Lecture Notes Flashcards
(133 cards)
1
Q
explain the:
what? who? where? when? why? so what?
of the epidemiological approach
A
what - case definition who - person where - place when - time why - causes/determinants
so what - prevention and control
2
Q
define “epidemiology”
A
the study of the distribution and determinants of disease frequency
(in human populations)
3
Q
what different factors come under “distribution of disease”?
A
time
place
person - age, gender, social class, ethnicity etc.
4
Q
differentiate between a suspected, probable and confirmed case of an infectious disease (e.g. a measles outbreak)
A
this will be defined by the case definition.
suspected - clinical features (e.g. fever and rash)
probable - clinical features + contact with confirmed case
confirmed case - clinical features and positive microbiology, serology etc.
5
Q
what is a case definition?
A
set of standard criteria for deciding whether or not a person has a particular disease or health related event
6
Q
explain the “sufficient, necessary and component” causes model
A
causation is multifactoral - individual ‘causes’ = components.
if enough components are combined, disease is expressed.
combination of component causes = sufficient cause.
some factors may be “necessary” causes (i.e. cannot experience disease without them), but often need more than just one necessary cause to express disease
7
Q
define prevalence
A
a measure of the proportion of the population that has a given disease, condition or characteristic at a given time (or time period)
8
Q
define point prevalence
A
no. cases at a point in time, compared with the total population
9
Q
define period prevalence
A
no. cases identified over a period of time, compared with no. people in the population over this time period.
NB - not NEW cases (that’s incidence!), just existing cases
10
Q
define incidence
A
frequency of NEW cases in a defined population in a specified time period
11
Q
define cumulative incidence (RISK)
A
no. NEW cases occurring over a given period of time in the population at risk at the BEGINNING of the time period
12
Q
what 4 different relative measures come under the term “relative risk”
A
prevalence ratio
risk ratio
rate ratio
odds ratio
13
Q
what does relative risk measure?
A
measures the strength of association between exposure and disease
14
Q
how do you interpret relative risk?
A
avoid using “more” or “less”
it’s a ratio telling us “how many times more likely” the outcome is in the exposed group
e.g. “the exposed group has a RR 1.25 times the risk of the unexposed group”
or
the exposed group were 1.25 times more likely to (have the outcome) than the unexposed group
15
Q
what are the 4 different measures of impact of a risk factor?
A
- attributable risk (aka excess risk)
- attributable risk fraction (or %)
- population attributable risk
- population attributable risk fraction (or %)
16
Q
what is attributable risk?
A
the excess incidence of our outcome that can be attributed to the exposure
17
Q
what does using an attributable risk fraction adjust for?
A
the fact that the exposed group would have had some disease anyway - AR fails to take into account the underlying, background rate.
i.e. - not all illness, even in the exposed group, will be due to the exposure
18
Q
what does the attributable risk fraction tell us?
A
what proportion of disease IN THE EXPOSED GROUP is attributable to the exposure
19
Q
what does population attributable fraction tell us?
A
what proportion of disease in the POPULATION that is attributable to the exposure
e.g. interpret PAF of 0.96 as “96% of (outcome) in the population are attributable to (exposure)
20
Q
why is age standardisation useful?
A
allows comparisons to be made between two populations with different age distributions
let’s you adjust for the confounder of age
e.g. can compare rates of CVD in two populations, even if one is a significantly older population
21
Q
what is a crude death rate?
A
death rate for the whole population, with no age (or other factor) adjustment
22
Q
describe how DIRECT age standardisation works
A
calculate death rates for each age group in your population
apply these rates to the same age groups in a “standard population”
produces expected deaths for each age group, and can total these to get the DSR (directly standardised rate) per 1000.
e.g. total population = 1000
total expected deaths = 38.5
DSR = 38.5 deaths per 1000 population
23
Q
describe how INDIRECT age standardisation works
A
take a set of “standard death rates”
apply these rates to your population
this produces expected deaths per age group
then you get the ratio of observed to expected deaths:
SMR = O/E
(SMR = standardised mortality ratio)
24
Q
what is a cross-sectional study?
