Professor Merdemy studied risk factors for shoulder injury. He conducted a study to assess whether swimmers were more likely to sustain a shoulder injury compared to weightlifters.
The study had 1,800 participants – 700 swimmers and 1,100 weightlifters. Both groups were followed for 12 months and there was no loss to follow-up. During follow-up, any participant with a new episode of shoulder pain had an ultrasound of the painful shoulder and a radiologist confirmed whether or not there was a new injury to the shoulder. Of the 1,100 weightlifters, 58 had a new shoulder injury during the study. Of the 700 swimmers, 39 developed a new shoulder injury during the study.
Doing push-ups was suspected of increasing the risk of shoulder injury. Dr Merdemy was concerned that this might interfere with the study results because he knew doing push-ups was a common activity for swimmers and weightlifters. Of the swimmers, 200 reported regularly doing push-ups and 15 of these swimmers developed a new shoulder injury during the study. Of the weightlifters, 800 reported regularly doing push-ups and 47 of these weightlifters developed a new shoulder injury during the study.
1) Calculate the crude relative risk of developing a new shoulder injury in swimmers compared to
weightlifters. To do this you will need to draw a 2*2 table.
(i) Show this table. (5 marks)
(ii) Show your working. (5 marks).
2) Interpret the result you obtained (in Question 1) in words that someone who has not studied epidemiology would understand. (5 marks)
3) Do you think regularly doing push-ups has confounded the estimate of whether swimmers are more likely to develop a new shoulder injury than weightlifters? Justify your answer, both in terms of the relative risk estimates and the criteria for confounding. Show any relevant working. Assume for the purposes of this question that doing push-ups is NOT on the causal pathway between the type of athlete (swimmer or weightlifter) and developing a new shoulder injury. (20 marks)
4) Is there evidence of effect modification? Explain your answer in one or two sentences and show any relevant working. (4 marks)
5) What proportion of swimmers who do push-ups develop a new shoulder injury because they are swimmers rather than weightlifters? (2 marks)
6) How many people who do push-ups would have to change from being weightlifters to swimmers for ten extra people to develop new shoulder injuries (4 marks)?
Professor Merdemy’s colleague, Dr Samoht, conducted another study of risk factors for shoulder injury, comparing swimmers to weight-lifters. He was aware that the main known risk factors for shoulder injury were age (risk increases with age), sex (risk higher in males compared to females), occupation (risk higher in people with a manual occupation compared to people with a professional occupation) and regularity of doing push-ups (risk higher in people who regularly did push-ups compared to people who didn’t). Therefore, in this study, Dr Samoht chose his subjects carefully. He made sure that at the beginning of follow-up in the study, the swimming group and the weight-lifting group had the same proportions of females, the same age distribution, the same proportions of manual workers and the same proportion of people who regularly did push-ups.
People were recruited to the study through advertisements on popular social media applications. From the swimmers who answered the advertisements, six hundred were randomly selected to be considered for the study. Similarly, from the weightlifters who answered the advertisements, six hundred were randomly selected to be considered for the study. The actual participants were chosen from these 1,200, attempting to ensure equal distribution of the risk factors described above. At the beginning of the study, there were 500 people in the Swimming group and 400 people in the Weightlifting group. The characteristics at baseline of these 900 participating subjects are shown in Table 1.
Subjects were followed for one year. At the end of the follow-up, the outcome was able to be assessed in 450 of the Swimming group and 240 of the Weightlifting group. The characteristics of the 690 subjects who completed the study are shown in Table 2.
Of the 50 people lost from the Swimming group, 30 didn’t respond to repeated contacts from Dr Somaht, 10 had changed job and were no longer available and 10 moved away from the city. Of the 160 people lost from the Weightlifting group, 44 didn’t respond to repeated contacts from Dr Somaht, 96 had changed job and were no longer available and 20 moved away from the city.
Table 1. Characteristics at baseline for the commencing participants by study group
Swimming (n= 500) Weightlifting (n=400)
Age (mean, standard deviation) 35.2 (8.3) 35.0 (7.1)
Gender (% female) 42 42
Manual occupation (%) 40 39
Regularly did push-ups (%) 30 30
Table 2. Characteristics at end of the study for remaining participants by study group
Swimming (n= 450) Weightlifting (n=240)
Age (mean, standard deviation) 40.2 (7.5) 31.1 (8.7)
Gender (% female) 45 46
Manual occupation (%) 42 32
Regularly did push-ups (%) 29 21
1) Draw a flow diagram that summarises what happened in the study. (15 marks)
2) SELECTION ISSUES (40 MARKS)
(i) Do you think the way the subjects were selected for the study and into their study groups is likely to have introduced important selection bias? If so, how might that have affected the study results?
(ii) Does Table 1 suggest important selection bias at the beginning of this study was likely? If so, how might that have affected the study results?
(iii) Do you think it is likely drop-outs introduced important selection bias? If so, how might that have affected the study results?
(iv) Does Table 2 suggest important selection bias was likely by the end of the study? If so, how might that have affected the study
