Brainscape's Knowledge GenomeTM

Browse over 1 million classes created by top students, professors, publishers, and experts.


See full index

MPH Epidemiology > Carneiro Notes > Flashcards

Carneiro Notes Flashcards

1
Q

what is prevalence?

A

no. existing cases in a defined population, at a defined point in time, divided by the total no. people in that population at that time

aka proportion of people in a population that have the disease/outcome

can never be greater than one, and has no units
(but is normally expressed as a %)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

what is incidence?

A

the frequency of NEW cases in a defined population during a specified time-period

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

what is risk and how is it calculated?

A

aka cumulative incidence

= no. new cases in a specified time period
//////
total no. individuals at risk in the population at the start of the time period

a proportion, so can’t be >1 and has no units

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

how can you interpret risk?

A

the likelihood (‘risk’) that an individual will develop an outcome during the specified time period

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

what is the secondary attack rate of a disease?

A

a specific form of risk.

the ‘risk’ that a contact of a case will develop the outcome during a specified time-period

= no. new cases among contacts in a specified time-period
/// total no. contacts of a primary case in that time-period
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

what are the odds of an outcome? how are odds of outcome calculated?

A 
= no. new cases in a specified time-period
//// 
no. who did not become a case during that time period

odds is a ratio of two proportions, so can be >1

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

what is a dynamic (aka open) population?

A

one in which people enter and exit the “population at risk” at different time points - so people are exposed to risks for different lengths of time

this is when you’d use person-time to calculate incidence rate.

(compared to a closed population where everyone is at risk for the same, fixed length of time)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

how do you calculate incidence rate and what must be included when reporting it?

A 
= no. new cases in a specified time-period
///
total person-time at risk during that time period

must specify time units e.g. person-months, person-years, 1000 person-years at risk etc.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

what can you do to calculate incidence rate if you have a large population, in which calculating specific person-time at risk would be too difficult?

A

use the population at the mid-point of the time-period of interest, multiplied by the length of the time period

gives an estimate of person-time at risk - provided it’s a relatively unchanging population

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

what 4 measures are included under the broad term “relative risk” and what are they measures of?

A

prevalence, risk, odds and incidence rate ratios.

they estimate the strength of an association between an exposure and an outcome.

indicate how much more likely it is that an exposed individual will develop an outcome compared with an unexposed individual.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

what does it mean if a relative risk is:
>1
=1
<1

A

> 1 –> exposed individuals are at greater risk than unexposed individuals
=1 –> there is no difference in risk between exposed and unexposed
<1 –> exposed individuals are at lower risk (exposure is a protective factor!)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

what studies can you calculate prevalence ratio from?

A

cross-sectional

population surveys

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

how do you calculate prevalence ratio?

A

prevalence of outcome in exposed group
///
prevalence of outcome in unexposed group

(a/a+b) divided by (c/c+d)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

what studies can you calculate risk ratio from?

A

ecological, cohort or intervention studies

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

how do you calculate risk ratio?

A

risk of outcome in exposed group
///
risk of outcome in unexposed group

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

what studies can you calculate odds ratio from?

A

ecological, cohort or intervention studies

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

how do you calculate odds ratio?

A

odds of outcome in exposed group
///
odds of outcome in unexposed group

(a x d) / (b x c)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

how do you calculate odds ratio of exposure?

when is it typically used?

A

odds of exposure in those with the outcome
////
odds of exposure in those without the outcome

used in case-control studies as individuals have been selected on the basis of their outcome status

according to Ilona, maths means that it will be the same as the odds ratio of outcome - but DON’T use the terms interchangeably

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

what studies can you calculate incidence rate ratio from?

A

ecological, cohort or intervention studies

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
20
Q

how do you calculate incidence rate ratio?

A

= incidence rate of outcome in exposed group
///
incidence rate of outcome in unexposed group

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
21
Q

if you have a very rare outcome, are different relative risk measures likely to be quite similar or quite different?

