Feldmand. Module 1 – Intro to Epidemiology & Measures of Disease Flashcards
Definition of epidemiology
• From Greek
- – Epi (among, upon)
- – Demos (the people)
- – Logos (knowledge, doctrine)
• The study of the distribution and determinants of health-related states or events in specified populations, and the application of this study to the control of health problems (Dictionary of Epidemiology)
Goals of Epidemiology
- Elaboration of causes that explain patterns of disease occurrence
- Determine extent of disease
- Study natural history of disease
- To promote, protect and restore health
- Provide foundation for developing public policy and regulatory decisions
Typical Epidemiologic Approach
- Determine existence and magnitude of problem
- Describe WHO has the problem (animal/person, place, time)
- Develop hypotheses about WHY problem is happening
- Test the hypotheses using appropriate study designs and statistical tests
- Develop interventions based on findings
- Evaluate effectiveness of interventions
Basic Tenet of Epidemiology
• Disease does not occur randomly in a population
- – Disease occurrence is related to environment of species being studied
- – Environment includes physical, biological, sociological, meteorological, & management characteristics
• Epidemiology triad
Epidemiologic Triad
• Disease is result of forces within a dynamic interaction between
– Agent
– Host
– Environment
Measures of Disease: Ratios
- An expression of the relationship of 2 quantities
- Numerator is not in the denominator
- With a dimension – # of dogs owned / 100,000 population
- Without a dimension – # 2nd year vet students / # 3rd yr vet students
Measures of Disease: Proportions
- Ratio in which numerator is contained in the denominator
- Dimensionless
- Ranges from 0 – 1
- Tells us what fraction of population is affected
vets sitting ACVPM exam / Total # vets
Rates
- Ratio in which there is a relationship between the numerator and the denominator
- A true rate is instantaneous change in one quantity per unit change in another quantity (usually time).
- Tells us how fast disease occurs in a population
tests taken in vet curriculum / person-years in vet curriculum
Measures of Morbidity: Prevalence
• Prevalence (point prevalence, prevalence rate) – Proportion of pop’n with disease at a specific time
of subjects with disease at a point in time / Population at the same point in time
• Period prevalence – Frequency of disease for a given time interval
of subjects with disease for given time interval / Population at mid-interval
Interpretation of prevalence
Probability of having disease at a particular point in time.
Measures of Morbidity: Cumulative incidence, incidence proportion
– Proportion of subjects who develop disease during a certain time period
– Unitless, interpret in the context of time period
– Measure of average risk for a population
events during a period of time / Population without disease at beginning of period
Measures of Morbidity: Incidence • Incidence rate, incidence density
– Occurrence of new event per unit time
– The numerical value has no interpretability because it depends on the arbitrary selection of the time unit
new events / Total person-time at risk in population
Measures of Morbidity: Incidence • Attack rate
– Cumulative incidence used for particular populations observed for limited periods of time, as in an outbreak
– Usually expressed as a percent
events of dz during epidemic time period / Population at risk at start of period
Interpretation of cumulative incidence
Risk of developing disease over given time period
Interpretation of incidence density
Rapidity with which new new cases develop over given time period
Relation between Prevalence, Incidence & Duration
- Prevalence is not a measure of risk: Does not take into account the duration of the disease
- If prevalence (P) is small and incidence rate (I) and duration (D) are constant over time then P ≈ IxD
Measures of Mortality
- Mortality rate: Total # deaths from all causes in 1 year / Population at mid-year
- Disease-specific mortality rate: Total # deaths from specific dz in 1 year / Population at mid-year
- Case-Fatality Rate (CFR) # deaths after dz onset or diagnosis / # individuals with the dz
- Proportionate Mortality # deaths from specific dz in time period / Total # deaths in that time period
Mortality: when can it be a good index of the risk of disease?
