Statistics and Evidence Flashcards
Why is decision making important in medicine? (3)
- doctors make decisions very often
- these decisions have effects in patients, their families and wider society
- understanding the medical decision making process and the role if evidence can improve medical practice
what is the hypothetico-deductive model of decision making?
coming up with a hypothesis, and trying to find evidence to prove/disprove
name 4 factors influencing an evidence based decision
- evidence from research
- clinical expertise
- available resources
- patient preferences
name 3 reasons why we need EBDM
- limited time to read
- inadequacy of traditional sources of infromation
- disparity between diagnostic skills/clinical judgement and up-to-date knowledge/clinical performance
Name 5 ways in which evidence based decision making is possible
- development of strategies for effectively identifying and appraising evidence
- creation of systematic reviews and summaries
- creating of evidence based journals and clinical guidelines
- creation of information systems
- identification and application of strategies for lifelong learning
What types of research question are the following study designs suitable for:
- cohort studies
- case controls
- RTCs
- qualitative approaches
- Diagnostic and screening studies
- Systematic reviews
- prognosis; cause
- cause
- treatment interventions. Benefits and harm
- patient perspective
- identification
- summary of evidence for a specific question
- what are background questions?
2. what are foreground questions?
- questions querying general knowledge of disorder (e.g who what where when)
- questions querying specific knowledge about managing patients with disorder (PICO)
What does PICO stand for?
P - population
I - intervention
C - comparator
O - outcome
What is a diagnostic test
any kind of medical test performed to aid the diagnosis or detection of disease
Define:
- Sensitivity
2. Specificity
- the ability of a test to correctly identify all those with disease
- the ability of a test to correctly exclude all those without disease
How do you calculate:
- Sensitivity
- Specificity
- no of true positives ÷ all those with disease
2. number of true negatives ÷ all those without disease
Define:
- Positive Predictive Value
2. Negative Predictive Value
- the chance of having disease with a positive test result
2. the chance of not having disease with a negative test result
How do you calculate:
- Positive predictive value
- negative predictive value
- number of true positive ÷ number who test positive
2. number of true negatives ÷ number who test negative
- Which aspects of test performance are affected by prevalence?
- How do they change as prevalence increases
- predictive values
2. PPV rises and NPV falls
What are likelihood ratios?
how the belief about the chance of disease has changed as a result of the test
What is screening?
the systematic application of a test, to identify individuals at sufficient risk of a specific disorder to warrant further investigation or direct preventative action, amongst people who have not sought medical attention on account of symptoms
What type of prevention is screening thought as?
secondary prevention
What are the criteria for the condition being screened for? (3)
- must be an important health problem
- epidemiology and natural history must be adequately understood
- Cost effective primary prevention must have been implemented
What are the criteria of the screening test (4)
- should be a simple, safe, precise and validated screening test
- a suitable cut off for the test value should be agreed
- test should be acceptable
- There should be an agreed policy on further management
What are the criteria for the TREATMENT of the condition being identified by screening? (3)
- there should be an effective treatment, with evidence of early treatment leading to better outcomes
- there should be agreed policies covering who should be offered treatment
- clinical management of the condition should be optimised prior to the implementation of a screening programme
What are the criteria for the screening programme? (5)
- there must be RCT evidence that the programme is effective in reducing morbidity and mortality
- there should be evidence that the whole programme is acceptable to professionals and the public
- benefit should outweigh the harm
- opportunity cost should be economically balanced
- there should be a plan for quality assurance
- Why is selection bias a particular problem with assessing screening programmes
- What is sojourn time?
- What is length bias?
- What is lead time bias?
- the people who will opt in to screening programmes are often very health aware and therefore likely to have good outcomes
- the length of time of disease before it causes morbidity/mortality
- screening is disproportionately likely to detect linger progressing disease, which tend to have better outcomes (making the screening programme look better)
- screening makes disease life longer because it is found earlier.
Which study designs are useful for investigating the following areas:
- diagnosis
- aetiology
- prognosis
- treatment
- patient views
- cross sectional study
- cohort study
case control - cohort study
- RCT
- qualitative research
Name 6 benefits of systematic reviews
- contain all the available evidence to answer a question
- includes unpublished research or that published in non-english language journals
- increase the total sample size, and so increase levels of certainty and precision
- indicate heterogenicity (variation) among studies
- permit subgroup analyses
- permit sensitivity analyses
Define Bias
the systematic introduction of error into a study that can distort the results in a non-random way
Name 4 things that appraisal assesses a study for
- bias
- applicability
- limits
- value
What are the three discrete steps of critical appraisal?
- are the results valid?
- what are the results?
- Can I apply the results to this patient’s care?
Describe confounding
confounding happens when a relationship between an exposure and an outcome is distorted by their shared relationship with something else (the confounding variable)
- What is a cohort study?
