Upgrade to remove ads
Terms in this set (62)
Assess, Access, appraise, act
Hierarchy of evidence
o Risk factor/intervention in question occurred outside investigator's control
o Observing the consequences of a natural experiment
o Risk factor/intervention is deliberately introduced and is within investigator's control
o Rarely if ever ethical where a known harmful risk factor is involved
A detailed report of an unusual disease in a single patient
A detailed report of an unusual condition in several patients
Provide an important interface between epidemiology and clinical medicine
A study in which info is collected in a planned way in a defined population at one point in time.
Frequency of a particular behaviour or opinion.
Uses data from population groups to compare disease frequencies
In same population at different times OR between populations at same time
Study starts with identification of the disease.
Study of multiple risk factors in relation to a single disease.
Subjects are selected based on whether they have disease (cases) or not (controls).
Exposures are measured retrospectively.
Exposures in cases and controls are compared.
Study starts with identification of the exposure.
Study of multiple diseases in relation to a single risk factor.
Subjects are selected based on whether they have been exposed to a factor, or not.
Outcome has not yet developed.
The 2 groups are followed up over time.
Incidence of outcome in exposed and non-exposed is compared.
Randomised Control Trial
Experimental study where participants are randomised either to receive the new intervention being tested or to receive a control treatment
− Collect all studies addressing a particular health question
− Base conclusions on cumulated results
− Presence of methods section indicates systematic review
o Statistical process of combining the results from several studies
Q: Frequency + rate/ demographics
Aetiology (cause) + risk factors
Patient or professional experience
Which is the best study design to investigate an epidemic of limb malformations in newborns?
Case control study
Is cytisine effective as a smoking cessation therapy?
Do coffee drinkers suffer from lower rates of depression?
How common is TB in Birmingham and who is affected?
Is suicide in teenagers related to antidepressant prescribing?
Case control or ecological study
Why do young people have unprotected sex?
Qualitative, cross sectional
were specific risk factors like maternal smoking and bed caring more likely to be present for babies experiencing SIDS in your area than for other babies of the same age?
What effects do HRT, Diet and alcohol have on the development of cancer in middle aged women?
Case control advantages
Efficient for rare diseases or diseases with a long latency period
Good for studying dynamic populations in which follow-up is difficult
Case control disadvantages
Subject to selection bias
Inefficient for rare exposures
Info on exposure is subject to observation bias
Indicate the temporal sequence between exposure and outcome
Allow you to calculate incidence of a disease
Useful for evaluating effects of rare exposures
Enables examination of multiple outcomes
Have to follow large sample for a long time
Not good for rare diseases or diseases with a long latency period
High internal validity
Control of exposure
Randomisation ensures balance of factors that could influence outcome
True measure of efficacy
Limited external validity
Artificial environment that includes strict eligibility criteria and is conducted in specialised centres, which limit generalizability
Difficult to conduct
Limited scope due to ethical considerations
What is epidemiology?
− The study of patterns of health + illness + associated factors at the population level
− Helps in identification of risk factors for disease + determining optimal treatment approaches
Different types of clinical questions
- Effectiveness (drugs; surgery; care pathways)
Does this intervention do more good than harm?
How likely is a patient to have a condition on the basis of test results (History & examination & other tests)
How common is this condition in this population?
- Prognosis (natural history)
Prediction of future outcome...."How long have I got, doctor?"
- What is the patient experiencing... experience of illness; acceptability of treatments; discomfort associated with treatment; quality versus length of life
- Aetiology/risk factor
-Does this 'factor' cause this disease? And how do we target this appropriately
o Formulate a clear clinical question
o Identify study design that would provide the most accurate answer
o Need to know what we're looking for - sometimes referred to as defining your clinical question
o Search for + retrieve research
o Need to get hold of the information ie how to search
Accessing evidence: seeing if anyone else has researched this topic before: PubMed, BMJ, MEDLINE, conchrane library (rcts)
o Identify the strengths + weaknesses of the research
o Understand what the research results mean
o Need to understand what we've got- positive or negative results
- Does the study address a research question which is relevant to my clinical problem - is there a match between my PICO & the study question?
