Insert answers under each question
Here, we describe a cohort study used to examine the association between smoking and Parkinson’s disease in older men and women in Iceland.
Overall 5,250 people aged 65 years volunteered to be in the study, all of whom were given a physical exam. There were 250 volunteers with signs of Parkinson’s disease.
The remaining 5,000 were enrolled in to the cohort because they were determined to be free from Parkinson’s disease. Their smoking status was determined at enrollment, and the participants were categorized as 1) never-smokers, 2) past smokers, or 3) current smokers. They were then followed for 10 years.
- Based on these data what is the estimated prevalence of Parkinson’s disease among 65-year olds living in Iceland? Show formula and calculations.
- There are 2,000 current smokers among the 5,000 participants. Their 10-year cumulative incidence (risk) was 10%.
There are 1,000 never-smokers among the 5,000 participants. Their 10-year cumulative incidence (risk) was 20%.
Use this information to draw a well-labelled 2×2 table of exposure (treat current smokers as exposed and never-smokers as unexposed). Show calculations used.
- What is the risk ratio of Parkinson’s disease among the current smokers compared to the never smokers?
- If there was a dose-response relation between smoking status and Parkinson’s disease what might you expect the cumulative incidence to be among the past smokers?
- The risk ratio for smoking and Parkinson’s disease is different in men and women. What is this called in epidemiology?
SECTION B
Insert answers under each question
BMI as the main exposure of interest; the outcome [cases] is incident coronary heart disease (CHD deaths and non-fatal myocardial infarction). The other variables of interest are smoking status and physical activity.
Table 1. Incident cases of CHD and person-years of follow-up by BMI, physical activity and smoking status (from table 4 in the original paper)
- Using the data in the table, create a new table to explore the crude association between BMI and cases [CHD]. Calculate the crude incidence rate ratio (IRR) comparing those with BMI > 25 (exposed) to those with BMI between 18.5 and 24.9 kg/m2 (unexposed [sometimes called reference group]).
Based on the crude analysis above, describe in words the association between BMI and CHD
- Calculate the crude incidence rate difference (IRD) comparing the incidence rates of cases among the study participants with BMI>25 (exposed) to those with BMI between 18.5 and 24.9 kg/m2 (unexposed [sometimes called reference group]).
Based on the crude analysis in (c), describe in words the association between BMI and CHD
- Based only on your knowledge, why might smoking status be a confounder of the association between BMI and CHD? What properties must a potential confounder have? In describing the properties use smoking as the confounder, use BMI as the variable of interest (exposure) and use CHD as the cases.
- Using the data provided in Table 1, create a new table to assess the relationship between smoking status and CHD among individuals with “normal range” BMI, 18.5-24.9 kg/m² (here considered the unexposed). For the purposes of this exercise, combine never and past smokers into one group, comparing them to current smokers. Is smoking positively or inversely associated with CHD among the unexposed?
- Using the data provided in Table 1, create a new table to assess the relationship between smoking status and BMI in the study base (i.e. the person-time that gave rise to the cases). Again, consider never and past smokers as one group, comparing them to current smokers. Calculate this association and indicate whether smoking status is positively or inversely associated with BMI in the study base.
- Should we adjust for smoking when considering the relationship between BMI and CHD?
SECTION C
Insert answers under each question
Use the table in section II to complete the questions below.
- If you were to include three different categories when controlling for smoking status (never, past, and current), and simultaneously stratified on physical activity as a two-category variable (<1 hour per week, ≥1 hour per week), how many strata would you have? Alternatively, if you simultaneously control for smoking as a three-category variable (never, past, current) and physical activity as a three-category variable (<1, 1 – < 3.5, ≥3.5 hours per week), how many strata would you have?
- Complete this table
Incidence rate ratio and incidence rate difference for the association between BMI and CHD for each stratum defined by physical activity (3 levels) and smoking status (3 levels) simultaneously.
The association between BMI (kg/m2) and CHD jointly stratified by physical activity and smoking status
| IRR | IRD (cases/10,000 person years) | |||||
| Physical activity (h/week) | Smoking status | |||||
| <1 | Never | |||||
| Past | ||||||
| Current | ||||||
| 1 – <3.5 | Never | |||||
| Past | ||||||
| Current | ||||||
| ≥3.5 | Never | |||||
| Past | ||||||
| Current |
- Is there evidence of effect measure modification on either the additive (IRD) or multiplicative scales (IRR) by smoking and physical activity jointly? Explain briefly.