A
a study in which data are collected on each study participant at a single point in time
a SNAPSHOT
aka prevalence study
25
what are the two types of cross-sectional study?
descriptive and analytical
26
what do descriptive cross-sectional studies do?
what do they measure?
collect info on frequency and distribution of health-related exposures or outcomes, in a defined population.
measure point or period prevalence of the outcome OR exposure
27
how are data typically collected for a cross-sectional study?
surveys
28
what do analytical cross sectional studies do?
investigate the association between exposure to risk factors and the outcome of interest
(NB - the info is collected simultaneously on each individual - no temporality!)
29
what are the differences between an analytical and a descriptive cross-sectional study?
descriptive cross-sectional studies basically just find the prevalence of an exposure or outcome, whereas analytical cross-sectional studies look at both exposures and outcomes to investigate the association between the two
30
what types of bias are cross-sectional studies particularly susceptible to?
recall bias - if asked about exposures that occurred a long time ago
non-response bias - always an issue with surveys - look at what the response rate is in the study
31
what measures are used in analysis of a cross-sectional study?
prevalence (of disease OR exposure)
| prevalence ratio and prevalence odds ratios - for outcomes or exposures
32
list some advantages of a cross-sectional study
- quick, cheap and easy (ish!)
- provides prevalence of risk factors and disease in a defined population
- useful for health service planning
- repeated studies can monitor changes over time
33
list some disadvantages of a cross-sectional study
- exposure and disease info collected simultaneously = problems with temporal sequence - disease may modify exposure etc
- studying prevalent cases = can miss out cases with quick recovery, or short survival
- bias - recall, non-response
- not useful for rare conditions
34
what is an ecological study?
a study carried out at the population (or group) level rather than at the individual level
35
what is a multi-group ecological study?
aka ecological correlation study.
| compares different groups (or areas) at a point in time
36
what is a time-trend study?
a type of ecological study, aka a time series study.
examines data in a population over time.
investigates if changes in incidence correlate with changes in exposures over time.
can be long (e.g. seasonal variation) or short (e.g. daily variation)
37
give examples of information that might be available at a population level, but not an individual level?
to study these, we do an ecological study
```
pollution
income
GDP and other national statistics
per-capita consumption
climate
diet
etc. etc.
```
38
give some reasons to study groups/populations
| ecological studies
- to investigate differences between populations
- to study group-level effects (e.g. seat belt law only works at a group level!)
- convenience and availability of group level data (e.g. air pollution data is only available at a group level)
- quick and cheap study design!
39
give 4 reasons why ecological studies must be interpreted with caution
1. confounders - often, you can't adjust for these due to lack of data
2. bias - data may be collected using different methods or definitions over time or in different places
3. ecological fallacy
4. migration of populations between groups can dilute differences
40
what is the "ecological fallacy"
cannot assume that group level associations between exposure and outcome will also apply at the individual level
e.g. increased meat consumption and breast cancer rates - is it the meat-eaters who get the cancer?
41
what is a cohort study?
a "follow up" or "observational" study.
a cohort = a group of individuals sharing a common characteristic.
cohort studies take exposed and unexposed cohorts and follow them up over time, measuring incidence.
may be prospective or retrospective.
exposure is decided before outcome is observed.
42
what does a cohort study do?
compares INCIDENCE of an outcome in individuals with different exposure to a risk.
useful for investigating rare exposures, and/or several outcomes.
43
what measures does a cohort study provide?
risk ratio, odds ratio or rate ratio.
derived from incidence that is measured over the course of the study.
can use these to calculate AR and PAR
44
what are the two main types of cohort study?
prospective and retrospective.
prospectives start now and follow-up into future.
retrospective use existing data on exposures and outcomes.
