exam 2 Flashcards
how we define normal
Abnormal as unusual:
Gaussian: Mean +/- 2 standard deviations
Percentile: 2.5 th to 97.5 th percentile
-abnormal associated with disease: diagnostic comparison with gold standard.
problems with gaussian and percentile definitions of normal
-not all diagnostic tests fit gaussian distribution
-both methods assume all diseases have the same prevelance.
-leads to the “diagnosis of nondisease” where 95% of normal subjects fall within the normal reference range and 5% do not.
diagnostic test
-may include
any technique that differentiates
healthy from diseased individuals
or between different diseases
Accuracy
Degree of agreement between the estimated value (test result or measurement) and the true value.
Accuracy is the quality of a test or measurement reflecting its validity (lack of bias) + reproducibility (precision or repeatability)
accuracy = validity + reliability
Validity
Ability to measure what it is
supposed to measure,
without being influenced by other sources of systematic
errors.
VALID = UNBIASED
but does not ensure accuracy.
Valid not always repeatable
Reliability
-The tendency to give the
same results on repeated
measures of the same
sample.
A reliable test gives repeatable
results, usually over time,
locations or populations, but
does not ensure accuracy
sources of false pos and neg results
-Laboratory error: depends on both analytical accuracy and precision.
Improper sample handling
Recording errors
sources of false pos and false negs (test quest)
False negative results
improper timing of test
wrong sample
natural or induced tolerance
non-specific inhibitors
False positive results
group cross-reactions
cross contamination
how to know if test is valid
The accuracy of any diagnostic test
should be established by a “blind” comparison to an independent and
valid criterion for infection or
disease status - the gold standard
Pathognomonic tests
Absolute predictor of disease or disease agent
Can have false negatives
Eg: Culture of MAP
Eg: Culture of T. foetus
Surrogate Tests
Detect secondary changes that will hopefully predict the
presence or absence of disease or the disease agent
Can have false negatives and false positives
how to choose test for our purposes.
-when selecting a test need to know 2 things
- diagnostic validity of test and sensitivity and specificity.
understanding our test subject for choosing a test
What is the prevalence of this disease
in the source population for our subject ? or
What is the pre-test probability that our patient has the disease ?
Sources: signalment, history, and clinical examination, published literature and clinical judgement
Sensitivity
the proportion of subjects with the disease who have a positive test
indicates how good a test is at detecting disease
1 – False negative rate
SnNout: When using tests
with very high sensitivity,
Negative results help to
Rule-Out disease
-the more sensitive the test the less false negatives
Specificity
the proportion of subjects without the
disease who have a negative test
indicates how good the test is at identifying
the non-diseased
1 – False positive rate
SpPin: When using tests with very high specificity, Positive results help to Rule-In disease
relationship between sensitivity and specificity
To distinguish positive & negative test results we need to define a
cut-off value
The cut-off will
determine
the sensitivity and
specificity of the
diagnostic test
Prevalence (or True Prevalence)
The proportion of the population who
have the infection under study (or
disease) at one point in time.
-true provenance (a+c)/n
Positive Predictive Value
The proportion of patients with positive test results who have the
target disorder which acturally have the disease in question***
- Affected by sensitivity, specificity and prevalence
-PPV+ a/a+b
Negative Predictive Value
The proportion of animals with
negative test results who don’t
have the target disorder, true negative
Affected by sensitivity, specificity
and prevalence of disease
-NPV=d/c+d
best tests to rule out disease
Negative test with high sensitivity and NPV
best test to confirm (or rule in) disease
Positive test with high specificity and PPV
parallel testing
- 2 or more different tests are performed
and interpreted simultaneously. to increase our chances of finding disease. - An animal is considered positive if it reacts positively to one or the other or both tests.
Increased sensitivity and NPV
More confident in negative test results
Patient must prove it is healthy
-false negatives are decreased
Serial Testing
Test are conducted sequentially based
on the results of a previous test
-max specificity and improves predictive value
An animal is only classified as positive if
it is positive on both tests
-Patient must prove it has the condition!
-false positives are decreased
Repeat Testing (modified serial testing)
-Negative (herd) re-testing
- Test negative animals are re-tested
with the same test at regular intervals
Forms the basis of test and removal
programs designed to eradicate disease
Improves aggregate-level sensitivity
ex. johnes disease, heart worm.
best test for trying to rule out a disease
-use a test with a high sensitivity and high neg predictive value.
