# Statistics & Misc Flashcards

In a study of 100 people with hypertension screened for hyperlipidaemia, 40 were found to have hyperlipidaemia. Of those with hyperlipidaemia, 10 subsequently had a myocardial infarction and of those who did not, 3 had a myocardial infarction. What can be concluded from this study?

A. The relative risk of MI in those with hyperlipidaemia is 5

B. The attributable risk of hyperlipidaemia in MI is 20%

C. The attributable risk of hyperlipidaemia in MI is 5%

D. Hyperlipidaemia is not a risk factor for MI

E. A relative risk of greater than 1 implies hyperlipidaemia is a cause of MI

Answer: B

Attributable risk of hyperlipidaemia in patient with MI in this group is 23%.

The attributable risk is the difference in incidence of MI between those with and without i.e. 25% - 5% = 20%

The relative risk would be 0.25/0.05 = 5.

Range, standard deviation and variance are common measures of dispersion. Which one of the following is a correct description about measures of dispersion?

A. Variance is the standard deviation squared

B. Range is the most commonly used measure of dispersion of a mean value

C. In a normal distribution, 80% of the values will fall between one standard deviation above and below the mean

D. The range gives the width of the entire distribution and the pattern of distribution

E. In a skewed distribution, the extreme values will affect the median to a larger degree than the mean

Answer: A - Variance is the standard deviation squared

Which one of the following definitions for data interpretation of diseases is used correctly?

A. Prevalence is identical to incidence

B. Cumulative incidence rate is an alternative measure of prevalence

C. Incidence is the number of newly affected individuals in a study population over a given time period

D. Prevalence is the incidence multiplied by the population number

E. Lifetime incidence is the proportion of a population experiencing the disease at some point in their life

Answer: C - Incidence is the number of newly affected individuals in a study population over a given time period

A study has been testing a new drug ‘DONTCLOT’ compared to warfarin for treating DVT. The study shows that there was no statistically significant difference between the 2 treatments. A statistician examined the study and suggested that there was a type II error. What does this mean?

A. DONTCLOT should have been compared to placebo

B. Warfarin is better than DONTCLOT

C. The p value is incorrect

D. The study suggests there is no difference but the trial was too small to detect the real difference existing between the 2 groups

E. The absence of statistical difference is not true

Answer: D - The study suggests there is no difference but the trial was too small to detect the real difference existing between the 2 groups

A randomised control trial was undertaken to compare a new endoscopic procedure with a conventional method. The main finding is that a new procedure reduces the risk of persistent peptic ulcer bleeding from 8% to 4%. What does this study show regarding the new endoscopic technique?

A. Statistical significance B. Successful trial C. Incomplete response D. Effectiveness of new procedure E. Efficacy of new procedure

Answer: E - Efficacy of new procedure

A study investigated whether drug X is better as a single agent or in combination with a new drug Y for hyperglycaemia treatment. After randomisation of recruited subjects with type 2 diabetes, a few patients on both drugs (X + Y) dropped out of the study because of adverse effects; there were also several deaths. How should the data be analysed?

A. Exclude analysis of those who died

B. Exclude those who dropped out from analysis

C. Recruit more subjects to the X+Y cohort

D. Analyse the 2 comparison groups separately

E. Include the outcomes of those who dropped out in the drug X+Y group

Answer: E - Include the outcomes of those who dropped out in the drug X+Y group

Intention-to-treat analysis - will reduce the benefit seen to account for this bias

A new drug for cirrhosis has been evaluated in a randomised control trial. The 5 year mortality rate with the drug is 45% and without the drug is 50%. How many patients with cirrhosis have to be treated to prevent 1 extra death over the 5 year period?

A. 2 B. 5 C. 10 D. 20 E. 50

Answer: D - 20

NNT = 1 / Absolute risk reduction RRR

5% risk reduction -> 1/0.05 = 20

Rofecoxib is a COX-2 inhibitor approvaed by the FDA in May 1999 but then voluntarily withdrawn from market in September 2004 due to concerns about twice the risk of myocardial infarction and stroke with long-term and high dosage use. In which phase was the drug withdrawn?