45
how do retrospective cohort studies work?
all the events have already taken place, and records of them must exist.
a 'start date' for the study is in the past, and then records are checked to see what outcome(s) developed after the start date.
does NOT look back from an outcome to find the exposures!!
46
explain some factors that must be considered in selecting a study population for a cohort study
if it's a common exposure, select your population before classifying by exposure - if it's a rare outcome, you may need to recruit on the basis of exposure.
in selecting an unexposed group, you may choose either an internal or external comparison group, or compare with general population - but beware healthy worker effect
47
what are some important considerations to do with collection of outcome data in cohort studies?
might need a long follow-up period - loss to follow up (aka attrition) may be a serious problem!
data should be collected without knowledge of exposure status
48
how do you decide whether a risk or rate (ratio) is most appropriate for a cohort study?
if follow up times for all participants are similar, use risk.
if they vary, use rate so that person-time at risk is taken into account.
49
list 5 things that could explain an observed association between exposure and outcome in a cohort study
1. true association
2. bias
3. confounding
4. chance
5. reverse causality
50
list the 9 Bradford Hill criteria for causality
1. strength
2. consistency
3. dose-response
4. temporality
5. plausibility
6. reversibility
7. coherence
8. analogy
9. specificity
51
which of the 9 Bradford Hill criteria for causality do cohort studies do a good job of meeting?
temporality!
| one of the few study designs that definitely meet this criteria
52
define bias
any error that results in a systematic deviation from the true estimation of the association between exposure and outcome
(a systematic error which leads to a distortion of the truth)
53
list some of the important biases affecting cohort studies
loss to follow up (selection bias)
non-participation (selection bias)
classification of outcome and exposure (observer bias)
54
define confounding
situation where a factor is associated with the exposure of interest, and independetly influences the outcome, but does NOT lie on the causal pathway
e.g. grey hair and back pain are associated - is this a causal relationship?
no! age is a CONFOUNDER here.
55
list some strengths of cohort studies
- useful for rare exposures
- can study effect of exposure on a range of outcomes
- if accurate and detailed exposure assessment is carried out, can assess for a dose-response relationship
- data on potential confounders can also be collected (if prospective)
- meets temporality!!
56
list some limitations of cohort studies
- large sample size may be required
- can't use for rare outcomes
- costs (with propsective)
- time required for follow-up (with prospetive)
- retrospectives don't usually have as accurate and consistent exposure assessment, or data on confounders
57
what study design is useful for rare exposures?
cohort study
58
what study designs should be avoided for studying a rare outcome?
cross-sectional
| cohort
59
describe the study design of a case-control study
individuals with the outcome of interest (cases), and individuals without the outcome (controls) that match the cases on some demographic factors, are identified from a population.
then look back to identify how many individuals in each group were exposed/unexposed.
60
what things must be considered when deciding how to carry out case selection for a case-control study?
need a precise case definition, with clear inclusion and exclusion criteria set BEFORE selection begins.
will you be using prevalent or incident cases?
61
should you use incident or prevalent cases for a case-control study?
generally, incident cases preferred.
prevalent cases may be abnormally more or less severe (e.g. if less severe ones have recovered, or if more severe ones have died)
the disease influences how many cases picked up by prevalent or incident cases.
if you use prevalent cases, you can't distinguish between factors associated with the rates of the disease, and factors associated with the disease PERSISTING
62
what considerations must be made when selecting controls for a case-control study?
controls should represent exposure distribution of the population from which the cases were drawn.
need inclusion and exclusion criteria that are similar to that for cases (but obvs no disease!).
need to be careful where you source controls from e.g. if cases are hospital based, should controls also be hospital based? won't represent general population but would be more similar to cases...
