Skills Practised:
The dataset used here is the BCG dataset that comes from the HSAUR package. The dataset contains the results from 13 different studies that assess the effectiveness of the Bacillus Calmette–Guérin vaccine in preventing Tuberculosis. I explore the effectiveness of the vaccine by looking at the odds ratio by year and latitude.
There seems to be a slight upward relationship between year and odds of having received the vaccine given they have tuberculosis. We can see that from 1968 onwards, the odds of having received the BCG vaccine given they have tuberculosis is lower than years 1948 to 1968. However, in 1969 and 1980 (because their odds ratio is more than 1), the odds of having received the BCG vaccine given they have TB is higher than the odds of having received the BCG vaccine given they do not have TB.
If we look at the general trend, we can see that the higher the latitude the study was conducted in, the lower the odds ratio of having received the BCG vaccine given they have TB. This means that studies conducted at higher latitudes generally have lower odds of having received the BCG vaccine given you have TB. However, it is worth noting that the highest odds ratio of having received the vaccine given you have TB is in the middle latitude value (latitude = 33). Generally, the odds of having received the BCG vaccine given they have TB is [odds ratio] times lower than having received the vaccine given they do not have TB.
Given the above analysis, the BCG vaccine is effective as odds of having received the vaccine given they have tuberculosis is generally lower than odds of having received the vaccine given they do not have tuberculosis. This means that more people who had the vaccine did not get TB, as compared to those who did get the vaccine but still got TB.
The relative risk and odd ratio of each study seem to have a linear relationship. The higher the relative risk, the greater the odd ratio.
The dataset used here is the birthwt dataset from the MASS package. This dataset contains information about the mothers of children, and the birth-weight of those children. I explore the effect of different factors on the birth weight of children:
| variable | risk_ratio |
|---|---|
| smoke | 1.6076421 |
| hypertension | 1.9855769 |
| No_prenatalvisit | 0.7178527 |
| before20 | 0.9224599 |
Conclusion:
Mothers who smoke are 1.6076421 times more likely to have children with low birthweight than those who did not. Mothers who had hypertension are 1.9855769 times more likely to have children with low birthweight than those who did not. Mothers who did not attend any prenatal care visits are 0.7178527 times more likely to have children with low birthweight than those who did not. Mothers who gave birth before age 20 are 0.9224599 times less likely to have children with low birthweight than those who gave birth after 20.
The dataset used here is the bdendo dataset from the Epi package. In this section, I’ll use the epitools package to compute the odds ratio from study data.
Hypertension status:
The odds of having endometrial cancer for people who have hypertension is 0.7595308 times lower than those who do not have hypertension. The odds of having endometrial cancer for people who have gall bladder diseases is 0.2889246 times lower than those who do not have gall bladder diseases. The odds of having endometrial cancer for people are obsese is 0.4657895 times lower than those who are not obese.