2. What is a case control study?
- start with exposure and compare outcomes (prospective)
2. start with outcomes and compare exposures (retrospective)
Name four ways in which confounding can be addressed in design and analysis
- restriction
- matching
- stratification
- multiple variable regression
- What is restriction?
2. Name 2 disadvantages of using restriction
- limiting yourself to people who do not have the confounding variable
- less data
difficult if there are multiple confounders
- What is matching
- what type of study is matching most commonly used in?
- Is it a good way of accounting for confounding?
- Create a comparison group based on the possible confounder
- case control
- not really; still need to consider confounding in analysis
- What is stratification?
2. what is a problem with stratification?
- subdivide the population into different levels of exposure, and analyse the exposure:outcome association for different subgroups of the confounder
- numerous strata to take into account multiple confounding factors
Name two ways in which death rate can be adjusted for confounding variables
- standard mortality ratio - how many deaths compared to standard for each strata
- direct standardisation - weighted average of stratum specific rates
Name 4 reasons why we need research informed practice
- personal experience is biased
- research reports findings for more patients than we can see in personal experience
- research involves the application of scientific method to minimise bias
- recommendations have been assessed for their clinical and cost effectiveness
Name 5 gaps of getting evidence into practice
- identifying the need for knowledge
- discovery of that knowledge
- synthesis of knowledge
- application of knowledge
- development of routine clinical applications or policy
BARRIERS TO THE UPTAKE OF EVIDENCE
- Characteristics of the recommendation (3)
- characteristics of the adopters (4)
- Characteristics of the organisation and environment (3)
- are the recommendations easy to follow and compatible with exisiting norms and values?
us there a need for new skills
are the recommendations complex? - lack of awareness of the recommendation
doubts over credibility of source
over-reliance on convenient sources of information
perciebed resistance of patients - time and resources specified in recommendations
organisational culture
social influence - team norms; influential peers
- What is quality improvement?
- Name 6 things that quality assurance involves
- Describe the quality assurance cycle
- the uptake and continuing use of evidence based policy and practice, focussing on recurrent problems, with systems of care to improve performance, professional development and patient outcomes
- interactive and iterative
engage participants across organisational levels
foster an environment where improvement and innovation are norma
empower staff to strive for change
provide knowledge and methods to implement change
remove barriers to change - plan > do > study > act
- What type of Quality improvement is most effective?
- What is the CQUIN framework?
- What have been the focus of CQUIN in 2017-2019? (5)
- patient mediated
multifaceted approach to remove barriers to change
2.payment framework that links proportion of healthcare provider’s income to the achievement of local quality improvement goals
- improving staff health and wellbeing
reducing impact of serious infections
improving transitions out of children/young people’s mental health services
supporting proactive and safe discharge
preventing ill health by risky behaviours
What is decision analysis?
- Expected
- utility
- theory
- relates to future events
- based on values assigned to these future events
- aims to explain what will happen
- How is decision analysis carried out?
2. Name the advantages of decision analysis
- divides decision task into components
- uses decision trees to structure the task
- evidence is the used in the form of probabilities so risks associated with each option can be examined
- examines utility/cost associated with each option
- divides decision task into components
- assists in understanding of decision task
- incorporates values of decision components relating to health economics or patient values
- suggests the most appropriate decision for that particular situation
- assists in understanding of decision task
Describe the stages is decision analysis (4)
- structure problem in decision tree
- assess the probability of every choice branch
- assess numerically the utility of every outcome state
- identify the outcome that maximises expected utility
- conduct a sensitivity analysis to explore effect of varying judgement
On a decision tree:
- what does a square node indicate?
- What does a circle node indicate?
- What must values of probability for each branch equal?
- Give two examples of utility measures
- How is expected value calculated for each branch?
- How is the expected utility for each decision calculated?
- decision node (indicates choice between actions)
- chance node (indicates potential outcome of each decision; requires probability)
- 1.0 or 100%
- QALY; ED-50
- working right to left, utility is multiplied by probability
- Working right to left, expected values for each branch are added together
- Name 3 benefits of decision analyses
- Name 2 limitations of using probability estimates
- Name 3 limitations of utility measures
- makes all assumptions in a decision explicit
allows examination of the process making a decision
integrates research evidence - required data sets to estimate probability may not exist
subject to bias - individuals may be asked to rate a state of health they have not experienced
different techniques result in different numbers
subject to presentation framing effects
What do the following mean?
- positive risk difference?
- risk ratio <1
- negative risk difference
- Odds ratio >1
- risk ratio >1
- Odds ratio <1
- higher risk
- lower risk
- lower risk
- higher risk
- higher risk
- lower risk
- What are guidelines?
- What is compared in a meta analysis?
- What is plotted on a forest plot? (2)
- What does a forest plot indicate?
- statements that include recommendations intended to optimise patient care, that are informed by systematic reviews and assessment of benefit/harm
- risk ratios between intervention and comparitor
- risk ratios from all studies, and confidence intervals
- the side of the line that the plot is found indicates the condition in which is favoured