-works best when pre test probability of disease is low.
-SnNOUT
best test for trying to rule in a disease or confirm diagnosis?
Use a test with high specificity
and a high PPV.
Works best when pre-test
probability of disease is high
SpPIN
What is the cost of a false
negative test?
-need high sens tests even at the expense of specificity.
-false negs can have large consequences.
-avoid at all costs
-do multiple tests interpreted in parallel.
What is the cost of a false
positive test?
High treatment costs
Treatments that are potentially
dangerous
Euthanasia of valuable animal might be
possible
Use highly specific tests
Multiple tests should be interpreted in series
choosing a cut point
-this is a continous outsome for a diagnostic test. ex enzyme levels.
- It is impossible to find a cut-point which will
perfectly discriminate between the 2 groups
of animals
Select a low cut-off point and you get a good sensitivity
False negative results are not acceptable
False positives could be controlled by using a second more specific test on initial positive results
Consequences of false positives are not severe
Disease can be treated, untreated cases are fatal
Select a high cut-off point and you get a good specificity
False positive results are not
acceptable
False negative results are controlled by using a second test in parallel
False negative consequences are not
severe
Disease is severe but confirmation
has little impact in terms of therapy or prevention
cut off point values
Cut-off values are flexible!
May vary between populations and test purposes
May have an intermediate zone (grey or “fuzzy” zone)
Animals with results in this zone are re-tested after a certain time period
Receiver Operator
Characteristic (ROC) Curves
- Graph the true-positive rate on vertical axis (Sensitivity)
- Against false-positive rate on horizontal axis (1-Specificity)
- Point closest to top left corner will maximize sensitivity and specificity
-consider costs of false pos and negs because when we set our cut off points.
mass screening
-Sampling volunteers or a sample of the population to detect disease
-Seeking an early diagnosis when a client brings
animal to veterinarian for unrelated reasons.
-specificity is most important
Volunteer Effect
Clients that bring animals for screening tests are not the same as those that don’t
Better management, improved health status
Therefore, more likely to live longer etc.
Zero Time Shift or Lead Time Bias
Comparing survival times after early diagnosis to survival times after conventional diagnosis
The “zero point” for the survival time is time of diagnosis.
If early diagnosis takes place before conventional diagnosis, and lead time is not taken into account = Lead time bias
-does early diagnosis actually give the animal a longer life, not just a longer life from the time of diagnosis.
Length Time Bias
Long pre-clinical phase diseases usually have a long clinical phase
Short pre-clinical phase diseases usually have a short clinical phase
Therefore, those diseases that are more likely to be detected by screening tests will survive longer
than those that are not
Hazards of Early Diagnosis
Marketing our treatment on clients!
Need to be sure of efficacy!
False Positive Risk: Especially
important if there is a debilitating
treatment involved
Labeling: More important in human
medicine
Herd testing
Determine prevalence of infected
herds
Certify herds as disease negative
for eradication or trade
Examine risk factors for herd level
disease
Differences between herd and
individual tests
Uncertainty around individual Se
and Sp are amplified in HSe and
HSp
Bias in individual Se and Sp are
amplified in HSe and HSp
Impact of false results is
generally greater
Screening Tests on a Herd Basis
A positive test is not a positive
diagnosis
False positives can occur in clean
herds
Note that as disease prevalence drops
in the herd, the PPV of the screening
test gets progressively worse
Need tests with very high specificity
especially when prevalence is low
Pooled Samples (the lab pools samples)
Ideal in situations where within-herd prevalence is low
-when we are pretty sure there is no disease there but we want to be sure.
Pros:
-Decreased laboratory cost
-Increased HSe due to increased n
Cons:
- Risk of decreased Se due to dilution
- Logistical challenges of mixing sample
what to do when test has no gold standard: measure agreement
Comparing 2 tests agreement (neither of which is a gold standard)
Comparing agreement between 2
clinicians
Comparing agreement within clinicians
-want to know kappa statistic
-can also do latent class analysis
Kappa Statistic
-The proportion of agreement
measured beyond that expected
by chance alone
-higher the kappa % better the test is 70> is decent. above 80 is great.
