ld-archer
commented
1 year ago From Lang et al. (2007):
Measured Physical Performance Impairment (ELSA)
A physical performance score was calculated based on performance on three tests: balance, chair stands, and grip strength. This test is a modified form of the Short Physical Performance Battery to assess lower extremity function16 with gait speed replaced by grip strength.17 (This was necessary here, because gait speed in ELSA was assessed only in respondents aged 60 and older.) The test was scored as follows.
The balance component tested the respondent's ability to complete three tasks. Respondents scored 0 if they could not complete any of the tasks; only respondents who successfully completed a given stance attempted the following one. They scored 1 point for completing the first task, holding a side-by-side stance (standing with feet together) for 10 seconds. The second task was a semi-tandem stand (heel of one foot against side of the big toe of the other) and was also worth 1 point if held for 10 seconds. The third task was a full-tandem stance (feet aligned heel to toe); respondents scored no additional points if they could not hold this position for 3 seconds, 1 point if they held it for 3 to 10 seconds, and two points if they held it for more than 20 seconds.
The chair stand test involved a pretest and a timed test. The chair used was the respondent's own armless, straight-backed chair; beds, cots, folding chairs, garden chairs, and chairs with wheels or that swiveled were not used. For the pretest, respondents were asked to fold their arms across their chest and try to stand up once from a chair. Those unable to do this scored 0 and did not attempt the rest of this test. Those who were able to do this repeated this movement five times in succession as quickly as possible. Scores were calculated according to quartiles of time taken: the fastest quartile scored 4, the slowest scored 1. Those who had completed the pretest but were unable to complete the timed test within 1 minute were given a score of 0.
The grip strength test involved squeezing a handheld grip-strength meter. Grip strength was recorded for the dominant hand, and the mean of two attempts was used. Scoring was according to quartile, with the strongest quartile scoring 4, and the weakest scoring 1. Because of sex differences in grip strength, quartiles were calculated separately according to sex.
A trained clinician conducted all three tests. Scores were out of 12, and a score of 7 or less was taken to represent impaired performance.
Earlier in the paper they use the vigorous activity variable alone from the harmonised output (or actually from non-harmonised but its an equivalent variable). They then also use another variable for measured physical performance, defined above. I could recreate this variable, which could be useful in a number of ways but perhaps not directly for the loneliness and social isolation work. I will write this into it's own issue, and instead go with Shankar et al.'s binary physically active variable for simplicity. My hypothesis is that the two groups will have very different relationships to loneliness and social isolation.
I think we can create a physical activity 'score' based on some studies that use ELSA. Check out references from issue #94, links to different health outcomes. Shankar et al. created a binary score for physically active or not, which incorporated information on physical activity from the three questions (light, moderate, vigorous) as well as information on a respondents job (whether it was sedentary, active, etc.). This is a nice and simple thing to predict in the FEM so would be a good starting point (also very easy to incorporate into other models) but the references from the issue linked could indicate a better way.
Reminder: