Exam Study Flashcards
Evidence Based Practice
Practice supported by scientific evidence, expertise and client questions
- best research evidence
- clinical expertise
- patient values
A combination of these three to achieve evidence based practice
Allocation Bias
(intervention bias)
Difference between treatment and control groups of the start of the experiment
- allocation bias reduced by random allocation
Detection Bias
(intervention bias)
Difference in how treatment and control groups are assessed/measured
Performance Bias
(intervention bias)
Events other than intended treatment
Attrition Bias
(intervention bias)
Some types of participants leave study, setting up unwanted differences between groups in the background characteristics of participants
Measurement Bias
(intervention bias)
Outcomes measured inaccurately
Publication Bias
(systematic reviews)
Studies researching unpopular research topics or treatments don’t get published, unavailable to reviewers
Long Lag Bias
(systematic reviews)
Delay on publication prevents research being found by practitioners/reviewers in time for their reviewers
Duplicated Publication Bias
(systematic reviews)
Same results from same studies repeatedly published, suggesting there’s more evidence than really is
Outcome Reporting Bias
mainly desirable/expected/statistically significant results get published, even through other results equally valid/informative
Citation Bias
Study cited by many other authors, reviewers are more likely to find that research compared with studies that are rarely cited/not at all
Database Inclusion Bias
Studies more easily found if available from online database
Language Bias
preference among reviewers for studies published in language they understand, commonly English
Reviewer’s Personal Bias
Reviewers’ unfairly exclude an article because they don’t like topic/results, even though valid and relevant
Level 1: Systematic Reviews of RCT
Evidence obtained from a systematic review of all relevant control trials
Reviews combines results of selected original studies to arrive at a summary conclusion
Advantages
> less costly to review rather than create a new study
> more reliable and accurate than individual studies
Disadvantages
> very time-consuming
> may not be easy to combine studies
Level 2: Randomised Controlled Trial
Randomly assigns participants into an experimental group or a control group
As the study is conducted, the only expected difference between the control and experimental groups in a RCT is the outcome variable being studied
Advantages
> results can be analysed with well known statistical tools
> good randomisation will ‘washout’ any population bias
Disadvantages
> expensive in terms of time and money
> volunteer biases - population in participants are from may not be completely representative
Level 3.1: Pseudo-randomised Controlled Trial
Same as RCT, but participant allocation to treatment/control not genuinely random, could be approximately random
Advantages
> less effort into random allocation
Disadvantage
> risk of allocation bias
Level 3.2: Cohort - Study
One or more samples (cohorts) are followed prospectively and evaluations with respect to disease or outcomes are conducted to determine which exposure characteristics (risk factors) are associated
e.g. start with group of healthy people (no disease) and follow over time to determine risk factors associated with getting disease
Advantage:
> standardisation of criteria/outcome is possible
> easier and cheaper than RCT
Disadvantage:
> no randomisation, meaning imbalances in patient characteristics could exist
Level 3.3: Case-Control Study
Compares patients who have a disease/outcome of interest (cases) which patients who don’t have the disease/outcome (controls)
looks retrospectively to compare how frequently the exposure to a risk factor is present in each group to determine relationship between risk factor and disease
Advantage:
> good fro studying rare conditions or disease
> lets you simultaneously look at multiple risk factors
Disadvantage:
> more problems with data quality because they rely on memory (recall bias)
> hard to find suitable control group
Level 4: Cross-Sectional Study
Captures information on a single treatment group only, at a single point in time
Advantage
> not costly/time consuming
> used to prove/disprove assumptions
Disadvantages
> doesn’t help determine cause/effect
Correlation and Correlation Coefficient
Describes the size/direction of relationship between 2 or more variables
CORRELATION COEFFICIENT (r) Measures the strength/direction of linear relationships between 2 variables on scatter plot - value between 1 - -1
Probability Values
Enable us to quickly determine whether or not a relationship between variables is statistically significant
- lower p-value = less likely result due to chance
- p-value = p<0.05, statistically significant
PICO and PEO Formats
PICO > Quantitative
- P: population, patient, problem
- I: intervention
- C: comparison
- O: outcome
PEO > Qualitative
- P: population, patient, problems
- E: exposure
- O: outcomes and themes
Nominal
Used for labelling variables without quantitative value
Labels
Ordinal
Order of values what’s important/significant, the difference between the values is unknown
Interval
Numeric scales, know order and the exact difference between values
Systematic Errors
Come from measuring instruments, data handling system, or instrument used incorrectly
Random Errors
Caused by unknown/unpredictable changes in environment
May occur in measuring instruments or environmental conditions
Sample and Sampling
Sample:
Group of participants who have been chosen to be apart of current study
Sampling:
Process of selecting participants so researchers can attempt to generalise their results back to a theoretical population
Theoretical Population and Study Population
Theoretical Population (target population): the larger group that the researcher wants to generalise their findings to
Study population (accessible population):
population the researcher has access to draw participants from
subset of theoretical population
Sampling Errors (random errors, systemic errors)
Participants chosen is inadequate or not random
Random Errors:
common/occur randomly as result of under/over-representation of certain groups
Systemic Errors:
result of inconsistencies/errors in sampling frame
Non-probability Sampling
Doesn’t involve randomisation
Process that doesn’t give all participants in the population an equal chance of being selected
Used to disprove hypothesis rather than prove
Probability Sampling
Fundamental characteristics is random selection of participants from population
Doesn’t ensure generalisability of findings, does ensure differences are due to chance
Type 1 Errors
When the p-value says that the results are statistically significant (the intervention works) but in reality it doesn’t
e.g. p=0.01: technically statistically significant but due to chance
Type 2 Errors
When the p-value says that the results aren’t statistically significant but in reality they are
e.g. p=0.06: technically not statistically significant but actually is
Simple Random Sampling
Every participant has an equal chance of selection
There are several methods (e.g. statistical software, random number tables)
Advantages:
> easiest method and most commonly used
> high generalisability
Systemic Random Sampling
Participants are systemically selected from a list, selected at intervals pre-determined by researcher
Generally every Xth number until desired sample size in reached
Advantages:
> very easy to use
Disadvantages
> systemic biases possible
> can only be random if the list is ordered randomly
Stratified Random Sampling
A population is divided into groups known as ‘strata’ and then continue by either implementing simple random sampling or systemic random sampling
Advantages:
> ensures adequate sample size for subgroups in the population of interest
Disadvantages
> problematic is stratas aren’t clearly defined
> analysis is typically complicated and the technique is time consuming
Cluster Random Sampling
When the population is divided into a cluster, then you randomly sample the cluster
Advantages:
> cost effective
Disadvantages:
> less efficient as you need a larger sample
Multi-Stage Random Sampling
Sampling techniques that is carried out in various stages
Sample has a primary population followed by sub-populations
Advantages:
> used when simple random, systemic or stratified sampling would be to complex/expensive
Convenience Sampling (aka. accidental/haphazard)
Participant chosen because of convenience (e.g. close proximity)
Advantages:
> easy access to participants
> cost effective
> can provide rich qualitative data
Disadvantages:
> doesn’t produce sensitive samples
> results hard to replicate
Snowball Sampling
Begin by identifying someone who meets the criteria for inclusion in your study
Ask them to remember others who they know meet the criteria
Advantages:
> used for hard-to-reach participants that would typically be hard to access
> cost effective
Disadvantages:
> not used for generalisations - except for similarly hard-to-locate participants
