Epi Flashcards
What is epidemioglogy?
The study of the frequency, distribution, and determinants of health-related states in populations and the application of such knowledge to control health problems.
Food security
Safely feeding the growing human population
AMR
Antimicrobial resistance (Super bugs)- MRSA, cancer and infectious diseases, HIV-AIDs
EIDs
Emerging infectious diseases (mutation, mixing, and trafficking of pathogens; encroachment on wild animal habitat)
epi triad of interacting causal factors
Infectivity
Ability of an agent to infect. In a population of infected animals, there may be subclinical infection, mild, moderate, and severe clinical signs or even death
Virulence equation
Virulence (%) = count of severe and dead cases/ all infected
Case Fatality Rate (CFR)
100 cases of disease, 80 dropped dead- 80%
Pathogenicity
Infectivity and virulence
What is the iceberg effect related to epi?
Within a population, subclinical/ asymptomatic disease is often the big problem
What is meant by the natural history of disease?
Hosts
Animal capable of being infected. Replication/ development of the agent typically occurs in a host.
Carriers
Infected host without clinical signs potential source for others
Reservoir
Source where agent normally lives (wildlife, soil, water)
Vector
Animate object that transmits infection (insects/ vector-borne viruses)
Vehicle
Inanimate object that transmits infection (transmisson on fomites)
TADs
Transboundary animal diseases
EADs
Emergency animal diseases
What is an EID?
Newly recognized in a population, known for some time but rapidly increasing in incidence or range OR was maintained in a reservoir poopulation waiting to “spill-over”
* Infectious diseases cause > 25% of global human deaths
* 60% of EIDs are zoonoses
* 72% of these are from wildlife reservoirs
What is a Transboundary animal disease?
Rapid spread over national borders, serious socio-economic or public health consequences, major importance in trade of animals/animal products
What are emergency animal diseases?
An animal disease that requires an emergency response.
* Exotic- has penetrated quarantine barriers (rabies, FMD, equine influenza)
* Emerging- start within Australia, previously unrecognized (Hendra, Aust. Bat Lyssavirus)
* Re-emerging- known to occur in Australia, but spreading more widely than previously (anthrax, HPAI, bluetongue)
Sporadic
No pattern
* a reservoir host and only infrequently comes in contact with this host
OR
* carriers (hosts with inapparent infection)
e.g. Hendra Virus
Endemic
Always present
* predictable long term pattern
* stable long-term balance between host and agent and environment
* source of major ongoing losses in animal production systems (internal & external parasites, mastitis, abortions)
* environment invariably important, disease in one region and not another
* incidence is not truly constant over time: - long term trends, seasonal variation, and cyclical variation
e.g. Ross River Virus- endemic mosquito borne virus in Victoria
Epidemic
More than expected.
* a new strain (mutation)
* naive host population (viral trafficking)
* environmental shift (climatic change)
Pandemic
Spreading globally
Point source epidemic
* Very steep up-slope, gradual down-slope
* subsides as no further susceptible animals at risk
Incubation periods effect on a propagating epidemic curve
Delays commencement and prolongs outbreak, waves may correspond to incubation period
Herd immunity
If >75% of population resistant to infection, disease won’t propagate, cycling each time susceptible increase
Density
influences number of contacts, effectiveness of those contacts
Does virulence make sense for the agent?
Over time the relationship between agent and host often moves from parasitic to commensal- balance between host and agent in a given environment.
If illness impairs disease transmission, it doesn’t favor the agent. Not always the case. Key factor is mode of transmission
Does virulence make sense for Rabies?
The pathogenic process changes behaviour and increases transmission
Does virulence make sense for anthrax?
sit and wait strategy- bacteria discharged from dead animals and form resistant spores in the soil- waiting for the next animal to come along
Does virulence make sense for Influenza?
Evolving and evading immunity- RNA viruses mutate rapidly and have the capacity for antigenic drift to contiually evade immunity. Occasional re-assortment of genes, can lead to changes in infectivity, host range and virulence.
H5N1
Highly infectious and ebola-like virulence in chickens, carried in ducks. Low infectivity but high virulence in humans. Infectivity in humans relates to ability of virus to bind to sialic acid on cells (N protein)- agent and host factors- familial association with spread. Question still remains if this virus mutates for effective human to human transmission.
