Stats and epi Flashcards
Ratio
Differences in rank order plus equal intervals, plus true zero
Interval
Differences in rank order plus equal intervals
parameter
the mean value within a whole population
statistic
the mean value within a sample
Regression analysis
A type of statistical model that examines the predictive relationship between one or more predictor variables and an outcome variable.
‘Independent variable’ and ‘dependent variable’ are sometimes used as alternative terms, but are best avoided.
aim: 1. To determine how well a pre-selected set of predictor variables predicts values of an outcome variable.2. To identify whichset of predictor variables will best predict values of an outcome variable.3. To predict specific values on the outcome variable from specific values on a set of predictor variables.
Standardized regression coefficients
A regression coefficient can be standardized so that coefficients of predictors measured on different scales can be compared. It measures the change in Y in SD units for a one-SD increase in X. Useful for comparing predictors within a model, but not for comparing a model across samples.
Model fit
R^2
Usually measured by R2. This is the proportion of variance in the outcome variable explained by the predictor variable(s) in the model. Adding more predictors to a model will always improve its goodness of fit.
The adjusted R^2 takes into account the number of predictors and the sample size.
R^2 can be tested for significance; this tells you about the significance of the whole model.
Interaction OR
In regression analysis you can include an effect moderator (e.g. sex) as an interaction term.In a previous example, this could tell you how much larger or smaller the odds ratio for survival would be if the animal were male rather than female.
e. g
1. 47 (OR for females) x 1.06 (OR for interaction term)= 2.53( OR for males)
Descriptive epidemiology
examine the distribution of disease in a population
observing the basic features of its distribution
what are Two broad types of epidemiology
. Descriptive epidemiology
. Analytic epidemiology
Analytic epidemiology
investigate a hypothesis about the cause of disease by studying how exposures relate to disease
The epidemiologic triad/triangle
an external agent, a susceptible host, and an environment that brings the host and agent together.
Exposure
risk factor being investigated, and may or may not be the cause
Outcome
the disease/event/health-related state, we are interested in
randomised control trial
an investigator assigns exposures
the exposures are randomly allocated
the goal is to invesigate prevention and treatment
Experimental, analytical
Prospective cohort design
Starts at the time of exposure (intervention) – follow-up until outcome occurs
Key features:
Control arm (no exposure)
Random allocation of exposure to intervention groups: similar baseline characteristics; similar distribution of confounders
Blinding of participants (e.g. owners) and clinicians (where possible)
Strong evidence for temporal associations
Can investigate multiple outcomes
Low risk of selection bias and confounding
Blinding: reduce risk of information bias
Not suitable for rare outcomes
Not suitable for harmful exposures
Can take a long time (depending on length of follow-up)
non randomised control trial
the investigator assigns exposures
the exposures are not randomly allocated
the goal is to investigate prevention and treatment
cross sectional (prevelance study)
an investigator did not assign exposures
it is a descfriptive study
theres no comparison group
it shows burden and impact
Cross-sectional study design is a type of observational study design.
In a cross-sectional study, the investigator measures the outcome and the exposures in the study participants at the same time.
ecological study
an investigator did not assign exposures it is a descriptive study theres no comparison group it shows burden and impact an observational study defined by the level at which data are analysed, namely at the population or group level, rather than individual level.
