Session 3 - reading & designing studies

lecture 3

layout of a typical paper

title

abstract

introduction

methods

results

discussion

conclusion

Title

concise, informative, and accurately reflect the content of the study

abstract

summarise study’s purpose, methods, findings

introduction

summarises context of study

methods

contains all details of how study was conducted

results

states findings of study

often uses tables and figures

discussion

results in context of published literature

logical interpretations, extrapolations and possible generalisations

conclusion

summarise main findings

emphasise significance of results in context of study’s objectives

when can an RCT be used?

must be possible to assign the exposure being studied

must be clinical equipoise (genuine uncertainty about which treatment is better

placebo/nocebo effect

knowledge of being in the trial/taking the trial intervention can cause a clinical effect

different care (other than intervention)

care that the intervention arm receives is different from the control arm

differential drop out

participants who drop out may be different from the ones that complete

intention to treat analysis used - outcome compared based on initial group assignments not on treatment eventually received

bias from investigators

risk of (subconscious) bias from the investigators measuring the outcome if they know which intervention the participants are recieving

blinding

those involved in the trial are not able to discern which study group participants have been assigned to

may be aided using a placebo control

single blinding

concealed from participants

double blinding

concealed from investigators & participants

RAMbo mnemonic meaning and use

Randomised (random/concealed allocation, baseline similarity of groups?)

attrition (adequate follow-up, intention-to-treat analysis, equal treatment of groups?)

measurement (blinded assessors of objective measures used?)

Use: when assessing an RCT for bias

Observational study

observes exposures and outcomes and looks for associations

intervention/exposure not assigned by the experimenters

Why might an observational study be used?

not possible to assign the intervention/exposure due to practicality, ethics, logistics

4 types of observational study focused on in FEBP

cohort

case control

cross-sectional

ecological

Cohort study

data obtained from groups who have been exposed/not exposed to new technology or factor of interest (found in a database?)

No allocation of exposure made by researcher

Best use: studies effect of predictive risk factors on an outcome

Advantages of cohort study

ethically safe

subjects can be matched

timing and directionality of events can be established

eligibility criteria and outcome assessments can be standardised

administratively cheaper and easier than RCT

Disadvantages of cohort study

controls may be difficult to identify

exposure could be linked to a hidden confounder

blinding is difficult

randomisation is not present

large sample size and long follow up may be necessary - not always possible for a rare disease

Case control study

Patients with outcome/disease and controls without outcome/disease are carefully chosen.

Information obtained as to whether the participants have been exposed to the factor in question

Advantages of a case control study

quick and cheap

only feasible method for rare disorders/those with long lag time between exposure and outcome

fewer subjects needed than cross-sectional studies

Disadvantages of a case control study

reliance on recall or records to determine exposure status

confounders

selection of control groups is difficult

potential bias in participants selected/participant recall

cross sectional study

examines relationships between diseases and other variables of interest in a specific population and a specific time

Best use: quantifying prevalence of a disease/risk factor and quantifying accuracy of a diagnostic test.

Advantages of cross sectional study

cheap and simple

ethically safe

Disadvatages of cross sectional study

establishes association at most, not causality

susceptible to recall bias

confounders may be unequally distributed

only includes those currently diagnosed/alive at time of study

group sizes may be unequal

ecological studies

measures exposures and outcomes at a population level

Advantages of ecological studies

quick and cheap

good to identify exposures and health outcomes at a broader level

Disadvantages of ecological studies

measures exposures and outcome at a single timepoint, so direction of effect cannot be measured

Ecological fallacy: risk associations between different groups of people may not accurately reflect the true association between individuals within those groups.

Retrospective study

looks back at existing data, e.g. medical records

all case-control studies

Prospective study

follows participants forward in time, so collect new data

Retrospective vs prospective studies

Retrospective studies quicker and cheaper, as data is already available, but may be gaps/inaccuracies in the data collected

Prospective studies there is more control over the nature and accuracy of the data collected.