RESS

Question 1

What is incidence?

Answer 1

Measures number of new cases that occur over a specified time period in a population at risk. Number of new cases ÷ number of new people at risk during the specified time period. If time period is one year: ‘annual incidence’.

Question 1

Why sample nigga

Answer 1

• Cost - Utility – might not want the entire pop (some may be more informative), may have sufficienct power sampling a subset, may want to reuse samples elsewhere and the two are incompatible. - Accessibility – might not be able to sample entire pop, some people unable/unwilling to partake, some people’s data will be incomplete.

Question 1

How to calculate odds ration?

Answer 1

Odds ratio for two groups = odds of event for exposed individuals ÷ odds of event for unexposed individuals.

Question 2

For populations/samples n

Answer 2

Use the students’ t distribution (slightly fatter bell curve) - we must estimate the population SD

Question 3

What is cause specific mortality?

Answer 3

Cause-specific mortality (proportion of deaths attributable to a specific cause) = number of cases with this disease who die ÷ total number of any cause deaths

Question 3

Pros and cons of prospective data collection?

Answer 3

 Helps ensure that you record all the variables you need, using best techniques available, applied consistently.  Helps enseure all participants provide data on all variables (reducing missingness).  CON - Time-intensive  CON - Resource intesnvie  CON - Tendency to collect data on more variables than you actually need or can use in your analyses.

Question 4

What is the usual accepted cut off for something to be significantly different from the null hypothesis?

Answer 4

p

Question 4

Why calculate power? (before conducting study and after concluding that no effect found)

Answer 4

 Be sure study is well designed  To (thereby) justify study to grant reviewers etc.  Demonstrate you were sufficienctly well-powered to detect an effect  To estimate that, if there is an effect, how small it must be for it to be missed.

Question 4

What are the key principles of ethical research?

Answer 4

 The project is conducted in compliance with an explicit protocol.  An authorised research ethics committee has approved the protocol.  Further research ethics committee approval is requested if: - There are any substantial changes to the project protocol - The project has any unexpected deleterious effects

Question 5

What is risk?

Answer 5

Risk = ‘probability’: the proportion of a population or sample with the outcome or exposure

Question 6

Ways of summarising and describing variables?

Answer 6

Frequency tables, bar charts/pie charts, average values, distribution of values, histograms

Question 8

what is an audit?

Answer 8

quality improvement process that seeks to improve patient care and outcomes through a systematic review of care against explicit criteria (evidence-based clinical standards).

Question 9

What are histograms?

Answer 9

Very similar to bar charts but can only be drawn for continuous variables or ordinal categorical variables. Offer visual representation of distribution of frequency as value of continuous (or ordinal categorical) variable increases.

Question 9

What is the study sample?

Answer 9

the units/participants drawn from the target population that constitute our data set.

Question 10

What is a variable?

Answer 10

a characteristic or ‘parameter’ for which a ‘sample statistic’ can be calculated as an estimate of the population parameter.

Question 11

What is overall mortality?

Answer 11

Overall mortality (proportion of pop who die) = number of deaths (regardless of cause) ÷ number of people in population

Question 11

Possible sources of measurement error

Answer 11

 Inherent variation/instability in the variable  The use of a proxy for a latent variable  Poor quality measurement equipment/technique  Self-reported/interviewer-generated information  Response bias – what interviewer wants/gets  Prestige bias – what appears favourable  Recall bias – what is accurately remembered  (Multiple) transcription from source(s) to proforma

Question 12

Three questions to identify study design?

Answer 12

Q1. What was the aim of the study? - Combine findings from more than one study – Meta-analytical - Describe a population - Descriptive - Quantify relationships between variables – Analytical (go to Q2) Q2. Was the intervention/exposure allocated by researcher? - Yes - Trial - No - Observational study (go to Q3) Q3. When was/were the outcome(s) determined? - Some time after the exposure/intervention? – Cohort - At same time as exposure/intervention? – Cross-sectional - Before the exposure/intervention determined? - Case-Control

Question 13

What does qualitative research aim to do?

Answer 13

Access the experiences of others

Question 14

As prevalence declines what happens to OR and RR?

Answer 14

RR changes, OR stays the same

Question 15

What is fatality?

