforehand.
Selecting the study population
Selection of study subjects.
Selection of comparison groups.
 Internal comparison ---Common exposure Ex: smoking, tobacco
chewing etc
 External comparison---rare exposure Ex: Industrial chemicals (
different industries), Ionizing radiation
 Comparison with general population.
General
population
Special
groups
Select
groups
Exposure
groups
Formation of cohort
Cohorts/ subjects selected must be:-
Free from Outcome under study.
Must be equally susceptible for the outcome.
Both the groups should be comparable in respect of
all the possible variables, which may influence the
frequency of outcome.
Must be amenable for follow-up with least possible
attrition.
Healthy worker effect
 When exposed group is working population and
Comparison group is general population.
 Working population is healthier
 General population includes- Sick to work.
How to reduce???
• Same workforce – comparison group with different
level of exposure
• Comparison group in different industry
 External comparison group is general population
 people with unhealthy lifestyle refuse to participate
 Volunteers as comparison group.
What if not convinced with comparison group????
 Two or more comparison group
 Industrial chemical and cancer
 Exposed is – workers in industry
 Unexposed – General population, another industry.
Steps
 Defining the study question
 Selecting the study population
 Measure exposure
 Follow up of participants
 Ascertaining outcome
 Analysis and results
Measure exposure
 Interviewing cohort members
 Medical records ,Employment records
 Medical examination/diagnostic procedures.
 Environmental surveys.
 Effect of Ionization radiation on workers
Exposed are workers in the unit.
Unexposed are – those residing in vicinity but not working
in unit
Unexposed group is genuinely unexposed to exposure
 Accurate measurement- to avoid Misclassification
 Exposure status may change during follow up period
 Reassess at periodic interval
Information bias
 Inadequacy of information derived from medical or
other records.
 Difference in Quality and extent of information
obtained for exposed and non exposed groups
 Limited sensitivity and specificity of the diagnostic
tests involved
 Applying tests of differing sensitivities, differentially
to study groups
Steps
 Defining the study question
 Selecting the study population
 Measure exposure
 Follow up of participants
 Ascertaining outcome
 Analysis and results
Follow up of participants
 Periodic medical examination of each member of the
cohort.
 Reviewing physician and hospital records
 Routine surveillance of death records
 Mailed questionnaires, telephone calls, periodic
home visits –preferably all three on an annual basis
 Potential weakness.
Biases
 Lost to follow up due to
 Death
 change of residence
 Migration
 withdrawal of occupation
How to reduce??
 Ensure completeness of follow up – develop
mechanism for that.
 Special group of population can be selected.
 Contact relatives to collect information.
Steps
 Defining the study question
 Selecting the study population
 Measure exposure
 Follow up of participants
 Ascertaining outcome
 Analysis and results
Ascertaining outcome
 Records:-
 Uniformity
 completeness
 method adopted may differ between time periods.
 Interviews/questionnaires:-
 Validity
 reliability
 social desirability.
 Direct measurements:-
 variations
 measurement error.
Sources of data for outcomes
 Outcome fatal – death certificates
 Routine surveillance
 Autopsies
 Medical Records
 Registries
 Insurance data
 Periodic direct medical examination
Bias in assessment of the
outcome
If the investigator who assess outcome is aware of
exposure status of subjects, then the judgment as to
whether the disease developed may be biased by
that knowledge.
 Masking the investigator regarding exposure status.
 Newer modalities of outcome ascertainment
Steps
 Defining the study question
 Selecting the study population
 Measure exposure
 Follow up of participants
 Ascertaining outcome
 Analysis and results
Analysis and results
The data is analyzed in terms of :-
1. Incidence rates of outcome among exposed and
non exposed
2. Estimation of risk
 relative risk
 attributable risk
Incidence rates
 Estimation of disease incidence rates (λ)
 Calculate total number of events observed among all
individuals (D)
 Calculate the sum of the individual observation times (Y)
– person years at risk
 Calculate λ = D/Y  incidence rates
 Rate ratio = λ1/ λ0
 Rate difference = λ1- λ0
Individual’s observation time
 starts when subject joins the study
 stops when
 – Subject develops the disease
 – Subject is lost to follow-up
 – The follow-up period ends
Relative risk
 RR= incidence of disease among exposed
incidence of disease among non exposed
It is the direct measure of the strength of association
between suspected cause and effect.