A

similar.

if you have common diseases, the different relative risks will all be quite different, but for rare diseases they will be very similar

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
22
Q

what is attributable risk?

A

the excess incidence of an outcome that we can ‘attribute’ to the exposure

(assuming a causal relationship!)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
23
Q

how do you calculate attributable risk?

A

incidence in exposed - incidence in unexposed

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
24
Q

why use attributable risk, instead of just relative risk measures?

A

relative risk measures only tell you how strongly an exposure is associated with an outcome.

they don’t tell you the real impact of exposure on the incidence of an outcome in a specific population

e.g. may have outcome A, with incidence rate in exposed of 6, and in unexposed of 1 (IRR = 6)

then outcome B, with incidence rate in exposed of 30, and in unexposed of 5 (IRR also = 6).

but, attributable risk (AR) shows is 5 for A and 25 for B, so gives you a better idea of how much the exposure affects the population!

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
25
Q

what is attributable risk fraction?

A

the proportion of an outcome in exposed individuals that can be blamed on the exposure

(a proportion, rather than simple numbers per person-years at risk or whatever)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
26
Q

how do you calculate attributable risk fraction?

A

AR / incidence in exposed

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
27
Q

how does attributable risk fraction relate to relative risk?

A

attributable risk fraction = (relative risk - 1) / relative risk

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
28
Q

what is population attributable risk (PAR)?

A

this is the attributable risk of an exposure, when applied to a population

allows you to apply measures of relative risk from a study to a real population

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
29
Q

how do you calculate PAR?

A

incidence in population - incidence in unexposed

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
30
Q

what is population attributable fraction (PAF)?

A

the proportion of an outcome in the real world population that is attributable to an exposure, as calculated using incidences found in a study

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
31
Q

how do you calculate PAF?

A

PAR / incidence in population

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
32
Q

what defines an ecological study?

A

analyses the relationship between outcome and exposure at a population (group) level.

the data on each is not linked to individual study participants - unit of analysis is the group.

e.g. compare rates of outcome in two groups with the proportion of each group that experience the exposure

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
33
Q

list 4 main reasons for undertaking an ecological study

A
  1. data only available at group level
  2. data difficult to measure at an individual level
  3. to study group level interventions e.g. health policies, health promotion interventions e.g. seat belt law
  4. data quicker and cheaper to collect at a group level
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
34
Q

what is a multi-group design of an ecological study?

A

an ecological study comparing groups

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
35
Q

what is a time-trend design of an ecological study?

what might make this design difficult to interpret?

A

an ecological study comparing the same population/group at different points in time.

difficult to interpret if an outcome has a long/unknown latent period.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
36
Q

what is a mixed design ecological study?

A

an ecological study comparing groups and time

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
37
Q

give some examples of routine data sources that may be used for an ecological study

A

vital registration (births and deaths)

demographic data - population censuses and household surveys

information on chronic conditions through outcome-specific registries (e.g. QOF)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
38
Q

how does direct age standardisation work?

A

you take death (or other outcome) rates for each age group in your study population, and apply these rates to a standard population.

generates expected death (or other outcome) rates for each age group in the standard population, you total these to get your DSR

(DSR = directly standardised rates)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
39
Q

how does indirect age standardisation work?

A

takes standard death (or other outcome) rates for each age group, and applies them to your study population.

the expected number of deaths/cases is then compared to the observed number of deaths/cases, generating standardised mortality ratio:

SMR = O/E

NB - you should compare SMRs to the standard population, not between 2+ study populations

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
40
Q

what 4 things must be considered in interpreting an ecological study?

