Table 4.
Ordered logistic regression model demonstrating the univariable and multivariable association of demographic and occupational factors with increasing financial concerns at follow-up
| Variable | Unadjusted odds ratio (95% CI) | P-value | Adjusted odds ratio (95% CI) | P-value |
|---|---|---|---|---|
| Age | 0.73 (0.69–0.77) | <0.001 | 0.77 (0.72–0.81) | <0.001 |
| Sex assigned at birth | ||||
| Male | Reference | – | Reference | − |
| Female | 1.17 (1.01–1.35) | 0.03 | 0.98 (0.84–1.14) | 0.77 |
| Ethnicity | ||||
| White | Reference | – | Reference | – |
| Asian | 1.02 (0.86–1.22) | 0.80 | 1.08 (0.90–1.31) | 0.41 |
| Black | 0.97 (0.65–1.44) | 0.88 | 0.79 (0.52–1.20) | 0.27 |
| Mixed | 0.81 (0.58–1.13) | 0.21 | 0.81 (0.57–1.15) | 0.24 |
| Other | 0.93 (0.52–1.66) | 0.80 | 0.96 (0.54–1.70) | 0.89 |
| Occupation | ||||
| Medical | Reference | – | Reference | – |
| Nursing | 2.47 (2.06–2.98) | <0.001 | 2.28 (1.84–2.82) | <0.001 |
| Allied health professionals | 1.95 (1.66–2.29) | <0.001 | 1.64 (1.37–1.97) | <0.001 |
| Dental | 2.50 (1.86–3.37) | <0.001 | 1.48 (1.09–2.02) | 0.01 |
| Admin/estates/other | 2.02 (1.49–2.73) | <0.001 | 1.41 (1.02–1.94) | 0.04 |
| Index of Multiple Deprivation quintile | ||||
| 1 (most deprived) | 1.30 (1.01–1.68) | 0.04 | 1.08 (0.82–1.42) | 0.01 |
| 2 | 1.12 (0.90–1.38) | 0.31 | 1.02 (0.82–1.27) | 0.34 |
| 3 | Reference | – | Reference | – |
| 4 | 0.71 (0.58–0.86) | <0.001 | 0.81 (0.67–0.99) | 0.04 |
| 5 (least deprived) | 0.66 (0.55–0.80) | <0.001 | 0.81 (0.67–0.98) | 0.03 |
| Degree of concern about future financial situation at baseline | ||||
| Not at all | Reference | – | Reference | – |
| A little | 6.18 (5.28–7.23) | <0.001 | 5.62 (4.80–6.59) | <0.001 |
| Moderately | 16.65 (13.03–21.26) | <0.001 | 15.50 (12.10–19.85) | <0.001 |
| Quite a bit | 53.09 (37.65–74.86) | <0.001 | 44.71 (31.51–63.44) | <0.001 |
| Extremely | 172.31 (103.5–286.78) | <0.001 | 166.09 (99.02–278.58) | <0.001 |
The table shows univariable and multivariable ordered logistic regression with an outcome of financial concerns at follow-up (five-level ordinal variable). Parallel odds assumption was checked with the Brant test. This was significant at P < 0.001. We plotted odds ratios derived from logistic regression for each level of the outcome variable and determined that the parallel odds assumption was likely violated by inclusion of baseline score. We tested this hypothesis by removing baseline score from the model and performing the Brant test again, which returned a non-significant P-value (see Supplementary Appendix 1). Given the importance of adjusting for baseline score, but the lack of importance placed upon the relationship of this variable with the outcome, we elected to continue with the planned analysis of using ordered logistic regression and adjusting for baseline score (accepting that the odds ratios for baseline score may not be accurate for all levels of the outcome variable). Odds ratio for age is per decade increase. The nursing category includes healthcare assistants, nursing associates and midwives. The allied health professionals category includes healthcare scientists, ambulance workers, pharmacists and those in optical roles.