HADPOP

MEASURING DISEASE IN POPULATIONS: INCIDENCE v PREVALENCE
INCIDENCE:
 Number of new cases (events)
 INCIDENCE RATE: Number of new cases, per year, per head of population.
 IR= Number new cases
Pop size x timescale
MORTALITY RATE:
 Is a special case of IR where the event is death rather than onset of disease.
 Is expressed in terms of: “per 1000 person years”
 MR and IR can be compared between populations to see whether individuals in one
population are at higher risk than another pop.
POINT PREVALENCE:
 Is a measure of the number of people who currently have the disease
 Calculated as: number of suffers
Number at risk
 Is influenced by death rate, cure rate and incidence rate.
AGE-SEX CONFOUNDING:
 Age and sex are non-modifiable risk factors for disease that often get in the way of
seeing useful modiefiable risk factors.
 To get a single “overall” rate ratio which takes account of this systematic variation,
we calculate an age sex standardised mortality ratio (SMR).
 This is a summary figure describing mortality experience of a local population
against a standard reference population which takes into account the confounding
influence of age and sex.
 Interpreting SMR’s: SMR’s are usually expressed as a % against a reference
population which is usually whole country.
If SMR = 100: then local population mortality is same as reference pop.
If SMR = +100 = higher mortality than reference pop.
If SMR = -100 = lower mortality than reference pop.
SOURCES OF INFORMATION ON MORBIDITY, MORTALITY & POP SIZE:
SOURCES OF ROUTINE DATA
1) The Census:
 Since 1801, the UK has undertaken a census of its residents and households
every 10 years.
 It is a legal requirement to complete the census.
 In 2001, the census was undertaken by the office for national statistics
(ONS).
 The census provides information on demographics (age,sex), cultural
characteristics, state of health / long term illness, qualifications,
employment etc.
 It provides “base populations” data for calculation of fertility / mortality
rates.

Deaths. 000 households in UK). leprocy. haemoglobin & ferritin. 3) National Health Survey  Running since 1991 it is based on a random sample of 1. people may be classified into “wrong” death classification or “fit” into several catagories. 4) Register of Births.2) General Household Survey (by ONS):  Has been undertaken since 1971 and is conducted on a small sample of population (aprox 13.e. Is performed annually to collect info similar to census to help with planning of services and allocation funding/resources. projection.g. Death by diseases of resp system: J00-J99 MAIN SOURCES ERROR IN MORTALITY DATA a) Numerator  Death certification & ICD coding – i. TB ect) which Doctors have a statutory responsibility to notify local authority. 4) Abortions .g. estimation. Consists of a health and socio-economic survey. Local health authorities collect info from birth notifications and make monthly returns to ONS. SOURCES OF MORTALITY DATA 1) Census 2) Death Registration 3) International Classification of Diseases (ICD)  Keeps records of exactly what people die from by assigning deaths to a specific category based on a prescribed code / definitions of death e. Marriages  Performed statutory by local authority’s then passed onto ONS for comparisons. Malaria. SOURCES INFO ON MORBIDITY DATA 1) Cancer Registration  Hospitals have a non statutory responsibility to register cancer patients with regional registers who send info to ONS. 3) Congenital malformations  A voluntary national scheme covers malformations detected in 1 st 7 days of life. physical health measurements and blood sample tested for cholesterol.600 adults age 16+. cholera. 2) Communicable Diseases  There are 30 notifiable diseases in England & Wales (e. b) Denominator  Census.

e. It is a statutory requirement to notify Chief Medical Officer within 7 days of performing an abortion. comparing risk in one group of people (or cohort) with another group – e. rare diseases).  It can be impossible to determine wether disease causes exposure or vice-versa (‘reverse causality’) 3) Cohort Studies  Is a type of observational study.g. Strengths:  Quick and relatively inexpensive  Possible to look at a lot of different exposures in detail. risk in one group exposed to a potential hazard (such as asbestos) with another group not exposed.  Prone to information bias (e.  Compare groups using an Odds Ratio.  Good for rare outcomes (e. Weakneses:  Not good for rare exposures. people known to have a disease)  Get controls (up to 6) for each case.  Investigate exposures of interest in the past (usually from records).  They are comparitive studies. . recall bias) and selection bias. 2) Case-Control Studies  Start with cases of disease (i.g. EPIDEMIOLOGICAL STUDY DESIGNS There are four main study designs: 1) PREVALENCE SURVEYS 2) CASE CONTROL STUDIES 3) COHORT STUDIES 4) RANDOMISED CONTROL TRIAL OBSERVATIONAL STUDIES EXPERIMENTAL STUDIES 1) PREVALENCE SURVEYS 4) RANDOMISED CONTROL TRIAL ANALYTICAL STUDIES 2) CASE-CONTROL STUDIES 3) COHORT STUDIES 1) Prevalence Survey  Are simple cross sectional surveys of the prevalence of a disease or diseases in a population.g. We compare the odds of having been exposed in the cases WITH the odds of having been exposed in the controls.

