These authors reviewed the activity levels of 46 patients having primary reverse total shoulder arthroplasty. In comparing 17 patients younger than 65 years and 29 patients older than 65 years at least one year after surgery, they reported no "significant differences" for range of motion, strength, or number of activities.
Patients younger than 65 years were more likely to require narcotic pain medication and to have disability.
Comment:
The statement "no significant differences" needs to be made in the context of statistical power in order to avoid what is known as a type II error: failing to detect a difference when their actually is one.
As an example, consider the following hypothetical study showing activity levels for young and old patients. We have three patients in each age category. While the means are different, the p value is not less than .05. Do we conclude that there is no significant difference between the groups?
Another concern here is the apparent inconsistency in the data as presented. Contrast the statement in the text, "For individual activities, significant differences were found between groups for low-demand baking (17 bakers aged > 65 years compared with 9 bakers aged < 65 years, P = .013) and high-demand wheelbarrow use (12 patients aged > 65 years compared with 1 patient aged < 65 years, P = .018)", with that reported in the table:
Comment:
The statement "no significant differences" needs to be made in the context of statistical power in order to avoid what is known as a type II error: failing to detect a difference when their actually is one.
As an example, consider the following hypothetical study showing activity levels for young and old patients. We have three patients in each age category. While the means are different, the p value is not less than .05. Do we conclude that there is no significant difference between the groups?
It turns out that we were able to add three more patients to each group with exactly the same activity levels as those in the original group.
The means stay the same, the standard deviations drop a bit. But all of a sudden the p value becomes less than p<.05 so the difference is 'significant'. From this example, we see that the chance of avoiding a type II error (the statistical power) increases with sample size. A concern with this paper is that the sample sizes may be too small to detect a difference that may actually be present. A calculation of the sample size needed to achieve reasonable statistical power would seem in order.
Another concern here is the apparent inconsistency in the data as presented. Contrast the statement in the text, "For individual activities, significant differences were found between groups for low-demand baking (17 bakers aged > 65 years compared with 9 bakers aged < 65 years, P = .013) and high-demand wheelbarrow use (12 patients aged > 65 years compared with 1 patient aged < 65 years, P = .018)", with that reported in the table:
If we take the data in the table from the paper (above) and calculate the percents of patients in each age group that were performing each of the "high demand" activities, it appears that the 17 younger patients were more active than the 29 older ones (p<.02).
===
Use the "Search" box to the right to find other topics of interest to you.
You may be interested in some of our most visited web pages including:shoulder arthritis, total shoulder, ream and run, reverse total shoulder, CTA arthroplasty, and rotator cuff surgery as well as the 'ream and run essentials'
You may be interested in some of our most visited web pages including:shoulder arthritis, total shoulder, ream and run, reverse total shoulder, CTA arthroplasty, and rotator cuff surgery as well as the 'ream and run essentials'