Matching can be done at individual or group level (e.g. match individual controls to a case, or e.g. if case group is 70% male, ensure control group is also 70% male)
Beware overmatching - when matching variable is closely related to exposure variable
63
what is a nested case-control study and why do one?
a case-control study carried out within a full cohort study.
e.g. cohort study of outcome A following exposure A, could also carry out a nested case-control looking at exposures B, C and D in relation to outcome A. (i think??)
avoids selection bias, as cases and controls arose from same population.
also avoid info bias as all the same data collection is carried out on all participants.
avoids reverse causality
64
what measure is used in analysis of a case-control study?
odds ratio of exposure
65
list some biases affecting case-control studies
observer bias - ideally, researcher collecting exposure should not know case-control status
reporting (response/recall) bias - cases may remember exposures better than controls e.g. cases of asbestosis more likely to remember that an old work place has asbestos than controls!
66
give some strengths of case-control studies
- relatively cheap and quick
- useful for rare outcomes
- can study effect of multiple exposures on disease risk for a single disease
67
give some limitations of case-control studies
- potential for selection/information bias
- problems with temporality - possible reverse causality
- no good for rare exposures
- no good for multiple outcomes for a single exposure
- can't estimate incidence or prevalence of a population
68
what type of study design is useful for studying rare outcomes/diseases?
case-control
69
what is an intervention study?
a study in which participants are actively allocated an intervention by the investigators
i.e. an experiment
70
what is a randomised control trial?
Randomised allocaation to intervention or control.
Controlled - use of a contemporary comparison arm, with participants given nothing/placebo/usual treatment.
Trial - an experimental study
71
why do an intervention study?
observation studies are more subject to bias and confounding.
RCTs provide GOLD STANDARD for causality
72
what are the two main considerations when allocating participants to intervention or control in an RCT?
randomisation and blinding
73
give some different types of randomisation that might be used in an RCT
stratified randomisation
blocked randomisation
systematic randomisation
simple randomisation
74
what are the benefits of blinding in an RCT, and who should be "blind"?
avoids measurement/reporting/analytical bias.
| investigator, participant and statistical analyst should all ideally be blind.
75
explain the difference between and efficacy and an effectiveness trial?
in an efficacy trial, you are testing the maximum effect of an intervention if used in a closely monitored setting (e.g. the effect of a drug when patients are supervised taking it each day, or given lots of reminders etc).
in an effectiveness trial, you are looking at what the effect of an intervention is in routine clinical practice - this is more generalisable.
i.e. in an efficacy trial a drug might have X effect, with patients receiving daily reminders to take the drug at correct dose, time etc.
but in an effectiveness trial, the drug will then have Y effect (lower than X), because some patients will take it at the wrong time of day, or miss doses etc
76
what is a cluster randomised trial and when might it be used?
allocation of groups rather than individuals e.g. entire schools, or GPs.
needed when the intervention is at group level e.g. free milk in schools.
also useful if contamination between intervention and control groups is likely.
BUT - need larger sample size
77
explain the difference between intention-to-treat analysis and per protocol analysis, as used in an RCT
ITT should always be the primary analysis of an RCT - outcome is compared between study groups according to their initial allocation, regardless of any loss to follow up, or switching between groups.
this ensures comparability and avoids potential selection biases that might arise.
a per-protocol analysis can tell you the true potential effect of an intervention, if e.g. compliance was improved.
only participants who received intervention according to protocol are included in analysis.
78
what measures might be used in analysis of an RCT?
risk ratio, rate ratio, relative risk reduction.
absolute risk reduction
numbers needed to treat (NNT)
79
briefly list some ethical issues associated with RCTs
control group actively denied the intervention
use of placebo - generally, give the current "usual" treatment, rather than replacing that with a placebo
informed consent
80
what are "stopping rules" in relation to RCTs
to meet ethical requirements, RCTs will usually have predefined stopping rules to ensure that if a trial is showing clear harm or benefit early on, it is not continued.
