Evidence

Question 1

What is epidemiological transition?

Answer 1

Global change in the LEVELS and CAUSES of mortality, summarised as a decline in total mortality and a significant reduction in infectious/deficiency disease, alongside an increase in the role of non-communicable diseases like cancers, cardiovascular disease, chronic respiratory disease and diabetes.

Question 2

Explain the reason for the current epidemiological transition? (x5 and x5)

Answer 2

DECREASED INFECTIOUS DISEASE (AND DECREASED MORTALITY): better sanitation, better healthcare, better housing, technological change, better living conditions. INCREASED DEATHS FROM NON-COMMUNICABLE DISEASE: more smoking, poorer diet, more air pollution, less physical activity, live longer so can die from chronic disease.

Question 3

Why is evidence-based medicine important in clinical medicine?

Answer 3

Evidence-based medicine looks at all the research that there is about a disease or treatment. Research and evidence are a better gauge about what’s good for the patient = better care and safety. This is because, when researchers look at whether a treatment works, they look at many more patients than a single doctor would ever treat.

Question 4

What is the definition of hierarchy of evidence in study design?

Answer 4

The hierarchy ranks all types of study design so that those with the highest power, least bias and strongest quality of evidence are at the top, and those with the least are at the bottom.

Question 5

List the hierarchy of evidence in study design. (x9)

Answer 5

Systematic Review/Meta-Analysis. Randomised Controlled Trial. Cohort study. Case-control study. Ecological study. Descriptive/Cross-sectional. Case Series. Case report. Expert opinion.

Question 6

What is a systematic review/meta-analysis?

Answer 6

Designed to answer a specific question – these COMBINE and EVALUATE the data and results of MANY RANDOMISED CONTROL TRIALS to see if the consensus of trials supports the studied intervention. Meta-analysis combines published estimates to generate a pooled risk estimate.

Question 7

What is a randomised controlled trial?

Answer 7

Entry criteria are specified, and volunteers are RANDOMLY ASSIGNED to a treatment and placebo group. The outcomes of each group are compared to see if those receiving the new treatment are significantly better.

Question 8

What is a cohort study?

Answer 8

This can be retrospective or prospective. It is an OBSERVATIONAL study, taking a cohort of HEALTHY VOLUNTEERS and recording information about exposure over the study period, whilst also recording outcomes in order to calculate the RELATIVE RISK of outcomes for exposure. [Look at clinical trial notes for a better definition].

Question 9

What are cohort studies used for?

Answer 9

Used to study the impact of risk factors on the risk of getting a disease.

Question 10

What are randomised controlled trials used for?

Answer 10

Study the effectiveness of a drug, treatment, intervention or technique.

Question 11

What is meant by retrospective and prospective cohort studies?

Answer 11

RETROSPECTIVE: looks at historical data for a group – some have developed the disease, and some have not. PROSPECTIVE: population does not have the disease when the study commences. Data analysis occurs after a period has elapsed.

Question 12

What is the main advantage and disadvantage of retrospective cohort studies?

Answer 12

Cheaper, but higher selection bias.

Question 13

What is the main advantage and disadvantage of prospective cohort studies?

Answer 13

Expensive, but less bias. Time-consuming.

Question 14

What is a case-control study?

Answer 14

Retrospective, OBSERVATIONAL study that studies two populations – one with the disease and one without. Exposure to something is studied between the two groups. If something has greater exposure in the diseased group than the control group, it could be a risk factor for that disease. [Look at clinical trials notes for better definition.]

Question 15

What is an ecological study?

Answer 15

Observational study. Looks for associations between the occurrence of a disease and exposure to suspected causes. Similar to case-control studies, though the study occurs at population level.

Question 16

What is an ecological study used for?

Answer 16

Geographical comparisons of a disease.

Question 17

What is a disadvantage of an ecological study?

Answer 17

Must take care when interpreting as confounding sex, age and healthcare seeking behaviours may explain differences.

Question 18

What is a descriptive/cross-section study?

Answer 18

OBSERVATIONAL study. Describe the distribution of a disease with respect to the person, place and time, using ROUTINE DATA, already collected without a research question in mind. Cross-sectional looks at absence/presence of disease and exposure at same point in time.

Question 19

What are cross-sectional studies used for?

Answer 19

Compare outcomes in different areas or populations, and for allocation of resources.