Drivers of spread
Reproductive ratios and herd immunity and vaccination
Reproductive ratio calculation
Reproductive ratio
* a measure of the transmission potential of an infection
* the average number of secondary cases from one primary case in a 100% susceptible popultion
*R0 >/ 1 disease will spread in a population
R0\< 1 then disease will be eliminated in a population
* disease control programs aim to reduce R<1
* it is not static or uniform across a population
* it is influenced by Agent, Environment, Host (triad)
Host factors
Genetics, predisposing factors, age, immunity
Agent factors
Stain virulence, transmissibility
Environmental factors
Persistence of pathogen, population density, and amount of contact between animals
Effective reproductive ratio equation
Herd immunity threshold equation
Good case definition specifies:
- population at risk
- what distinguishes cases from the rest of the population
Measures of disease (3)
Measures of morbidity
* prevalence (not a rate, a proportion)
* Incidence risk (not a rate, a proportion)
* Incidence rate (the main measure of morbidity)
* Attack rate (incidence estimate in outbreak setting)
Measures of mortality
* mortality rate (the main measure of mortality)
* case fatality rate (not a rate, a proportion)
* proportional mortality rate (comparing only amongst the dead)
Morbidity answers what question?
How much disease is there in the population?
Prevalence
The proportion of cases measured once at a specific point in time (includes new and old cases)
Incidence
* a true rate measurement- in units of time
The frequency of new cases of disease observed in a specified population in a specified time period
* measures how frequently susceptible individuals become cases over time
* a risk (or rate) derived from three pieces of info: numerator, denominator, and referent time period
Example: The incidence risk of dystocia in a herd over the previous twelve months was 5 calves per 100 calvings
* a quick approach might be looking at the beginning and the end and take the average- this might be useful for large numbers/ hard to track if the host was there and at risk the whole time OR any new host coming in pretending they came in at 1/2 time (divided by 2)
Measures of morbidity equation
Incidence equations
Case fatality rate equation and definition
Not a rate, no time dimension
Proportional mortality rate
Causation vs. association
Single cause outcomes are the exception rather than the rule.
* presence or absence of disease depends on a complex interplay of factors
* epi helps us understand what factors are involved in a causal pathway to disease
* the relative importance of each factor as a determinant of disease
* allows interventions to be targeted at the most important determinants (more efficient use of resources)
Cause
An event, condition, or characteristic without which the disease would not have occurred
- must precede the effect
- can involve host or environmental factors
- can be either positive (presence of an exposure causes disease) or negative (protective, e.g. vaccination)
Confounding
An alternative explanation for the association between a factor and an outcome
Sufficient Causes
The whole pie, a set of conditions without any one of which the disease would not occur, not usually a signle factor, often several
Necessary causes
A is present in both (all) causal complexes
One that must be present for the disease to occur (the most important piece). Component causes may not necessarily be involved in every causal complex and may not be sufficient on their own to caue disease (not everyone who smokes gets lung cancer and not everyone who gets lung cancer smoked)
Component Cause
(A, B, C, D) Make up the pieces of the pie. Factors such as high cholesterol, smoking, lack of exercise, genetics and the presence of concurrent disease are all component causes of coronary heart disease in humans.
