Statistics and Evidence (yellow)

Question 1

1. what is evidence?
2. what is decision making?
3. how are medical decisions uncertain? (4)
4. how is evidence used in medical decision making?
5. name 4 examples of evidence sources

Answer 1

1. an observation, fact, or organised body of information, offered to support or justify inferences or beliefs in the demonstration of some preposition or matter at issue
2. a choice of action
3. • different people act differently to the same decision
     - there may be insufficient information to assist with accurate diagnosis or estimation of treatment effects
     - you may not know what the patient wants
     - it may be difficult to apply available knowledge to a specific patient
4. as a way of reducing uncertainty
5. personal experience; experience of others; views of “experts”; information from books or journals

Question 2

1. what sources are made when making a clinical decision (4)
2. why is evidence based decision making important? (6)
3. how is evidence based decision making implemented?

Answer 2

1. patient preferences; available resources; research evidence; clinical expertise
2. • deals with uncertainty
     - medical knowledge is incomplete/changing
     - patients will receive the most appropriate treatment
     - constant need for innovation and improvement
     - improving efficiency of healthcare services
     - reduces practice variation
3. evidence based clinical guidelines
   summaries of evidence provided to practicioners
   access to reviews of research evidence
   practitioners evaluating research for themselves.

Question 3

1. define diagnosis
2. define prognosis
3. name and describe the 3 theories of decision making

Answer 3

1. the process of determining the nature of a disorder by considering the patient’s signs, symptoms, medical background and test results
2. assessment of future course and outcome of a patient’s disease
3. normative - what should you be doing according to social or professional norms
   descriptive - what are you doing?
   prescriptive - how can we improve what we are doing?

Question 4

1. describe the 4 components of the hypothetico-deductive model of diagnostic reasoning
2. who uses this model of diagnostic reasoning?
3. describe the prospect theory of diagnostic reasoning
4. who uses this model?

Answer 4

1. cue acquisition. hypothesis formation. cue interpretation. hypothesis evaluation
2. unexperienced clinicians and experienced clinicians with a less familiar diagnostic problem
3. framing and editing - preliminary analysis of decision problem
   phase of evaluation - framed prospects are evaluated and the prospect with the highest value selected
4. experienced clinicians

Question 5

1. what is broad evidence?
2. what is narrow evidence?
3. what is the heirarchy of evidence?
4. what increases going up the pyramid?
5. why should the heirarchy be used critically (3)
6. where can good evidence be found?
7. what is bias?

Answer 5

1. any factor that can and should influence clinical decision making
2. results of rigorous clinical trials and observational studies
3. lists the type of study design ranked in order of their perceived ability to provide evidence for use in practice.
4. less potential for bias therefore more predictive power
5. study designs are suited to answering different research questions
   there are good and bad studies of any type
   the pyramid does not include qualitative research
6. cochrane database
   evidence based journals
   medline
7. systematic error in measurement

Question 6

1. what is interval data?
2. what is ordinal data?
3. what is nominal data?
4. what is descriptive statistical analysis?
5. what is correlational statistical analysis?
6. what is inferential statistical analysis?
7. which 2 of the above can test for significance?

Answer 6

1. quantitative, discrete data, where only certain values are possible (e.g number of admittances to hospital) or continuous, where any value is possible (e.g. age, weight)
2. qualitative, but ordered where there are more than 2 categories that have logical order (e.g. service satisfaction rated, poor, good, excellent etc)
3. qualitative, multinominal, with more than 2 categories which are not ordered (e.g. single, married, divorced, widowed etc)
4. organising, summarising, and describing data (mean, SD etc)
5. examining relationships between data
6. using stats tests to make generalisations about a population
7. correlational and inferential

Question 7

1. what percentage of data is expected to be found within 1SD of the mean?
2. what percentage of data is expected yo be found within 2SD of the mean?
3. describe a negative skew
4. describe a positive skew
5. name 3 measures of location
6. name 3 measures of dispersion

Answer 7

1. 68%
2. 95%
3. peak found to the right. mode>median>mean
4. peak found to the left. mode

Question 8

1. what does a probability of 1 suggest?
2. what does a probability of 0 suggest?
3. what does a null hypothesis state?
4. what does an alternate hypothesis state?
5. what P value usually indicates statistical significance? What does this mean?

Answer 8

1. certain to happen
2. certain not to happen
3. that there is no difference/any difference observed is due to chance
4. that there is a difference/difference seen is due to manipulation
5. 0.05

Question 9

what is the difference between statistical significance and clinical significance?

Answer 9

statistical significance states that there is a difference
clinical significance questions whether this difference is worth it
- yes it improves life expectancy but how much by

Question 10

1. how do you calculate standard deviation?
2. what is standard error?
3. how do you calculate standard error?

