Title: Measures of disease frequency
1Measures of disease frequency
2Measures of Effect and Disease Frequency
- Aims
- To define and describe the uses of common
epidemiological measures of disease frequency and
effect
3Epidemiology in a nutshell
Is change in exposure distribution temporally
related with change in disease distribution?
- Aim
- does exposure cause disease?
- does drug treat disease?
Statistical power calculation (type-1, type-2
error, prevalence of disease in unexposed,
minimum detectable effect)
Design study
- Can I randomise?
- Ethical?
- Clinical equipoise?
Randomised study Report (RR)
Yes
No?
Observational study
Rare disease? One outcome?
Case-control (report OR)
Rare Exposure? Many outcomes?
Cohort (report RR)
Define case and exposure status
4Table 1
Check missing data, duplicates, data range,
bivariate scatterplots and lowess curves
Are there systematic differences between exposure
and unexposed groups (confounding)
Yes (shouldnt be in RCT!)
Are they adjusted for in the analysis if
confounders?
Population divided by exposure status?
What population is the study sample drawn from?
Is it representative of underlying population or
is there likely selection bias?
5Results Analysis
Check data distributions Transform?
Outcome variable?
Continuous
Categorical
Report crude or univariate measures of
association (OR/RR/HR)
Chi-square or Fisher exact test if cell counts lt5
t-test
- Confounders? Review scientific literature is
there likely to be a - Shared common cause of exposure and disease?
Logistic regression and or stratification
Multiple linear regression
If difference between crude and adjusted gt10,
then Statistical evidence of confounding
Report adjusted measures of association (OR/RR)
6Interpret study results
Estimate OR/RR and 95 C. I.
Hypothesis likely false
Is there an association between exposure and
outcome?
Is P lt0.05 or 95 CI for measure of association
contain null value (1)?
No
Consider type-2 error confounding, bias, other
studies
Yes Exposure is associated with disease
Is there another explanation?
Confounding
Bias
Type-1 error (consider strength of association)
Information (recall)
Shared common cause of exposure and disease?
Selection (survivor loss to follow up, hosp.
controls)
How does my study compare with others?
Regression or stratified analysis
Could study design be improved?
7Discussion
Is the association I have detected causal?
Bradford Hill criteria Temporality (cohort
study? Not cross sectional or case-control which
do not separate exposure and disease) Strength of
association (odds ratio or relative risk, does
it indicate gt50 increase) Dose response is
there increasing association with increased
exposure? Biological plausibility (are there any
laboratory studies to support your
assertions?) Consistency (do other studies using
different methods, with different groups come up
with similar findings?) Experimental evidence
(Any randomised studies?) Analogy (Any similar
findings from related fields of
science?) Specificity Is exposure to the cause
reliably followed by disease? Also are there
any other competing explanations? Are there any
studies which shed light on these? If not then
Calculate Risk difference, NNT and PPAR.
Yes (on balance) Exposure causes disease
8How is a disease diagnosed?
9Methods
- Clinical
- History (symptoms)
- Physical examination (physical signs)
- Laboratory tests
- Radiology
- Microbiology
- Epidemiology
- Hospital discharge codes, drug use, lab use,
attendance at outpatient clinics, diagnosis
codes, autopsy.
10Proportions vs Rates
- Proportions concerned with (disease) states
- Useful for assessing health status of population
- Planning health care services
- Quality of care
- Screening for asymptomatic disease
- Rates concerned with (disease) events
- Useful for assessing causation
11Incidence and Prevalence
Emigration
?Incidence
?Excess Death rate
Prevalence
Immigration
12Populations
- Closed
- No new members
- Loses members to death
- Approx when prevalence assessed over short time
frame
- Open
- Gains new members
- Immigration
- Birth
- Emigration
- Long time period
13Prevalence
- The proportion with disease (at given instant)
Adult Population
Diabetes
14Incidence Proportion/Cumulative incidence/Attack
rate
per year
0.5
0
Adult Population
Adult Population 2010 (no diabetes)
Diabetes Year 2014
Incidence proportion 0.5 Incidence rate
impossible to calculate (once diseased, stop
being at risk).
15Incidence Rate
Get disease
Death
16(No Transcript)
17Incidence rate
- Average rate of disease occurrence over time
interval - Assumes rate doesnt change over time
- Reciprocal of waiting time
- Years until disease
- Usually only first event counted even if
biologically, second event may be independent.
18Prevalence, incidence and duration of disease
If disease is rare
19Duration of disease
- E.g. Systemic lupus erythematosis
- Prevalence 0.5
- Incidence 6 to 35/100,000 people per year
- Say 20.5/100,000
- Duration of disease 0.005 /(20.5/100,000)
- 24 years
20Prevalence CHD (men) in NZ
21Some questions
- You are about to design a programme to measure
how many people with diabetes are taking
appropriate drugs. - Would the population of interest consist of
prevalent or incident cases of diabetes?
22Incidence proportion is
- a) Useful for assessing the health status of the
population - b)The proportion of initially disease free people
who, over a specified period of time, develop
disease. - c) Measures survivors only.
- d) A rate.
- e) Measured at one point in time.
23Prevalence is
- a) Useful for assessing the health status of the
population - b)The best measure to use when identifying risk
factors for a disease - c) Unaffected by deaths in the population
- d)Unaffected by migration
- e) A rate.
24Comparisons
- Involves time difficult to measure
- Measured as either rate or proportion
- Less affected by other influences.
- Numerator count of people who develop disease
during follow-up - Denominator
- (prop.) People at risk and disease free
- (rate) Person-time at risk.
- One point in time easy to measure
- Proportion or
- Affected by many factors apart from those causing
disease. - Numerator count of people with disease
- Denominator count of total population at risk
- No time component
25Summary
- Prevalence (among population, proportion with
historic diagnosis of disease). - Incidence proportion (disease free, follow,
proportion that develop disease in defined
period) - Incidence rate (disease free, follow, number of
cases that develop disease divided by time at
risk).