Exam Flashcards
What are the 5 steps of EBP?
- Access
- Ask
- Acquire - search lit in publically accessable databases
- Appraise - assess rigour/value
- Apply
What frame work would you use to ask an answerable question?
PICO!
Eg. What is the effectiveness of physiotherapy intervention for lumbar disc herniation in adults
P = adults I = physiotherapy intervention C = no treatment O = pain & functional movement
What does PICO stand for?
P = patient/problem eg. Elderly with ankle sprain I = Intervention eg. "Does ice and rest reduce swelling" C = Comparison eg. Alternate treatment or treatment vs. no treatment
What does PECOT stand for??
P = patient/problem E = exposure (intervention) eg. Smoke C = Comparison eg. No smoking O = Outcome eg. Lung cancer T = timeframe eg. 5-10 years
Define sensitivity: and state equation
Probability diagnostic test is positive in patients who DO have the disease (TP)
TP = (TP/TP+FN) X 100
Define specificity: and state equation
Probability the diagnostic test is negative in patients that DO NOT have the disease
TN = (TN/TN+FP) x 100
What does a sensitivity of 0.75 mean??
There is a 75% chance that if a test is positive the individual has the condition
What does a specificity of 0.85 mean?
There is a 85% probability that if the test is negative the person does not have the disease
What is the mode?
- The most commonly occuring value in a data set
- works for both categorical and numerical data sets
What is the mean?
- arithmetic average
- continuous and discrete but NOT categorical
- most severely influenced by outliers
What is the median?
Middle value in distribution arranged in ascending or descending order
- not affected by outliers
What are the steps in hypothesis testing?
- State the research question(PICO)
- Specify null and alternate..assume null is true, try and find evidence to support alternate
- Decide on significance - alpha = 0.05 ..”how much are you letting chance play a role”
- Calculate test statistic(z, t or other)
- Reject or fail to reject the null (usually through use of p-value)
- State conclusion
State the null:
There is no significant difference between specified groups/populations
State the alternate
There is significant difference between groups/populations
Meaning sample observations are influenced by non-random cause -> research hypothesis
Draw the errors in decision making table:
Describe quantitative research:
- emphasis on control and objectivity, produce rigorous and generalisable results
- required to describe cause and effect
- tests a theory (deductive)
- variables are controlled
- numerical forms of data representation
- conclusion is stated with predetermined degree of certainty(alpha)
Describe qualitative research:
- describes in depth a phenomena
- aiming to generate a theory -> inductive
- data representation in narrative and verbal forms
- it asks “why” and “how many” through looking @ social and cultural factors
What are the misconceptions of qualitative research?
- unscientific and anecdotal
- lack scientifically rigorous methods
- personal perspective dominant
- question applicability and relevance
Why are the misconceptions of qualitative research not true?
Clinical practice more than science, underpinned by personal observation, reflection and judgement
- when looking at view points it does not have focus on minimisation of chance, so p-value plays no role
Types of data collection in qualitative research?
- Focus group
- Interviews (clinical = diagnostic, opinion poll = deductive, qualitative research = inductive)
- Participant observation
- Document analysis
Why can you not use library data base?