- Did they use the study design most likely to provide a valid answer to the clinical question ie for the effectiveness of a therapy (was it an RCT?)
- Was the study done well / is the study trustworthy ? (Were steps taken to ensure validity / reduce bias)?
- If the study was done well (valid), what were the results?
- randomisation to intervention / comparator groups was indeed random
- allocation to groups was concealed- those consenting participants do not know what the person is to receive- the intervention or the comparator
- intervention and comparator groups were similar at start of study- - before intervention or comparator applied
- follow-up was complete (minimal drop out of patients in treatment/ comparator groups) and sufficiently long- eg lasting reduction in pain not immediately following treatment only
- patients were analysed using an 'intention to treat' analysis
- outcomes were measured blind of knowledge to which group patient had been allocated to
ASIDE from the treatment, what other factors may impact on how likely an individual is to respond to treatment or have an adverse event
Age, co-morbidity, severity of illness, genes: Perhaps some are older, or have more severe disease, or have other illnesses e.g. diabetes. There are probably also unknown risk factors leading to differences.
So to compare the effect of a treatment on 2 groups we would want the groups to be similar at the start of the trial.
To prevent bias, what should we do with the patients who swap trial arms?
Patients should be analysed according to original group they were randomised to (ignoring where they end up)
INTENTION TO TREAT ANALYSIS
Comparison is then made between the consequences of a policy decision to treat in a particular way, rather than simply the consequences of treatment, (how drug performs in the real world rather than in a research setting). It answers the question of 'what is the best treatment policy'.
Any factor that results in a systematic deviation of results from the truth....
Bias and precision
− Selection bias
o The selection of individuals, groups or data for analysis such that proper randomization is not achieved, thereby ensuring that the sample obtained is not representative of the population intended to be analyzed
− Allocation bias
o Bias resulting from a systematic difference (other than the intervention) between experimental and control groups in a clinical trial. Allocation bias can be avoided by randomization.
− Attrition bias
o A kind of selection bias caused by attrition (loss of participants), discounting trial subjects/tests that did not run to completion. It includes dropout, nonresponse (lower response rate), withdrawal and protocol deviators.
Systematic differences in the way outcomes are ascertained between comparison groups in an analytical study.
Systematic differences in care provided apart from the intervention under study in an analytical study.
− Positive results bias
o When authors are more likely to submit, or editors accept, positive than negative/inconclusive results- e.g. if I get a positive result e.g. this drug causes this, more likely to be published than a drug that causes no benefit or correlation
− Outcome reporting bias
o When several outcomes within a trial are measured but are reported selectively depending on the strength and direction of those results
o Apply the evidence, taking into account its limitations, alongside other info eg. Patient preference
o Need to apply the evidence and decide if and how information is going to change our course of action
− Relative risk RR
o Example: risk of no pain in chiro group divided by risk of no pain in physio group
- No difference between treatment and comparator
- More outcomes in treatment group compared to comparator group
- Less outcomes in treatment group compared to comparator group
Risk (absolute risk) of outcome in treatment group = a/(a+b)
Risk (absolute risk) of outcome in control/comparator group = c/(c+d)
Relative risk (RR) = [a/(a+b)] / [c/(c+d)]
Therefore if a CI for a RR includes 0 then the true effect includes 'no DIFFERENCE'.
− Risk difference RD
o Example: risk of no pain in chiro group - risk of no pain in physio group
- No difference between treatment and comparator
- More risk of outcome in treatment group
- Less risk of outcome in treatment group
Attributable risk or risk difference (RD) = [a/(a+b)] - [c/(c+d)].
Therefore if a CI for a RR includes 0 then the true effect includes 'no effect'.
− Number needed to treat NNT
o Number of patients you would need to treat to obtain 1 additional beneficial outcome compared to comparator
NNT is the number of participants that need to be treated for 1 additional participant in the intervention group to avoid an adverse outcome over the stated period of time compared to the comparison group
OR the number of participants that need to be treated for 1 additional participant in the intervention group to benefit compared to the comparison group.