RR=1 indicates no association
RR>1 indicates positive association
RR<1 indicates negative association
Attributable risk
Absolute difference between the two risk estimates. It estimates the
absolute excess risk associated with a given exposure
 Percent Attributable Risk= Iexp - Inon exp x 100
I exp
 AR% = (RR –1) / (RR) x 100
Population attributable risk
 1953- Levin - first described this measure (smoking
& lung Ca)
 It gives the measure of disease risk in the total
population associated with the exposure
 AR= Iexp – Inonexp
 PAR= Ipop – Inonexp
PAR estimates the amount by which the disease
could be reduced in the population if suspected
factor is eliminated or modified.
Types of cohort study
 Prospective cohort study
 Retrospective cohort study
 Retro-Prospective cohort study
 case cohort study
Prospective cohort study
 More expensive
 Requires long time follow up
Retrospective cohort study
 Here outcomes would have occurred before the start
of the investigations
 The investigator goes back in time, select the study
groups from the records, and traces them forward
through time up to the present.
 Faster, less expensive
 Useful for disease with long period between
exposure and outcome
 Data on exposure and confounders – may be less
accurate
Retro prospective study
 Both the retrospective and prospective elements are
combined.
 Cohort is identified from past records, and is
followed up prospectively for the assessment of
outcome
Case cohort study
Advantages
 Incidence can be calculated
 Several possible outcomes related to exposure can
be studied simultaneously.
 Provide a direct estimate of relative risk
 Dose response ratio can also be calculated.
 Bias can be minimised
 PAR can be calculated.
Disadvantages
 Involve large number of people- unsuitable for
investigating uncommon diseases
 It takes long time to complete study and obtain results
 Administrative problems are inevitable
 Lost to follow up
 It is difficult to introduce new knowledge and
technologies.
 Expensive
 Ethical issues.
 Study itself may alter peoples behavior.
Summary
 Cohort study constitutes the design of reference for all
other designs in observational epidemiology. The
advantage of cohort design is the possibility of studying
the occurrence of several events in addition to that
initially planned.
 Cohort studies have proven to be extremely valuable in
many areas of medical research including cardiovascular
disease, cancer, infectious diseases. Data from these
studies is of great use for the development of public
policy, such as treatment guidelines.
References
1. Park K. Cohort study . Parks text book of
Preventive and social medicine. 23rd edition.
Jabalpur, M/S Banarsidas Bhanot Publishers.
January 2015 page 75-80
2. Leon Gordis. Cohort studies. Epidemiology 4th
edition. Elsevier. 2009 page 167-176.
3. Roger detels. Oxford textbook of public health. 5th
edition. Oxford 2009 page 508-525
4. Textbook of public health and community medicine
1st edition 2009. AFMC page 147-150
Cohort study

Cohort study

  • 1.
    Dr Sridevi NH MBBS(MD) Community medicine chikmagalur
  • 2.
    Contents  Introduction  Hierarchyof scientific evidence  Indications for cohort study  Elements of a cohort study  Types  Advantages and disadvantages  Summary  References
  • 3.
    Introduction  Cohort studyis a type of analytical study.  Cohort is defined as a group of people who share a common characteristic or experience within a defined time period.
  • 4.
    what is cohortstudy???  Compare - occurrence of events /outcomes among groups of people whose exposure status differs ‘naturally’.
  • 7.