Weakneses:  The study and reference populations may not be comparable due to selection bias e.  Large.  Then assess both groups using same criteria.   They start with disease free individuals and follow them up. thereby minimising the effects of these confounding factors on the results. 4) Randomised Control Trials A Randomised Control Trial is a clinical trial (experimental trial) in which interventions are studied in contrast to observational studies (e.g. no morbidity data so comparisons usually on basis of mortality which does not reflect how an exposure may be affecting health during life. sex and social class. despite cohort studies defining cohorts on basis of exposure and then following them forward to see who develops disease (prospective cohort study).  Then allocate one group to be given treatment. the other to be given standard treatment.  Then follow up the two groups equally and aim to maintain all participants in trial. often over a long period to see which develop disease.g.  Good for rare exposures and establishing temporal sequence. outcomes and confounders. (This is the idea that in order to work you have to be healthy.e. that is – we can be confident that disease has occurred chronologically after exposure. Strengths:  Allows detailed and prospective assesment of exposures. recruit.   Why use Randomised allocation to treatment groups? Randomly allocating individuals to two groups ensures that you have similar age.. Randomisation minimises confounding factors (as both groups should have same confounders!) . Therefore. some cohorts are recruited historically – that is: can use existing records (retrospective cohort study). Comparisons between groups can be made using either IRR (comparison within cohort – internal comparison) or SMR (comparison with reference population – external comparison). However. time consuming and resource intensive (costly)  Not very good for rare diseases. consent and maintain two comparable groups of participants. cohort)  Are used to asses question: “Is the new treatment better or worse than the usual (standard) treatment?”  Steps involved in an RCT are:  Identify. allowing comparison of risk disease based on personal exposure. so occupational cohorts are often healthier than general population  Often limited data on reference population: i. “the healthy worker affect”.  Then compare outcomes in both groups to see how big is difference in outcome & if difference is attibutable to treatment.

Large scale trials minimise this. In a single blind trial. cared for more carefully or from being on a new “special” treatment.  BLINDING: Ensures the knowledge of treatment allocation is not available to: 1) The allocating clinician 2) The patient 3) The clinician assessing the effects of the treatment. In a double blind trial. A Placebo is used to eliminate the Placebo effect. Chance: Randomisation only achieves perfect balance in long run. two of these people are blind to treatment status. so small RCT’s may be influenced by this. One treatment is more effective than the other: By minimising above factors through large scale randomisation we can make conclusions about the effectiveness of one treatment over another. Intention to treat analysis is when all 100 are analysed together. Blinding clinician to treatment minimises this. v. Blinding assessor minimises this. Blinding patients to treatment minimises this. This is the only way to ensure that the trial will give a realistic indication of the likely impact of routine clinical use of the drug. what can cause one treatment group to be more successful than the other? i. A Placebo is an inert substance which looks. iii. Treating Clinician knows treatment: This may lead to different choice of secondary treatment given to patient. tastes and is packaged identically to the comparison drug. 100 people may have been randomised into a “new treatment” group but only 80 complete the course of drug therapy because other 20 withdrew due to adverse side affects. Clinical trials should be analysed on an intention-to-treat basis.  INTENTION TO TREAT ANALYSIS: Is when all the people in a trial group are included in the analysis regardless of whether they took a treatment or not. . only one of these people is blind to treatment status. iv. ii. In an RCT.  THE PLACEBO EFFECT: The placebo effect is the psychological benefit that derives from being looked after. Assesor / investigator knows treatment: This may lead to a different approach and sensitivity to the outcomes. thus preserving true randomisation and eliminating confounding factors. For example. Patient know’s treatment: This may lead to amendment of patients behaviour & distort results trial.

temporal sequence. biological plausibility etc. specificity of association. Bias invalidates any conclusions that might otherwise have been drawn. E. but it is possible that people who are alcoholics tend to look for work in bars.CAUSALITY: Cause or merely association? If we observe an association between two factors e. Alcohol is associated with lung cancer.  Confounding A confounding factor is something that is associated with both the outcome and the exposure of interest. true associations other than proposed causal link can result from:  Reverse causality  Chance: The 95% CI of a result is used to eliminate affects of chance. consistency of association. or processes leading to such deviation. Age and sex are also confounders that produce systematic variation in the results. analysis. For example. The direction of causality is very important. but only because drinkers tend to smoke more.g. publication or review of data that can lead to conclusions that are systematically different from the truth. Bias can arise from any trend in the collection. .g. can we be sure it is causal? E. exposure and outcome (as in cohort study). A confounder can distort results so that it appears that there is a causal link between two factors when in fact the association is between the confounder and the exposure/outcome. It can take several forms including selection bias and information bias.  Reverse Causality It is possible that sometimes a cause-effect relationship exists in the opposite direction.  Bias: Bias is deviation of the results or inferences from the truth.  True causality If you eliminate / minimise the affects of bias. but is not on the causal pathway between exposure and outcome. Can we be sure that exposure has caused outcome? Answer: NO! Apparent associations can result from:  Chance  Bias  Confounding In addition. interpretation.g. bar work is associated with alcoholism. You can then use Bradford Hill’s criteria for inferring causality to evaluate the evidence for a cause-effect relationship: Strength of association. confounding and chance then you can make conclusions and generalisations that the association between two things is a true cause-effect relationship.

 Uses odds ratio  Two identical groups are recruited & randomly allocated into a “new treatment” and “standard treatment” groups. Allows prospective assessment of exposures & outcomes Establishes temporal sequence of disease after exposure Good for rare exposures.g.rare diseases).Cohort study Method Case-control study RCT’s Starts with disease free individuals and follows them to see who dev disease – allowing comparison of risks based on exposures. Follow up and assessment identical so only treatment differs.  Randomisation eliminates confounding factors  “Blinding” minimises treatment bias.  Selection bias may occur due to “healthy worker effect”  Often limited data on reference population  Time consuming & resource intensive  Not good for rare diseases.  Quick and cheap  Possible to look at many exposures in detail  Good for rare outcomes (e.  Not good for rare exposures  Prone to information bias and selection bias  Can be difficult to eliminate reverse-causality  Large scale RCT’s can be costly & labour intensive   Advantages    Disadvantages . Can be prospective or retrospective.  Starts with individuals with disease and then looks back retrospectively to identify exposures.   A well run RCT can demonstrate true causeeffect relationships.  Is “gold standard” study design.