Latent class analysis
- None of the tests are treated as the
imperfect gold standard - Maximum Likelihood Estimation (MLE)
– Independence between tests required
– Observed data - Bayesian
– Allows for correlated tests
– Prior information + observed data
outbreak and outbreak investigation
-a series of events clustered in time and space
A systematic procedure
to identify causes (risk factors) of disease
outbreaks and impaired
productivity
objectives of outbreak investigation
- Halt the progress of disease
- Determine reasons for the outbreak
- Recommend procedures to reduce the chance of future outbreaks
procedures for investigating herd outbreaks
the 5 W
DEFINE THE PROBLEM –
“WHAT” ( Establish the existence of the outbreak
or “sub-optimal” productivity problem)
DEFINE THE GROUPS - Describe
“WHO” was affected
“WHEN” and “WHERE”
Collect samples
-establish WHY
-TAKE ACTION
-FOLLOW UP
data gathering in disease control
-gather data to make good case definition to compare cases to non-cases
Comparing clinical cases to sub-
clinical cases
identifying important groups/ collecting samples in disease control
Establish herd inventory:
*Body condition score
*Establish pregnancy status
*Record group affiliation
-need to collect samples from enough animals to recognize patterns across groups. (acute vs chronic, young/old, ect.)
-timing of samples can be critical
The 7 S’s for SAMPLING
SUCK= blood
SCOOP =poop
SWAB =nose, eyes, etc.
SLICE =necropsies
SPOON =feed
SIPHON =water
SPECIFY =identify
lab sampling can establish..
-or verify the pathological and etiological diagnosis
Recognize that in many cases only establishing a definitive pathologic or etiologic
diagnosis does NOT solve the producer’s problem.
establish risk factors
Identify important groups and look
for patterns of disease
Look for patterns and natural
experiments
Orient by subject, place, and time, location
Epidemic curves (could show point source of epidemic or if sporadic) endemic if steady.
attack rate tables to evaluate risk factors
-attack rate tables: compare % of sick animals across suspected risk factors (exposed vs unexposed)
-get info by examining herd records, finding data is the hard part.
-incidence= # new cases in period/ pop at risk (during a given period of time) = risk or ATTACK rate.
-include all relevent risk factors on attack table. the risk factor with the HIGHEST RELATIVE RISK is where you start looking for cause of the problem.
prevalence
= # existing cases/ population at risk (disease AT some point in time)
incidence = attack rate= risk
Cumulative Incidence = # new cases in period / total population at risk
(during a given period of time)
case control studies and key determinants
-when doing a case control comparison to find disease use ODDS RATIO not RR.
- Identify the key determinants for disease:
Key determinants are those risk factors causing the problem that CAN BE MODIFIED on this premises.
taking action in disease investigation
Create a list of action items
Be very specific
Recommendations must be
appropriate for individual herd management
-Written report with recommendations for action, include plans for follow-up.
causation def and why be concerned?
Why be concerned with cause?
– So that we can intervene and prevent disease
Basic definition of cause
– Any factor that produces a change in the severity or frequency of the outcome.
Do not need to understand all causal factors to prevent or at least control disease
experimental evidence
-Traditionally the “gold standard” for causation was an experiment.
In experiments, we randomize individuals to receive a factor and some to receive nothing (or
a placebo or standard treatment).
We know factor precedes disease and other variables accounted for by randomization.
We contrast outcomes in treatment and control group.
-we know it works but don’t know if it will work in a real world situation, difficult to duplicate realistic dose, exposure pathway.
observation evidence
In observational studies, we estimate the outcome differences between individuals that happen to vary in their exposure status.
-real world applications
Use matching and restriction where appropriate to minimize differences between groups.
Measure association between changes in exposure and outcome.
-best ones: meta analysis and systematic reviews. then controlled.
interpreting observational studies
-there’s an exposure, and a disease or outcome and were looking for the association.
-need to compare: to nothing, to treatment, to care.
Cohort studies (observational)
- We start by defining groups
(cohorts) of animals according
to the exposure of the animals
in the groups to the factors of
interest. - We then follow these groups
forward in time to see which
animals develop the disease
under investigation
Compare risk in exposed an
unexposed groups and report as
the relative risk.
Can look at more than one
disease resulting from a specific
type of exposure.
Observational study type closest
to the RCCT, easiest to interpret
ex. exposure + cows of BSE and exposure - cows of BSE= what % have disease.
Case-control studies
- Define groups of diseased and
healthy (or control) animals - Then assess whether the animals in the two groups have
differences in past exposure to
different risk factors
We calculate the odds ratio to indirectly estimate relative risk
Good for studying rare diseases.
Can assess more than one exposure in the same study
-was exposure before initiation of disease?? hard to prove
Statistical significance
-dose not = causation but helps us put pieces together
Demonstration of statistically
significant association does
not prove a factor is causal.
To “prove” causal association
we need to describe a chain
of events, from cause to
effect, at the molecular level
confounding
Confounding is the effect of an outside variable that can wholly or partly account for an apparent association between variables in an investigation.
Confounding can produce a spurious association between study variables, or can mask a real association
A confounder must:
- Be associated with the response
variable - Be associated with the risk factor (exposure or treatment) of interest
- Not be an intervening or
intermediate step between the risk
factor and response.
Ex: A New Zealand study revealed
that wearing an apron during milking was associated with an increased risk of contracting leptospirosis, apron was not the variable it was a large herd size. larger herd size workers wore more aprons. larger herd size is associated with leptospirosis and was the confounder.
Component model of causation
ALL disease is multifactorial.
A cause is described as sufficient
if it inevitably produces an effect.
A sufficient cause virtually always comprises a number of
component causes.
A particular disease may be
produced by different sufficient
causes
If a risk factor is a component
of every sufficient cause then it
is described as a necessary
cause
Causal complement
– The shared component causes
that make up a sufficient cause.
* ie. equilibrium disorder,
slippery walkway,
no grip shoes,
no handrail,
strong wind,
osteoporosis,
and other unknown factors
= fall and broken hip
interaction among causes
Two or more component causes
acting in the same sufficient causes interact causally to produce disease.
goal of causation models
Removal of one or more
components from a
sufficient cause will then
prevent disease produced by
that sufficient cause
causal webs
Direct and indirect causes may also be thought of as representing a chain of actions with indirect causes activating direct causes producing a “web of causation”
-things have to happen in a particular order.
Direct causes are often the proximal causes emphasized in therapy.
Indirect cause is one where the effects of exposure are mediated through one or more intervening variables
-ex. nutrional def-> dystocias-> weak calves, no colostrum-> disease.
Hill’s criteria for causality good for in court and proving causation
-Temporality
-Strength of association
-Biological gradient or dose
response
- Coherence or plausibility
Consistency
Specificity
Analogy
Experimental evidence
time sequence for causailty (hills)
– Cause must always precede
effect in time
– But the same factor could occur
again after disease in some
individuals
– Often difficult to establish time
sequence, especially with surrogate exposure measures
causality strength of association
– A strong statistically significant
association between a factor and
disease increases the likelihood
that the factor is causal
– Assumes that it is less likely
residual confounding could
explain the result
Hill’s criteria for causality
Strength of association
– Problem is strength of association
depends on distribution of other
components of the sufficient cause
– Important weak associations have
been considered causal:
environmental tobacco smoke
and lung cancer
– Some strong associations are due to
confounding: birth order and Down’s
syndrome
Hill’s criteria for causality
Consistency:
– Repeated observations of an association
in different populations under different
circumstances
– Associations can be causal under
unusual circumstances
– Statistical significance should not be
used to assess consistency
cause inference
-There must be a mathematical
association between the exposure
and the hypothesized effect or
outcome
– In most cases, the outcome (disease)
have a monotonic association with the
increasing exposure.
Besides temporality, there are no
criteria that are either necessary or
sufficient.
