Epidemiology

1
Epidemiology

• HSTAT1101 27. oktober 2004
• Odd Aalen

2
Measuring disease occurrence

• The aim of epidemiology is to map disease
  occurrence statistically, so that the disease may
  be better understood and perhaps prevented
• This requires measures of disease occurrence. Two
  major measures
• prevalence
• incidence rate

3
Types of study

• Cross-sectional study
• assessing the situation at one specific time
  (example how many smokers and non-smokers have
  asthma at the present time)
• Cohort (or follow-up) study
• looking ahead in time (e.g. follow-up of smokers
  and non-smokers to observe occurrence of asthma)
• Case-control study
• looking back in time (e.g. patients with asthma
  are compared with control group to look for
  previous risk factors, e.g. smoking)

4
Epidemiology 200415 653659 Lönn et al
5
Prevalence

• Prevalence The proportion of a population that
  has a certain condition at a specified time

• Example
• Prevalence of asthma in Norway 2.4
• Prevalence of multiple sclerosis in Norway 100
  per 100,000
• (Note Sometimes another basis number than
  100,000 may be used, e.g. 1 million)

6
Estimating prevalence

• Need an estimate of the population size
• Need an estimate of the number of cases of
  disease. Cross-sectional design is sufficient
• Requires definition of case. This is often not
  obvious
• example asthma (dyspnea, wheezing, cough,
  spirometric measurements)
• sometimes an apparent increase in prevalence is
  due to a changing definition (or increased
  awareness) of disease

7
Incidence rate

• Incidence Rate of new cases per year of a
  certain condition

• Examples
• Incidence of multiple sclerosis in Norway
• 5 per 100,000 person years
• Incidence of HIV infection in Oslo in 2000
• 11 per 100,000 person years

8
Estimating incidence

• Need an estimate of the population size or
  person-years
• Need an estimate of the number of new cases of
  disease over some time period (e.g. one year)
• requires definition of when the disease started
  (e.g. time of first diagnosis by a medical
  doctor)
• Preferably a cohort (follow-up) study

9
Prevalence vs. Incidence

• Incidence measures risk of disease
• Prevalence measures burden of disease
• The burden may increase because the risk
  increases, or because the disease lasts longer,
  e.g. if mortality of disease decreases

10
Illustration of basic concepts
Incidence
Prevalence
Recovery
Death
11
ExampleHIV-infection

• With new treatments progression to AIDS or death
  has been strongly decreased
• No complete recovery takes place
• The incidence of HIV infection is largely
  unchanged
• This results in considerably increased prevalence
  of HIV infection

12
Computing an incidence rate by the person-years
method

• The incidence rate is estimated as
• By person-years we mean the sum of the
  observation times for all individuals

13
Example

• From the Cancer Registry of Norway
• During 1983-87 there were 460 cases of breast
  cancer among women in the age group 30-39 years
• The population in this age group in 1985 was
  302,501. Number of person-years are 302,501 5
• The incidence rate is

14
Example

• On the next slide is presented incidence of
  malignant melanoma in Norway, a disease which has
  become much more common over the last few decades
• The incidence is age-adjusted, to correct for
  changing age-composition. This is done by
  standardization

15
Incidence of malignant melanoma among women in
Norway 1956-1995
16
Population and sample

• The population consist of all the individuals we
  want to study. Examples
• All people between 20 and 60 years of age in a
  city
• All people in the country suffering from
  tuberculosis
• People in a profession e.g. bus drivers
• The sample consist of those individuals that are
  actually included in the study

17
Sampling
Total population
Random sampling
Study population
18
Association and causation

• Epidemiology gives us statistical associations
• Example smokers have much higher risk of lung
  cancers than non-smokers
• Example People with high blood pressure have
  increased risk of heart disease
• Association does not necessarily imply that the
  factor is a biological cause

19
Confounding

• Example Cigarette smoking in mothers is
  associated with sudden infant death (SIDS). Is
  this causal?
• Smoking could be an indicator of other lifestyle
  factors that influence the risk of SIDS.
• Such other factors that could explain an
  association are called confounders

20
Survival analysis

• Studying durations
• duration of disease
• duration of remission
• duration of marriage
• age at breast cancer diagnosis
• Durations are important clinical and
  epidemiological outcome parameters
• do patients live longer
• does the remission period last longer
• can we postpone disease

21
Censoring

• Special problem of duration data incompletely
  observed times (censored data). Causes
• study is terminated
• withdrawal
• observation ceases
• Basic assumption No selective censoring
• the individuals which get censored at any given
  time shall not differ, on the average, from those
  that are under observation but not censored at
  that time
• can be modified for Cox-regression
• Censoring precludes the use of ordinary
  statistical methods for measurement data

22
Small example

• Data set
• 26, 17, 7, 41, 34, 9, 13, 25, 37, 18
• censoring time
• The same data ordered
• 7, 9, 13, 17, 18, 25, 26, 34, 37, 41

23
Graphical presentation

• Survival curve Describing proportion that
  survives up to some time
• Hazard rate Describing risk of event (death,
  relapse etc) as function of time

24
Example Hazard rate (incidence rate) of divorce
in Norway
25
Survival of marriages contracted in 1960, 1970
and 1980
26
Treatment of acute myocardial infarction

• Analyzed by Cox model, adjusted hazard ratio 2.31
• Propor-tionality?