Introduction to Epidemiology Basic Principles 2

1
Introduction to Epidemiology Basic Principles 2
Martin Frisher Department of Medicines
Management Keele University
2
Learning Objectives

• This morning,
• Risk and Outcomes, Measures of Risk, Cohort and
  Cross Sectional Studies, Confounding/Bias
• This afternoon,
• Research Design
• Reliability and Validity
• Statistics and Data Analysis
• Epidemiology and Causality
• Substance Misuse Epidemiology

3
Types of Research Design

• Laboratory, experimental
• Non-experimental intervention (RCT/non RCT)
• Population survey
• Subgroup survey
• Qualitative study
• Case study
• Review (systematic, selective)
• Secondary database analysis

4
Deductive Logic Of David Hume
ACCEPT AS
5
Inductivist Philosophy (J.S.Mill)

• Inductive reasoning based on intuition that each
  event is followed by an effect
• Observations drawn from hypotheses are called
  inductions
• Conclusions are drawn from inductions

6
Reasoning Examples

• Throughout history people repeat the same
  mistakes, so we can conclude that mistakes will
  be made in the future.
• The whale is a mammal, so all killer whales are
  mammals.
• All killer whales are mammals, so the whale is a
  mammal

7
Hills Checklist For Judging Causality

• Temporality
• Strength
• Specificity
• Consistency
• Coherence
• Biological Plausibility
• Analogy

8
MAIN CONSIDERATIONS

• Temporal sequence of the association i.e.
  whether the cause is preceding the effect or
  not, has to be searched first. If it is
  present, it is more in favor of causal
  association.
• Then the strength of the association (the
  relative risk/ odds ratio and dose response
  relationship) which decides the power of the
  association between the cause and effect has to
  be determined. If relative risk is high, the
  association is more likely to be a causal one.

9
J. S. Mills (1856)Methods Of Induction

• Agreement if a factor is common to a number
  of different circumstances, that are associated
  with the presence of a disease, that factor may
  be the cause of disease that means there is an
  agreement between the factor and the disease
  under different circumstances.
• Difference if the frequency of a disease
  is markedly different under two different
  circumstances and some factors can be identified
  in one circumstance not in other, then the factor
  or its absence, may be the cause of disease.
• Concomitant variation factor whose
  frequency or strength varies with that of the
  disease, it may be the cause of the disease.

10
Qualitative vs. Quantitative
11
OBSERVATION VS. EXPERIMENTS

• Advantages
• less intrusive
• less likely to create artificial behaviour

• Disadvantages
• control over extraneous variables are less rigid
• harder to establish cause and effect

12
Research Paradigms
Cohen, L, Manion, L. Morrison, K.(2000).
Research Methods in Education, 5th Ed, Routledge
Falmer.
13
Research Paradigms

• Positivist
• focus on facts
• look for causality and fundamental laws
• reductionist
• hypothetico-deductive logic

• Interpretive
• focus on meanings
• look for under-standing in context
• integrationist
• inductive logic

14
NON EXPERIMENTAL DESIGNS

• Experiment may create behaviour which is too
  artificial.
• Observational data may be audio, visual or
  written.
• Classification of behaviour rating scale.
• Sampling of behaviour time, point and event
  sampling.
• Measures must be operationalised, i.e. clearly
  defined rating scales.
• Ensure reliability (e.g. inter-observer
  reliability).
• Analysis of qualitative data can and should be
  done using explicit, systematic, and reproducible
  methods

15
Examples of Pharmacy M.Sc. projects
16
Levels of measurement

• Nominal data
• Names indicate a classification, groups are
  discrete
• There is both no overlap between classifications
  and no intermediary values
• Examples gender (male or female) blood group
  (O, A, B, AB).
• Ordinal data
• Objects of a set can be rank-ordered
• Consumer satisfaction low medium high"
• 1
  2 3
• The numbers do not indicate absolute quantities
  nor do they indicate that the intervals between
  the numbers are equal nor do they have a zero
  point
• Examples most psychological or drug dependence
  scores

17
Types of Data - Continuous

• Interval
• Numerically equal distances represent equal
  distances in the property being measured
• A B C D E
• 1 2 3 4 5
• i.e., level of school achievement (5-23)
  (4-13)
• but we cannot say that the achievement of D is
  twice that of B
• Ratio
• Has an absolute or natural zero (i.e. the object
  has none of the property)
• Examples money, height, number of overdoses (?)

18
What type of scale?

• In memory experiments, the dependent variable is
  often the number of items correctly recalled.
  What scale of measurement is this?
• You could reasonably argue that it is a ratio
  scale.
• First, there is a true zero point some subjects
  may get no items correct at all. Moreover, a
  difference of one represents a difference of one
  item recalled across the entire scale. It is
  certainly valid to say that someone who recalled
  12 items recalled twice as many items as someone
  who recalled only 6 items.

19
Number of Items

• Number-of-items is a more complicated case than
  it appears at first. Consider the following
  there are 5 easy items and 5 difficult items
• Half of the subjects are able to recall all the
  easy items and different numbers of difficult
  items while
• The other half of the subjects are unable to
  recall any of the difficult items and remember
  different numbers of easy items.

20
Measures of Central Tendency

• MODE
• is the most frequently occurring value (or
  values). It is generally used for categorical
  (nominal) data
• MEDIAN
• when the data is placed in order, it is the
  middle value it is generally used for ordinal
  data, since it is based on ranking information.
  Also used for data which has a skewed
  distribution
• MEAN
• is the sum of all observed values, divided by
  the number of values. It is generally used for
  numerical data from symmetrical distributions

21
Skewed distributions

• Skewness refers to the asymmetry of the
  distribution

• A positively skewed distribution is asymmetrical
  and points in the positive direction.
• A) mode lt B) median lt C) mean

A) Mode 70,000 B) Median 88,700 C) Mean
93,600
A
B
C
22
Measures of Central Tendency
Measurement Scale Best Measure of the
"Middle Nominal (Categorical)
Mode Ordinal Median Interval/ Symmetrical
data Mean Ratio Skewed data Median
23
Normal Distribution-Central Tendency
24
T-Test
The t-test assesses whether the means of two
groups are statistically different from each
other. This analysis is appropriate whenever you
want to compare the means of two groups.
25
T-Test
t difference between groups
sampling variability (within groups)

• When the value on the top of the equation is
  large, or the value on the bottom of the equation
  is small, the overall ratio will be large.
• The larger the value of t, the farther out on the
  sampling curve it will be, and, thus, the more
  likely it will be significant

26
Variance
27
Purpose of Statistics

• In any comparison in a medical context,
  differences are almost bound to occur. The
  problem is separating real effects from random
  variation.
• The researcher must decide how much variation
  should be ascribed to chance and how much is a
  real effect.

28
Interpretation of Probabilities
29
Steps in Quantitative Analysis

• Step 1 quantitative research design
• (E.G. Hypothesis, sampling, samples sizes)
• Step 2 Your data - (entering and coding data)
• Access to SPSS
• Entering data, some issues an example
• Step 3 Analysing (from data to information)
• Why use statistics? Descriptive versus
  inferential.
• How to choose a statistical test?
• Examples of parametric (t-test, ANOVA,
  regression, correlation)
• Examples of non-parametric tests (Rank
  correlation and Chi-squared)

30
Using SPSS to calculate Odds Ratio and Relative
Risk
31
Qualitative Research Exploring patients and
practitioners' beliefs about the causality and
expectations for treatment of chronic
musculo-skeletal pain

• Patients and health professionals experience
  and some previous research suggests that
• patients with persistent musculoskeletal pain can
  be dissatisfied with the care that they receive
  for their pain and
• that health professionals can be dissatisfied
  with what help they are able to offer their
  patients with persistent musculoskeletal pain.
• A review of previous studies that have explored
  these issues
• A postal questionnaire survey to a random sample
  of 5,940 people in the Southern England and
• Interview studies with a) patients who reported
  persistent pain in their questionnaire and b)
  with the health professionals (both NHS and
  private) consulted by these patients.

http//www.mrc-gprf.ac.uk/maindocs/research
32
Qualitative Research Sample Size and
Justification

• 1 Around 20 of the population suffer from
  chronic musculoskeletal pain
• 2 Approximately three-quarters of our subjects
  will come from those individuals with pain grades
  II-IV on the Chronic Pain Grade Questionnaire
  (CPG)
• 3 To ensure that enough individuals are
  identified with more severe problems and who are
  willing to participate, we require a sampling
  frame (for the qualitative study) of 60 potential
  research subjects in CPG grade I and 180 in CPG
  grades II-IV

33
Sample Size and Justification

• 4 Assuming that half of those identified in CPG
  grades II-IV are suitable for the study and
  interested in participating, around 5 of
  questionnaire respondents will be potential
  subjects for the interview study
• 5 If the response rate to the postal
  questionnaire is 70, for each 1000 people
  approached 35 potential research subjects in CPG
  grades I-IV will be identified.
• 6 Thus if we approach 5,400 adults, 300 from each
  of 18 GPRF practices, we should be able to
  identify 189 potential subjects in CPG grades
  II-IV

34
Reliability

• When a Measurement Procedure yields consistent
  scores when the phenomenon being measured is not
  changing.
• Degree to which scores are free of measurement
  error
• Consistency of measurement

35
Reliability
36
Validity

• The extent to which measures indicate what they
  are intended to measure.
• The match between the conceptual definition and
  the operational definition.