* act far apart in time
* involve the presence of a causative exposure or lack of a preventive exposure
* prevents the completion of the sufficient cause and therefore prevents the occurence of disease by that pathway
* completion of a sufficient cause is synonymous with occurrence (althought not necessarily diagnosis) of disease
Types of causes
Predispose (age, sex, previous illness); enable (low income, poor nutrition); precipitate (exposure to a specific disease agent- tipping over the edge); reinforce (repeated exposure- may aggravate an established disease or state); interact- the effect of two or more causes acting together is often greater than would be expected on the basis of summing the individual effects
Koch’s postulates
* the agent has to be present in every case of the disease
* the agent has to be isolated and grown in pure culture
* the agent has to cause disease when inoculated into a susceptible animal and the agent must then be able to be recovered from that animal and identified
Evan’s unified concept causation
* criteria for judging whether exposures cause disease
* if association, to determine if it is causal:
- proportion of individuals with disease should be higher in exposed than not exposed
- exposure should be more common in cases than non-cases
- new cases should be higher in exposed to not exposed as shown in prospective studies
- the disease should follow exposure (temporal sequence)
- biological specturm of host responses (dose-effect)
- the disease should be reproducible experimentally
- preventing/modifying the host response should decrease disease
- elimination of the putative cause should decrease disease
Risk
The likelihood (probability) of experiencing disease in a defined time period, estimated from previous incidence
Risk Factors
Those characteristics of some individuals taht, on the basis of epidemiological evidence, are associated with increased risk of disease
Measures of strength
Risk ratios and odds ratios
Measures of effect
attributable risk (ARexp), attributable fraction (AF)
Measures of total effect in the population
- population attributable risk (PAR)
- population attributable fraction (PAF)
Attributable Risk
The incidence risk of disease in the exposed that is attributable to exposure
Attributable fraction
The proportion of disease in the exposed that is due to exposure
Populations attributable risk (PAR)
The incidence risk of disease in the population attributable to exposure
Population attributable factor (PAF)
the proportion of disease in a population that is due to exposure
What is surveillance?
The systemic ongion collection, collation and analysis of info related to animal health and the timely dissemination of info to those who need to know so that action can be taken
Monitoring vs. Surveillance
Monitoring: Regularly observing the health of a population to observe trends in disease
e.g. Incidence and Prevalence- compare disease events in different areas- need to know population size to workout out disease per 100 animals. Cannot do absolute comparisons with say Tasi and VIC because unless you do per 100 animals at risk
Surveillance: observing the health/disease status of a pop and taking action when disease reaches a specific threshold
KEY: you are acting
Why is surveillance important?
- work out which diseases are important
- estimate prevalence and therefore set priorities for use of resources
- plan, implement, and evaluate control programmes
- detect emerging disease syndromes
- meet international reporting requirements
- demonstrate disease freedom to trading partners
Components of a surveillance system
- clearly defined objectives
- the hazard or health state under surveillance
- case definition (who, what, when where why)
- target pops- region, species, farms, animals
- timing of sampling intervals
- data mgmt- capture, collation, and cleaning
- methods for data analysis and triggers for action
- feedback and dissemination of results
e. g. Bovine spongiform encephalopathy- shut down the meat and bonemeal industry- 5 year incubation takes a long time to tell if it worked
Where does the analysis happen?
AHA- NAHIS database- national database (Central Animal Health Database)
AND
DAFF in Canberra (reports to OIE)
Notifiable disease in VIC
Early recognition is key
World Organization for Animal Health (OIE)
DAFF in Canberra reports to them
Key to set up trade with other countries
Types of surveillance systems
* Active vs. passive
* Scanning vs. Targeted
* Risk based surveillance
* Syndromic surveillance
* Sentinel surveillance
* Participatory surveillance
Active vs. Passive surveillance
Active: main users of info make active efforts to collect data. Surveys, meat inspection, diagnostic lab reporting
Passive: main users of info take no action to initiate data collection- they wait for data to come to them. Makes use of secondary data.
Scanning vs. Targeted surveillance
Scanning: also called general- keeping a finger on the pulse; monitors an animal pop to detect the undefined and unexpected
Targeted: answers a specific question about a defined disease or condition to be detected
Risk Based Surveillance
Two meaning to the word risk in this context
- Preferential surveillance for hazard that have more serious consequences
- Preferential testing in sub-pops that have a higher risk of having disease
e. g. TSE surveillance in AUS. Targeting downer cows on farm or abbatoirs- looking at their brains for BSE.
OR Avian Influenza in Bhutan (in the Himalayas)- Highly Pathogenic Avian Influenza (HPAI)- surrounded by countries with avian influenza. Few resources. Risk factors: border areas, areas where commercial and non-commercial poultry pop are kept, bodies of water- rivers and wetlands (wild birds), road networks. CREATE A MAP taking each of the risk factors. Layering the maps together. Each map is a grid- so 0-1 scores to classify each cell of the map. Presence or absence (1 or 0). Then add them up- consider which factors are more important than others (multiply by a factor)
**Blue is four risk factors present** Better defined where to look for Avian Influenza
Syndromic Surveillance
Automated data collection methods- statistical alerts that allow to monitor in real time or near real time
e.g. seasonal flue surveillance using google search (running nose- they type into google how do I treat a runny nose)
Sentinel surveillance
- sentinel flocks/ herds of animals in the pop of interest are regularly tested for disease or antibodies to disease
- sentinels intially free of disease of interest
- when test positive, it is known that the disease is circulating in that general animal population
- testing focussed on just the sentinel animals so cost of testing is reduced
e. g. put animals out into areas and look for disease in the animals
Victorian arbovirus disease surveillance. Sentinel chickens- checked regularly for arbovirus.