Question 20

What is a case series?

Answer 20

Description of a small number of cases of a disease, including the same information as a case report. Can be used to inform on the management of a novel disease.

Question 21

What is a case report?

Answer 21

Description of a single case of a disease.

Question 22

What are case series and case reports useful for?

Answer 22

Understanding rare conditions.

Question 23

What are the SR/MAs or randomised controlled trials useful for?

Answer 23

When making CLINICAL DECISIONS about an individual patient.

Question 24

List the possible causes for an observed association? (x4)

Answer 24

Chance, bias, confounding, causation.

Question 25

What is an observational study?

Answer 25

Draws inferences from a sample to a population where the independent variable is not under the control of the researcher.

Question 26

How does chance lead to observed associations?

Answer 26

If a sample is used, normal variation within the sample due to chance may lead to apparent correlations that are not found in the population as a whole.

Question 27

How can chance be assessed?

Answer 27

Look at confidence intervals and the p-value to determine if the result is likely due to chance. E.g. if it falls outside the confidence intervals.

Question 28

How can association from chance be limited?

Answer 28

Increasing sample sizes.

Question 29

How does bias lead to observed association?

Answer 29

Systematic errors leading to an incorrect estimate of exposure’s effect – consequence of study design.

Question 30

How can bias be limited?

Answer 30

Has to be done before study – need to change the study design.

Question 31

What is confounding?

Answer 31

Factors that may have a real effect on the risk of disease and the risk factor studied e.g. age, sex…

Question 32

What is selection bias?

Answer 32

Systematic differences between those selected for study and those not. i.e. the sample is not representative of the population. Refers to distortion from the method of collecting samples e.g. non-random sample of a population causing some members of the population to be less likely selected than others (e.g. selected by social media takes away a lot of the elderly).

Question 33

What is measurement bias?

Answer 33

Inaccurate measurements of classifications.

Question 34

What is recall bias?

Answer 34

Failure of a participant to accurately recall the level of exposure, either deliberately or by accidence e.g. social acceptability bias.

Question 35

What are the Bradford-Hill criteria for establishing causation? (x9)

Answer 35

STRENGTH: strong association is more likely to be causal. CONSISTENCY: other studies should consistently show the same result. SPECIFICITY: the exposure should increase the risk of ONE disease, not many others as well i.e. the exposure results in a SPECIFIC outcome. AND, a specific outcome results from a single cause. TEMPORAL RELATIONSHIP: the exposure must occur before the disease. DOSE-RESPONSE: increased exposure leads to more rapid/significant disease onset. PLAUSIBILITY: needs to make biochemical sense. COHERENCE: the causal relationship does not conflict with other facts regarding the disease and its natural history – similar to plausibility. EXPERIMENTAL EVIDENCE: should be available that indicate that when the exposure is removed, the disease risk decreases. ANALOGY: there are similar cause-effect relationships e.g. if it is accepted that certain drugs increase chance of birth defects in pregnancy, then it is easier to accept that another drug does similar things.

Question 36

What is relative risk?

Answer 36

The ratio of a probability of an outcome in an exposed group, compared to an unexposed group e.g. smokers are X times more likely to develop lung cancer than non-smokers.

Question 37

What is attributable risk?

Answer 37

The DIFFERENCE in rate of a condition between an exposed and unexposed population i.e. the increase attributed to the exposure e.g. if the baseline level of lung cancer is 5/100, and in smokers it is 20/100, the attributable risk is 15/100.

Question 38

What is odds ratio?

Answer 38

Likelihood of having the exposure if you have the disease, relative to the likelihood of having the exposure if you do not e.g. those who have lung cancer are X times more likely to have smoked than those that do not have lung cancer.

Question 39

How is relative risk, attributable risk and odds ratio calculated?

Answer 39

RELATIVE RISK: incidence in exposed group/incidence in unexposed group. ATTRIBUTABLE RISK: (incidence in exp – incidence in unexp)/100. ODDS RATIO: odds cases/odds controls. The top table represents values that should be used to calculate the risks and odds ratios. The bottom table shows how to calculate relative risk and odds ratio using these values. Don’t remember what \<1, 1, \>1 means for RR and OR in this flashcard. KNOWING HOW TO CALCULATE RR AND OR IS A SOFIA POINT.

