0611 - Pop Health Revision

Question 1

What are the measures of disease frequency?

Answer 1

Always define what you are measuring (clear case definition).

Prevalence – Total people with the outcome (either new or existing) – Given as either percentage or per 1000 population
Incidence Rate – Rate at which new outcomes are occurring – cases per person-day/week/month/year
Cumulative incidence – How many new cases per 1,000 population?

Question 2

How do you calculate prevalence, what are its key features?

Answer 2

Number of people with disease over total number of population

Looks at total cases, not new cases. Used in cross sectional studies, sometimes cohort and RCTs. Does not provide evidence of causality.

Question 3

How do you calculate incidence rate? What are its key features?

Answer 3

Number of people who develop disease in specified period over number of person-time when people were at risk of getting the disease.

It is the rate at which new events occur. Used in cohort studies and sometimes RCTs. People ‘drop out’ once they get the outcome. Not as simple as cumulative incidence.

Question 4

How do you calculate cumulative incidence? What are its key features?

Answer 4

Number of people who develop a disease in a specified period over number of people at risk of getting disease at start of period.

Measures population at risk at start of period only, generally given as new cases per 1,000 population. Used in cohort studies.

Question 5

What are they key measures of association?

Answer 5

MoA link exposure to disease, comparing the outcome in groups with and without exposure.

Risk Ratio and Odds ratio most important
(Also rate ratio and prevalence ratio)

Question 6

What is the risk ratio? How do you calculate it?

Answer 6

Ratio of risk of occurrence of disease among exposed vs not exposed.
Used in cohort studies and RCTs

Cumulative incidence in exposed over cumulative incidence in unexposed
(a/(a+b))/(c/(c+d))

Question 7

What is the odds ratio? How do you calculate it?

Answer 7

The odds of having exposure in cases over the odds of having exposure in controls.

(ad)/(bc)

Question 8

What is attributable risk? How do you calculate it?

Answer 8

Number of cases attributable to that exposure. Usually given as increase in cases per 100 population.

Incidence in exposed-incidence in unexposed
(a/(a+b))-(c/(c+d))

Question 9

What are the four main study types?

Answer 9

Cross Sectional – everything measured at same time.

Case Control – start with cases(and controls) and work back to ascertain exposure

Cohort – Start with NO outcome ->identify exposure status->follow-up for outcome

RCT – start with defined population ->random allocation of exposure -> follow up for outcome.

Question 10

What are the key features, advantages, and disadvantages of a cross-sectional study?

Answer 10

Cross Sectional - Everything is measured at the same ‘snapshot in time’.

Advantages – quick; can easily cover a whole population (thus representative); good for measuring prevalence and generating hypotheses; good for multiple exposures and outcomes in one study.

Disadvantages – based on self report (accuracy?); no ability to establish a causal sequence (no element of time); impractical for rare diseases; confounding can be hard to control for.

Question 11

What are the key features, advantages, and disadvantages of a case control study?

Answer 11

Case Control – Start with cases (and controls) -> work backwards to establish previous exposure.

Advantages – relatively cheap and quick way of finding causal factors; can estimate RR and OR; longitudinal in nature; useful for rare diseases and multiple exposures; suitable for diseases with long induction periods.

Disadvantages – can have recall bias; controls may not be equivalent to cases; can’t determine incidence or prevalence; selection bias (only looking at survivors).

Question 12

What are the key features, advantages, and disadvantages of a cohort study?

Answer 12

Cohort – start with no outcome->identify exposure status->follow up.

Advantages – Directly estimate risk and prevalence; optimal for short induction periods; suitable for multiple outcomes and studying progression of disease; good for causality (risk factors are ascertained prior to disease).

Disadvantages – not as suitable for rare exposures or outcomes; requires large populations; expensive and time consuming; attrition in participants.

Question 13

What are the key features, advantages, and outcomes of an RCT?

Answer 13

RCT – start with defined population ->randomly allocate exposure ->follow up

Advantages – random allocation; protection against confounders; directly estimate risk; compare multiple outcomes.

Disadvantages – limitations of types of interventions (ethics/practicalities); not suitable for rare outcomes; may not be generalisable to population; very expensive.

Question 14

What is the difference between a confounder and an effect modifier?

Answer 14

While confounders can be an independent causal exposure for the outcome, effect modifiers can change the strength of association between exposure and outcome, but can’t independently cause the outcome.

Question 15

How do you control for confounders?

Answer 15

Design phase – randomise

Analysis phase – stratify results.

Question 16

What are the three types of causality

Answer 16

Component Causality - when exposure is linked to disease but insufficient to cause it on its own.

Sufficient causality – when exposure ‘inevitably’ leads to outcome.

Necessary causality – when outcome cannot exist without the exposure.

Question 17

What are the 9 (really 10) Bradford Hill Criteria?

Answer 17

Strength – what is the strength of association between cause and effect (relative risk)
Specificity – is the exposure associated specifically with a single effect?
Coherence – are the results conflicting with current understanding and natural history of disease.
Temporal Relationship – cause must precede outcome
Experimental evidence – does the data show the association?
Dose-response – does increased exposure increase the outcome?
Consistency – have similar results been shown? (replicability)
(Reversibility – does removal of possible cause reduce disease risk?)
Plausibility – is this consistent with other knowledge?
Analogy – do other similar exposures cause similar outcomes

Some Scientists Caution That Every Disease Can’t Really be Plausibly Associated

Question 18

What are the two types of data?

Answer 18

Categorical Data (to categorise things) – Ordinal (logical progression but you can’t manipulate it algebraically – primary school=1, high school=2, undergrad=3, postgrad=4) or nominal (number purely in place of name - male=0, female=1)

Continuous – Ratio or Interval – Ratio has a meaningful zero (i.e. 0=no height or no time), whereas interval has a zero that is no different from any other number (0 does not mean no temperature).

Question 19

What is standard deviation? Why is it useful?

Answer 19

Measure of how far scores are from the mean. Useful to tell you how tightly distributed your scores are around the mean.

Question 20

What is standard error? Why is it useful?

Answer 20

It correlates your sample mean compared to what would be expected in repeated studies of the broader population. Useful for calculating confidence interval.

Question 21

What are Type I and Type II errors?

Answer 21

Type I error occurs when you reject the null hypothesis, when it is, in fact, true. It can be reduced by increasing the alpha value, though this increases the chance of a type II error.

Type II error occurs when you accept the null hypothesis, when it is, in fact, false. It can be reduced by increasing power (including sample size, and decreasing alpha value).

KNOW the 2x2 Table of this