Answer 15

Fatality = number of cases who die ÷ number of cases with the disease

Question 15

What should a project protocol provide sufficienct evidence to do?

Answer 15

 Enable project team members to conduct the project  Prospective project participants to assess whether to volunteer  Research ethics committee members to approve, refer or reject.

Question 16

What are high risk ethical issues?

Answer 16

 Vulnerable study participants  Covert data collection  Sensitive topics  Administering drugs/food/fluid  Conducting potentially harmful/painful procedures  Collecting biological body materials  Abnormal psychological stress  Prolonged or repetitive testing  Material or financial participation incentives  Data transfer outwith the European Union  Conflicts of interest  Potential risks to researchers, non-participants or the environment  Animal (Scientific Procedures); Human Tissues; Data Protection Acts

Question 18

What are 3 basic person characteristics that can determine disease risk?

Answer 18

Sexxxxxxxxx, class, age

Question 19

What's a fucking sample?

Answer 19

A collection of data drawn from a population

Question 20

What is odds? What is odds risk?

Answer 20

Odds similar to risk when proportion of the population/sample with the outcome or exposure is low. BUT odds risk is stable as outcome prevalence changes.

Question 20

What is a type I error?

Answer 20

You find an effect when there is no genuine effect - oh what fun!

Question 20

What is power?

Answer 20

Probability of finding an effect if there is one to be found

Question 20

When are confidence intervals smaller?

Answer 20

We have less variation in our data We have a larger sample size

Question 21

Potential sources of measurement bias?

Answer 21

 Information bias – extent of information varies amongst participants. Priori hypothesis and primary data collection.  Observer bias – influenced by prior knowledge or belief. Blinded, independent, objective or validated.  Recall/response/prestige bias – influenced by prior knowledge or belief. Blinded.

Question 23

Problem with pie charts?

Answer 23

can be visually complex when there are large numbers of classes or when the frequency of some classes is very low.

Question 25

What is person time?

Answer 25

Number of people at risk can vary during the time period. Epidemiological concept that multiples number of people at risk at various points across time period and uses this as the denominator in any calculation of incidence.

Question 26

How do you assess which covariates are important?

Answer 26

what are the (likely) causes of the exposure? What are the (likely) causes of the ‘outcome’? Which are (likely to be) caused by the ‘exposure’? Think conceptually and operationally about which covariates we would like/be able to measure.

Question 27

What is a sample?

Answer 27

a subset of a population, intended to be representative of that population, from which we can explore the population by ‘inference’

Question 28

What does a mediator do?

Answer 28

 Causes the outcome  Is caused by the exposure  Must not be adjusted for because it is on the causal pathway

Question 29

Is qualitative research time consuming?

Answer 29

Yes

Question 30

Types of projects that do require formal ethical approval?

Answer 30

 Non-human subjects/material covered by the Animals (Scientific Procedures) Act 1986  Service Evaluations involving vulnerable participants or ethically sensitive topics  Projects involving a new intervention – Experimental Research  Projects collecting new information on more than only service delivery – Non-Experimental Research

Question 31

What should a project protocol aim to explicitly describe

Answer 31

 WHY the project is necessary/justified (i.e. it is an important issue, that has not been examined before or needs verification)  WHAT the project will involve (i.e. how data will be collected, how, when and from and by whom; and how these data will be analysed).  HOW key ethical, legal and governance issues will be avoided, addressed or mitigated in the design and conduct of the project.

Question 32

3 types of clinically relevant studies

Answer 32

Audit, service evaluation, research

Question 32

What is population?

Answer 32

every member of a defined group of interest

Question 33

5 types of potential stakeholders?

Answer 33

1. Other researchers/evaluators/auditors 2. Research/evaluation/audit funders 3. Clinical practitioners/policy-makers 4. Service users 5. Society at large

Question 33

Types of qualitative study?

Answer 33

 Existing sources – examined post hoc. Text (website, leaflet, newspapers), audio (radio, film scores…), video (youtube, documentary, cct).  New sources – examined de novo. Individuals (one to one interviews), groups (focus groups, ethnographic observation).

Question 35

For chronic diseases will prevalence or incidence be higher?

Answer 35

Prevalence will be higher than incidence for diseases of long duration – chronic diseases.

Question 36

What is a latent variable?

Answer 36

Missing variable.