• Can indicate severity of disease and also be an index of the risk of disease
- – If a disease is not typically fatal, mortality is not a good index of incidence • E.g., West Nile virus
- – When the case fatality rate is high and duration of disease is short, mortality is good reflection of risk • E.g., Rabies
Risk
- The probability of a disease-free individual developing a given disease over a specified period, conditional on that individual not dying from any other disease during the period
- Risk is without units, ranges from 0 to 1
- Risk=Attack rate in outbreak settings: # events of disease during time period / Population at risk at start of period
Odds
• The probability of an event occurring compared to the probability of that event not occurring
Probability event occurs / Probability event does NOT occur
2-by-2 tables
Disease Present Absent
Exposure +
Exposure -
Note that the disease status is across the top and the exposure status is on the left
(use of) Ratio Measures of Association
- Assess the strength or magnitude of the statistical association between the exposure and disease of interest
- In cohort studies, use relative risk
- In case-control studies, use odds ratio
Identification of Risk Factors
Epidemiological studies are conducted to identify risk factors through the comparison of incidence or prevalence between groups exposed and not exposed to a risk factor.
Probabilities of disease occurrence can be compared using:
- measures of strength of association: RR, OR
- measures of potential impact: attributable risk, attributable fraction.
Measures of strength of association and information provided
Involves calculation of ratios such as relative risk and odds ratio which measure the magnitude of a statistically significant association between risk factor and disease.
They are used to identify risk factors, but do not provide information on absolute risk.
Relative Risk (RR)
- The relative risk is the excess risk in the exposed group, compared to the unexposed (background, expected group)
- Expressed as a ratio
- RR = Riskexposed / Riskunexposed
- Riskexposed = # events of dz in exposed group / pop’n at risk in exposed group = a / (a + b) = a / h1
- Riskunexposed = # events of dz in unexposed group / pop’n at risk in unexposed group = c / (c + c + d) = c/h2
- RR = Riskexposed / Riskunexposed = (a/h1) / (c / h2)
- The RR will be > 1.0 when risk is greater in the exposed group than in the unexposed group – i.e., when exposure is risk factor for disease
- The relative risk will be <1.0 when risk in the exposed group is less than risk in the unexposed group – i.e., when exposure is protective
- When no association between exposure and disease, RR = 1
Odds Ratio (OR) - (type of studies to use, why)
- In case-control studies, you cannot directly calculate the risk of disease because you do not have a denominator population
- You can calculate the odds of exposure among cases and controls
- The OR compares the odds of exposure in cases to the odds of exposure in controls
Measures of Potential Impact: what are those and why are important
- Reflect the apparent contribution of an exposure to the frequency of disease in a population
- Important for policy makers and funding sources to understand impact a prevention program might have
Measures of potential impact and information provided
Measures of potential impact include differences such as the attributable risk or fractions such as the attributable fraction.
These allow quantifying the consequences from exposure to a risk factor, and are used to predict, quantify the effect of prevention and to plan control programs
Risk Difference
Measure of potential impact.
• Also called excess risk, attributable risk:
Risk Difference = Riskexposed – Riskunexposed
• Excess risk in those exposed
Attributable Risk Percent
- Only appropriate if RR > 1
- Proportion of cases in the exposed group attributable to exposure
- The most that we can hope to accomplish in reducing risk of disease if we completely eliminate the exposure
(Riskexposed – Riskunexposed) / Riskexposed = (RR – 1) / RR
Population Attributable Risk Percent
• Proportion of cases in the entire population (exposed and unexposed) attributable to exposure
(Riskoverall – Riskunexposed) / Riskoverall
Prevented Fraction in the Exposed - Vaccine Efficacy
- Comparable measure to attributable risk percent for a protective factor, such as vaccination
- Only appropriate if the RR<1.0
- Proportion of potential new cases which would have occurred had the exposure been absent (the proportion of potential cases prevented by the exposure)
- Prevented fraction in the exposed = (Riskunexposed - Riskexposed) / Riskunexposed = 1 - RR
Formula for True Prevalence (from Apparent Prevalence)
TP = (AP + Sp - 1) / (Specificity + (Sensitivity - 1)
Formulas to calculate PPV
PPV = a / (a + b)
PPV = TP / (TP + FP)
Formulas to calculate NPV
NPV = d / (c + d)
NPV = TN / (TN + FN)
Definition of infectivity
Ability of an agent to establish itself in a host (ID50 = numbers of agents required to infecto 50% of exposed susceptible animals under controlled conditions)
Definition of pathogenicity
Ability of an agent to produce disease in a range of hosts under a range of environmental conditions
Definition of virulence
Measure of the severity of disease caused by a specific agent. It is commonly quantified using the LD50 (= numbers of agents required to kill 50% of exposed susceptible population under controlled conditions)
Definition of carrier state
A true carrier state is characterised by an infected host who is capable of dissemination of the agent, but typically does not show evidence of clinical disease.
