Errors

1
Putting Theory into Practice

Lecturer Alison Thornton
(Research Fellow) Course, Module BSc,
Epidemiology/Statistics Module Date
14th December 2007
2
Overview of Session

• Putting Theory into Practice
• Part I
• Reminder of epidemiological study designs
  (observational)
• Background for group task Gulf War Syndrome
• Group task
• Part II
• Presentation of group protocols
• Examples from literature
• Methodological Challenges
• Conclusions

3
Putting Theory into Practice

• PART I

4
Epidemiology

• The study of the occurrence, distribution and
  determinants of disease (and other health-related
  conditions) in populations.

5
Epidemiology

• The study of how often diseases occur in
  different groups of people and why.
• (Coggon, 2003)

6
Study Designs in Epidemiology
STUDY DESIGNS
Descriptive
Analytical
Case report
Case series
Experimental
Observational
Cross-sectional

• Cross-sectional
• Cohort
• Case-control
• Migrant
• Ecological
• Nested case-control

Clinical trial
Field trial
7
Observational Epidemiology

• Observational epidemiology involves the
  comparison of groups of people.
• What factors distinguish people with and without
  disease?
• Compare those with and without risk factors
• and/or those with and without the disease.

8
Observational Study Designs

• Cross-sectional
• Case-control
• Cohort
• Migrant
• Ecological
• Nested case-control

9
Main Observational Study Designs I

• Cross-sectional survey
• An analytical epidemiological study in which the
  exposure and disease status (outcome) are
  assessed simultaneously at a fixed point in time.
• e.g.
• Psychological distress and low back pain
  (questionnaire).

10
Main Observational Study Designs II

• Case-control study
• An analytical study in which participants are
  defined according to their disease status.
• Known disease status (outcome).
• Compare prior exposures.
• e.g.
• Risk factors for Sudden Infant Death Syndrome
  (SIDS).

11
Main Observational Study Designs III

• Cohort study
• An analytical study in which participants are
  defined according to their exposure status.
• Known exposure
• Assess (future) disease status (outcome).
• e.g.
• Study of pain in children (multiple exposures and
  outcomes).

12
Direction of inquiry
FUTURE
PRESENT
PAST
FUTURE
PRESENT
PAST
Time
Assess exposure
Cross
-
sectional study
Cross
-
sectional study
and outcome
Known
Assess
Assess
Case
-
control study
Case
-
control study
outcome
exposure
exposure
Known
Assess
Assess
Prospective cohort
Prospective cohort
exposure
outcome
outcome
Known
Assess
Retrospective cohort
Retrospective cohort
exposure
outcome
13
Observational Study Designs

• Nested case-control Study
• Cases and controls drawn from the pre-defined
  cohort.
• Cases participants who developed the outcome.
• Controls random sample of participants who did
  not develop the outcome.
• e.g.
• Cancer and exposure to electromagnetic fields.

14
Group Task

• Ill-Health Amongst Veterans of
• The Gulf War
• Gulf War Syndrome ?

15
Group Task - Background

• The Gulf War 1990
• August 1990 - Iraqi invasion of Kuwait
• UN Security Council demand immediate withdrawal.
  Trade boycott.
• UN Security Council ultimatum (November 1990).
  Iraq to withdraw from Kuwait by 15 Jan 1991.

16
Group Task - Background

• The Gulf War 1991
• Iraq ignores ultimatum 15 January 1991
• Operation Desert Storm
• Launched 17 January 1991
• 580,000 allied troops
• Saddam Hussein orders Iraqi troop withdrawal from
  Kuwait 26 February 1991
• The 100 hours war
• UK involvement
• 52,300 UK troops deployed
• Majority withdrawn by April 91

17
Reports of Ill-Health UK Veterans

• UK VETERANS
• Spring 1993 the MoD first become aware of
  possible unexplained illnesses amongst UK Gulf
  War Veterans.
• The symptoms which these early presenting
  veterans complained of included
• Tiredness
• Muscle and joint pains
• Irritability
• Sleep disturbances
• Short-term memory loss

18
Reports of Ill-Health UK Veterans

• UK VETERANS
• Government set up a Medical Assessment Programme
  (MAP). Available to Gulf War veterans who were
  concerned that they may be suffering ill-health
  as a result of service.
• Veterans attributed their ill-health to several
  exposures encountered prior to, or during the
  course of, the war
• immunisations
• nerve agent prophylaxis
• organophosphate pesticides

19
Reports of Ill-Health US Veterans

• US VETERANS
• Reports of ill-health also arose in United States
  veterans.
• 62 different categories of symptoms.
• Most common symptoms similar to those in UK.
• The US Department of Defence initiated a
  programme of
• Medical investigation programmes
• Preliminary epidemiological studies
• Independent reviews
• Comprehensive Clinical Evaluation Program (CCEP)
  June 1994