37
Reliability and Validity
38
Rising Life Expectancy
Source United Nations (U.N.) Population
Division, Demographic Indicators, 1950-2050 (The
1996 Revision) (U.N., New York, 1996).
39
Health Transition in Sweden

• Year Life Expectancy Infant MR/1000
• 1780 37 187
• 1900 53 100
• 1935 65 36
• 1996 79 4

40
Factors Influencing Changing Pattern

• Improvements due to industrialization
• Nutrition
• Environmental
• Sanitation
• Water supply
• Housing
• Medical advancements
• Antibiotics
• Immunization
• Disease surveillance programs

41
UK trends in cardiovascular disease mortality
42
(No Transcript)
43
Autism and MMR vaccine
(2-5 year old males)
Source Kaye et.al. BMJ 200132202
44
Reaction to the MMR/Autism Study

• ...MMR is unlikely to be the sole cause of the
  huge increase in autism that has occurred over
  the last twenty years or so.
• those who suggest that there may be a link
  between MMR and autism are not necessarily making
  such a suggestion.
• ...strong anecdotal evidence that MMR may be the
  trigger to autism in some cases.
• I have spoken to the parents of three children
  in whom the MMR vaccination was followed by an
  immediate, quite severe, reaction and a sudden
  subsequent descent into autism".

P Allmark, Sheffield University,
bmj.bmjjournals.com/cgi/eletters/322/7284/460
45
Attributable Risk
46
Variation in Disease Pattern

• Consider the possible reasons why variations in
  disease pattern might be an artefact rather than
  real. (You
  may find 7-10 reasons).

47
Variations

• Chance
• Errors of observation
• Changes in the size and structure of the
  population
• The likelihood of people seeking health care and
  hence being diagnosed
• The likelihood of the correct diagnosis being
  reached
• Changes in the clinical approach to diagnosis
• Changes in data collection methods
• Changes in the way diseases are diagnostically
  coded
• Changes in the way data are analysed and presented

48
Explanations for real changes in disease
frequency

• What explanations are there for real change in
  disease frequency?
• Host e.g. genetics, behaviour
• Agent e.g. virulence, introduction of a new
  agent
• Environment e.g. housing, weather

49
Definitional Problems

• Drug abuse/dependence is a behaviorally defined
  disease or disorder
• No pathogens or biological indicators of the
  condition

50
The Natural History of Opiate Dependence

• opiate dependence stems from a physiological
  medical disorder in the human brain that causes
  the addicted individual to crave and continue to
  use the substance despite the risk of physical or
  psychological harm.
• There is consistent evidence that medical
  treatment can be utilized to effectively manage
  this disorder and that treatment can provide
  substantial positive benefits to the addicted
  patient and society
• Effective Medical Treatment of Opiate Addiction,
  National Institute of Health Consensus Statement
  1997

51
Evidence That Opioid Dependence Is a Medical
Disorder

• Despite varying cultural, ethnic, and
  socioeconomic backgrounds, there is clear
  consistency in the medical history, signs, and
  symptoms exhibited by individuals who are
  opiate-dependent.
• There is a strong tendency to relapse after long
  periods of abstinence.
• The opioid-dependent person's craving for opiates
  induces continual self-administration even when
  there is an expressed and demonstrated strong
  motivation and powerful social consequences to
  stop.
• Continuous exposure to opioids induces
  pathophysiologic changes in the brain.

52
Defining Addiction from the Patient's
Perspective

• The defining characteristic of addiction is
  compulsive, out-of-control drug use despite
  serious negative consequences. . . . "Effective
  management depends on conceptualizing addiction
  as a chronic, relapsing medical illness. . . . .
  Tolerance and dependence are neither necessary
  nor sufficient for addiction.
• Indeed, withdrawal symptoms from cessation of
  addiction drug use tend to resolve within days to
  weeks and therefore cannot account for the
  profound persistence of relapse risk, which has
  been well documented in addicted populations
• A 28-YEAR-OLD MAN ADDICTED TO COCAINE, Nov. 28,
  2002, JAMA (vol. 286, No. 20,pp 2586-2594).

53
Prevalence - NHSDA, 2000 Illicit Drug Use
Percent Admitting
Any Illicit Drug Use by Gender
54
Patterns of cannabis/cocaine use use among
experienced users in Holland, 1995
55
Are defective brains to blame?
When maturational dyssynchrony (e.g. incongruity
in timing and sequencing among hormonal,
physical, psychological, and social processes)
occurs during late childhood and early
adolescence, homeostatic activity of the
hypothalamicpituitaryadrenocortical (HPA) axis,
the hyyothalamicgonadal (HPG) axis, and the
mesotelencephalic dopaminergic pathways is
perturbed. These changes are posited to influence
both the timing of puberty and brain reward
mechanisms, thereby increasing the risk for
substance use.
Developmental sources of variation in liability
to adolescent substance use disorders Drug and
Alcohol Dependence 61 (2000) 314
56
Advisory Council on the Misuse of Drugs
"On strong balance of probability, deprivation is
today in Britain likely often to make a
significant causal contribution to the cause,
complication and intractability of damaging kinds
of drug misuse...We want now and in the future to
see deprivation given its full and proper place
in all considerations of drug prevention
policy".187
57
Reasons for drug use

• Environmental factors
• Drug availability
• Some people are more susceptible
• Effects of drugs on individuals

58
Summary

• Research Design
• Reliability and Validity
• Statistics and Data Analysis
• Epidemiology and Causality
• Substance Misuse Epidemiology