INTERPRETING GRAPHS AND DIAGRAMS
(NB All the figures and outcomes are made up.)
Question 1 and 2 refer to the diagram below
The figure below is a L’Abbe plot displaying the result of 3 trials of diazepam for social phobia
- What is the NNT for study A
- 1
- 1.25
- 1.5
- 2.5
- Not possible to calculate from the information presented
- What is the significance of the fact that the circle for study C crosses the diagonal line
- The result is statistically significant
- The result is statistically insignificant
- The confidence interval is the largest of the three studies
- Treatment is harmful
- None
- All of the following could be featured on the y axis of a funnel plot EXCEPT
- Standard Error
- 1/se
- Sample size
- Log sample size
- Effect size
- Funnel plots are analysed via
- Bellow’s test
- Egger’s test
- Roth’s test
- Mailer’s test
- Updike’s test
+ An extra bonus smartarse point if you know what links all 5 answersand my deliberate mistake
- The funnel plots above tabulate the same 5 RCTs of the effect of diazepam in the treatment of performance anxiety, with 2N on the y axis on the left and 1/se on the right. Taken in concert their appearances could be plausibly attributed to all of the following EXCEPT
- Selection bias in some studies
- Publication bias
- Clinical Heterogeneity
- Statistical heterogeneity
- Methodological heterogeneity
Questions 6 to 11 refer to the Forest plot below of 5 placebo controlled studies of Moclobemide in social phobia.
- Standardised mean difference is standardised to what?
- Expert consensus
- Expected effect size based on previous work
- Standard deviation of the study
- The International reference unit of shyness that Morrissey keeps in a vault in his bedroom
- The level of shyness in the US population in 1940
- Where is the line of equivalence in the diagram?
- -2
- -1
- 0
- 1
- 2
- Which study has been rated as most reliable
- Regarding study A; taking the point estimate of effect size what proportion of subjects receiving IPT improved more than the mean of the control group?
- None
- Most
- 50%
- 84%
- 95%
- What kind of heterogeneity is being tested for?
- Behavioural
- Clinical
- Methodological
- Sample size
- Statistical
- Statistical heterogeneity is present as indicated by
- P value < 0.05 for chi squared test
- 3 studies show no effect
- Discrepancy of CI for fixed and random effect analyses
- Differing weights
- Smaller studies having wider confidence intervals
Questions 12 to 15 refer to the diagram below
The diagram shows a Kaplan Meier Curve plotting time to first relapse for a sample of 20 patients with schizophrenia following discharge from hospital.
- What are the little hatched lines in the horizontal lines
- Significance levels
- Relapsing patients
- Confidence intervals
- Censored values
- Centiles
- ‘Censored’ values would include
- Patients who left the area and were no loner available for follow up
- Patients who had a relapse
- Patients who refused consent prior to study onset
- Patients who relapsed after the end of the study
- All of the above
- What is the median survival time to relapse?
- 100 days
- 180 days
- 300 days
- 360 days
- Impossible to assess without complete follow up data
- If a patient survived without relapse for 100 days, what would be the probability of their surviving a further 100 days? (as in all the best exams you only get full marks if you show your working)
- 1
- 0.9
- 0.8
- 0.77
- 0.7