A
  1. ecological fallacy
  2. bias - data may be collected in different ways for different groups
  3. migration between study populations - can dilute differences
  4. confounding - data are often collected for other reasons so you’re often missing data on potential confounders
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
41
Q

explain “ecological fallacy”

A

the problem that occurs when you try to draw individual-level associations from a group-level study

you can’t assume that the individuals who had the outcome in your study were the same individuals that experience the exposure

e.g. could see a lot of association in different populations between car ownership and breast cancer rates, but you don’t know that the people who own multiple cars are the people suffering breast cancer

42
Q

what defines a cross-sectional study?

A

data on outcomes and/or exposures are collected on each study participant at one point in time.

a “snapshot”

43
Q

what is the main advantage of a cross-sectional study?

A

quick and easy.

44
Q

explain the difference between a descriptive and an analytical cross-sectional study

A

a descriptive cross-sectional study = a prevalence survey, just used to estimate the prevalence of outcomes OR exposures.

an analytical cross-sectional study collects data on both outcome and exposure (at same time point) to try and identify an association

45
Q

give some examples of biases that cross-sectional studies are subject to

A

non-response bias
recall bias

  • cross-sectional studies are often surveys, and these biases are common to most surveys
46
Q

which measure of association is most appropriate for analysis of a cross-sectional study?

A

prevalence ratio

47
Q

what defines a cohort study?

A

a study which measures the exposures of interest, and then follows up study participants over time to measure incidence of an outcome.

can be prospective, or retrospective.

48
Q

which of Bradford-Hill’s criteria for causality is most likely to be met by using a cohort study design?

A

temporality - guaranteed that the exposure has occurred before the outcome

49
Q

are cohort studies useful for study rare exposures, or rare outcomes?

A

rare exposures - able to select your study participants based on exposure status

case-control studies are useful for rare outcomes.

50
Q

what type of study design is particularly useful for rare exposures?

A

cohort - able to select your study participants based on their exposure status, so can seek out those with rare exposures

51
Q

what type of study design is particularly useful for rare outcomes?

A

case-control

able to seek out individuals with the rare outcome, for inclusion in the study

52
Q

what are the main disadvantages of a cohort study?

A
  1. large no. participants required
  2. follow-up = expensive data collection
  3. length of study increases loss to follow up, and expense
53
Q

what differentiates a retrospective cohort study from a case-control study?

A

cohort is defined based on EXPOSURE status rather than outcome status.

54
Q

how do retrospective cohort studies work?

A

pre-existing data on both exposures and outcomes are used

e.g. look back in records for those who had exposure, and then look through the records up to present day to assess outcome status.

55
Q

what are the pros and cons of doing a retrospective cohort study rather than a prospective one?

A

quicker and cheaper, as don’t have to deal with a long follow up time.

but, routine data that you have to use may well be poorly collected, with inaccurate or missing data

56
Q

what are the advantages of using a workforce (aka occupational) cohort for a cohort study?

A

higher participation and follow up than general population cohorts

doesn’t need to be representative of general population, as long as the exposed and unexposed cohorts are comparable

57
Q

what must you be aware of when using workforce cohorts?

A

healthy worker effect

if exposure is workplace-based, must compare to either an internal or external comparison group, rather than general population

58
Q

what is the “healthy worker effect”?

A

a form of selection bias - tends to underestimate the risk associated with an occupation, by comparing it with the general population (which will include those who are too ill to work)

59
Q

in a descriptive cohort study, what measures of the frequency of an outcome might you use?

A

risk, or rate

use risk if follow up times for all participants are similar, use rate if they vary (takes into account person-time at risk)

60
Q

in an analytical cohort study, what are the appropriate measures of association to use?

A

risk or rate ratios

might use these to calculate AR, ARF, PAR and PAF

(rate is used where follow up times vary, as rates take person-time at risk into account)

61
Q

what defines a case-control study?

A

“cases” of an outcome are identified, as well as corresponding controls.
the previous exposure status of individuals within each group are then determined, looking for association between exposure status and outcome status.

62
Q

what are the advantages of using a case-control study, over a cohort study?

A

cheaper and faster.
can use a smaller sample size.
avoids lengthy follow-up.

can be used for rare outcomes.