20
Symptom Frequency (CCEP)
Chief complaint
Any complaint


Joint pain
11
49
Fatigue
10
47
Headache
7
39
Memory loss
4
34
Sleep disturbance
2
32
21
Primary Diagnoses (CCEP)
total
Healthy
9.7
Tension Headache
3.4
Fatigue, not specified as chronic
3.3
Depressive disorder not cl. elsewhere
2.9
Prolonged posttraumatic stress disorder
2.8
Headache
2.7
22
Possible Reasons for Ill-health amongst Gulf War
Veterans
23
Reports of Ill-Health from Medical Assessment
Programmes

• Heterogeneous group of veterans different
  complaints and different exposures?
• Self-referred subgroup cannot extrapolate to
  the wider veteran population.
• No suitable comparison group within the
  programme.
• In a large cohort of service personnel, many
  would have become ill in the period after the war
  anyway, regardless of deployment.

24
Group Task

• Outline a protocol (which may involve more than
  one study)
• To determine whether there is excess mortality or
  morbidity amongst personnel who served in the
  Gulf War.
• And
• If an excess is present, to ascertain the nature
  and possible causes of ill-health.

25
Hint

• Which may involve more than one study
• What is the research question (or questions)?

26
Things to think about

• Descriptive or analytical does it involve
  comparisons?
• Appropriate study design (s)
• Define outcome.
• Define exposure.
• Possible confounders.
• Study population, sampling frame, sampling
  methods.
• Data collection methods and sources - how to
  measure exposure and outcome?
• Tracing subjects, maximising participation.
• Possible biases?

27
Putting Theory into Practice

• PART II

28
Putting Theory into Practice

• Group Protocols

29
Research Questions

• Compared to similar non-deployed personnel, is
  there excess mortality amongst Gulf War Veterans?
• Compared to similar non-deployed personnel, is
  there excess morbidity amongst Gulf War Veterans?
• What is the nature of the excess morbidity? (Is
  there a unique identifiable syndrome?)
• What are the causes of the excess morbidity?

30
Mortality

• Compared to similar non-deployed personnel, is
  there excess mortality amongst Gulf War Veterans?

31
Excess Mortality?

• Macfarlane GJ, Thomas E, Cherry N.
• Mortality among UK Gulf War veterans.
• Lancet 2000 35617- 21
• Aims To explore whether Gulf War veterans have a
  higher mortality rate than similar
  non-deployed personnel
• Design Retrospective cohort study
• Subjects 53,462 UK Gulf War veterans (Gulf
  cohort)
• Non-deployed reference group (Era cohort)
• Matched for age, sex, rank, service, fitness for
  service
• Exposure Served in Gulf War
• Outcome Deaths (all causes) after end of war
• 8 years (1 April 1991 31 March 1999)
• ONS / NHS Central Register

32
Excess Mortality?

• Results
• Cohorts matched for sex, age, rank and service
• 97 male
• 72 lt 30 years
• 70 army
• 18 officers other ranks
• Mortality
• Gulf cohort
• 395 deaths
• Mortality rate 0.74 (94 deaths per 100, 000
  person years)
• Era cohort
• 378 deaths
• Mortality rate 0.71 (90 deaths per 100,000
  person years)

33
Excess Mortality?

• Mortality Rate Ratio (MRR)
• 1.05 (95 CI, 0.91 1.21)
• Mortality in Gulf cohort higher for external
  causes, but lower for disease-related causes.
• No significant difference in cancer deaths.
• Discussion / Conclusions
• Slight but non-significant excess in mortality
  amongst Gulf War veterans
• Not disease-related
• Similar findings to US Gulf veterans (Kang et al,
  1996)
• Similar to patterns of mortality after other
  conflicts

34
Excess Mortality?

• Limitations?
• Use of routine data
• Length of follow-up
• Further research
• Continued follow-up to monitor mortality

35
Excess Mortality Cancer?

• Macfarlane GJ, Biggs AM et al
• Incidence of cancer among UK Gulf war veterans
  cohort study.
• BMJ 20033271373-5
• To determine whether cancer incidence rates are
  higher in UK Gulf War veterans than in
  non-deployed personnel
• Further follow up of same cohorts to 31 July 2002
• Outcome Incident cancers (NHS central registers)
• Results
• Incident cancers
• Gulf War cohort 270
• Era cohort 269
• Incidence Rate Ratio (IRR) 0.99 (95 CI 0.83
  1.17)

36
Excess Mortality Cancer?

• Adjustment for confounders
• Smoking, alcohol
• Did not alter results
• Specific exposures explored
• Multiple vaccinations, pesticides, depleted
  uranium
• No association between these exposures and risk
  of cancer
• Conclusion No excess risk of cancer overall or
  of site specific cancers
• Further research Continued follow-up (long
  latent period)

37
Morbidity

• Compared to similar non-deployed personnel, is
  there excess morbidity amongst Gulf War Veterans?
• What is the nature of ill-health?