- not for evidence based information
- non-replicable, other health professionals will be unable to replicate your search strategy and obtaining similar results using the same databases eg. Medline or CINAHL
- search must be transparent and reproducible
Describe case report/case series + pros and cons
- Usually a rare disorder
- describe experience of single patient or group(report) or report on series of patients with an outcome of interest
PRO: richness of info
CON: isolated observations collected in uncontrolled unsystematic manner => cannot generalise to population
Describe crossectional study + pros and cons
- ASSOCIATION between possible causal factors and a condition..done so by determining exposure to factor
Eg. Recently given birth + cleft plate + drug during pregnancy
PRO: easy, inexpensive, ethically acceptable
CON: does not establish cause & effect, only association and exposure, depends on accuracy of recall
Describe case-control + pros and cons
- people with condition “cases” matched to people WITHOUT “control”
- retrospective…look back in time to determine proportion of people in each group exposed to suspected causal factor
PRO: quick, inexpensive, best for rare long time exposure & outcome
CON: recall basis, medical records inaccurate or incomplete
Describe cohort study + pros and cons
- treated or exposed vs. non-treated or non-exposed
- prospective..see how many develop disease or outcome
PRO: less expensive and easier to administer than RCTs, more ethically acceptable because potentially beneficial treatment is not withheld, and possibly harmful treatment is not given
CON: can never perfectly match cohorts, social/occupation factors
what is RCT + pros and cons
- The gold standard of experimental quantitative research
- involves blinding, randomisation, sample size justification and statistical analysis together have the greatest ability to minimise bias
Name the types of bias
- Sampling/selection = randomly and methodically
- Allocation = to treatment or control
- Maturation = occurs naturally over time
- Attrition = drop outs, how did it effect study?
- Measurement = eg. wrongly calibrated tool => standardisation
- Placebo = participants believe intervention will result in improvement
- Hawthorne effect = experience changes due to attention..pleasing researcher
What is the importance of EBP in health care quality and safety?
- Improve quality, effectiveness and appropriateness of clinical practice
- Shares decision-making with patients
- Substantiates the care provided to patients
- Reduce variations in practice patterns resultant from geographical difference or gaps in current knowledge!
What is the central limit theorem??
It states that the distribution of the mean is approximately normal if sample size is large enough, regardless of underlying distribution of original measurements
–> DRAW NORMAL DISTRIBUTION CURVE
Venn diagram what 3 things combine to make EBP
- research evidence
- clinical expertise
- patient values
Describe non-parametric testing:
- no assumption on distribution or variance
- ordinal, nominal data sets
- spearman correlation coefficient used
- median no mean used, due to catagorical data
What is a systematic review??
- focuses on a single issue within a topic
- replaced literature review, should have same level of rigour reviewing research evidence as producing evidence in the first place
- focuses on quantitative
- methods to reduce/eliminate bias, reproducible results
What is a meta-analysis??
- a product of a systematic review
- combine data of several studies that address a set of related research hypothesis
- gain true effect of treatment
When is meta-analysis useful?
- studies that report different effects
- small sample sizes
- single studies rarely provide definitive conclusions, need a body of evidence
- MUST HAVE: same pop, same intervention administered in same way, measure the same outcomes in same way, homogeneity
What are clinical practice guidelines:
- systematically developed statements to assist health professionals and patient make decisions about appropriate health care for specific circumstances
Aims of Clinical Practice Guidelines?
- improve quality of health care, increase chance of getting well as quickly as possible
- provide recommendations for treatment and care of people by health professionals
- develop standards to assess practice of health professionals
What to remember about Participatory Action Research:
- involves disempowered/marginalised individuals
- by involving them in research you will be able to create stratagies to help them
What to remember about Ethnography?
- focuses on the social and cultural influences on a persons view point
Which can also influence their health decisions
What to remember about grounded theory?
- specific focus on theory development
- generates new theory about particular phenomenon
- interplay between data collection and data analysis
- becomes a sort of circle
What to remember about Phenomenology:
Focus on the INDIVIDUAL every day world from the perspective of the person who is experiencing it
Eg. Child abuse, rape survivors
What to remember about feminist research:
Female health focus
Decisions for womens health not made by women
What to remember about discourse analysis:
- Uncovering dominant and marginalised discourses and locating inconsistencies
- what drove the decision?
Eg. White aus policy
What to remember about GENERIC qualitative/quantitative description:
- basic superficial over view
What to remember about case study(qualitative)
Looks at particular case/community
What are the key components which underpin health care quality?
Safety = no harm comes to patient
Effectiveness = if it has a sustainable and durable outcome
Timeliness = how quickly a person gets access to care
Efficiency = value for money
Patient centeredness = patient plays active role in decision making
Equity = everyone gets same care regardless of social/cultural factors
When talking similarities and differences between studies what do you talk about?
