Lecture 3: Understanding the Data-- Measures of Frequency and Association
Lecture Overview
Subject: Understanding the data -- Measures of frequency and association
Course: KHPM324 Chronic Diseases of Modern Society
Instructor: Hannah Oh, ScD
Department: Division of Health Policy & Management, College of Health Sciences, Korea University
Use of Epidemiologic Data
Distribution
Key Questions:
How many diseases are there?
How often do diseases happen (frequency)?
Where do they happen (places)?
When do they happen or how do they change over time?
Who has them and who does not (which population groups are at higher risk)?
Determinants
Key Questions:
Why do diseases happen?
What determines/causes diseases or health-related events?
What works to reduce the burden or risk of the disease? What is effective?
Measures of Disease Frequency
Definitions
Prevalence (유병률):
Definition: The proportion of the population with a particular health condition or disease at a specified time (or period).
Incidence Proportion (발생률) (also known as Risk):
Definition: The proportion of the population newly developing a disease over a specified period of time.
Note: "Incidence proportion" and "risk" are often used interchangeably.
Comparison: Prevalence vs. Incidence
Prevalence:
Concept: Viewed as a pool of disease in a population.
Incidence:
Concept: Describes the input flow of new cases into the pool of disease.
Outputs: Death and recovery reflect the output flow from the pool.
Example Calculations
First Example
Population: 100 individuals followed for one year.
Initial Cases: 20 had the disease at the start; 8 new cases developed during the year.
a) Prevalence at the start of the year:
rac{20}{100} imes 100 = 20 ext{ %}b) Incidence proportion for the one-year period:
rac{8}{100 - 20} imes 100 = 10 ext{ %}
Second Example
Population: Similar to the first, but starting cases were 10.
a) Prevalence at the start of the year:
rac{10}{100} imes 100 = 10 ext{ %}b) Incidence proportion for the one-year period:
rac{8}{100 - 10} imes 100 = 10.88 ext{ %}
Factors Affecting Prevalence
Effective treatment can make a disease MORE common. Therefore, prevalence depends on the duration of the disease.
Comparing Risk between Groups
Questions to Interpret Data
Which population is healthier?
Which population has a higher risk of heart disease?
Population A vs. Population B
Population A: Risk of heart disease = 30%
Population B: Risk of heart disease = 60%
Important Note: Comparisons are not meaningful without a specified time period. The 30% risk has different implications over 12 months compared to 10 years.
The risk (or incidence proportion) is cumulative over time and reaches 100% if follow-up is long enough.
Case Fatality Rate (CFR)
Definition
CFR: Proportion of people who die from a specified disease among all individuals diagnosed with the disease over a certain period of time.
Formula:
ext{CFR} = rac{ ext{Number of deaths}}{ ext{Number of confirmed cases}}
Importance of Estimating CFR
Understanding disease severity and mortality during the infection period.
Based on cases detected through surveillance.
Age Group Data on CFR
Age Group | Confirmed Cases | Deaths | Case Fatality Rate (%) |
---|---|---|---|
80+ | 767 | 162 | 21.12 |
70-79 | 1,417 | 95 | 6.7 |
60-69 | 2,840 | 42 | 1.48 |
50-59 | 3,535 | 16 | 0.45 |
40-49 | 2,627 | 4 | 0.15 |
30-39 | 2,437 | 2 | 0.08 |
20-29 | 4,233 | 0 | - |
10-19 | 1,122 | 0 | - |
0-9 | 422 | 0 | - |
COVID-19 CFR Across Countries
Variability of estimates between countries:
Country | Deaths | Cases | Case Fatality (%) | 95%-CI |
---|---|---|---|---|
France | 29209 | 154188 | 18.94 | (18.75 to 19.14) |
Belgium | 9619 | 59437 | 16.18 | (15.89 to 16.48) |
Italy | 33964 | 235278 | 14.44 | (14.29 to 14.58) |
UK | 40597 | 287399 | 14.13 | (14.00 to 14.25) |
Hungary | 550 | 4017 | 13.69 | (12.64 to 14.79) |
Others Included | … | … | … | … |
Measures of Disease Frequency
Definitions
Incidence Rate:
Definition: The number of newly developing cases per unit of person-time during follow-up.
Measured by total time contributed by all individuals in the follow-up study (time at risk).
Odds:
Definition: The probability of developing a disease divided by the probability of not developing a disease.
Prevalence Odds:
ext{Prevalence odds} = rac{ ext{Prevalence}}{1 - ext{Prevalence}}Incidence Odds:
ext{Incidence odds} = rac{ ext{Incidence proportion}}{1 - ext{Incidence proportion}}
Examples
Risk vs. Rate
Two Populations Followed for 1 Year
Important to differentiate between risk and rate:
Population A's risk vs. Population B's risk.
Rate comparisons can be more accurate as it accounts for actual time of follow-up until disease occurrence.
Measures of Association
Key Metrics
Risk Difference:
ext{Risk difference} = ( ext{Risk in the exposed group}) - ( ext{Risk in the unexposed group})Risk Ratio:
ext{Risk ratio} = rac{ ext{Risk in the exposed group}}{ ext{Risk in the unexposed group}}Rate Difference:
ext{Rate difference} = ( ext{Rate in the exposed group}) - ( ext{Rate in the unexposed group})Rate Ratio:
ext{Rate ratio} = rac{ ext{Rate in the exposed group}}{ ext{Rate in the unexposed group}}Odds Ratio:
ext{Odds ratio} = rac{ ext{Odds in the exposed group}}{ ext{Odds in the unexposed group}}
Criteria for No Association
No association exists if:
Risk difference = 0
Rate difference = 0
Risk ratio = 1
Rate ratio = 1
Odds ratio = 1
Example: Risk Ratio
5-Year Follow-up Data
Unexposed Group vs. Exposed Group:
Calculation of risk ratio using the following data.
Details Needed of disease present and absent for the final calculation.
Post-Test Instructions
Visit www.socrative.com for the post-test.
Student log-in required.
Provide your name for participation points.
Room details provided in class.
Closing Remarks
Next class scheduled for Wednesday!