Definition of incubatory carrier
Incubatory carriers are infected, disseminate the agent but are in the pre-clinical stage.
Definition of convalescent carriers
Convalescent carriers are infected, disseminate the agent and are in the post-clinical stage.
Definition of antigenic variation
Refers to biological situations where an agent evades the host defence by changing its antigenic characteristics.
Definition of incubation period
Defined as the time between infection and the first appeareance of clinical signs.
Definition of period of communicability
Time during which the infected host is capable of transmitting the agent
What is a natural reservoir of infection
A species is considered a natural reservoir of infection if infection can be maintained within the species population without requiring periodic re-introduction.
Unified concept of causation criteria:
Developed by Evan, is accepted for identifying cause-effect relationships.
- The proportion of individuals with the disease should be higher in those exposed to the cause than in those not exposed.
- exposure to the cause should be more common in cases than in those without the disease
- number of new cases should be higher in those exposed to the case than in those not exposed (as shown in prospective studies)
- temporally, the disease should follow exposure to the cause
- There should be a measurable biological spectrum of host responses
- The host response should be repeatable following exposure to the cause
- The disease should be reproducible experimentally
- preventing or modifying the host response should decrease or eliminate the expression of disease
- elimination of the cause shoud result in a lower incidence of the disease
- the relationship should be biologically and epidemiologically plausible
What are necessary and sufficient causes of disease?
Cause of disease can be categorised into:
- necessary causes: must be present for a disease to occur (e.g. distemper virus in canine distemper)
- sufficient causes: set of minimal conditions and events inevitably producing disease.
False Negative Rate (β)
beta = type II error
beta = 1 - Se
beta = False Neg / (True Pos + False Neg)
False Positive Rate (alpha)
alpha = type I error
alpha = 1 - Sp
alpha = False Pos / (False Pos + True Neg)
Positive likelihood ratio calculation
Positive likelihood ratio (LR+) = Se / (1 - Sp)
[LR+ = True Pos Rate / False Pos Rate]
Interpretation: likelihood ratio for a positive test = 3 this means that a positive test (in this case PFL>=4) is 3 times more likely to come from an animal with disease (e.g. SL) than from an animal with NSL.
Negative likelihood ratio calculation
Negative likelihood ratio (LR-) = (1 - Se) / Sp
[LR- = False Neg Rate / True Neg Rate]
Accuracy calculation
ACC = (True Pos + True Neg) / Population
Diagnostic Odds ratio (DOR)
DOR = LR+ / LR-
(positive likelihood ratio / negative likelihood ratio)
calculation of True Positives (a) with prev and test parameters
True Pos = True Prev * Se
calculation of False Positives (b) (with prev)
False Pos = (1 - True Prev) (1 - Sp)
calculation of True Negatives (d) with prev and test parameters
True Neg = (1 - True Prev) * Sp
calculation of False Negatives (c) with prev and test parameters
False Neg = True Prev * (1 - Se)
Strategies for selection of an appropriate test to rule out disease and to find evidence of disease
- If the objective of diagnostic testing is to rule out disease, it means that a reliable negative result is required and therefore the test should generate few false negatives (=high sensitivity).
- In contrast, in order to find evidence (=rule in) of true disease and minimise false positive results, a reliable positive result is required with few false positives (=high specificity).
Methods for choosing normal / abnormal criteria (i.e. criteria for deriving cut-off values)
- Gaussian distribution method
- percentile
- therapeutic
- risk factor
- dianostic or predictive value
- culturally desirable
