Table 3.
Results of linear mixed-effects models predicting total WEMWBS and global PSQI using an indicative subset of PROMs.
| Y | Variables | Model coef. | 95% CI | t-statistic | p-value |
|---|---|---|---|---|---|
| Total WEMWBS = f(PROMs) | (intercept) | 50.649 | [29.104, 72.195] | 4.612 | < 0.001 |
| Gender male | − 0.555 | [− 2.286, 1.175] | − 0.63 | 0.529 | |
| Age | 0.008 | [− 0.369, 0.386] | 0.044 | 0.965 | |
| Day into study | − 0.004 | [− 0.006, − 0.002] | − 3.36 | < 0.001 | |
| Work sector: manufacturing | 1.202 | [− 2.372, 4.776] | 0.66 | 0.509 | |
| Work sector: social care | − 1.06 | [− 4.729, 2.609] | − 0.567 | 0.571 | |
| Work sector: finance | 0.04 | [− 2.407, 2.488] | 0.032 | 0.974 | |
| BJSQ 1: much work | 0.237 | [− 0.179, 0.652] | 1.117 | 0.264 | |
| BJSQ 14: friendly atmosphere | − 0.459 | [− 0.769, − 0.149] | − 2.908 | 0.004 | |
| BJSQ 17: worth doing | − 0.408 | [− 0.807, − 0.010] | − 2.011 | 0.045 | |
| BJSQ 23: irritable | − 1.591 | [− 2.760, − 0.422] | − 2.669 | 0.008 | |
| BJSQ 24: tired | 0.897 | [− 0.229, 2.023] | 1.563 | 0.118 | |
| BJSQ 27: tense | − 1.468 | [− 1.860, − 1.076] | − 7.352 | < 0.001 | |
| BJSQ 46: can’t sleep | − 0.988 | [− 1.344, − 0.632] | − 5.443 | < 0.001 | |
| WWQBI 1: stressed | 0.612 | [0.279, 0.945] | 3.605 | < 0.001 | |
| WWQBI 10: work-life balance | 1.253 | [0.759, 1.746] | 4.975 | < 0.001 | |
| Global PSQI = f(PROMs) | (intercept) | 9.839 | [− 1.212, 20.890] | 1.747 | 0.081 |
| Gender Male | − 0.339 | [− 1.202, 0.525] | − 0.770 | 0.442 | |
| Age | − 0.052 | [− 0.245, 0.141] | − 0.532 | 0.595 | |
| Day into study | 0.001 | [− 0.000, 0.001] | 1.125 | 0.261 | |
| Work sector: manufacturing | 1.938 | [0.061, 3.815] | 2.026 | 0.043 | |
| Work sector: social care | − 0.547 | [− 2.378, 1.284] | − 0.586 | 0.558 | |
| Work sector: finance | − 1.044 | [− 2.279, 0.191] | − 1.658 | 0.097 | |
| BJSQ 1: much work | 0.296 | [0.128, 0.465] | 3.446 | < 0.001 | |
| BJSQ 14: friendly atmosphere | 0.070 | [− 0.055, 0.195] | 1.092 | 0.275 | |
| BJSQ 17: worth doing | 0.183 | [0.017, 0.350] | 2.158 | 0.031 | |
| BJSQ 23: irritable | − 0.455 | [− 0.930, 0.020] | − 1.878 | 0.061 | |
| BJSQ 24: tired | 0.651 | [0.193, 1.109] | 2.789 | 0.005 | |
| BJSQ 27: tense | 0.180 | [0.021, 0.340] | 2.218 | 0.027 | |
| BJSQ 46: can’t sleep | 0.525 | [0.380, 0.670] | 7.116 | < 0.001 | |
| WWQBI 1: stressed | − 0.198 | [− 0.333, − 0.063] | − 2.882 | 0.004 | |
| WWQBI 10: work-life balance | − 0.383 | [− 0.584, − 0.181] | − 3.727 | < 0.001 |
CI stands for Confidence Interval. See the subsection ‘Linear mixed-effects models’ in the “Methods” section for the generic mixed-effects model formula and further details. The actual variables used in the models are presented in the second column of the Table. We restricted explorations to the work-related questionnaire items to illustrate work-health relationships with total WEMWBS and global PSQI. Bold entries indicate statistical significance at the p = 0.05 level; for convenience we followed the convention that low computed p-values were marked as < .001. See the “Questionnaires_in_SHAW.xlsx” supplementary file for details of each item and its interpretation (e.g. if higher is better). The total WEMWBS lies in the range 14–70, higher being better; the global PSQI lies in the range 0–21, lower being better.