-the observed association must NOT be due to error or chance or systematic error in the design of the study or data.
biological gradient
– A dose-response relationship between a factor and disease increases the plausibility of a factor being causal
– Exceptions to linear change: threshold effects
– Most should have a monotonic – a gradient that never changes direction
– Exception: alcohol consumption and death – J shaped curve
for infection to occur we need:
A susceptible host
Effective contact with an infectious host
Probability of contact with an infectious host depends on number of contacts with others in the population and prevalence of infection in that population
The likelihood of transmission given contact depends on the number of organisms to which the animal is exposed, the characteristics of the infectious agent, and route of transmission (presence of innate resistance or natural barriers)
biosecurity
Precautions taken to reduce the risk of
exposure to disease
Preventing introduction of infectious
disease
Minimizing the risk of disease
transmission
Biosecurity A-RITS
Assess – take look at what can go wrong, do often, want to reduce, control and eliminate risk
Resist – resistance
Isolate –
Traffic –
Sanitation-
resistance
Resistance refers to the animal’s disease defense (immune system) mechanisms having the ability to not become infected if exposed.
Increase resistance to infectious diseases
implement a strategic vaccination program
reduce stress on animals from other diseases, poor nutrition, housing and lack of consistency in management.
-ex. getting colostrum reduces risk
isolation
Prevent the introduction of infected animals
Keep a closed herd.
A herd is not closed if:
animals are purchased or boarded
animals share a fence line
bulls are purchased, borrowed or loaned
animals are transported by someone else or in someone else’s vehicle
isolation preventing risk induction of infected animals
-prevent introduction of infected animals
-test to prevent
-transport purchased animal in herd owned truck
-quarantine newly purchased animals
-minimize comingling and movement of infected animals within a facility and of established groups within cattle operations.
-do all in all out management
-separate risk groups to decrease exposure to disease
traffic control
-traffic onto and off the operation
-includes more than vehicles
All animals and people must be considered.
Animals other than cattle include dogs, cats, horses, wildlife, rodents, and birds.
Pest control should be reviewed
3 ways to control disease in populations
-Remove agent:
Removal or treatment of infected hosts
-Stop transmission:
Direct contact with infected host
Indirect contact with contaminated environment
Contact with vectors
-Enhance host resistance
Inherent
Acquired
Methods to control disease in
populations
Selective slaughter*
Depopulation
Quarantine
Mass treatment*
Mass immunization*
Environmental
control*
Education
Applied ecology
Genetic improvement
selective slaughter
-Test and slaughter”
-Deliberate killing of a minority of infected animals to protect the health majority
-Usually involves a method of case finding (ie: a diagnostic screening test)
-Works well early in disease outbreaks and in slowly spreading diseases
Was used early in Brucellosis eradication in
Canada – Would NOT be used now
mass treatment
-Treating all (sick and well)
-Combats diseases occurring at very high
prevalence where depopulation and
slaughter are not economical or viable
-Need safe, cheap and effective
therapeutic agent
-potential problem of disease resistance
mass immunization
-Creating immunity in population which
limits spread and impact of disease
Has been successful in past
Canine distemper, parvo virus,
Basic Reproductive Ratio (R0)
-R 0 (“R Zero”): The average
number of susceptible individuals
that are infected by each infected
individual when all others are
susceptible
-This is a measure of the ease of
transmission of an infectious agent.
R 0 = pcD
p=prob of infection on contact
c = rate of contact
D = duration of infectiousness
Basic Reproductive Number (R 0 )
-For communicable infections to
establish in a population, on average
each infected individual must infect
one or more susceptible individual.
If each infects > 1, the outbreak will
take off.
If on average each infects < 1, the
outbreak will die out.
R* (“Effective R “)
-The average number of susceptible individuals that are infected by each infected individual in the current epidemiologic context.
Depends on
Probability of contact
Probability of transmission given contact
Duration of infectiousness
% of population that is susceptible
Effective Reproductive Number (R*)
-Over the long run, an R* 1 is
required for an infectious agent
to survive in a group.
-The goal of control and
prevention strategies for
infectious disease is to
reduce R * < 1 if not to zero.
critical fraction
fc > 1 - 1/R0
To achieve herd immunity or prevent
an outbreak from progressing we
need to create immunity in this
proportion of the population.
Environmental control
-Utilization of the classic host/agent/
environmental triad to control disease
-Includes management, environmental
control, feeding, husbandry etc.
-Many health management programs
revolve around environmental hygiene
eg: ventilation management in barns,
laminitis control in dairy cows
-Disinfection of fomites: surgical
sterilization etc
Environmental Factors potentially
affecting disease control programs
-population density
-housing
-environmental conditions