Participatory surveillance
Going out and asking people what is happening
e.g. HPAI H5N1 in Indonesia- mortalities in chicken flocks
10 step process to outbreak investigation
- Prepare for field work
- Verify the outbreak
- Verify the diagnosis
- Define, ID, and count cases
- Describe outbreak according to invdividual, place, and time
- Develop hypothesis
- Evaluate and test hypothesis
- Follow up investigations
- Implement controls
- Communicate your findings
** sum up in three sentences and 3 key things**
Background
An outbreak is a series of disease events clustered in time
May involve part of a herd, whole herd, or an entire country
e.g. clinical mastitis in a group of dairy heifers, botulism in a feedlot, BSE, FMD, and classical swine fever
* During an outbreak the investigator asks:
- what is the problem?
- can something be don to control it?
- can future occurrences be prevented?
Outbreaks come on tiny cat feet- what does this mean?
Detection of many epidemics relies on the acute astute clinician
What helps working with groups of animals?
We can compare affected animals, both clinical and subclinical, with unaffected animals, in both cross-section (at one point in time) and over time to determine the differences between both the animals themselves and the factors affecting them
What is meant by outbreak investigations being multithreaded and recursive?
Meaning that a number of problems may be occurring at once
Recursive- meaning that new information causes a previous step to be revisited
Stevenson’s First Law of Veterinary Science
Multiple clinical signs in an individual animal tend to have a single aetiology (single pathological process)
In a population though, you have the reverse, a single syndrome in a population (e.g. production inefficiency, disease) tends to have multiple aetiologies (presents in different ways in individual animals)
Three types of problems
Acute- e.g. management or husbandry error
Additive or cyclic- cumulative errors over time e.g. combination of management or husbandry errors over time
Chronic- e.g. long action of managment or husbandry errors that required the passage of time before the consequences became sufficient magnitude to be recognized
Step 3 in an outbreak: Verify the diagnosis
Make a provisional diagnosis if you can
- clinical exams, necropsy, samples, testing etc
Define ID and count cases
Establish a case definitition- establishing a definitive pathological and aetiologic diagnosis, possible to resolve outbreak problems without having a precise aetiologic diagnosis
e.g. Case definition for acute salmonellosis in dairy herds in NZ.
Probable case
Confirmed case- laboratory confirmation
What is meant by enhance surveillance?
Intensive surveillance reqd to accumulate enough cases on to draw conclusions. Get the press involved raising awareness.
What tests are used that fall under the category “measuring or detecting”?
* A physiological or production parameter (WBC count or milk volume)
* the agent, its antigens or nucleic acid
* the animal’s immune response to infection (e.g. Abs) or some other challenge
What is a binary test?
Yes or no. Positive or neg. diseased or not disease.
Binary tests were originally continuous likely.
What is a continuous test?
E.g. blood pressure, blood chemistry, serum enzymes, serology
For continous tests and binary tests, what do we have to do?
Set a cut-off that distinguishes between positives and negatives (normal or abnormal).
What is one of the most common ways of determining the limits of normality? What have most normal references ranges been arbitrarily defined as?
Therefore in a population of 100 normal dogs, what would happen with the reference range?
Use of statistics.
The range of values which captures 95% of all values observed in a population of healthy animals.
95 dogs would have a WBC count for example within the “normal range” and 5 of the normal dogs would have a WBC count outside of that range.
How are references ranges estimated?
From the mean and standard deviation of the population, using the rule that, in normally distributed populations- 95% of the observations lie within 1.96 standard deviations of the mean.
What is a major limitation of the arbitrary statistical approach to describing normality?
A measurement that is outside a reference interval only indicates unusualness, it is NOT a definitive sign of disease. We should expect 5% of healthy animals to fall outside the reference range of any given test.