    Indications for cohortstudy  There is good evidence of association between exposure and disease.  Exposure is rare, but the incidence of disease among exposed is high.  When attrition of study population can be minimized.  When ample of funds are available.
  • 8.
    Elements of cohortstudy  Defining the study question  Selecting the study population  Measure exposure  Follow up of participants  Ascertaining outcome  Analysis and results
  • 9.
    Defining the studyquestion  A study question is framed.  Important terms should be defined.  The diagnostic and eligibility criteria of the disease must be defined beforehand.
  • 10.
    Selecting the studypopulation Selection of study subjects. Selection of comparison groups.  Internal comparison ---Common exposure Ex: smoking, tobacco chewing etc  External comparison---rare exposure Ex: Industrial chemicals ( different industries), Ionizing radiation  Comparison with general population. General population Special groups Select groups Exposure groups
  • 11.
    Formation of cohort Cohorts/subjects selected must be:- Free from Outcome under study. Must be equally susceptible for the outcome. Both the groups should be comparable in respect of all the possible variables, which may influence the frequency of outcome. Must be amenable for follow-up with least possible attrition.
  • 12.
    Healthy worker effect When exposed group is working population and Comparison group is general population.  Working population is healthier  General population includes- Sick to work. How to reduce??? • Same workforce – comparison group with different level of exposure • Comparison group in different industry
  • 13.
     External comparisongroup is general population  people with unhealthy lifestyle refuse to participate  Volunteers as comparison group. What if not convinced with comparison group????  Two or more comparison group  Industrial chemical and cancer  Exposed is – workers in industry  Unexposed – General population, another industry.
  • 14.
    Steps  Defining thestudy question  Selecting the study population  Measure exposure  Follow up of participants  Ascertaining outcome  Analysis and results
  • 15.
    Measure exposure  Interviewingcohort members  Medical records ,Employment records  Medical examination/diagnostic procedures.  Environmental surveys.
  • 16.
     Effect ofIonization radiation on workers Exposed are workers in the unit. Unexposed are – those residing in vicinity but not working in unit Unexposed group is genuinely unexposed to exposure  Accurate measurement- to avoid Misclassification  Exposure status may change during follow up period  Reassess at periodic interval
  • 17.
    Information bias  Inadequacyof information derived from medical or other records.  Difference in Quality and extent of information obtained for exposed and non exposed groups  Limited sensitivity and specificity of the diagnostic tests involved  Applying tests of differing sensitivities, differentially to study groups
  • 18.
    Steps  Defining thestudy question  Selecting the study population  Measure exposure  Follow up of participants  Ascertaining outcome  Analysis and results
  • 19.
    Follow up ofparticipants  Periodic medical examination of each member of the cohort.  Reviewing physician and hospital records  Routine surveillance of death records  Mailed questionnaires, telephone calls, periodic home visits –preferably all three on an annual basis  Potential weakness.
  • 20.
    Biases  Lost tofollow up due to  Death  change of residence  Migration  withdrawal of occupation
  • 21.
    How to reduce?? Ensure completeness of follow up – develop mechanism for that.  Special group of population can be selected.  Contact relatives to collect information.
  • 22.
    Steps  Defining thestudy question  Selecting the study population  Measure exposure  Follow up of participants  Ascertaining outcome  Analysis and results
  • 23.
    Ascertaining outcome  Records:- Uniformity  completeness  method adopted may differ between time periods.  Interviews/questionnaires:-  Validity  reliability  social desirability.  Direct measurements:-  variations  measurement error.
  • 24.
    Sources of datafor outcomes  Outcome fatal – death certificates  Routine surveillance  Autopsies  Medical Records  Registries  Insurance data  Periodic direct medical examination
  • 25.
    Bias in assessmentof the outcome If the investigator who assess outcome is aware of exposure status of subjects, then the judgment as to whether the disease developed may be biased by that knowledge.  Masking the investigator regarding exposure status.  Newer modalities of outcome ascertainment
  • 26.
