Module 3: Measures Flashcards
What is prevalence? What is it also known as?
Presence of disease in a group.
“Burden” of disease.
Prevalence has no explicit consideration of time except for what?
Time over which prevalence calculated.
What is incidence? What is it also known as?
New occurrence of disease.
“Risk” of disease.
Incidence refers to what transition?
Non-diseased to diseased state.
The COVID-19 dashboard we viewed refers to what three measures?
Active cases = prevalence
Total cases reported daily chart = incidence
Histogram = “epidemiological curve” - visual representation of disease risk over time.
There are 10 students in a tutorial. 2 have a cold. Report the prevalence in proportion and in odds.
Proportion: # of colds / # of ppl. in tutorial = 0.2 = 20% (20 out of 100 ppl. in tutorial).
Odds: # of colds / # of ppl. without colds = 0.25
What are units used when reporting proportion?
Units are people, no time element.
When thinking about proportion and odds, what happens when prevalence is low?
The proportion and odds are similar.
What is point prevalence? Provide an example.
The amount of disease in a population at a point in time, snapshot; the most common option, assumed if not specifically stated otherwise.
“Do you currently feel depressed?”
What is period prevalence? Provide an example.
The amount of disease in a population during a time period/frame.
“In the past year, have you felt depressed?”
What is lifetime prevalence? Provide an example.
The amount of disease in a population at any point during the lifetime of those alive in the population.
“Have you ever felt depressed?”
If a professor keeps track of all colds in their tutorial for one year, they count 5 colds. What is the incidence PROPORTION of colds in their tutorial during the academic year?
Number of new colds / # of people at risk = 0.5 = 50% (50% risk of getting a cold over one year).
What is the incidence rate? What does Patten liken it to?
The rate at which people become ill; time explicitly included.
Patten likens to velocity, km/h.
What is “person-time”? What does it mean when somebody refers to 1 or 5 person-years?
The sum of the time periods of observation of each person who has been observed for all or some of that time period.
1 person-year = one person at risk who is observed for one year.
5 person-years = one person at risk who is observed for five years OR 5 people each at risk for 1 year.
If a professor keeps track of all colds in their tutorial for one year, they count 5 colds. What is the incidence RATE of colds in their tutorial during the academic year?
Number of new colds / total person-time at risk = 5 per 10 person-years, expressed as 0.5 person-year−1.
What is the difference between incidence proportion and rate?
Proportion: New events / Those at risk in the relevant period. No explicit time in proportion.
Rate: New events / Total person-time at risk. Time explicitly included.
What are attack rates? When are they used?
Attack rates are types of incidence proportions (never referred to as “attack proportions”).
When a particular disease corresponds with a particular risk interval, like food poisoning or flu season, attack rates used to describe and compare different outbreak scenarios.
What is cumulative incidence? Why is it used?
Proportion of cases that accumulate over an interval of observation in an at-risk cohort.
People who are ill in earlier time periods are not at risk of illness in later time periods, so we use an exponential function to calculate the true proportion of individuals who will be sick after multiple time periods.
How is cumulative incidence calculated?
Cumulative incidence, n years = 1 − (1 − Annual Incidence Proportion),n
What is the relationship between prevalence and incidence?
Prevalence = Incidence Rate × Mean duration of disease
Prevalence Odds = Incidence Rate,t−1 × Mean duration of disease,t
How is mortality rate calculated and expressed?
Number of deaths per year / The mid-year population in the same year
Usually expressed as “per 100,000” by multiplying the above by
100,000.
For mortality rates, the mid-year population is meant to reflect what?
Person-time at risk that year.
How are specific mortality rates calculated?
Cause-specific mortality rates: # of deaths due to disease X per year / The mid-year population in the same year
Sex-specific mortality rates: # of deaths among females per year / The mid-year female population in the same year
Age-specific mortality rates: # of deaths in age group X per year / The mid-year population of age-group X in the same year
Common age-specific mortality rates refer to babies. What are these rates?
Perinatal mortality: # of deaths age 0-6 days or still birth of 28 weeks + gestation / # of total births
Infant mortality: # of deaths less than 1 year / # of live births
Neonatal (first 28 days of life); Post-neonatal (>28 days, <1 year); Under 5; all use live births as denominator.
What is case-fatality rate (or case-fatality proportion)? How is it calculated?
Percentage of people who die when they get a disease.
Number died while having disease / # of people with disease x 100
With case-fatality rates, “rate” is used but with what caveat?
Risk interval characterized by illness rather than specific time interval.
“Potential Years of Life Lost” (PYLL) refers to what? Provide an example.
A measure of premature mortality or early death, referring to a standard reference age (e.g., 75).
Death at age 20 = 75 - 20 = 55 PYLL
Mortality rates/causes and PYLL will give different estimates of the importance of different causes of death. Provide an example of this.
Suicide - low cause-specific mortality rate, ranks highly in causes of PYLL because people who commit suicide are typically young.
What is life expectancy? What does it use?
An estimate of future survival experience of contemporary cohort, a projection.
Uses current age-specific mortality rates to guess how many years, on average, a baby born this year can be expected to live.
Life expectancy reacts to changes in _____ _____.
Current mortality.
What are the 5 traditional measures of mortality and morbidity?
Life expectancy.
Mortality rate.
Number of chronic conditions.
Prevalence of a disease.
Incidence of a disease.
Newer measures of health try to get at quality of life. Take health status like health-related quality of life - what does that mean?