Question 37

When is there greater power?

Answer 37

Same factors that affect size of confidence intervals - mean effect is bigger, variation in effect is smaller, sample is bigger

Question 38

What is the epidemiological triad

Answer 38

Person, time, place

Question 38

In what studies is risk used?

Answer 38

Cohort (incidence)

Question 39

What does priori registration do?

Answer 39

- Helps reduce publication bias and selective reporting - Helps reduce unnecessary duplication and fosters better collaboration (including prospective meta-analysis) - Helps identify gaps in planned/ongoing research - Helps facilitate recruitment of participants - Helps screen for errors/weakness in research design.

Question 41

For what kind of variables can you estimate confidence intervals?

Answer 41

Normally distributed

Question 43

Three types of samples?

Answer 43

- Complete samples – entire study population - Unstratified random sampling – every member of the target population has same chance of being samples. - Stratified random sampling – randomly sample from the target population within strata (groups). Every member of the target population within each stratum has the same chance of being sampled. Usually numbers in each stratum in proportion to size of strata but may over sample (e.g. comparing two groups).

Question 44

What is prevalence?

Answer 44

Measures number of cases that exist at a specified time point in a population at risk

Question 45

As prevalence increases what happens to OR and RR?

Answer 45

OR \>\> RR

Question 46

For incurable diseases will prevalence or incidence be higher?

Answer 46

Will increase at same rate

Question 47

What does a confounder do?

Answer 47

 Causes the outcome (dependent variable)  Causes the exposure (independent variable)

Question 48

What does missingness do?

Answer 48

reduces number of participants on whom you have complete data, and therefore the ‘realised’ sample size that can be achieved in any analyses containing these variables

Question 50

For populations/samples n\>200 what do we use for confidence intervals?

Answer 50

use the ‘normal’ distribution and its z statistic value of 1.96. (we know the population SD)

Question 51

What is a stakeholder?

Answer 51

Stakeholder = people who have an interest in the results of research. Important to communicate research to them as effectively as possible.

Question 53

Frequency distribution histograms take what sort of curve?

Answer 53

Normal/bell curve

Question 54

Three descriptions of a 'causal' relationship?

Answer 54

 Functional (no contraception --\> teenage mother)  Theoretical (teenage grandmother --\> teenage mother)  Speculative (teenage grandfather --\> teenage mother)

Question 55

What does a competing exposure?

Answer 55

 Causes the outcome  Is unrelated to the exposure

Question 56

What is checking?

Answer 56

Whether sampling is stratified or not there are usually features that you hope will be ‘representative’ of the target population. These can be used to check sample is representative.

Question 57

What's the target population?

Answer 57

total finite population we wish to know about and from which the sample is drawn.

Question 58

Pros and cons of retrospective data collection?

Answer 58

 Data collection is less time-intensive and less resource-intensive  CON – not role in choosing which variables are measured, in choosing the number of participants, in the consistency/inconsistency of measurement, in the level of ‘missingness’ tolerated.

Question 59

Odds ratio = 1?

Answer 59

No effect bitch

Question 60

In what studies is odds used?

Answer 60

Case-control/trial (prevalence)

Question 61

What is a bar chart?

Answer 61

simple visual summary of frequency/relative frequency of cases/events that can also include confidence intervals.

Question 62

What do you need to know to calculate power?

Answer 62

 The effect size in your population.  The variability of the effect in your population  What sample size will/can you collect  Then calculate the probability that the observed effect in your sample will be significant (i.e. Confidence interval does not cross zero).

Question 64

What is relative frequency?

Answer 64

counts expressed as a percentage of the total number of counts

Question 66

For diseases of short duration will incidence or prevalence be higher?

Answer 66

Incidence will be higher than prevalence for disease of short duration – infectious diseases.

Question 67

What are the four main types of variable?

Answer 67

Categorical - nominal, ordinal Numerical - discrete, continuous

Question 68

Possible source of dissemination bias?

Answer 68

 Publication bias – eventful results more likely to be published. Priori registration and meta-analysis.

Question 69

Mnemonic for defining search terms used in literature searches?

Answer 69

PECOS - Patient/Participant/People - Exposure/Event/Experimental intervention - Comparison (if appropriate) - Outcome - Study design (appropriate for question)

Question 70

What is ethics/governance/consent?

