A short introduction to epidemiology Chapter 2a: Conducting a cohort study

1
A short introduction to epidemiologyChapter 2a
Conducting a cohort study

• Neil Pearce
• Centre for Public Health Research
• Massey University
• Wellington, New Zealand

2
Chapter 2 (additional material)Cohort studies

• This presentation includes additional material on
  conducting a cohort study
• It particularly focuses on occupational studies,
  because these often have good historical exposure
  data
• Brief mention is also made of other types of
  cohort studies
• More information on data analysis is given in
  chapter 9

3
Chapter 2 (additional material)Cohort studies

• Defining the cohort
• Defining exposure
• Follow-up
• Data analysis

4
Birth
End of Follow up
Death other death lost to follow up
non-diseased symptoms severe disease
5
A Hypothetical Cohort Study
6
Defining the Cohort

• Cohort studies are most frequently conducted in
  two different contexts
• Studies based on a particular community (e.g.
  Framingham, birth cohort studies)
• Studies based on a particular occupational group
  (e.g. lung cancer in asbestos workers)
• Each type of study may involve an external
  comparison (e.g. with national mortality rates)
  or an internal comparison

7
Defining the Cohort

• What is the appropriate comparison population for
  an external comparison?
• What is the most appropriate geographical area?
  (national, regional, local)
• What is the most appropriate population within
  this area? (all persons, all employed persons,
  same social class)

8
The Healthy Worker Effect

• William Ogle (1885)
• Some occupations may repel, while others
  encourage the unfit at the age of starting work.

9
Employed
Non-employed
10
The Healthy Worker Effect

• Healthy individuals are more likely to obtain
  employment
• Unhealthy individuals are more likely to leave
  employment
• The health worker effect is particularly strong
  for heart disease and non-malignant respiratory
  disease, and is generally weaker for cancer

11
The Healthy Worker Effect

• Checkoway et al (1985). Mortality in energy
  research laboratory workers
• Cause Observed Expected SMR
• Cancer 194 250.0 0.78
• Heart disease 344 459.9 0.75
• Respiratory 42 69.2 0.61
• Other 386 540.9 0.71
• Total 966 1320.0 0.73

12
Defining an Occupational Cohort

• All workers ever employed in one factory
• Workers from multiple plants, engaged in the same
  industrial process
• Members of a trade union or professional
  organisation
• Registered cases of occupational disease (e.g.
  asbestosis)

13
Defining an Occupational Cohort

• Cohort enumeration
• Plant personnel records
• Union membership listings
• Other data sources

14
Example of an Employment Record
15
Defining an Occupational Cohort

• Cohort restriction
• All employees
• Workers first employed on or after a particular
  date
• Gender, ethnicity
• Workers with a minimum employment duration (e.g.
  one month)

16
Defining the Cohort
17
Chapter 2 (additional material)Cohort studies

• Defining the cohort
• Defining exposure
• Follow-up
• Data analysis

18
Exposure and dose

• Exposure the presence of a substance in the
  environment external to the worker
  (external/environmental)
• Dose The amount of a substance that reaches
  susceptible targets in the body (internal)

19
Measures of exposure

• Intensity of exposure
• Duration of exposure
• Cumulative exposure

20
Types of exposure data used in occupational
studies

• Quantified personal measurements
• Quantified area- or job-specific data
• Ordinally ranked jobs or tasks
• Duration of employment in the industry
• Ever employed in the industry

21
Chapter 2 (additional material)Cohort studies

• Defining the cohort
• Defining exposure
• Follow-up
• Data analysis

22
Start of Follow-up

• Follow-up starts on the first date that each
  worker satisfies the eligibility criteria for the
  cohort, I.e. the most recent date of
• Start of employment (plus minimum employment
  period, e.g. one month)
• Date of start of study

23
Vital Status

• Deaths ( or e.g. cancer registrations) can be
  identified through national records
• For non-deceased participants, vital status can
  be identified through sources such as
• Employment records
• Superannuation records
• Electoral rolls
• Drivers license records
• Postal questionnaire

24
End of Follow-up

• Follow-up ends on the last date that each worker
  satisfies the eligibility criteria for the
  cohort, I.e. the earliest date of
• Date of death
• Date of emigration
• Last date known to be alive
• Date of finish of study

25
Follow-Up
26
Employment History of a Worker

• Year 60 61 62 63 64 65 66 67
• Age 27 28 29 30 31 32 33 34 (at
  risk)Employment 1 1 1 1 1 0 0 0 status
• Follow-up 0 1 2 3 4 5 6 7

27
Accumulation of Person-Years
28
Chapter 2 (additional material)Cohort studies

• Defining the cohort
• Defining exposure
• Follow-up
• Data analysis

29
Data Analysis

• Each worker accumulates person-years of follow-up
  in categories of age, calendar period (and
  gender)
• These are used to calculate the expected numbers
  of deaths from all causes and from specific
  causes
• These are then compared to the observed deaths in
  the cohort

30
Data Analysis
31
Data Analysis

• We wish to compare the death rates in this cohort
  with those in some standard external comparison
  population (usually the national population)
• This involves calculating the expected number of
  deaths in the cohort if it had had the same death
  rate as the comparison population

32
Data Analysis
33
Strategies of Data Analysis

• Overall cohort analysis
• assumes that all members of cohort are exposed
• cohort can be stratified by duration of employment

34
Strategies of Data Analysis

• Subcohort analysis
• job categories
• duration of employment within job categories
• ordinally ranked job categories
• cumulative exposure

35
Ever Employed in the IndustrySMRs Amongst While
Male Phosphate Industry Workers 1949-78
(Checkoway Et Al, 1985)

• Cause of death Observed Expected SMR
• All causes 1,620 1,623.8 1.00
• Lung cancer 117 95.9 1.22

36
Duration of EmploymentLung Cancer Amongst White
Male Phosphate Workers 1949-78 (Checkoway Et Al,
1985)

• Years of work Lung Cancers SMR
• 1-4 29 1.36
• 5-9 17 1.18
• 10-19 29 1.09
• 20-29 25 1.05
• 30 17 1.62

37
Unranked JobsRespiratory Disease Mortality in
Metal Trades Occupations (Beaumont and Weiss,
1980)

• Job category Lung Cancers SMR
• Welders 53 1.31
• Ship Fitters 12 0.57
• Burners 11 1.57
• Boilermakers 14 1.98

38
Ordinally Ranked JobsLung Cancer Amongst
Amosite Asbestos Factory Workers (Acheson Et Al,
1984)

• Exposure Lung Cancers SMR
• Background 11 1.06
• Low 12 1.34
• Medium 41 2.25
• Heavy 8 4.25

39
Cumulative ExposureLung Cancer Amongst
Chrysotile Asbestos Textile Workers (Dement Et
Al, 1983)

• Cumulative exposure Lung SMR(fibers/cc x
  days) Cancers
• lt1,000 5 1.33
• 1,000-9,999 10 2.69
• 10,000-39,999 7 3.20
• 40,000-99,000 11 10.00
• 100,000 2 14.93

40
A short introduction to epidemiologyChapter 2a
Conducting a cohort study

• Neil Pearce
• Centre for Public Health Research
• Massey University
• Wellington, New Zealand