Some enzymes, for example, are only abnormal if elevated. For example, Amylase, Lipase, and CK (only abnormal if above a certain number). What changes in regards to statistically determining normality?
Within top 5%- > mean + 1.65 SD (as opposed to ~2 SD of the mean)
When are false positives particularly important?
What is the probability of a false positive?
What does this tell us as clinicians?
False positive become particularly important if multiple tests are performed on a healthy subject.
Probability of a false positive = 1- (0.95)^n, where n= the number of tests
ONLY test if there is a likelihood the animal has the disease. Test wisely!
What does fitness for intended purpose refer to with tests?
The test is accurate (precise and reliable)
What does it mean if a test is optimised?
The test has an efficient protocol
What does it mean if a test is standardized?
The test is calibrated having “best” practice “standard” results available to compare against
What does it mean for a test to be robust?
The test’s results are unaffected by small changes in test situation
What does it mean for a test to be repeatable?
Test results are repeatable within and between replicated runs in the same lab
What does it mean for test results to be reproducible?
Test results are repeatable between replicated runs in different labs
What does analytic sensitivity mean in regards to a test?
The limit of detection (smallest detectable amount) of the analyte tested for
What does analytic specificity refer to with a test?
The test distinguishes the target from other compounds.
What does diagnositc seNsitivity refer to?
The test produces few false negatives when tested in (reference) animals of known positive infection status (positive controls) representative of the target population of animals where the test will be applies.
What does diagnostic sPecifity refer to in regards to a test?
The test produces few false positives when tested in (reference) animals of known negative infection status (negative controls) representative of the target population of animals where the test will be applied
What is precision vs. what is accuracy?
Precision is the ability of a test to give consistent results in repeated tests. The result may be wrong, but if it turns up repeatedly the test has a high precision. A precise test is highly repeatable.
Accuracy is the ability of a test to give a true measure of the condition that is being tested for. A test can be precise without being accurate but it cannot be accurate without being precise.
What are pathognomic tests?
Tests that detect a sign, substance, response or tissue change is an absolute predictor of the presence of the disease of disease agent. For example, a positive culture of Brucella abortus from cow’s milk is pathognomic for Brucella infection.
What are surrogate tests?
Tests that detect secondary changes, which it is hoped will predict the presence or absence of disease or the disease agent. For example, Testing for antibodies to Brucell is a surrogate test since it is not measuring the presence of Brucella abortus per se but rather the body’s reaction to Brucella organisms or cross-reacting antigens.
What is DSe?
The probability of a test correctly identifying those animals that are infected or have a specified condition (disease or cases).
Recall: few false negatives= better at identifying animals that have the disease
It says nothing about the test’s accuracy with non-diseased animals. DSe can range from 0 to 1 (0% to 100%)
False negatives is indicative of a test with poor sensitivity (e.g. DSe= 65%)
What is DSp?
The probability of a test correctly identifying those animals that are not infected or which do not have the specified condition (non-diseased or non- cases)
Recall few false positives= better at identifying animals that are not diseased.
If there is poor specificity, then lots of falst positives.
How are diagnostic sensitivity and diagnostic specificity related?
Inversely related. Where test results are measured on a continous scale, DSe and DSp can be varied by changing the cut off value: increase DSe and therefore decrease DSp and vice versa.
If the priority is on finding diseased individuals, you need a test with what?
high DSe (i.e. you don’t might getting a few false positives)
If the priority is to make sure that every test positive is “truly” diseased, what do you need?
A test with a high DSp (i.e. you don’t mind getting a few false negatives)
How do you estimate diagnositc sensitivity and diagnostic specificity? Talk through it.
2 x 2 table
What is true vs. apparent prevalence?
True prevalence is the actual prevalence of disease in the population. Apparent prevalence is the estimate of disease prevalence worked out on the basis of an imperfect test. Apparent prevalence can be more than, less than, or equal to true prevalence.
How do you determine the DSe and DSp of a test? What is the golden rule in order for DSe and DSp to be valid?
Testing a number of animals of known status (reference samples).
The golden rule: In order for diagnositic sensitivity and specificity to be valid, the group of animals selected for the evaluation study should be representative of the population to which the test will be applied. (e.g. the diseased population should contain early and chronic cases as well as subclinical cases, also the “non-affected” population (negative controls) should include animals that may have conditions that could possible cross react the test)
** DSe and DSp are considered to be constant characteristics of a test once the cut-off has been set**
When is a highly sensitive test of most use?
