Introduction to statistical modelling – theory and practice
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit
Introduction to statistical modelling - theory and practice
Final assessment - individual part
Question 1.
A randomised controlled trial of new antihypertensive treatment found that before treatment, blood pressure was similar in the treatment (n=1249) and control (n=1251) groups. After 6 months of treatment, the mean (SD) systolic blood pressure was 138 (19.9) and 145 (20.5) mm Hg, respectively, and over the next 5 years, the risk of adverse cardiovascular event (stroke, myocardial infarction or sudden cardiac death) was 18% and 23%, respectively.
1. Was the difference in systolic BP between the two groups statistically significant? Conduct appropriate test to assess the effectiveness of the new antihypertensive treatment. Please, state a null hypothesis for such test, give a formula for the test and calculate the test statistics. What does the value of the statistics suggest?
2. Please calculate the relative risk for adverse cardiovascular events.
Question 2.
A study using aggregate data on 32 OECD countries in 2002 found that unemployment rate was negatively associated with happiness measured by “happiness index” (r = -0.12, p= 0.003). The regression coefficient was -0.7 (standard error 0.2) per 1% increase in unemployment. After additional controlling for per capita gross domestic product (GDP) and the mean years of education, it was -0.6 (SE 0.2).
a) Please explain what correlation and regression coefficients tellus
b) What do the results above suggest about the effect of unemployment on happiness?
a. Is the association statistically significant at 5% level of significance? Why?
b. Is it a strong association?
c. Is it confounded by GDP and education? Why?
Question 3.
A longitudinal study was carried out in the USA to investigate the association between breastfeeding and obesity at 5 years of age. Singleton term babies born between January 1998 and December 1999 in a single hospital were recruited with parental consent. Infants were weighed and their length or height measured at birth and 5 years of age. Mothers were interviewed at birth and at 6, 12 and 18 months to determine whether they breastfed their child, and to ascertain the duration of exclusive breastfeeding (no food given other than breast milk). Data on duration of breastfeeding and on weight and height at 5 years of age were available for 9142 children. Results were as follows:
Duration of breastfeeding |
Number of children obese at 5 years |
Never breastfed (n=4022) |
181 |
Ever breastfed (n=5120) |
142 |
|
|
Those who have ever been breastfed (n=5120): Duration of breastfeeding: |
|
< 2 months (n=2084) |
79 |
3-6 months (n=2052) |
47 |
7- 12 months (n=863) |
15 |
>12 months (n=121) |
1 |
a) Calculate risk ratios for obesity according to the duration of breastfeeding (use “never breastfed” as a reference category).
b) Further analysis of the data showed 41% of the children who were never breastfed had had parents with at least 10 years of education. Among children who were ever breastfed, this proportion was 67%. When the data were stratified according to parental education the results were as follows:
Duration of breastfeeding |
Risk ratios |
|
|
Parental education <10 yrs |
Parental education >10 years |
Never breastfed (n=4022)* |
1.00 |
1.00 |
Ever breastfed (n=5120) |
0.83 |
0.80 |
|
|
|
Never breastfed (n=4022)* |
1.00 |
1.00 |
Exclusively breastfed for: |
|
|
<=2 months (n=2084) |
1.01 |
0.98 |
3-6 months (n=2052) |
0.77 |
0.76 |
7- 12 months (n=863) |
0.64 |
0.62 |
>12 months (n=121) |
0.42 |
0.39 |
*reference category
Using this new information discuss whether confounding or interaction were present. Explain your conclusions.
2023-08-01