case report/series
an investigator did not assign exposures
it is a descriptive study
thers no comparison group
Careful, detailed description of a single case or series of cases (typically by observant clinician(s))
Analysis: narrative description, simple descriptive statistics (case series)
May be the first clues of new diseases, outbreaks, impact of a condition, unsuspected adverse effects, possible exposures
No comparison group – unable to test hypothesised association between exposure and outcome
could be random finding
Lack epidemiological quantities – not chosen from a representative population sample
Publication bias – journals mostly favour positive outcome findings
Overinterpretation – temptation to generalise when there is no clear justification
When less rigorous methodology for research on rare disorders required
When ethical constraints hinder experimental research
Make it possible to make changes in clinical practise – e.g withdrawal of drug from the
market
cohort study
the investigator did not assign exposures
it is an observational study
there is a comparison group
its an analytical study
investigates causes and prognosis
direction: exposure> outcome
Careful, detailed description of study population and exposures (risks)
Starts at the time of the exposure – follow-up until outcome occurs
a type of research design that follow groups of people over time
Stronger evidence for temporal associations
Can investigate multiple outcomes
Lower risk of selection bias and information bias
Not suitable for rare outcomes
Can take a long time (depending on length of follow-up)
Risk of information bias due to attrition (loss to follow-up)
case control study
the investigator did not assign the exposures
there is a comparison group
theres an analytical study
direction: outcome> exposure
Compares cases (diseased animals) and controls (non-diseased animals) with respect to their level of exposure to a suspected risk factor
Starts with the disease (or outcome of interest) and looks back at prior history of exposures
“all the effects are already produced before the investigation begins”
streanghts: Efficient: well-suited to rare diseases
Ideal when long latency between exposure and disease
Relatively quick and inexpensive
Investigate multiple exposures
Limitations
Susceptibility to bias:
Selection bias, information bias
Temporal association difficult to establish
cross sectional study
an investigator does not assigng an exposure it is an observational study
there is a comparison group it is and analytical study
it minvestigates causes and prognosis
a type of observational study design. In a cross-sectional study, the investigator measures the outcome and the exposures in the study participants at the same time.
Relatively quick and inexpensive
Investigate multiple exposures or outcomes
Susceptibility to bias high
Temporal association (nearly always) impossible to establish
Two main bias domains
Selection bias
Information bias
Selection bias
The study sample is not a good representation of the population of interest
Selection bias: selection or participation in a study is related to outcome or exposure
Information bias
Exposures and outcomes are not measured well, or not in a similar way in all study participants (animals)
Information bias: assessment of exposure varies depending on risk of experiencing the outcome / disease status
misclassification bias- arises when a study participant or is categorised into an incorrect category altering the observed association between study categories and the research outcome of interest.
observer bias- Bias that arises when the process of observing and recording information includes systematic discrepancies from the truth.
recall bias-Recall bias is a systematic error that occurs when participants do not remember previous events or experiences accurately or omit details: the accuracy and volume of memories may be influenced by subsequent events and experiences
Avoiding chance findings
Generate sufficiently precise estimates of the strength of an association: sample size!
Use robust statistical methods
parallel testing
the animal is positive if one or more tests are positive
the greatest predictive value is a negative test result
used to rapidly asses individuals
important if there is a penalty for missing the disease
series testing
animal disease positive if all tests are positive-
maximises sp and se and ppv- more confident disease is really present
screening + conformation testing
screening to test every animal (low test cost, high sensitivity) then confirmatory trst on positives that is higher cost and more specific)
used in disease controll programs
positive predictive value
probability that the animal tested positive is truly positive
negative predictive value
possibility that the animal tested negative is truly negative
aggregate testing
sampling and testing groups of animals with the same test
most control programs use this
as prevelance decreases proportion of false positives increase
sensitivity most valued here
negative herd re-testing
positive animals are removed and negative animals are sampled and retesed again
finds missed infections
used in TB
sequential testing
used in experimental studies
repeatedly sample and test animals to detect sero-conversion
powerful as does not rely on single result
looking for significant change in test result
labour intensive
using different tests for different diseases in the same animal
common in small animal- blood paramenters before anasthesia
used in dairy to produce metabolic herd nutritional status
simple random sampling
list all the sampling units in the sampling frame and select at random
systematic sampling
select sampling units at a predefined equal interval e.g. randomly start at no. 17, and then select every 17th animal/herd/flock
stratified sampling
divide the sampling frame into logical groups (strata) and make random selections from within all strata
Cluster sampling
divide the sampling frame into clusters (space or time), and randomly select clusters (one-stage) or also within clusters (two-stage)
Bradford-Hill’s “aspects to consider” when trying to infer causality from an association
- Strength. Very strong associations will generally be harder to explain away by confounding or bias.
- Consistency. An association that is repeatedly observed by different research teams under different circumstances may be less likely to be produced by confounding or bias.