When you want to find infected animals
* there is a high cost involved with calling a positive animal negative (don’t let the killer out of jail)
* There is a high probability of the condition. If the prevalence of disease is very high, most animals will be infected, few animals will be negative so the test needs to work well in a population dominated by positive animals.
When is a highly specific test of most use?
When you want to make sure that a positive test result is really positive.
* There is a low probabily of the condition. If the disease is low prevalence, the test needs to work well in a population dominated by uninfected animals.
* There is a high cost involved in calling a negative animal positive. Therefore negative animals must be identified as negative (don’t convict the innocent)
Why should no test be taken as absolutely correct? (based on what we just learned)
No test has a diagnostic specificity of 100% and no test has diagnostic sensitivity of 100%. E.g. a positive test does not mean that the animal has the disease in question.
What question does positive predictive value (PPV) answer? And NPV? How can you calculate these?
If my patient returns a positive result, what is the probability that it really has this disease?
NPV: If my patient returns a negative result to the test that i use, what is the probability that it is really free of this disease?
What is serial testing?
The animal has to be positive to all tests to be called a positive animal. Serial testing improves diagnostic specificity (but decreases DSe, increasing the chance of a false negative)
What is parallel testing?
If the animal is positive to any one of the tests, then it is called a positive animal. Parallel testing improves diagnostic sensitivity (but decreases DSp, increasing the chance of a false positive)
What is the effect of prevalence (or probability of infection) on PPV and NPV (with DSE = 95% and DPS=95%)?
As prevalence increases PPV increases and NPV decreases (inverse relationship)
Many tests are not worth doing unless you are above a certain percentage of certainty that the condition exists? What is the percentage? Explain
>50%
Using a test when there is low probability of that condition existing will just throw up false positive results.
When is screening done? What is screening?
Screening is done with the objective of early case detection. So best that screening tests have a high predictive value. So screening tests should be highly sensitive in the early stages of the disease control program (when the prevalence is high) and to be highly specific later in the program when the prevalence of disease is low. Because of this change in requirement as the control program advances, it is often necessary to change the test as the prevalence decreases.
What are prevalence surveys?
Random survey testing is often done to determine the prevalence of a disease. Results will only give the apparent prevalence. True prevalence can be calculated.
Describe outbreak according to the individual
Compare affected animals with unaffected animals (age, production level, etc.)
Clustering
An aggregation of disease events in space
e.g. Ross River Virus you’ll see a lot more in Melbourne, however it is because the population at risk is massive- you are actually at less risk in Melbourne
Point Source, Continuous, Propagating (examples and general definition)
Pneumonia: Propagating arching up and down “waves”- creeping up gradually
Legionnaire’s Disease- Point Source- AC- steep up rise and steep down slope
Continuous- series of point sources
Eradication vs. Control
Control implies that you’re doing something to reduce the morbidity and mortality due to disease
Eradication implies that you’re doing something to completely remove the disease from the population, reduce the prevalence of disease to such a level transmission cannot occur
How do you control disease?
Movement restriction (animals, ppl, equip)
Quarantine (length of time proportional to incubation period, import and export restrictions)
Biosecurity (cleaning and disinfection entry protocols, isolate animals or only those sourced from a tested clean population)
Outbreak surveillance and investigationg
Vaccination (blanket or strategic- ring)
Treatment and improved husbandry (reduced clinical phase and reduce shedding)
Criteria for eradication
* Biological and technical feasibility- absence of reservoir, low probability of spread, ability to mount an immune response, ease of diagnosis, effective practical intervention available (e.g. vaccination)
* Political will and popular support- consensus on perceived community benefit, synergy with other interventions, need for eradication rather than control, compensation and financial support for short-term losses
* Preconditions- sound epidemiological understanding of the disease, clear evidence of community benefits (economic analyses), safe effective and feasible diagnostics and vaccine/cure, appropriate control options, availability of resources
Comparing Rinderpest Eradication with Johne’s Disease eradication (which would essentially be impossible)
Cohort Study
Follow two or more groups forward from exposure to outcome over a course of time. Or you can do a cohort retrospectively. Cohort studies are defined based on the outcome.