Answer 70

- Ethics = ensuring that projects minimise risks and maximise benefits - Governance – ensuring that projects have appropriate permission(s) - Consent = ensuring participation is voluntary and without cost

Question 72

Type II error?

Answer 72

Fail to find an effect when there is a genuine effect - boo hoo!

Question 74

What is a service evaluation?

Answer 74

evaluates a proposed service or current practice with the intention of generated information to inform local decision-making.

Question 75

What is the p value?

Answer 75

the probability of seeing the result we have observed or something more extreme if the null hypothesis is true – this probability is the p-value

Question 76

Potential sources of sampling/selection bias? And solutions?

Answer 76

 External validity: non-representative samples. Use stratified/random sampling, and/or multi-site  Confounding – selection influences exposure and outcome. Use matching, adjustment or random sampling.

Question 77

Potential sources of analytical bias?

Answer 77

 Loss to follow-up: specific participants excluded. Intention to treat.  Omitted variable bias – imprecise adjustment for confounding. Priori hypothesis and primary data collection.  Attributional bias – interpretation of causality. Priori design.

Question 78

Aims of research ethics committees?

Answer 78

 Aim to ensure that the projects they approve…  Minmise the risk of harm to project staff, participants and society  Complies with relevant legal and regulatory requirements  Complies with requirements of professional indemnity insurance

Question 79

The less residual variation (noise) on a line graph...

Answer 79

the better the fit of the model and the more confident we will be (so the more statistically significant the correlation between the two variables will be).

Question 80

What don't you want in a questionnaire proforma?

Answer 80

 Ambigious questions – “Would you normally feel uncomfortable about your diabetes?”  Leading questions – “In what ways is diabetes a problem for your family?”  Multiple questions – “Have you found diabetes to be a limiting, longstanding illness?”

Question 82

Possible sources of missingness?

Answer 82

 Non-routinely collected viarbles  (Clinically) unimportant/irrelevant variables  Variables that are difficult/costly to collect  Variables that require invasive or potentially sensitive measurement (that undermines acceptability/measurement complicance)

Question 83

What is frequency?

Answer 83

number of cases/events expressed as if they were ‘counts’

Question 84

What does a DAG identify?

Answer 84

- Identify any measured/able ‘covariates’ acting as ‘confounders’ – so these can be adjusted for. - Identify any measured/able ‘covariates acting as ‘competing exposures’ – so these can be adjusted for - Identify any measured/able covariates acting as ‘mediators’ – so these are not adjusted for.

Question 85

What is research?

Answer 85

Derives generalizable new knowledge by addressing clearly defined questions with systematic and rigorous methods.

Question 86

Types of projects that do not require formal ethical approval?

Answer 86

 Data/material already published/publicly available – Secondary Research  Projects on non-human subjects/materials – Non-Human Research  Projects using only existing service management data – Audit  Projects collecting new information only on service delivery – Service Evaluation

Question 87

Methods of dissemination?

Answer 87

 Researchers/evaluators/auditors – journals (more prestigious = broader readership), conferences.  Funding bodies – regular reports.  Clinicians/policy makers – journals, conferences, professional associations.  Service users – via clinicians, focus groups/participants, press, online, leaflets.

Question 88

What are parameters that can influence disease risk?

Answer 88

Time of day, week, month, year, life (age

Question 89

What are adjustments?

Answer 89

Corrections made to sample statistics to make them... - Better estimates of true population values (e.g. when sample is not truly representative of population of interest). - Comparable to other samples/populations with different baseline characteristics (different age/sex structures).

Question 90

What does adjustment involve?

Answer 90

Adjustment involves calculating ‘stratum-specific’ rates by key characteristics (at least by age and sex) and weighting these rates in proportion to the distribution of these characteristics in the real or a standard population.

Question 91

How do we engage with stakeholders during study planning, implementation and dissemination?

Answer 91

 Study planning – patient steering groups, (peer) advisory committees, pilot studies.  Implementation – continued feedback, monitoring  Dissemination

Question 92

What is a dag?

Answer 92

A pikey dog. Nah not really... type of ‘causal path diagram’ with unidirectional (‘causal’) arrows linking variables and no circular paths. - Summarise the functional, theoretical and speculative relationships between variables pertinent to our question.