Tuesday, 28 January 2014

Tutorial 27 January 2014

Tonight we had a tutorial by Julie Morris on medical statistics.
The questions she discussed were:

Statistics Questions

1.      Weight gain in pregnancy was measured in a sample of 1000 women. Summary statistics of the weight gain were calculated.  (Answer yes/no to each statement)-

A     The median is the weight gain value which occurs most frequently
B     If weight gain was found to have a positively skewed distribution, the mean will be less than the median.
C     The interquartile range is unaffected by outliers
D     A large positive outlier will increase the mean
E     The median is affected by outliers

2.      For a cohort of babies, birthweight was considered to be Normally distributed with a mean of 3000g and a standard deviation of 500g. It follows that: (Answer yes/no to each statement)-

A     All birthweights must be less than 4500g
B     The distribution of birthweights is skewed
C     About 50% of birthweights are above 3000g
D     About 2.5% of birthweights are below 2000g
E     About 95% of birthweights are between 2500g and 3500g

3.      Theme: Summary statistics
A     mean                                               E          interquartile range
B     standard deviation                           F          percentage
C     range                                               G          mode
D     median                                            H          standard error

For each type of data description given below choose the SINGLE most appropriate summary statistic. Each option may be used once, more than once, or not at all.

1.     The variability of length of hospital stay for 230 mothers following a caesarean, where length of stay is Normally distributed.
2.     The extremes of length of hospital stay for 230 mothers following a caesarean, where length of stay is Normally distributed
3.     The elective caesarean section rate in a hospital for 2011

4.      To assess the effectiveness of a new antenatal screening test, 100 pregnant women over 45 years of age, were screened.  25 women subsequently had babies with mild congenital problems. 20 of these 25 women had a positive screening test result. 65 of those with normal babies had a negative test result. (Answer yes/no to each statement).

A     The predictive value of a positive test result  as a marker for congenital problems is 20/35
B     The sensitivity of the test will not depend upon the prevalence of congenital problems in the cohort of women to which it is applied
C     The sensitivity of the screening test is 20/100
D     The specificity of the screening test is 65/75
E     Specificity=1-sensitivity

5.      In a study looking at the relationship between blood pressure during pregnancy and fetal growth, 130 women had their 24-hour blood pressure readings taken in mid-pregnancy. The size of the baby was recorded at birth. The Pearson correlation between blood pressure and birthweight was found to equal -0.70 with an associated p-value of  p<0.001.

A     Blood pressure  explains 70% of the variability in birthweight
B     We can conclude that increasing blood pressure is a cause of low birthweight
C     A correlation of zero is interpreted as showing no association between two variables
D     The correlation between blood pressure and birthweight can be different from the correlation between birthweight and blood pressure
E     There is a significant negative linear  relationship between blood pressure and birthweight

6.      In a study of 75 infants with median gestational age 30 weeks (range 25 to 42 weeks), lung area was calculated from chest radiographs and lung volume from measurements of FRC. The aim of the study was to see whether lung area would predict lung volume in neonates.  (choose just one).

A     Chi-square test
B     Correlation coefficient
C     Mann-Whitney U-test
D     Regression analysis
E     Two-sample t-test

7.      A cohort of women experiencing nausea and vomiting in pregnancy were given acid-reducing medication. The Pregnancy-Unique Quantification of Emesis and nausea (PUQUE) score was recorded before and after treatment. The clinical effect of the medication could be evaluated using: (Answer yes or no to each statement):

A     The Student paired t-test
B     The Spearman correlation coefficient
C     The Wilcoxon matched-pairs test
D     The Sign test
E     The Mann-Whitney U-test

8.      Theme: Statistical significance tests
A     Chi-square test                                F          Spearman correlation
B     Two-sample t-test                            G          Mann-Whitney U-test
C     Pearson correlation                         H          Wilcoxon test
D     Fisher’s Exact test                           I           Paired t-test
E     Linear regression                            J          ANOVA

For each of the studies described below choose the SINGLE most appropriate statistical test. Each option may be used once, more than once, or not at all.

1.      A comparison of two audits (2011 vs 2006) of smoking habits of pregnant women looking at the percentage of current smokers.
2.     In a cohort of 500 pregnant women, maternal cotinine levels (an indicator of exposure to cigarette smoke) were measured. The aim of the study was to see whether high maternal cotinine levels were related to subsequent birthweight. It was assumed that cotinine levels would have a highly skewed distribution.
3.     Adrenaline concentrations were measured in 30 mechanically-ventilated babies at 1 hour and 24 hours to assess whether there had been a significant change. The adrenaline concentrations cannot be assumed to follow a Normal distribution.

9.      In a study of diet and epithelial ovarian cancer, 128 women with cancer and 251 women without cancer were asked about their consumption of dairy products. 50% of cancer patients and 10% of controls reported consuming yoghurt regularly: (Answer yes or no to each statement):

A     This is a cohort study
B     The odds ratio is 9
C     The odds ratio is 5
D     The relative risk can be estimated by an odds ratio if the disease is rare and the cases and controls are random samples from the same population.
E     An odds ratio of 0 indicates no difference between cancer patients and controls in the consumption of yoghurt.

10.     In a study of the effect of a new antibiotic therapy for bacterial vaginosis in pregnancy, the infection response rate for women on the new antibiotics was 0.4, whereas it was 0.2  in the control group (p=0.03). (Answer yes or no to each statement):

A     The infection response rate was significantly lower in the control group.

B     The NNT is the difference in the number of infection-free women in the treated group compared to the control group
C     The NNT=5
D     The NNT=2
E     The NNT is the number of women needed to receive the intervention to see one additional occurrence of the outcome in the treatment group

We then discussed two essays.

1. With regard to female genital mutilation (FGM).

1.   what are the key aspects of the law in the UK relating to FGM.                        2 marks.
2.   what are the responsibilities of the doctor who suspects that a child may be subjected to FGM?
2 marks.
3.   how is FGM graded?                                                                                       3 marks.
4.  outline the management of woman found at booking to have had FGM.   13 marks.

A 30-year-old woman para 1+1 presents at 34 weeks gestation with right loin pain, frequency of micturition, dysuria and pyrexia.
1. Justify the history you will take.                                                                6 marks.

2. Justify the investigations you will arrange.                                                  6 marks.

3. Justify your management, assuming the diagnosis to be pyelonephritis.     12 marks.