A. Phase I B. Phase II C. Phase III D. Phase IV E. Phase V

Answer: D - Phase IV

Phase IV is post marketing surveillance and studies

In order to evaluate a new test for hepatitis C, a cohort of 2000 subjects is investigated. This test is positive in 76 of the 2000 subjects. Hepatitis C infection defined by conventional ‘gold standard’ RNA testing was present in 40 of the 2000.

Of the 76 with a positive from the new test, 36 were shown to have hepatitis C by conventional testing and 40 were false positives without the disease. The new test was negative in 1924 of the 2000 subjects; four of the 1924 were false negatives; 1920 did not and were truly negative. Which one of the following statements is correct with regards to the new test?

A. Cohort prevalence is 10% B. Sensitivity is 80% C. Specificity is 98% D. Negative predictive value is 90% E. False negative rate is 2%

Answer: C - Specificity 98%

In a city with a population of 10 000, the results of a new diagnostic test for tuberculosis are given below. Which one of the following is the positive predictive value?

TB present TB absent Positive 1800 4000 Negative 200 4000

A. 50 B. 31 C. 90 D. 18 E. 45

Answer: E - 45

A new laboratory test has a sensitivity of 90% and specificity of 80%. Which one of the following is true concerning this new test?

A. 90% of patients who do not have the disease will have a negative result

B. If the test if positive, the probably of the disease being present is 90%

C. 10% of patients who have the disease will have a positive result

D. If the prevalence of the disease in the population is 1%, in screening this population over 90% will be false positives

E. Increased laboratory error increases the predictive value of the test

Answer: D - If the prevalence of the disease in the population is 1%, in screening this population over 90% will be false positives

EMQ Statistical test

A. Log rank correlation B. Spearman's rank correlation C. Wilcoxon signed-rank test D. Mann-Whitney U test E. Chi-squared test F. Logistics regression analysis G. Paired student's t-test H. Unpaired student's t-test

Pick the most appropriate statistical test

- A study is testing the effects of a new drug A on hypertension. 2 groups of hypertensive patients are recruited. One group is treated with drug A for 3 months while the other is not. The study evaluated the reduction of BP after treatment between the 2 groups.
- A research fellow is assessing the effectiveness of a new therapy for SCC and recruits newly diagnosed SCC patients. She then randomises the participants to either existing treatment or the novel arm and performs the standard operation for melanoma excision on the first group (n = 420). She also performs the newly proposed therapy on the second group (n = 390). The treatment arms are double-blinded; 6 months after surgery they are seen to assess recurrence rates.

Data:

Disease: Yes No

Standard treatment 80 340

Novel treatment 60 330

- The frequency of attendance of 50 BPTs at a weekly education session was recorded by an observer over 1 year. They were assessed at the end of this period with an MCQ and tested with a score out of 100. Which statistical test is best to evaluate the effectiveness of trainee’s attendance on higher examination scores?

- Answer: H- unpaired t-test
- Answer: E - Chi squared test (for categorical variables)
- Answer: B - Spearman’s rank correlation

- Non-parametric measure of statistical dependence between 2 variables from -1 to +1

An Endocrinologist performs a study to assess whether a patient’s HbA1c level is correlated to their LDL level. Assuming both are normally distributed, which one of the following statistical tests is most appropriate?

A. Pearson's coefficient B. Chi-squared test C. Spearman's rank correlation coefficient D. McNemar's test E. Mann-Whitney test

Answer: A Pearson’s coefficient

A new screening test is developed for colorectal cancer. It is a blood test which detects a protein; the higher the protein the more likely they are to have cancer. If the cut-off for a positive test is increased, which one of the following will also be increased?

A. P-value B. Specificity C. Likelihood ratio for a negative test result E. Sensitiivity E. Negative predictive value

Answer: B - Specificity

Increasing the cut-off will decrease the number of false positive and therefore increase true negatives/specificity.