- Specificity. A cause leads to a single effect not multiple effects. [Not to be over-emphasised.]
- Temporality. We should be confident that the exposure preceded the outcome.
- Biological gradient. Is there a dose-response, such that higher levels of exposure have a greater effect?
- Plausibility. Is a causal connection biologically plausible [depends on the state of biological knowledge at the time]
- Coherence. Does a cause-effect interpretation seriously conflict with other established facts about the disease?
- Experimental evidence. Does removal of the cause prevent (some cases of) the disease? [may not be feasible or ethical]
- Analogy. Can we draw any parallels?
name the two catagorical variables
nominal
ordinal
nominal variables
2 or more catagories
no order
female-male
cat,dog,reabbit
name the two numberical variables
discrete
continuous
ordinal variables
ranked e.g disease severity 1= none 2= mild 3= moderate 4= severe
discrete variables
counts of event
e.g number o cattle, no. of visits to vet
continuous variables
take any value in range
weight, age, body temp
describe mean vs median
the mean- makes more use of data
is distorted by outliers or skewed distribution
good for normal distribution
the median-
makes less use of data
better for skewed data
less easy to analyse
both useful to produce histeogram
measures of variablity
with the mean- varience and standard deviation- essentially how different observations are form mean
with median-
range
inerquartile range
descriptive statistics for catagorical variable
frequancy distributins and percentages
desciptive statistics for numerical variables
mean and std deviation or median and IQR
describe 95% confidance intervals
need: point estimate (mean), measure of variability (SD of mean) and sample size
a range of values so defined that there is a specified probability (95%) that the value of a parameter lies within it
small sample size or large variability widens confidance interval because of more uncertanty
to measure the assosiation between two nominal varibles we use
chi- squared (X^2) test
eg: assosiation between canned cat food and feline hyperthyroidism
compares the observed count, to count that would be expected if there was no assosiation between variable and outcome
Expected count = (column total * row total) / grand total
X^2= sum of (observed- expected)^2/ expected
from this the p value can be obtained
(will be listed under asymptotic significance (2-sided)
odds ratio
a measure of the streangth of an assosiation
the odd of an event is the ratio of the probability of occurence of the event to the probability of non occurance
odds = (P/P-1)
odds ration is the ratio fo odds for group 1 to the odds of group 2
(p1/(1-p1))/(p2/(1-p2)
an odds ratio of one means there is no difference
no difference also means the confidence interval would include 1
An odds ratio of less than 1 implies the odds of the event happening in the exposed group are less than in the non-exposed group
how do we test for two independant groups of numerical parametric data
e.g compare mean weight of specific breeds of dogs between deprived and non-deprived areas
unpaired/ 2 sample t-test
regession analysis
a way of mathematically sorting out which of those variables does indeed have an impact
examines the predictive relationship between one or more predictor values and and outcome variable
aims to determine how well the predicter variables predicts the outcome variable
which ones best predict the values of the outcome variable
to predicd specific values on the outcome variable from secific values on a set of predictor values
multivariable regression
either linear or logistic regression where there is more than one predictor variable included, such as the effect of both food type and age on weight
Describe Y=a+bX
Y= the predictor value of outcome variable (e.g in comparing quality of life to the level of independence someone has, the ocv is quality of life) a= the constant (the inercept with the Y value on the graph) b= the coeffienct for x, the amout that y increases or one unit of x x= the value of x
Describe Y=a+b1X1+b2X2
Y= the predictor value of outocme variable (e.g in comparing quality of life to the level of independence someone has, the ocv is quality of life) a= the constant (the inercept with the Y value on the graph b= the coeffienct for x, the amout that y increases or one unit of x x= the value of x
this is done for both variables
standerdised regression coeffiecents
a regression coefficient can be standardised so that the coeffiencts of predictors measured n diffrent scales can be compared
measres the change in Y in SD units for one SD increas in X
Model Fit
R^2
the proportion of varience in the outcome variable explained by the predictor variables in the modle
assing mroe predictors improves goodness of fit
ajusted R^2 takes into account the no. of predictors and sample size
R^2 can be tested for significance which tells you the significance of the whole model
regression coefficients