If the NNT is positive, more participants in the intervention group will experience benefit for each participant in the comparator group.
If the NNT is negative fewer participants in the intervention group will experience benefit compared to those in the comparator group.
Example of rr, rd and nnt
- Risk Difference = +22%... There is a 22% greater ABSOLUTE increase in outcomes (risk of no pain) in the chiropractor group compared to the physio group. 22% more of the chiropractic group are likely to be pain free at 3 months.
e.g. 0.92 = 8% risk decrease
- Relative Risk = 1.35.... Risk of being pain free at 3 months in chiropractor group is 135% of that of the physiotherapy group
- Number Needed to Treat = 4.5..... Would need to treat an extra 5 patients with chiropractic treatment rather than physiotherapy to achieve one more patient with no pain at 3 months.
summary of best study designs
Confidence intervals illustrate the precision of a result, for example the size of a relative risk or the size of a risk difference. Wider confidence intervals indicate a less precise result (more uncertainty surrounds the effect size).
for example a risk difference of +16 might have a CI (+8 to +24). We interpret this in the following way: a study result gives a risk difference of +16 but the true risk difference could be as low as +8 or as high as +24. Smaller studies generate wider CI as with fewer participants there is more uncertainty about the true effect size.
A 95% CI can be interpreted in the following way: 'If we were to repeat this study 100 times, 95 out of 100 times the result would lie between the lower and upper 95% CI)
stages of carrying out a systematic review
− Define question
o PICO (population, intervention, character, outcomes)
− Write protocol
o Description of methods
o Ensures methods and problems are considered a priori
o Avoids introducing bias
− Search for evidence/studies
o Comprehensive identification of relevant studies
o Using bibliographic databases e.g. medline, embase, Grey literature (hard to identify)
− Select relevant studies from search
o Unbiased selection of studies for review
o Two reviewers independently using pre-defined criteria (based on question(PICO)) to select relevant studies from the search, this helps to reduce errors
− Appraise studies
o Unbiased assessment of methodological quality of each study
o Two reviewers independently using checklist (e.g. casp)
o Reduces errors
− Extract, analyse + summarise data
o Unbiased extraction of relevant data from studies in review
o Two reviewers independently using a predefined analysis of data
o Use of validated appropriate methods of numerical data synthesis (meta-analysis)
o Summary Effect of Intervention vs Control (or Comparator) on Outcome
− Interpret review results
o Considers the results in the context of the quality of the studies included in the review
o Unbiased interpretation
− Discussion + conclusions
o Considers implications of the review findings in the context of current practice and knowledge and outlines specific future research
− e.g. forest plot
o Statistical technique used to combine the quantitative results from a number of studies
Critical appraisal questions (10)
1. Did the review ask a clearly-focused question?
2. Did the review include the right type of study?
3. Did the reviewers try to identify all relevant studies?
- which bibliographic databases were used
- if there was follow-up from reference lists
- if there was personal contact with experts
- if the reviewers searched for unpublished studies
- if the reviewers searched for non-English-language studies
4. How did the reviewers decide which studies to include and did they assess the quality (validity) of the included studies?
- if a clear, pre-determined strategy was used to determine which studies were included in the review and how their quality was assessed
- defined selection criteria (based on PICO)
- defined validity process (e.g using CASP checklist)
- more than one assessor
5. If the results of the studies have been combined together (e.g. meta-analysis) was it reasonable to do so?
6. How are the results presented and what is the main result?
7. How precise are these results?
- if a confidence interval were reported. Would your decision about whether or not to use this intervention be the same at the upper confidence limit as at the lower confidence limit?
- if a p-value is reported where confidence intervals are unavailable
8. Can the results be applied to the local population?
9. Were all important outcomes considered?
10. Should policy or practice change as a result of the evidence contained in this review?
Descriptive v analytical studies
Sets with similar terms
Exam 1 Review
Nutr. 302 Exam 1 study guide
Other sets by this creator