Asking somebody how they would rate aspect of health.
Newer measures of health include health-adjusted life years (HALYs) and health-adjusted life expectancy
(HALEs). What are these?
Numbers that account for length of life and health-related quality of life summarized for a population.
The basic idea of HALYs and HALEs is what? Explain.
Not all time alive is of equal value - life years can be adjusted by health (life years x health-related quality of life).
If someone’s health-adjusted calculation was 80 x 0.75, what does that mean?
They lived until 80 years, their health-related quality of life was 0.75, therefore they had 60 perfect-equivalent life years.
The Health Utility Index (HUI) is commonly used in population health surveys. Each respondent answers questions regarding function in the HUI’s 8 dimensions. The answers are applied to the “standard Canadian
preferences” for that particular health status. What does that mean?
Researchers ask people what trade-off they would be willing to accept if moved to a lower stage of a health dimension. Implicit trade-offs that people feel represent equal trade-offs between different states of health.
A HUI = 0.75 for one individual means what?
Their scores are representative of a person with a health utility score of 0.75, not that the respondent preference indicates a weight of 0.75.
Explain how Quality Adjusted Life Years (QALYs) are calculated and what weights are used? What does 1 QALY mean?
Life years × health-adjusted quality of life per year
Health-related Quality of Life: A weight [0,1] where health utility derived from evaluating different health states.
1 QALY = 1 year full health
Explain how Disability Adjusted Life Years (DALYs) are calculated and what weights are used?
Years of life lost + years lived with a disability
“Disability weight” [0,1]
If someone’s living 10 years with a disability weight = 0.5, what does that mean?
5 years lost.
What is the measure of health for the Global Burden of Disease?
DALYs
With regards to disease and exposure, what question does strength of association ask vs. attributable risk?
Strength of association: To what extent is the disease associated with the exposure?
Attributable risk: How much of the disease can be attributed to the exposure?
Incidence of disease is a building block for other measures, it’s the _____ _____ of something occurring.
Absolute risk.
We often compare the absolute risk between what two groups?
Those with the exposure versus those without.
Those with the disease versus those without.
How is Relative Risk (also known as risk ratio) calculated?
Risk (incidence) in the exposed / Risk (incidence) in the unexposed
Interpret an RR = 1; RR > 1; RR < 1.
RR = 1: No association (risk in exposed = risk in unexposed)
RR > 1: exposure associated with increased risk (risk in the exposed > risk in the unexposed)
RR < 1: Exposure associated with decreased risk (risk in the exposed < risk in the unexposed)
We’re examining the risk of osteoporosis among those who are long-term users of corticosteroids as opposed to non-users. We get an RR = 2. Interpret this.
The risk of osteoporosis among those who are long-term users of corticosteroids is twice the risk of osteoporosis in non-users.
How is an Odds Ratio (also known as relative odds) calculated?
Odds in one group / Odds in another group
What two groups do Odds Ratios pertain to?
Exposed versus the unexposed - “What are the odds that the disease will develop in an exposed person?”
Diseased versus not diseased - “What are the odds that a person with the disease was exposed?”
Interpret an OR = 1; OR > 1; OR < 1.
OR = 1 : No association.
OR > 1: Exposure positively associated with disease.
OR < 1: Exposure negatively associated with disease.
We’re examining the risk of osteoporosis among those who are long-term users of corticosteroids as opposed to non-users. We get an OR = 2.02. Interpret this.
The odds of osteoporosis among those with long-term use of corticosteroids is approximately double the odds of non-users.
What two reasons make Odds Ratios particularly useful?
The exposure OR and the disease OR are equal.
The OR can approximate the RR when the disease is rare (i.e. numbers of exposed with disease and unexposed with disease are small).
How is the Risk Difference calculated?
Risk (incidence) in exposed − Risk (incidence) in unexposed
We’re examining the risk of osteoporosis among those who are long-term users of corticosteroids as opposed to non-users. We get a RD = .01. Interpret this (hint: can’t use %. Why is this?).
The risk of osteoporosis is 0.01 higher among long-term users of corticosteroids compared to non-users.
Do not use % in risk differences because we are talking about absolute magnitudes of probability units.
Numbers needed to treat/harm assume causality. The number needed to treat and the number needed to harm pertain to what outcomes?
NNT: positive outcome (cure)
NNH: negative outcome (adverse event)
Numbers needed to treat/harm use what as part of their calculation?
Risk differences.
In a clinical trial, cure for the treatment group = 66% and cure for the control group = 33%. What is the risk difference and what is the implication?
Risk Difference = 0.66 − 0.33 = 0.33
The treatment increases the probability of being cured by 0.33.
In a clinical trial with a risk difference of 0.33, what is the number needed to treat?
NNT: 1 / 0.33 = 3
3 people must be treated for 1 of them to have been cured by the treatment.
We’re examining a rare medical event - in the general population: 1 in 100,000 in one month;
in people using a new drug: 5 in 100,000 in one month. What is the risk difference and what is the implication?
Risk Difference = 5 − 1 = 4
The new drug increases the risk of an adverse event by 4 in 100,000.
We’re examining a rare medical event, possibly related to a new drug, with a risk difference of 4 (i.e., 4 in 100,000). What is the number needed to harm?
NNH: 1 / 4/100,000 = 25,000
5,000 people need to be treated with this drug to create 1 adverse event beyond the usual rate.