value of the coefficient is the relationship between the predictor and the outcome variable when the other predictors are held constant (controlled for)
Avian influenza
Highly significant zoonose
diffrerent subtypes more prevelant in certain region than others
severity of the disease in poulrty depends on whether it is HPAI or LPAI
spread by wild birds - high during october through winter and in summer recead as birds migrate
tade offs between managment and welfare as poultry must be housed indoors
free range status of eggs in uk lost- econimic impact
implications for human heath- important source of protien
zoones
african swine fever
NOT a zoonose
pig meat however is a massive source of protien and hence threats to it effects human health risk and larg economic loss
contaminated pork products can spread it
matter of time till hits uk
losses in pig productin no vaccine socioeconomic burder mortality can be 100% a threat to and spread by wild pigs (ticks or direct contact) not zoonotic resistant in environment concern for biodiversity
bovine TB
enormous global disease threat end point of immunosupressed patients significan TB in uk form infected milk in past bad for cattle welfare diagnosis influences spread as it is challenging big iniciatives to control TB movemnt of cattle is a risk wildlife can be resovour
chronic bacterial infection
complicated by persistent infection of wild animals
zoonose- very serious
transmitted by direct contact, ingesting of contaminated material
slow course of diseases- can infect others before clincal signs show
estimated to account for up to 10% of human tb cases
improve testing, reduce transmission between animals and humans, improve collaberation
Rabies
threat from animals coming in to the country quarentine good vaccines good tests sucsessful eradication in UK
viral disease effects nervous system transmitted via saliva- bite zoonotic non specific symptoms incubation weeks to months fatal goal of elimination- vaccine very effective
q fever
notifiable in uk from the last 12 months vaccine available 20% of uk dairy herds seropositive fertility issues debilitating disease for humans that catch it unspecific symptoms fignsed by pcr, serology can survive in enviroment for long time bacterial infection
leptosporosis
bacterial infection " in practice" approach cattle infertility -establish herd lepto status - control and eradicate - vaccination pros and cons can beconme endemic in herds and cause low grade chronic repro losses in niave herds can cause substantil loss common differetail for repro problems passed in aborted fluids and urin so can spread to dairy workers vaccination available
different strains in different parts of the world
possibly returning as vaccination drops
good tests available- good survalence
- dont graze with sheep, understanding danger of watercourses, not testing before trading,
vaccination vs treatment
before lepto vaccine, whole herds were treated with antibiotics
vaccines are now effective- more responsible
eviromental role in one healt
energy security
food securiy
green energy soloutins with in vet practice
Animal healthsurveillance
a tool to monitor disease trends, to facilitate the control ofinfectionorinfestation, to provide data for use inrisk analysis, for animal or public health purposes, to substantiate the rationale forsanitary measuresand for providing assurances to trading partners.’
Data collection for MOSS
Active:
Systemic and regular recording of cases
Population defined by location and/or time
All the population or a sample of?
Depends on objective, expected prevalence, diagnostic tests
Random or targeted sampling
Passive (Scanning):
Relies on notification of disease suspicions and cases – less control
Active data collection: Population under scrutiny
Accurate information for potentially every individual animal
Labour-intensive – lot of sample collection
Expensive – field work, lab diagnostics, administration
Used for control/eradication programmes
(e.g. brucellosis in N. Ireland – see reference below)
Active data collection: Random sample
Estimate of disease prevalence / incidence and to describe temporal trends
Sample size depends on: Expected disease prevalence in population Population size Required precision of estimate Sensitivity and specificity of tests
Expensive if the disease is rare
Active data collection: Targeted sample
Focuses on high-risk population in which specific and commonly-known risk factors exist
Appropriate if:
Disease is less common in general population
Specific risk factors are known
Have knowledge of the epidemiology of the disease
Problem:
Undetected cases may occur in other segments of the population
Passive: Reporting cases
Relies on farmer and vet knowledge and willingness to report and sample – submit to a diagnostic laboratory
Limitations: Availability of diagnostic tools Inconsistency of data Under/over-reporting Lack of central recording Farmer generally has to pay – an inhibitor to submission – offer incentives?