This avoids undue risk to participants, depriving control group of an effective intervention, or continuing an ineffective intervention
81
give some strengths of RCTs
- minimise risk of bias and confounding (especially if properly randomised and blinded)
- can study multiple outcomes
- can measure "incidence" of outcome
- provides strong evidence of causal relationship
82
give some limitations of RCTs
- EXPENSIVE
- may need long follow up
- risk of high drop out rates
- ethical concerns
- RCTs may end up with conflicting results anyway
83
what is the aim of primary prevention?
to prevent a disease from occurring.
| carried out when no disease is present, done by reducing exposure or risk factor levels.
84
give some examples of primary prevention
lifestyle changes to reduce CVD risk.
fluoridation of drinking water to prevent tooth decay.
childhood imms to prevent communicable diseases.
85
what is the aim of secondary prevention?
detect early disease in order to alter the course of the disease.
also, prevention aimed at preventing a disease from recurring.
86
give examples of secondary prevention
SCREENING!
e.g. screening for breast cancer allows earlier treatment, altering course of disease.
also, treatment with aspirin to prevent recurrence of a heart attack.
87
what is the aim of tertiary prevention?
to minimise disability and prevent complications
88
give some examples of tertiary prevention
rehabilitation after a stroke.
| treatment to prevent death after a heart attack.
89
explain the prevention paradox
a preventative measure which brings a big benefit to the population, often offers little to each individual.
e.g. seat belt law - for every 1 life saved, 400 people have to wear their seatbelt everyday for 40 years.
so there's 399 people who have received no benefit to their survival from wearing a seatbelt daily for 40 years!
90
explain the population vs high risk approaches to prevention?
population approach means targeting the entire population with a preventative measure, offering each individual little benefit in exchange for a reduction in overall population risk.
high risk approaches target only those at high risk of an outcome.
91
give some advantages of a population approach to prevention
- potential to benefit the whole population
- "behaviourally appropriate"
? not sure what this means - maybe that it's seen as good to spend public money on strategies that can benefit everyone?
92
give some disadvantages of a population approach to prevention
- small benefit to individuals
- poor motivation of subjects
- poor motivation of physicians
- benefit-to-risk ratio may be low
93
give some advantages of a high-risk approach to prevention
- intervention is appropriate to the individuals targeted
- subjects and physicians are more motivated
- benefit-to-risk ratio is good
94
give some disadvantages to the high-risk approach to prevention
- screening costs (have to identify who is at high-risk, and this costs money/resources!)
- temporary effect
- limited effect
- "behaviourally inappropriate" ??
95
define screening
a process which sorts out apparently well people (i.e. those without symptoms) who PROBABLY have a disease (/precursors/susceptibility to a disease) from those who PROBABLY do not.
NB - not intended to be diagnostic
96
what are the main purposes of screening?
primary or secondary prevention.
secondary - e.g. screening by mammogram for breast Ca to treat it early
primary - screening to identify people with risk factors and reduce risk factor levels (e.g. NHS health checks, well man/woman checks)
97
give some aims of screening
- reduce risk of developing disease
- provide treatment
- provide information (e.g. pre-natal screening for genetic disorders)
98
define sensitivity
proportion of people with the disease who are correctly identified by the screening test
99
define specificity
the proportion of people without the disease who are correctly excluded by the screening test
100
define positive predictive value
the proportion of people with a positive test result who actually have the disease
101
define negative predictive value
the proportion of people with a negative test result who do not have the disease
102
which 2 of the 4 measures of effectiveness of screening are affected by underlying prevalence?
PPV and NPV.
sensitivity and specificity are specifically about the screening TEST - they will not change unless the test is altered.
103
if prevalence of a disease decreases, what happens to PPV and NPV?
PPV will decrease, NPV will increase
104
what three criteria for screening must the CONDITION meet?
1. condition should be an important health problem
2. natural history of the condition should be understood
3. there should be a detectable early stage
105
what three criteria for screening must the TREATMENT meet?