Question 40

What does relative risk of less than 1, 1, and more than 1 indicate?

Answer 40

Less than 1: exposure = lower disease risk. 1: exposure has no effect on disease risk. More than 1: exposure = higher disease risk.

Question 41

What does odds ratio of less than 1, 1, and more than 1 indicate?

Answer 41

Less than 1: those with the condition are less likely to have been exposed than those without the condition (so the factor is associated with reduced risk of developing the disease). 1: those with the condition are equally likely to have been exposed than without the condition (so no association with disease). More than 1: those with the condition are more likely to have been exposed (so associated with increased risk of the disease). Risk factors of a disease would have odds ratios of more than 1.

Question 42

How can the true underlying population risk be calculated from a sample?

Answer 42

Repetitive sampling will produce estimates centred around the true population risk.

Question 43

What is sampling?

Answer 43

Taking a representative group that is smaller than the whole population, and studying them to establish prevalence, incidence or risks of the population as a whole.

Question 44

What is sampling variation?

Answer 44

Due to variation within the population, and the random nature of sampling, chance will impact on any results, yet repetitive sampling will produce estimates centred around the true population risk.

Question 45

What is a p-value? And what can be determined about p-values in relation to significance of a result?

Answer 45

It determines the likelihood that something is down to chance. P = 0.05 suggests that the probability of a result being the result of chance is 5%. P \< 0.05 (less than 0.05) is used as a test of significance, whereby if p\<0.05 (less than), the probability the result is due to chance is less than 5%, and the result is significant.

Question 46

How are confidence intervals used when assessing relative risk?!

Answer 46

If the relative risk falls within the confidence intervals, you can be 95% sure the result is not due to chance.

Question 47

What type of data can confidence intervals an p-values be used?

Answer 47

NORMALLY DISTRIBUTED DATA only, where results deviate from a single point – the mean.

Question 48

What is the null hypothesis?

Answer 48

Denoted by H0. There is no significant finding. Findings are down to chance.

Question 49

What is the hypothesis?

Answer 49

Denoted by H1. There is a significant finding.

Question 50

How is the null hypothesis rejected?

Answer 50

In order to reject H0, we must be sure that the impact of chance is minimal, usually taken to be less than 5% (p\<0.05); overlapping confidence intervals suggest that chance may be responsible for any apparent difference and null hypothesis would otherwise be accepted.

Question 51

What is a confounder?

Answer 51

A factor e.g. age, sex, ethnicity, that is associated with both the exposure of interest and the outcome of interest. In other words, it affects both the dependent and independent variable.

Question 52

What is the problem with confounding?

Answer 52

Associations can sometimes by explained by confounders; where the confounder is actually responsible for the difference in outcome rather than the exposure studied e.g. when it gets warmer, shark attack incidence increases. This is not because the water is warmer, just that more people are in the water.

Question 53

What methods are there for dealing with confounding? (x3)

Answer 53

STRATIFICATION: splitting the analysis of data by subgroups (strata) to exclude the impact of these confounders to show if any difference is the result of exposure. STANDARDISATION: adjustment so that the rates of each factor are adjusted to the proportion of the total population that have that factor. REGRESSION: statistical analysis.

Question 54

What are the advantages and disadvantages of stratification for dealing with confounding? (x1 and x1) Calculation for number of factors.

Answer 54

Simple to perform. BUT, there is a limit on the number of factors that can be stratified before it gets too complex or sample size becomes too small – means not all confounders may be assessed. 2^n – where n is the number of factors.

Question 55

What is a critical appraisal?

Answer 55

Systematic examination of research to assess validity, results and relevance before using to inform a decision.

Question 56

What is generally considered when critically appraising medical evidence?

Answer 56

Question: hypothesis present? Relevant? Design: type of study and height in hierarchy? Appropriate? Population: sample size? Generalisability? Representative? Methods: measurements? Analysis: statistical tests? Chance? Confounding considered? Confounders: presence? Identified? Attempts to adjust/exclude? Bias: selection? Recall?... Ethics: ethical? Reviewed by panel/committee? Interpretation: correct? Bradford-Hill? Conclusion: Answers the question? Relevant to aims? Is it actually what the results show?

Question 57

What is PICOS?

Answer 57

Used to come up with a specific question an appropriate study design. Population? Intervention? Comparison? Outcomes? Study Design?