63
Q

what CAN’T you do with a case-control study?

A

estimate the frequency (e.g. prevalence, risk, odds or incidence rate) of an outcome in your target population

64
Q

what is a nested case-control study?

A

nested within a cohort study, cases are members of a cohort that have developed the outcome, and controls are members that haven’t.

allows automatic matching on factors common to all cohort members.

65
Q

should case-control studies use prevalent or incident cases?

A

use of prevalent cases makes results easier to generalise to the target population.

but, incident and prevalent cases may be different in a way that reduces validity if prevalent cases are used.

use of prevalent cases may mean you lose the more severe, rapidly progressive cases as they might die before they get the chance to be included in the study.

66
Q

explain the difference between individual and group matching, as used in case-controls studies

A

individual matching = finding 1-4 controls matched to each specific case for age, gender etc

frequency (or group) matching = ensuring the control group has similar demographics etc as the case group (e.g. both 70% male)

67
Q

how does reporting bias affect case-control studies?

A

knowledge of being a case can affect what people remember/report about events and exposures

e.g. cases of asbestosis are more likely to remember working in a building known to have asbestos

68
Q

what is the correct outcome measure for a case-control study?

A

odds ratio of exposure.

measures of frequency of outcome (e.g. prevalence, risk) are inappropriate because you’ve selected on the basis of outcome!

69
Q

name the ideal type of interventional study design

A

RCT

70
Q

why might you use a design other than RCT for an interventional study?

A

ethical issues.

RCT is also more complicated and expensive.

71
Q

what differentiates an interventional study from the other study designs discussed in this module?

A

the others are all observational.

in interventional studies, the researchers play an active role in assigning participants to different groups etc

72
Q

give 4 reasons interventional study designs are ideal for inferring causality

A
  1. meets temporality criterion as exposure is defined prior to outcome
  2. if intervention involve removing/reducing a suspected risk factor/exposure - meets the reversibility criterion
  3. randomisation of participants to being exposed or unexposed can reduce effect of confounding/selection bias
  4. it may be possible to conceal allocation of the intervention from the subjects and researchers, reducing information bias
73
Q

what are the disadvantages for interventional studies?

A

long and expensive.
can’t be used for very rare outcomes - would require a really long cohort-style follow up.
costs of intervention, allocation concealment, safety monitoring etc.
ethical issues with withholding an intervention from the control group if prev. shown to be safe and effective in another setting.

74
Q

name the WHO guidelines that govern the ethics of interventional studies

A

The Declaration of Helsinki

75
Q

explain the difference between efficacy and effectiveness studies

A

efficacy studies = measure effect of intervention under experimental conditions - when maximum effort is put into making sure it’s delivered correctly.

effectiveness studies = measure effect of intervention under more real world conditions. likely to be lower, as people forget doses etc

76
Q

what two things define a RCT?

A
  1. use of a control arm - contemporary comparison group of subjects who did not receive the intervention
  2. random allocation of subjects to intervention and control arms
77
Q

describe the design of a cluster randomised RCT

A

groups of individuals (“clusters” e.g. schools, GPs, villages etc) are randomly allocated to either intervention or control.

everyone in each cluster receives the same thing (intervention or control)

78
Q

when is cluster randomising useful for RCTs?

A

when there’s a large risk of contamination or migration between the intervention or control groups

79
Q

do cluster-randomised trails require larger or smaller sample sizes than individually randomised, and why?

A

larger

individuals within a group will share characteristics, and can’t be treated as independent - this reduces the statistical power, so you rely on an increased sample size to compensate

80
Q

what are some sources of selection bias in an RCT?