1. there should be an accepted treatment for the disease
2. facilities for diagnosis and treatment must be available
3. adequate health service provision should be made for extra clinical workload resulting from screening (e.g. need to be able to do the extra breast surgeries before you start screening people for breast Ca!)
106
what three criteria for screening must the TEST meet?
1. a suitable test should be devised for the early stage of a condition
2. the test should be acceptable
3. intervals for repeating the test should be determined (e.g. cervical screening every 3-5 years)
107
what two criteria for screening must the RISKS AND BENEFITS meet?
1. there should be an agreed policy on whom to treat
2. costs must be balanced against benefits
also - risks (physical and psychological) should be less than benefits!
108
who came up with the 10 (ish) criteria for screening?
Wilson and Jungner
109
what are the main biases that can affect studies evaluating screening?
selection bias
lead-time bias
length-time bias
110
what is the best study design to use to evaluate a screening programme?
RCT - individual or cluster
111
how does selection bias affect evaluation of screening programmes?
people who choose to participate in screening programmes may be different from those that don't bother
e. g. may be at higher risk - family hx of breast Ca = more likely to attend breast screening
e. g. may be at lower risk - women in higher SE groups are more likely to attend cervical screening, but are at lower risk
112
explain lead time bias
can occur when survival time is used as outcome to assess screening.
patients in screening group appear to have a longer survival time, as disease is being diagnosed earlier (at point of screening, rather than when symptoms develop) - however, survival time is actually the same.
113
explain length-time bias
a bias in which individuals with slower growing/progressing, less aggressive disease are more likely to be detected by screening, making screening appear to have better outcomes (as these individuals would have had a better outcome than those with a rapidly progressing, aggressive disease anyway)
114
give some examples of different types of screening
- population- based ("mass") screening programmes
- opportunistic screening e.g. GPs doing BPs when patients come in for any health issue
- Screening for communicable diseases
- Pre-employment and occupational medicals e.g. vision tests for commercial drivers
115
define bias
systematic deviation from a true estimate of the association between exposure and outcome
116
what are the two main categories of bias?
selection bias
| information (aka measurement) bias
117
define selection bias
a systematic error in selection of study participants, or in their allocation to different study groups
118
what type of selection bias affects cross-sectional studies?
non-response
119
what types of selection bias affects cohort studies?
healthy worker effect
| loss to follow-up
120
how can selection bias affect case-control studies?
selection of cases
| selection of controls
121
how can selection bias affect intervention studies?
systematic selection for intervention/control groups (rather than true randomisation)
122
define the validity of a measurement
the degree to which a measurement measures what it is meant to measure
123
define the reliability of a measure
the degree to which the results obtained by a measurement procedure can be replicated
124
give some examples of different types of information (measurement) bias
inaccurate measurement/classification of either exposure or outcome.
misclassification can be differential or non-differential
125
give some examples of sources of information bias
observer/researcher - observer bias
participant - recall bias
instrument - e.g. wrongly calibrated instrument
126
what does differential misclassification result in?
bias - in either direction.
e.g. measurement of BP in two samples by two nurses using two BPs
127
what does non-differential misclassification result in?
underestimation of any true association
(i.e. moves RR closer to 1)
this is a random misclassification, that's completely independent of exposure/outcome status
128
what steps can be taken to avoid information biases?
- blinding
- use of objective measures
- use of records rather than recall
- use of automated instruments rather than observers
129
what is confounding?
the situation where a factor is associated with the exposure of interest, and independently influences the outcome, but does NOT lie on the causal pathway
130
how can confounding be controlled for at the study design stage?
- restriction
- randomisation
- matching
131
how can confounding be controlled for in the analysis of a study?
- stratification
| - statistical modelling
132
what is the Comparative Mortality Ratio?
CMR = ratio of two directly standardised death rates
133
how do you interpret a CMR?
(for a CMR of 1.35)
after controlling for the confounding affects of age, the mortality in Country B is 35% higher than in Country A