38
Morbidity

• Cherry N, Creed F, Silman A, Dunn G, Baxter D,
  Smedley J, Taylor S, Macfarlane GJ.
• Health and exposures of United Kingdom Gulf war
  veterans I. The pattern and extent of
  ill-health. II. The relation of health to
  exposure.
• Occupational and Environmental Medicine
  200158291-306
• Aims Part I
• To compare health of UK Gulf War veterans with
  similar personnel not deployed
• To describe patterns and estimate extent of ill
  health in both groups
• (Looking for unusual cluster of symptoms
  specific to experience in Gulf)

39
Morbidity

• Design Retrospective Cohort Study
• Subjects 4795 Gulf War veterans (Main Gulf
  Cohort)
• 4793 Gulf War veterans (Validation Gulf
  Cohort)
• 4790 Non-deployed (Non-Gulf Cohort)
• Exposure Served in Gulf War

40
Morbidity

• Outcome
• Self-reported symptoms
• Postal questionnaire
• Mean symptom severity score
• Pain and numbness manikins
• Widespread pain? Toxic neuropathy?
• Cluster analysis
• Principal component factor analysis

41
Morbidity

• Results
• 85.5 response rate
• Mean symptom scores higher in Gulf veterans (3.0
  vs 1.7)
• Widespread pain more common in veterans (12.2
  vs 6.5)
• Symptoms suggestive of peripheral neuropathy more
  common in veterans (12.5 vs 6.8)
• Discussion / Conclusion
• No excess in signs of severe morbidity
  (hospitalisations etc)
• No group of symptoms unique to Gulf War veterans
• BUT
• 7 years post war, veterans more troubled by ill
  health than non-veterans.

42
Possible causes of ill-health?

• What are the possible causes of ill-health?

43
Possible causes of ill-health?

• Aims Part II
• To explore whether self-reported exposures were
  related to symptoms in a way that was consistent,
  specific and credible.

44
Possible causes of ill-health?

• Associations between exposures and symptoms
  generally weak. However
• Number of
• Immunisations
• Days handling pesticides
• Days exposed to smoke from oil fires
• were associated with severity of symptoms.

45
Possible causes of ill-health?

• Immunisations associated with high scores for
  skin and musculoskeletal complaints.
• Days handling pesticides associated with symptoms
  of toxic neuropathy.
• Considered to warrant further investigation
• ?Nested case control studies.

46
Methodological Challenges I

• Delay in commissioning of research
• High turnover in Armed Forces
• ? Difficulties tracing and contacting subjects
• Accuracy of exposure information?
• Recall
• Records e.g. immunisations
• Impact of media reports
• Exposure recall
• Symptom reporting
• (Hotopf M, Wessely S. 2005)

47
Methodological Challenges II

• Response rates
• Young men
• Mobile population
• Symptomatic (Gulf veterans) more likely to
  respond?
• Non-working more likely to respond?
• Comparison group
• Physical fitness of military personnel
• Risk-taking behaviours
• Smoking and alcohol
• ? Non military comparison group not applicable
• Healthy warrior effect (negative mortality
  findings)
• (Hotopf M, Wessely S. 2005)

48
Methodological Challenges II

• Defining outcome (case definition)
• Subjective self-reported outcome.
• Complex array of symptoms but no well-defined
  diagnoses.
• Is this a new disorder? Gulf War Syndrome?
• ? rate symptoms
• BUT
• Pattern of symptoms similar amongst veterans and
  comparison groups.
• (Hotopf M, Wessely S. 2005)

49
Conclusions

• No excess in disease-related mortality in Gulf
  War veterans compared to non-deployed controls.
• Higher prevalence of self-reported illnesses
  amongst Gulf War veterans, but symptoms do not
  constitute a single unique syndrome.
• (Hotopf M, Wessely S. 2005)

50
References

• Coggon D, Rose G, Barker D. Epidemiology for the
  Uninitiated (5th Edition). BMJ Publishing Group
  2003.

51
References (Gulf War)

• Hotopf M, Wessely S. Can epidemiology clear the
  fog of war? Lessons from the 1990 91 Gulf War.
  Int J Epidemiol. 2005 Aug 34(4) 791-800.
• Macfarlane G.J., Thomas E., Cherry N. Mortality
  among UK Gulf War Veterans. Lancet 200035617-21
• Macfarlane GJ, Biggs A-M, Maconochie N, Hotopf M,
  Doyle P, Lunt M. Incidence of cancer among UK
  Gulf war veterans cohort study. BMJ
  20033271373-5
• Cherry N, Creed F, Silman A, Dunn G, Baxter D,
  Smedley J, Taylor S, Macfarlane GJ. Health and
  exposures of United Kingdom Gulf war veterans I.
  The pattern and extent of ill-health. II. The
  relation of health to exposure. Occupational and
  Environmental Medicine 200158291-306