MOSS: Overall aims
Need to be able to:
Rapidly and reliably identify outbreaks of infectious disease;
Prove success of an eradication programme, or prove freedom (ongoing or recently achieved) from disease
Work within the budget available – limiting factor
EU ‘Animal Health Law’: Regulation (EU) 2016/429
On transmissible animal diseases – consolidates lots of previous EU animal health legislation - in force from 21 April 2021
Article 1:
‘This Regulation lays down rules for the prevention and control of animal diseases which are transmissible to animals or to humans.’
About:
‘the early detection, notification and reporting of diseases, surveillance, eradication programmes and disease-free status (Part II: Articles 18 to 42)’
Implication for UK: to enable ongoing trade with EU as 3rd country, more diseases had to be made notifiable in the UK, based on the listed diseases in Annex 2 of the regulation
15 of these diseases (10 are endemic) were not notifiable or reportable in GB up until then – have been added to domestic legislation
e.g. Included Johne’s disease (paratuberculosis), Q fever, infectious bovine rhinotracheitis (IBR), bovine viral diarrhoea (BVDV), porcine reproductive and respiratory syndrome (PRRS)
Onus on laboratories to report detection to the APHA – usually monthly laboratory reports to be returned, but may require immediate notification (e.g. Q fever – zoonotic)
Article 12: Responsibilities of veterinarians and aquatic animal health professionals
- Veterinarians shall in the course of their activities which fall within the scope of this Regulation:
(a) take all appropriate measures to prevent the introduction, development and spread of diseases;
(b) take action to ensure the early detection of diseases by carrying out proper diagnosis and differential diagnosis to rule out or confirm a disease;
(c) play an active role in:
(i) raising animal health awareness, and awareness of the interaction between animal health, animal welfare and human health;
(ii) disease prevention;
(iii) the early detection of, and rapid response to, diseases.
(iv) raising awareness of resistance to treatments, including antimicrobial resistance, and its implications;
(d) cooperate with the competent authority, operators, animal professionals and pet keepers in the application of the disease prevention and control measures provided for in this Regulation.
Article 13: Member States’ responsibilities:
‘1. In order to ensure that the competent authority for animal health has the capability to take the necessary and appropriate measures, and to carry out the activities, required by this Regulation, each Member State shall, at the appropriate administrative level, ensure that competent authority has:
(a) qualified personnel, facilities, equipment, financial resources and an effective organisation covering the whole territory of the Member State;
(b) access to laboratories with the qualified personnel, facilities, equipment and financial resources needed to ensure the rapid and accurate diagnosis and differential diagnosis of listed diseases and emerging diseases;
(c) sufficiently trained veterinarians involved in performing the activities referred to in Article 12.
- Member States shall encourage operators and animal professionals to acquire, maintain and develop the adequate knowledge of animal health provided for in Article 11 through relevant programmes in agricultural or aquaculture sectors or formal education.’
what are the diffrent levels Prevention/control/biosecurity interventions for infectious agents can occur
Individual- small animal cinic vaccination
Institution- farm, heard health programs, “healthy feet program”
Community/national - removing contamminated food from a production line, “horse meat scandal”, abbitours, prevention zones
International/global- bird flu, trade implications, countries being declared free of certain diseases, “transboundry animal diseases”, african swine fever
rabies
risk based intervations
tapeworm treatment for dogs coming back form cetain countries
rabies vaccine for traveling animals
disease managemtn strategies
Control
Prevention
Eradication
disease managemtn strategies- control
all steps to reduce frequency of disease in a population
Sick and healthy
Aim to decrease - communicability and contacts
Isolation, quarantine (in-contacts)
Limiting mixing
Running closed herds, all-in-all-out practices
Slaughter (+/- test)
Treatment of cases
Control measures can be difficult to apply in dynamic populations. Easier in homogenous populations. Early detection is key (flatten the curve!)