Answer 10

1. √(∑(x-x̄)2/(n-1) where x = each value, x̄ is the mean value of the sample, and n = number of samples
2. while standard deviation is a measure of spread, standard error is the measure of accuracy
   - describes how good a given estimate is
   - tells you how good your sample statistic is
   - looks at how accurate estimate of the mean is
3. standard deviation/√(n)

Question 11

1. describe 3 types of descriptive research

2. describe 2 types of analytic research

Answer 11

1. survey
case report
case series
2. experiment
observation - cohort study. case controlled study

Question 12

1. what is an experiment?
2. what is an observation?
3. what is a cohort study?
4. what is a case control study?
5. what is a historical cohort study?

Answer 12

1. interventions are assigned by the investigator (usually at random); if groups differ in terms of the intervention, then changes observed are a measure of the intervention. Provides powerful evidence about cause and effect.
2. no intervention by investigator. Analysis of spontaneously occurring events; group assignments are not random.
3. type of observational study. Subjects with a certain exposure are followed over time for outcome occurance
4. looks back to understand risk factors that lead to a particular disease (start with outcome; look back at exposures)
5. using information collected in the past and looking at outcomes.

Question 13

1. what is risk?
2. how can risk be quantified?
3. what are odds?
4. how can be odds be quantified?
5. how do we calculate risk?
6. how do we calculate risk difference?
7. how do we calculate risk ratio?
8. how do we calculate odds ratio?
9. what is relative risk?
10. what are relative measures of risk?
11. what is absolute risk?
12. what are absolute measures of risk?

Answer 13

1. the probability that an event will occur during a specified time
2. natural frequency (1 in 85)
   probability (1/85 = 0.012)
   percentage (1.2%)
3. the ratio of probability that something will happen to the probability that it won’t
4. 1/85 / 84/85
5. number of people who get thing/number in group

USING RISK OF CERVICAL CANCER IN THE PRESENCE OR ABSENCE OF ORAL CONTRACEPTIVE PILL AS EXAMPLE:

6. risk with pill - risk without pill
7. risk with pill / risk without pill
8. odds with pill / odds without pill
9. risk that is relative to a reference group (only makes sense if we know what it is relative to)
10. risk and odds ratios
11. risk that can be interpreted without a reference group
12. risk and odds themselves, and risk differences

Question 14

1. what is incidence?
2. What is prevalence?
3. How do we calculate incidence?
4. How do we calculate prevalence?

Answer 14

1. the measure of probability of occurrence of a medical condition within a specified period of time
2. the proportion of disease found to be affecting a particular population
3. number of new cases in a disease period/number of initially free cases
4. number of people with a disease at a particular point/total population

Question 15

1. what is a clinical trial?
2. Describe the following stages of clinical trials:
   a) Phase I
   b) phase II
   c) phase III
   d) phase IV
3. What is a randomised controlled trial?

Answer 15

1. a planned experiment involving patients, designed to determine the most appropriate treatment of future patients with a given medication
   2a) dose finding, safety and side effect. Carried out on healthy volunteers
   2b) initial investigation for efficacy. Carried out with patients
   2c) full scale evaluation - randomised and controlled
   2d) post marketing surveillance
2. an experiment in which consentong patients are randomly allocated into groups

Question 16

1. What is regression to the mean?
2. Name 3 reasons why we randomise studies
3. What is stratification?
4. name 4 ways in which bias can be introduced in RTCs
5. What can be done to reduce the potential for bias?

Answer 16

1. if a variable is extreme on its first measurement, it will tend to be close to the average on its second measurement. Occurs when a group are measured with an inexact measurement tool and then re-measured
2. • eliminates systematic bias in allocation of interventions
     - helps ensure balance across comparative groups for baseline factors that may affect outcome
     - differences in outcomes can be attributed to intervention
     ethical - treatment is not decided by a person.
3. the process that can be built into randomisation to ensure that important factors that affect outcome are balanced across the groups
4. selection bias - systematic error in creating intervention groups
   detection bias - distortion of results of a randomised trial as a result of knowledge of the group assignment by the person assessing the outcome
   performance bias - differences in care provided to participants in comparison groups
   Allocation Bias
5. Blinding

Question 17

1. What is inference?
2. What is a confidence interval?
3. What are the 2 numbers that make up the upper and lower ends of the confidence interval called?
4. How do standard error and confidence intervals differ?
5. What is a confidence level?

Answer 17

1. a conclusion reached upon the the basis of evidence and reasoning from a sample, and relating this back to the population
2. the range of values that are likely to encompass the true value corresponding to the population parameter
3. upper and lower confidence limits
4. standard error indicates how much the observed sample statistic may fluctuate in the same experiment (i.e. focusses on the sample) whilst confidence intervals indicate the range that is likely to contain the true population parameter (i.e. focusses on the population)
5. the probability that the confidence interval encompasses the true value. Expressed as a percentage