- observational/descriptive/experimental
- cause and effect/incidence/prevalence
How is sample size determined in qualitative research?
- data saturation: point in time when no more new info
- researcher runs out of time/money/resources
What are the 2 pairs of centrality and dispersion?
Mean + standard d = normally distributed
Median + interquartile range = skewed data
You get a question “based on p-values and confidence”… how do you answer?
- state alpha = 0.05
- state p value
- is p > or < alpha
- only 5% of studies will have the result due to chance
When justifying from sensitivity and specificity whether it is a good test or not?
No cut off, just want values to be as close to 1 as possible
Are qualitative research findings generalisable?
- No, as we’re not interested in distribution but rather just describing phenomenon of interest
- do not get representative sample, want to get depth of phenomenon
- not interested in role of chance
What are the strengths and limitations of literary review?
Strengths = answers background questions, good starting point for research, gets breath of topic, highlights gaps in knowledge, easy immediate access to broad info on topic
Limitations: no methodology, selective reporting
Limitations of systematic reviews
- publication bias, limited to whats being published..could be only positive studies
- inappropriate aggregation of studies
- heterogenous data = > meta-analysis
what denotes correlation?
- r is pearsons correlation coefficient, measure of linear association.
- scale -1 -> +1
describe confidence intervals
Confidence intervals provide information the precision of the estimate, range of values likely to a certain degree of confidence will contain populaiton parameter
- confidence level of 95% is usually selected
you get a question that says “based on confidence intervals” how do you answer?
- can do separate from p-value or together
- for “treatment” group, they are 95% confident that “what is being measured”(eg. disability/pain etc.) will be from 0.5->2.8, do same from control group
what does cohort and crosssectional studies actually measure?
cohort = incidence..measures at multiple points in time
cross sectional = prevalence..number of people at this point in time
when would you use PICO and when would you use PECOT?
PICO = singular focus on intervention PECOT = exposure, diagnosis/prognosis
what does confidence level of 95% mean?
if 100 samples were collected from population, confidence intervals calculated, 95 of these would contain the true pop mean
what is a type 1 error? give an example
rejecting the null if it is actually true
- saying a particular treatment has an effect when really it doesnt
whats the difference between parametric and non-parametric data?
parametric assumes normal distribution, data sets typically ratio or interval, can draw more conclusions from sample to population
non-parametric has no assumption of distribution or variance, ordinal or nominal data, more simple, less effected by outliers
what is a type 2 error?
why is it bad and in what kind of testing is it the worst?
failing to reject the null when it is false
- bad because it could mean withholding effective medication from people with disease
- diagnosic testing
what are some limitations of EBP?
- need time to develop new skills
- limited time & resources
- shortage of coherent, consistent scientific evidence
- difficulties applying evidence to care of individual patients
what are the misconceptions of EBP?
- promotes cookbook approach to patient management
- its just a cost cutting tool
- decisions made solely from research data
example of continuous data
nominal
height, weight, temp
example of discrete data
nominal
counts, number of students/children
example of ordinal
catagorical
grades, clothing size, mild-moderate-severe
example of nominal catagorical
gender, eye colour, buring-tingling-shooting
what is the hierarchy of evidence?
ranking system(levels) of various study designs like a ladder from most to least bias. Bias relating to internal and external validity
define internal validity:
relates to how you conduct research, so you’re sure effects seen is from exposure/intervention
define external validity:
generalisability of research findings from sample to reference population
what affects internal validity of study?
- Chance: random error, only minimised through sufficient sample size
- Bias: systematic error
- Cofounders: variables not taken into account, random allocation takes this into account by spreading cofounding events equally across groups
how can qualitative data be analysed?
- thematic: organise data into themes, code
2. content: systematic coding and categorisation eg. out of 10 ppl 8 said taste was most important