** USE RELATIVE RISK (have incidence)
Case Control Studies
A lot more common. More susceptible to bias, they are easier to do. Must define criteria for diagnosis of a case, controls should come from the same population as the cases, investigators should blind the data gatherers to the case or control status of participants.
** USE ODDS RATION dealing with prevanlence
What are the three types of studies?
Descriptive (case reports), Analytical (testing a hypothesis. Incidence in one group vs. another group. Cross Sectional- signal point in time, measuring exposure and outcome at the same point in time), Experimental (randomized clinical trials, community trials)
Case report vs. Case Series
Types of descriptive studies
case report- one unusual case
Case series- a series of cases
What kind of study is this an example of? Define this kind of study too? What is the measure of disease frequency?
Cross Sectional Study (type of analytical study)
* sample of individuals from a population is taken at a point in time (e.g. survey)
* indiv. are examined at the same time for the presence of disease and their status with regard to the presence or absence of specified risk factors
* PREVALENCE is what cross sectional measures
* Are they smokers? NOT HELPFUL because snapshot and they can’t smoke inside the airport
What are the advantages and disadvantages of a cross sectional study?
Advantages: quick and relatively inexpensive
Disadvantages: no information about incidence, difficult to investigate cause and effect relationships (establishing correct temporal sequence can be difficult), not appropriate for disease of short duration (milk fever in dairy cattle, claw amputation)
Cross Sectional Study (Analytical)
Cohort Study- what is it?
Compare disease incidence over time between groups (cohorts) that are found to differ in their exposure to a factor of interest
- either prospective or retrospective
- time invariant and time varying exposures can be studied
Cohort Study (analytical study)
What are advantages and disadvantages of cohort studies?
* Provide estimates of teh absolute incidence of disease in exposed vs. non-exposed individuals
* exposure status is recorded before disease has been identified: provides unambiguous info about whether exposure preceded disease
* particularly useful for studying rare exposures
* suitable for studying multiple outcomes
Disadvantages:
* require a long follow up period
* losses to follow up can be a problem (producing non response, migration, and loss to follow up biases)
* expensive
* May require very large samples
* not suitable for rare diseases
* not suitable for disease with long-latency
* sampling, ascertainment and observer biases still possible
Case Control Study Issue?
Key issue when designing a case control study is to ensure that cases and controls are similar in every way except for the exposure factors hypothesized to be associated with the disease of interest
What are some advantages and disadvantages of case control studies?
Advantages:
* quick, relatively inexpensive, multiple exposures can be examined, particularly useful for studying rare diseases, suitable when randomization is unethical (e.g. alcohol and pregnancy outcome)
Disadvantages:
* no information about incidence (we’re only working with a sample of disease negative population we don’t know the TOTAL size of the population at risk)
* can be difficult to investiage cause and effect relationships
* particularly vulnerable to biases arising from selection of subjects (most often of the control group) and measurement (estimation) of exposure (RECALL BIAS!!!)
* multiple outcomes can’t be studied
* if the incidence of exposure is high, can be difficult to show the difference between cases and controls
In a case control study, why is there no information about incidence?
We’re only working with a sample of the disease negative population- we don’t know the total size of the population at risk
What is this an example of?
Example of case control study (analytical)
The key here is to get the right exposed vs. unexposed groups
A flow diagram looking back, is an example of what type of study?
