Lec 2 & Tut 2 notes Flashcards
Cumulative incidence Calculation
Interpretation
Incidence rate
interpretation
Incidence prop or cumulative incidence or risk
# new events/ pop at risk
Interpretation: Probability the person dev condition during specified time period
Incidence rate/ incidence density
- interpretation: How fast new events occur
# new events/PY at risk
Measures of disease freq (N/D)
Incidence prop
Incidence rate
Pt prev
Period prev
Others
Age or sex specific incidence rate
Birth rate
Fertility rate
cause-specific mortality rate
case fatality rate
attack rate
Measures of disease freq
- Incidence prop:
o N: # new cases
o D: pop at risk at start of interval
- Incidence rate:
o N: # new cases
o D: PT contributed from at risk ppl during interval
- Pt prev
o N: # of current cases specified pt in tom
o D: pop at specified pt in time
- Period prev
o N: # of current cases
o D: mid-year pop
Other measures
- Age or sex specific incidence rate:
o N: # ppl in gp w/ event during time
o D: # of ppl in gp AT RISK during time
- Fertility rate
o N: # of live births
o D: women of reproduce age - Cause-specific mortality rate
o N: # of deaths due to specific cause
o D: pop - Case fatality rate
o N: # deaths from disease
o D: # ppl w/ disease - Attack rate: incidence of infectious disease outbreak
o N: # ppl w/ infectious event
o D: # ppl exposed
Based on the prevalence equation, what is the effect on prevalence if we
- increase incidence
- increase cure
- increase survival?
Measure comparison
- Absolute comparison:
- interpretation
- type of math operation
- 4 examples
- relative comparison
- interpretation
- type of math operation
- 4 examples
Factors that impact prevalence
- Prevalence = incidence x duration
o Increase incidence -> increase prevalence
o Increase cure (decrease duration) -> decrease prevalence
o Increase survival (increase duration) -> increase prevalence
Measure of comparison
o Absolute
What: PH impact of an exposure
How: subtraction; diff in measures
Examples:
• Risk diff
• Rate diff
• Attributable risk and %
• Pop attributable risk and %
o Relative
What: strength of relationship
How: division; ratio of measures
Examples
• Relative risk
• Risk ratio
• Rate ratio
• Odds ratio
Normal 2x2 table vs PT 2x2 table
Top: normal 2x2 tables
below: Table below looks at PT
people with no disease is not included (no b + d) cuz there is no PT for “no exposure”
Risk or rate difference formula
Interpretation
How to calculate based on the table
Re
Ru
RD = 0 meaning
RD > 0 meaning
RD < 0 meaning
Define NNT
NNT formula
NNT Answer needs to be …
Risk difference
- RD = Re – Ru
o Interpretation: Excess risk or rate due to exposure
Calculations
- Incidence in exposed (Re): a/a + b
o “in exposed” = the row with exposed
- Incidence in unexposed (Ru): c/c + d
- Risk difference: (a/a + b) – (c/c + d)
Interpretation
- RD = 0: risk in exposed = risk in non-exposed (IOW: no increased risk due to exposure)
- RD > 0: Risk in exposed > risk in non-exposed (IOW: excess risk due to exposure)
- RD < 0: Risk in exposed < risk in non-exposed (IOW: less risk due to exposure)
-
FORMULA: # need to treat (NNT) = 1/RD
o NNT: # ppl needed to be treated to prevent one more event
disease
o NOTE: NNT ALWAYS round UP!!!
What is Attributable risk percent or attributable fraction
2 methods or formula to calculate
Attributable risk percent or attributable fraction
- Express excess risk (RD) as a proportion
Method 1: [(Re - Ru)/Re] x 100
Method 2: AR% = (RR – 1)/RR
- Proportion of excess cases that can be due to the exposure
- Example
o RD of stroke in smokers vs non-smokers
Incidence smokers: 15%
Incidence nonsmokers: 5%
o RD: 0.15 – 0.05 = 0.10
o Attributable risk percent: (0.1/0.15) x 100%
67% of strokes among smokers can be prevented if they did not smoke
Relative risk or risk ratio formula
Risk ratio formula
Rate ratio formula
Define
RR = 1
RR > 1
RR < 1
2 ways to interpret it
Relative risk or risk ratio
- RR = Re/Ru
o A ratio b/w risk or rates in 2 pop
o Risk ratio: use cum inc
o Rate ratio: use incident rates
- RR = 1: risk in exposed = risk in non-exposed (no association)
- RR > 1: risk in exposed > risk in non-exposed (+ve association)
- RR < 1: Risk in exposed < risk in non-exposed (-ve association)
- X
- 2 ways to interpret it Interpretation
- #1: X times more likely
- #2: % increased risk
o % increase = (RR – 1) x 100
o
- E.g. Smokers 10x more likely to get cencer
o RR = 1.5
o Risk is 1.5x higher OR 50% higher
Based on the table, how to calculate
- Re
- Ru
- RD
- RR
Calculation
- - Incidence in exposed (Re): a/a+b
- Incidence in unexposed (Ru): c/c+d
RD: [a/(a + b)] - [c/(c + d)]
Relative risk: [a/(a + b)] / [c/(c + d)]
Relationship b/w RR, Baseline incidence, and RD
- define baseline incidence
- If RR is constant and BI increase, what happens to RD?