    Steps  Defining thestudy question  Selecting the study population  Measure exposure  Follow up of participants  Ascertaining outcome  Analysis and results
  • 27.
    Analysis and results Thedata is analyzed in terms of :- 1. Incidence rates of outcome among exposed and non exposed 2. Estimation of risk  relative risk  attributable risk
  • 28.
    Incidence rates  Estimationof disease incidence rates (λ)  Calculate total number of events observed among all individuals (D)  Calculate the sum of the individual observation times (Y) – person years at risk  Calculate λ = D/Y  incidence rates  Rate ratio = λ1/ λ0  Rate difference = λ1- λ0
  • 29.
    Individual’s observation time starts when subject joins the study  stops when  – Subject develops the disease  – Subject is lost to follow-up  – The follow-up period ends
  • 31.
    Relative risk  RR=incidence of disease among exposed incidence of disease among non exposed It is the direct measure of the strength of association between suspected cause and effect. RR=1 indicates no association RR>1 indicates positive association RR<1 indicates negative association
  • 32.
    Attributable risk Absolute differencebetween the two risk estimates. It estimates the absolute excess risk associated with a given exposure
  • 33.
     Percent AttributableRisk= Iexp - Inon exp x 100 I exp  AR% = (RR –1) / (RR) x 100
  • 34.
    Population attributable risk 1953- Levin - first described this measure (smoking & lung Ca)  It gives the measure of disease risk in the total population associated with the exposure  AR= Iexp – Inonexp  PAR= Ipop – Inonexp
  • 36.
    PAR estimates theamount by which the disease could be reduced in the population if suspected factor is eliminated or modified.
  • 37.
    Types of cohortstudy  Prospective cohort study  Retrospective cohort study  Retro-Prospective cohort study  case cohort study
  • 38.
    Prospective cohort study More expensive  Requires long time follow up
  • 39.
    Retrospective cohort study Here outcomes would have occurred before the start of the investigations  The investigator goes back in time, select the study groups from the records, and traces them forward through time up to the present.  Faster, less expensive  Useful for disease with long period between exposure and outcome  Data on exposure and confounders – may be less accurate
  • 40.
    Retro prospective study Both the retrospective and prospective elements are combined.  Cohort is identified from past records, and is followed up prospectively for the assessment of outcome
  • 42.
  • 43.
    Advantages  Incidence canbe calculated  Several possible outcomes related to exposure can be studied simultaneously.  Provide a direct estimate of relative risk  Dose response ratio can also be calculated.  Bias can be minimised  PAR can be calculated.
  • 44.
    Disadvantages  Involve largenumber of people- unsuitable for investigating uncommon diseases  It takes long time to complete study and obtain results  Administrative problems are inevitable  Lost to follow up  It is difficult to introduce new knowledge and technologies.  Expensive  Ethical issues.  Study itself may alter peoples behavior.
  • 45.
    Summary  Cohort studyconstitutes the design of reference for all other designs in observational epidemiology. The advantage of cohort design is the possibility of studying the occurrence of several events in addition to that initially planned.  Cohort studies have proven to be extremely valuable in many areas of medical research including cardiovascular disease, cancer, infectious diseases. Data from these studies is of great use for the development of public policy, such as treatment guidelines.
  • 46.
    References 1. Park K.Cohort study . Parks text book of Preventive and social medicine. 23rd edition. Jabalpur, M/S Banarsidas Bhanot Publishers. January 2015 page 75-80 2. Leon Gordis. Cohort studies. Epidemiology 4th edition. Elsevier. 2009 page 167-176. 3. Roger detels. Oxford textbook of public health. 5th edition. Oxford 2009 page 508-525 4. Textbook of public health and community medicine 1st edition 2009. AFMC page 147-150

Editor's Notes

  • #14 You have take a comparable group But not convinced about its comparability Feel some confounding factors are still there