A

at enrolment - some people may be more likely to enrol than others e.g. someone with asymptomatic hep B is less likely to bother enrolling than someone with abnormal LFTs that’s concerned about their health

follow-up - if there’s different lengths of follow-up between intervention and control arms - participants might be withdrawing because of side effects, or because they don’t think it’s working etc

81
Q

give some examples of types of random allocation that may be used in an RCT

A 
simple randomisation - uses random number tables
systematic randomisation
blocked randomisation
stratified randomisation
matched-pair randomisation
82
Q

explain the difference between analysing an RCT following intention-to-treat (ITT) or per-protocol analysis

A

ITT analyses based on the original allocation to intervention vs control arms, ensuring comparability between the two - doesn’t allow for participants dropping out, switching groups etc.

per-protocol analysis only included participants that actually received the intervention/placebo “per protocol” (as planned at the start of the study)

83
Q

what is the aim of primary prevention?

A

to stop an outcome developing - either by preventing or reducing exposure to a risk factor

84
Q

what is the aim of secondary prevention?

A

to interrupt progression from early to mid-stage of the outcome.

achieved by early detection and prompt treatment.

85
Q

what is the aim of tertiary prevention?

A

to reduce complications or severity/disability from an existing disease/outcome, by offering appropriate treatments and interventions

86
Q

explain the difference between screening and diagnosis

A

screening aims to identify small numbers of individuals at high risk of an outcome, whereas diagnosis is used to confirm the outcome

87
Q

what two measures tell you the validity of a screening programme?

A

sensitivity and specificity

88
Q

define sensitivity

A

the proportion of those who truly have the condition who are correctly identified

89
Q

how do you calculate sensitivity?

A

true positives / total no. individuals with the condition

90
Q

define specificity

A

the proportion of those who truly do no have the condition who are correctly identified

91
Q

how do you calculate specificity?

A

true negatives / total no. individuals without the condition

92
Q

how do you calculate PPV?

A

TP / TP+FP

93
Q

how do you calculate NPV?

A

TN / TN+FN

94
Q

what is positive predictive value?

A

proportion of positive test results that are correct

95
Q

what is negative predictive value?

A

proportion of negative test results that are correct

96
Q

explain lead time bias

A

this can occur when screening identifies an outcome earlier than it would otherwise have been identified, but doesn’t actually impact the outcome

survival time appears longer, but in actuality it’s the same, cases are just detected earlier

97
Q

explain length time bias

A

can occur for outcomes with longer latent periods - length time bias makes screening appear to improve outcomes, but actually the screening is just picking up slow progressing conditions that would have always had a better prognosis than a rapidly progressive, more severe case of the disease

98
Q

define bias

A

a systematic error that leads to an incorrect measure of association

99
Q

define confounding

A

when an apparent association between an exposure and an outcome is actually the result of another factor (a confounder).

confounding factors are independently associated with both the exposure and the outcome, but do not lie on the causal pathway.

100
Q

what are the three ways of avoiding confounding effects when designing a study?

A
  1. randomisation - of individuals to either exposure or control groups
  2. restriction - limits the study to people who are similar in relation to the confounder, although this can reduce generalisability
  3. matching - selects the two comparison groups to have the same distribution of potential confounders (mostly relevant for case-control studies)
101
Q

what are the two ways of controlling for confounding during analysis?

A

stratification and statistical modelling (e.g. regression analysis)

102
Q

what are the 9 Bradford-Hill criteria for determining causality?

A
  1. strength (of the association)
  2. consistency (reliability of results)
  3. temporality (does the exposure come before the outcome?)
  4. dose-response (does increased exposure lead to increased outcome)
  5. plausibility (is there a reasonable biological mechanism?)
  6. reversibility (does an intervention to remove/reduce exposure result in elimination/reduction of the outcome?)
  7. coherence (is there a logical consistency with other available info?)
  8. analogy (is there a similarity with other established cause-effect relationships?)
  9. specificity (is the relationship specific to the outcome of interest?)

MPH Epidemiology (3 decks)

Brainscape helps you reach your goals faster, through stronger study habits.
© 2024 Bold Learning Solutions. Terms and Conditions