- If RD is constant and BI increase, what happens to RR?
Will reducing an exposure with a high RR prevent many case of disease
- baseline incidence: unexposed ppl
Relationship b/w RR, Baseline incidence, and RD
- With a constant RR (eg. #1 and #3’s RR = 6), as baseline incidence increase (i.e. Ru 0.2 -> 10), RD increases
o If there’s more unexposed people; we need more exposed people to maintain the same RR
- With constant RD (eg. #1 and #2’s RD = 1), as baseline incidence increase (i.e. #1’s Ru = 0.2 -> #2’s Ru = 10), RR decreases (e.g. 6 -> 1.1)
Will reducing an exposure with a high RR prevent many case of disease
- It depends on the baseline incidence
- Eg see case #1 and #3
o Both have large RR = 6
o #1: has RR = 6, but we can only prevent 1 case per 100 people
o #3: has RR = 6, but we can prevent 50 cases per 100 people
o IOW, the prevalence of disease or baseline incidence matters
Attributable risk in exposed
- definition
- interpretation
- Formula
Pop attributable risk
- definition
- interpretation
- formula
Attributable risk in exposed: excess risk of disease in exposed indiv
o If causal (more to come), this is the % of disease that can be prevented if the exposure was elim
o Measure of potential benefit associated w/ removing an exposure among exposed indiv (i.e. prevention)
o 66% of “disease” in exposed indiv is sue to the “Exposure” and can be prevented if the exposure was elim
Formula: [(Re - Ru)/Re] x 100
o eg: 33% of cases in the pop are due to exposure
Pop attributable risk: excess risk of disease in the total pop (exposed + unexposed)
o Interpretation: disease in total pop that will be prevented if the exposure was elim
PAR is a function of both strength of association (RR) and prevalence of exposure in the PAR formula when you have RR and prevalence
[P x (RR - 1)]/ [P x (RR - 1) + 1]
What are some contemporary examples where it is very important to consider the absolute scale in the context of PH?
What are some contemporary examples where it is very important to consider the absolute scale in the context of PH?
- E.g. reducing person A’s cholesterol by 10 points seems not very impactful
- But if we reduce the populations’ cholesterol by 10 points, it can have a huge impact
- The severe consequences of COVID doesn’t seem to affect many people, especially in the 20s yo pop
- But if we multiply this risk across the population, it tiny risk adds up and has a huge impact
Define odds ratio
Probability of rolling 4 on a dice
Odds of rolling 4 on a dice
Convert prob to odds formula
Odds ratio
o Odds: ratio of the prob of an event to that of the non-occurrence (likelihood it happens vs it doesn’t)
Risk vs odds
- What is the probability of rolling a “3” on a dice? 1/6
- What are the odds? 1:5
- Convert prob to odds:
o Odds = P/(1 – P)
o E.g. If P = 0.167; odds: 0.167/0.833 = 0.2
- Convert odds -> prob:
o Prob = odds/(1 + odds)
o E.g. If odds = 0.2, P = 0.2/1.2 = 0.167
Based on the table
- Odds of exposure in those w/ disease
- Odds of exposure in those w/o disease
- odds ratio
Is odds = risk?
What happens to the odds and risk ratios when the outcome is rare?
What % is considered rare?
Calculation
- Odds of exposure in those w/ disease: a/c
- Odds of exposure in those w/o disease: b/d
- Odds ratio: (a/c) / (b/d) = ad/bc (cross product)
- X
- NOTE: we cannot interpret odds as risk
o E.g. the odds are 1.6 times bigger (NOT RISK)
Rare disease assumption
- If the outcome is rare, the odds ratio will approximate the risk ratio
o Consider less than 10% as rare???? (new cases/total pop)
Risk ratio vs odd ratio
- In earlier example
o RR = (84/3000) / (87/5000) = 1.61
IOW: smokers have 1.6x the odds of developing CHD compared to non-smokers
o OR = (a/c) / (b/d) = (84/87) / (2916/4913) = 1.63
- Is the outcome rare?
o 84 + 87 / 8000 = 2%