Case Control Study
Ecological study
* Group is the unit of interest (not an individual)
* summarize group level exposure and group level outcomes
* e.g. chlorination of drinking water and cancer mortality in Taiwan– problem with this kind of study– some groups might be older and have a higher prevalence of cancer anyway… association does not equal causation
* Cross level reference- inferencess are made an the individual (rather than a group level) (studies are often misinterpreted by media and lay people)
Advantages and Disadvantages of ecological studies
* Advantages: quick and cheap, data readily available (often collected for other purposes)
* Disadvantages: ecological fallacy: an association observed between variables at the aggregate level may not necessarily represent the association that exists at the individual level
Observational studies
Sit and watch what happens- Descriptive and analytical studies (vs. Experimental)
Experimental studies examples and what they are
Randomised clinical trials, community trials
* designed to test hypotheses between specific exposures and outcomes
* investigator has direct control over study conditions
Important concepts in the design of experimental studies
- Use of a control group- suitable control group- similar to treatment group in everyway except they don’t get the treatment (need the baseline for comparison)– Hawthorne effect- positive effect because of observation, placebo effect- positive effects because of inert intervention, natural improvements in disease conditions over time
- Randomization- randomly allocate individuals to either control group or treatment group– this ensures treatment and controls are similar…
- Admission criteria- uniform characteristics- person, place, time, demographic (age, sex, race for example), prior condition, risk factor restriction (non-smokers)
- Outcome ascertainment- remove bias by blinding– single, double, triple blind (blind subjects, evaluators, and statisticians)
- Ethics- you cannot knowingly expose a participant to known home- so if the treatment group is doing really well then you can stop the study early to treat the control group OR a negative outcome so now CONSORT statement so that there isn’t a bias for published studies– compulsory registration of clinical trials (once they are registered it is easier to ID which have positive and negative outcomes
Randomized Clinical Trial
* Individual (unit of interest)
* set eligibility criteria- non smoker, age, etc.
* process of randomization to asign eligible population into treatment and control groups
Advantages and Disadvantages of Randomized clinical trials
Advantages- randomization generally provides excellent control over extraneous variables (confounders), even factors that may be hard to measure or that may be unknown to the investigator
Disadvantages- for many exposures it may not be ethical or feasible to conduct a clinical trial (e.g. exposure to pollution)
- expensive
- impractical if long periods of follow-up required
Community Trials
*Instead of randomly assigning individuals to treatment or control groups, community trials assign interventions to entire group of individuals
* in the simplest situation one group (community) receives the treatment and another serves as a control
e.g. in Nepal giving supplement to community comparing data from before and after
Answers in lecture 11
Four steps involved in critical appraisal
- Describe the evidence (What relationship was being evaluated and what hypothesis was being tested? What were the exposure variables and what was the outcome variable? What was the study design?– case control- what are typical weaknesses common with these study types–? recall bias? selection of appropriate controls?….. what source population? time frames? Eligibility criteria? Did participation rates differ? Summarize main result in terms of association between exposure and outcome)
- Assess the internal validity of the study (describes the truthfulness of the inferences that we make about the study population (those that actually took part in the study… can we be confident the association we actually measured was valid? Is the relationship that we’ve seen, is it causal or just an association?)
- Assess the external validity of the study (how well can we translate the findings from this study to other populations)
- Compare the results of the study with other available evidence
Internal validity: Non-causal explanations?
Bias, confounding and chance
What are the types of bias? And define them.
- Selection bias: surveillance (dx more likely in animals that are under frequent medical surveillance), diagnosis, referral, non-response, length of stay, survival bias
- Misclassification (information) bias: recall, interviewer (e.g. film clip), prevarification (subjects in a study may have ulterior motices for overestimating exposure), improper analysis, obsequiousness (subjects systemically alter their responses in the direction they perceive to be desired by the investigator) (“Clever Hans” effect)
** if we have severe misclassification or selection bias the study is fundamentally flawed and there is nothing we can do to fix it**
Internal validity: Confounding (lurking variables)
quantify strength between smoking and cancer of the larynx. BUT many smokers are also drinkers- another risk factor. Drinking confounds the association between smoking and laryngeal cancer.
** If we found this, it is often not a show stopper. We can fix this. (unlike bias!)
Internal validity: Chance
Did we get the results of this study by a fluke?
Type I error: null hypothesis is rejected when, in reality, it is true.
Assuming you’re able to eliminate the non-causal explanations, you then move onto assessing the causal explanations:
- Is there a correct temporal relationship?
- Is the rel. strong?
- Dose response with relationship? the greater the exposure, the greater the risk of dx.
- Is it consistent? (if I applied to a whole range of subjects, similar outcome)
External validity
Can the results be applied to populations other than that which was studied? If the internal validity is poor, the answer is no!
If we get past internal validity, three aspects of external validity should be considered:
* Can the results be applied to the eligible population?
* Can the results be applied to the source population?
* Can the results be applied to populations OTHER RELEVANT than which was studied?
How can studies be ranked from highest to lowest in terms of reliability of information:
Comparison with other evidence
