Epidemiology• “The study of the distribution and determination of diseases in human populations”• Aim to understand – Precursors of a disease – Identify its causes – With an aim of finding a solution (prevention)
Sampling• Rarely examine entire population• Takes too long, not necessary and not enough resources• We use samples to provide estimates of population parameters.
Bias• During sampling consider Bias – Sampling bias is when the sample is selected in such a way that individuals chosen are not representative of the whole study population. – How do we avoid bias?
Rate of disease• Incidence – The rate of measuring the occurrence of new cases of disease• Prevalence – The rate of measuring existing cases of disease
Question• Is this Incidence or Prevalence? – Between 2009 and 2010, 300 individuals were diagnosed with Pneumonia – On the 24th of January 2010, it was estimated that 300 people have Pneumonia.• What is the difference between Clinical epidemiology and Public Health epidemiology?
The clinical iceberg• It relates to conditions such as diabetes, hypertension and coronary heart disease.• Why? – Only a select few with the condition are known to clinical services • Unaware • No access to local GP • Time-consuming • Think it might ‘go away’
Problem with the Clinical Iceberg• Infectious disease – Subclinical infection circulation of communicable disease – Highlights importance of herd immunity• Disease morbidity – Small risk distributed throughout the whole population is a clinical risk. – Contribute substantially to burden of CV morbidity.
Communicable disease• Endemic – They persist for long periods within a community – Example?• Epidemic – The occurrence of disease episodes increases substantially for a limited period of time. – Example?
Risk factor vs Cause• Risk factor – It is a characteristic that identifies a group at increased or reduced risk of a disease.• Cause – It is defined as something that increases the risk of a disease. Risk factors are correlational and not necessarily causal.
Question• A 74 year old Afro-Carribean female suffered a stroke.• She has high blood pressure and high cholesterol levels.• She has a family history of Stroke.• What are the modifiable and non-modifiable risk factors above?
Prevention• Depends upon: – Identification of modifiable risk factors – Causally related to the disease – And whose effects are reversible following intervention.
Primary, Secondary and Tertiary Prevention• Primary – Reduction of incidence of disease among healthy individuals e.g. vaccination.• Secondary – Early detection of preclinical disease and treatment to prevent progression e.g. screening.• Tertiary – Treatment of established disease to prevent complication.
Concept of disease• Follows the bio-psycho-social model of disease – Disease • Presence of a biological or functional abnormality – Disability • The perception of ill health – Handicap • The social consequence of disease
Population at Risk• The population at risk is made up of those person who, if were to develop the disease, would be included in the cases• Person time at risk is more useful when assessing disease incidence• Person time at risk = Sum of periods of time at risk contributed by each person in a population
Disease Frequency• Prevalence – What is it? – It is simply the state of having the disease – Which type of disease would have a low prevalence? – Therefore Prevalence = Incidence x Duration• Incidence – What is it? – It refers to the event of transition between health and disease.
Measurement of Incidence• Cumulative Incidence – Number of person diseased during follow-up period / Number of persons in the population at start of follow-up period• Incidence rate – Number of persons diseased during defined period / Total person time at risk during follow up period
Measurement of Prevalence• Point prevalence – Number of diseased person in a defined population at one point in time / number of persons in population at same point in time• Period prevalence – Number of persons with an episode of illness over a defined period of time / number of persons in the population over the same period.
Mortality and Disease Outcome• Mortality Rate – Incidence rate x Case Fatality• Case Fatality – Number of deaths among person with disease over defined period of follow-up/ Number of newly incident cases of disease at start of follow up period.
Risk and Odds• Odds of disease – Number of cases / Number of non-cases• Risk of disease – Number of cases / Catchment population
Risk, Cause and Prevention• Relative Risk – This is used in longitudinal studies – It is a ratio measure including Odds ratio, Risk ratio and Rate ratio.• Excess Risk – Rate in exposed group – Rate in unexposed group• Population attributable Risk – Rate difference x Proportion exposed.
Confusion between Relative Risk and Odds Ratio• Take the following example:
Association and Causation• Two variables are associated if changing the level of one alters the level of the other.• Association does not necessarily indicate a causal link• Causality can only be inferred when chance, confounding and bias have been eliminated.
Chance• A single set of observations, even in absence of sampling bias, may misrepresent the truth because of error arising from random variation.
Confounding• To be a confounding variable a factor must: – Be associated with disease – Be associated with, but not a consequence of the exposure• Providing that potential confounding variables are identified in advance, their effect can be controlled: – Design • Matching for confounding variables • Randomisation – Data Analysis
Bias• Selection Bias – Case-control studies – lack of comparability in the method of selection of subjects into two groups to be compared – Cohort studies – differential loss to follow-up in different exposure groups• Information Bias – lack of comparability in the information obtained for two comparison groups. This is particularly relevant in which studies?
Standardisation• This is a method for allowing for differences in characteristics such as age and sex composition of populations.• This is significant because they are significant risk factors in disease and therefore influence incidence / prevalence of most diseases.• There are two methods, but Indirect standardisation is the only one discussed in your book.
Study Designs• Observational – Three types • Cross sectional • Cohort • Case-control – Describing disease in a population• Interventional – Clinical Trial
Cross Sectional Study• This focuses on all persons in a defined population (or sample) and aims to determine their disease and/or exposure status at that point in time.• Vitally important to distinguish between Cohort and Case-control• It is either descriptive or analytical. – Descriptive – assessing the burden of disease in a population using – Point Prevalence – How? – Analytical – explain the observed pattern of disease by looking at possible aetiological factors using risk ratios.
Design Issues• Study Group – Study sample – Study population – Target population• Sample Size• Non respondents
Cross Sectional Study• Advantages – Multiple consequences of a given exposure can be studied – Disease prevalence is measured which is important in assessing health service burden• Disadvantages – Not suitable for rare diseases or diseases with short duration
Cohort Studies• This aims to answer the question: “Do persons with a characteristic develop the disease more frequently than those who do not have the characteristic”?• It is observational.• It can either be prospective or retrospective – Difference?
Measure of Risk from Cohort Study• Incidence Risk – Number of new cases of disease in a specified period of time / number at risk of acquiring disease at the beginning of the period• Incidence Rate • Number of new cases of disease in a specified period of time / Number of person years at risk during period.• Relative Risk – Incidence rate in exposed / incidence rate in non-exposed• Excess Risk – Rate in exposed minus rate in non-exposed
Example• A cohort study looked at shisha smoking as a risk factor for development of COPD.• Looking at the table, how would you calculate the Rate Ratio; Total Cases Non-cases Rate Rate Ratio Exposed 10000 100 9900 0.01 Unexposed 10000 10 9990 0.001
Cohort Studies• Advantages – Exposure is measured before disease onset, reducing potential for bias accuracy of causality – Multiple diseases can be studied for any one exposure – Incidence can be measured in both groups• Disadvantages – Slow, expensive and administratively difficult because needs large numbers over long time. – Losses to follow up may introduce selection bias
Cohort Design Issues• Selecting controls – Virtually identical• Measurement of exposure• Measurement of confounding• Measurement of disease outcome• Methods of follow-up• Study Size
Case Control Studies• This starts with identification of a group of cases (with a particular illness) and a group of controls (without the illness).• The prevalence of a particular exposure is then measured in the two groups and compared.• If the exposure is more common among cases than controls, it may be a risk factor, and if it is less common it may be a protective factor.
Cont’d• It is best represented in a 2x2 table; Exposed UnexposedCase A BControl C D
Advantages and Disadvantages• Advantages – Cheap, quick and efficient – Effective in looking at rare disease – Can investigate a number of exposures for rare diseases• Disadvantages – Prone to bias – No prevalence measure – Unsuitable for rare exposure.
Design Issues• Case Definition and Selection – Disease definition – Incidence v Prevalence – Exclusion criteria – Identification• Control Definition and Selection – Selection bias
Matching• Control for confounding factors.• Can either be individual or stratum• Improves statistical power, especially when variables that are matched for are strong confounders.• However, there are practical limitations on the number of variables that can be matched and a danger of over-matching.
So now…• A prospective, randomised, double-blind study was performed to compare the effects of propranolol and placebo on sudden cardiac death in a high risk group of patients who survived a myocardial infarction.”
Placebo• In intervention studies we want to be sure that any health benefits from an active treatment are truly from the active treatment and not because of the placebo effect.• The placebo effect is an alleviation of symptoms following an intervention which the patient believes to be effective against the disease, but which, in fact, is completely neutral e.g. sugar pill.• It is based on the person’s own self-healing capacity triggered by belief they are receiving medication.
Interventional Studies• This is a clinical trial designed to evaluate the effects of various treatments.• Individuals are randomly allocated to two groups.• Investigators are blinded to the allocation of individuals to minimise bias.• One group is assigned the preventative measure under investigation while the other group does not receive this intervention.
Randomisation and Blinding• Randomisation ensures that neither the observors nor the individual participating in the study can influence which is allocated to receive which intervention.• Blinding avoids prior expectations of trial participants or observors influencing the results.
Blinding and Bias• The effects of blinding of observor avoids: – Selection bias – decision to enter patient into trial. – Assessor bias – response to treatment is recorded – Bias in withdrawal of patient from a trial.• The effects of blinding of patient avoids: – Selection bias – whether to agree to participate – Response bias – way in which health outcomes are recorded by trial participants. – Bias in withdrawal from trial – When should the blinding code be revealed?
Cross Over Trial
Factorial Trial• Two or more interventions are carried out simultaneously.• Best explained with an example.
Example• Participants are randomised to receive either aspirin or placebo.• Participants are also randomised to receive behavioural intervention or standard care.• This is known as a 2x2 factorial trial.• The intention is to achieve two trials for the price of one.• The assumption is that the effects of the different active interventions are independent – no synergy.
Intention to Treat• This is an analysis in which all the participants in a trial are analysed according to the intervention to which they were allocated whether they received it or not.• Rationale: – Estimate effects of allocation of intervention in practice• Compare this to per-protocol analysis.
Per protocol analysis• An alternative is PPA aka OTA (on treament analysis)• This means limiting the analysis of the data to patients that comply with treatment.• The major problem with this is that those that comply with treatment (Hmm..) are often different in nature to those who do not.• Therefore it is difficult to decipher whether treatment effect is due to intervention or difference in make up of individuals
Number Needed to Treat• Assesses effectiveness of a health care intervention• It is the average number of patients who need to be treated to prevent one additional bad outcome.• It is described as the inverse of the absolute risk reduction.• The ideal NNT is 1 i.e. everyone improves with treatment and no-one improves with control.• They are time-specific
This is what you need to know• Absolute Risk Reduction (ARR) – Control Event Rate – Experimental Event Rate• NNT = 1 / ARR• NNT values are time specific so if a study ran for five years and the NTT value was 100 during the 5 year period, the NTT would have to be multiplied by 5 to assume the right NNT for a one year period.
Some things you need to cover…• Inequalities in health• The control of Communicable Disease in the UK
Tomorrow• IFP – Professionalism – Ethics – Gillick Competency – Battery / Negligence – Lots of theories! – Confidence Intervals / P values / Reference Ranges / Standard Error etc.