Summary of accelerated failure time* regression (‘log-logistic’) for job type: paramedic
| Variable | Estimate | 95% CI | P value† | Direction interpretation‡ | |
| Age | −0.0047 | (−0.013, 0.0039) | – | – | |
| Gender | Male (reference) | ||||
| Female | 0.061 | (−0.038, 0.16) | – | – | |
| Nationality | British (reference) | ||||
| Not declared | −0.21 | (−0.54, 0.13) | – | – | |
| Other | 0.018 | (−0.084, 0.12) | – | – | |
| Marital status | Single (reference) | ||||
| Divorced/legally separated/widowed | 0.14 | (−0.24, 0.52) | – | – | |
| Married/civil partnership | 0.031 | (−0.12, 0.18) | – | – | |
| Not declared | −0.16 | (−0.38, 0.05) | – | – | |
| Pay scale | Band 5 (reference) | ||||
| Band 6+ | −0.23 | (−0.44, −0.019) | <0.05 | … that time to leaving role decreases as pay increases | |
| Staff absence duration (ratio of month) | 0.04 | (0.03, 0.06) | <0.005 | … that time to leaving role increases as time lost to short-term absences increases | |
| Overtime (hours) | Payment in lieu of breaks | 0.28 | (0.23, 0.34) | <0.005 | … that time to leaving role increases for employees who work through breaks |
| Planned | 0.25 | (0.18, 0.32) | <0.005 | … that time to leaving role increases for employees who work planned overtime | |
| Unplanned | 0.14 | (−0.015, 0.3) | <0.1 | … that time to leaving role increases where employees work unplanned overtime | |
| Incidents (per shift worked) | 0.18 | (−0.13, 0.5) | – | – | |
| Percentage of incidents attended in a month by IMD of incident location§ | IMD: 1 (%) | 0.0049 | (−0.013, 0.023) | – | – |
| IMD: 2 (%) | 0.0042 | (−0.0066, 0.015) | – | – | |
| IMD: 3 (%) (excluded)¶ | |||||
| IMD: 4 (%) | 0.0083 | (−0.0031, 0.02) | – | – | |
| IMD: 5 (%) | 0.0006 | (−0.01, 0.012) | – | – | |
| IMD: 6 (%) | −0.0037 | (−0.015, 0.0072) | – | – | |
| IMD: 7 (%) | −0.0084 | (−0.021, 0.004) | – | – | |
| IMD: 8 (%) | −0.028 | (−0.038, −0.017) | < 0.005 | … that time to leaving role decreases for employees who respond to more incidents at IMD:8 locales | |
| IMD: 9 (%) | 0.014 | (−0.00081, 0.03) | <0.1 | … that time to leaving role increases for employees who respond to more incidents at IMD:9 locales | |
| IMD: 10 (%) | 0.034 | (0.0079, 0.06) | <0.05 | … that time to leaving role increases for employees who respond to more incidents at IMD:10 locales | |
| Job cycle time (hours per shift worked) | ‘Mobilisation’ | 1.4 | (−0.74, 3.6) | – | – |
| ‘Running’ | 0.29 | (−0.4, 0.98) | – | – | |
| ‘On scene’ | 0.18 | (0.052, 0.31) | < 0.01 | … that time to leaving role increases for employees who spend more time at the scene of incidents | |
| ‘To hospital‘ | −0.68 | (−1.10, −0.22) | <0.005 | … that time to leaving role decreases for employees who spend more time conveying patients | |
| ‘Arrived at hospital to patient handover’ | 0.17 | (−0.017, 0.36) | <0.1 | …that time to leaving role increases for employees who spend more time waiting at hospitals | |
| ‘Patient handover to clear’ | 0.2 | (−0.18, 0.58) | – | – | |
| Non-JCT | 0.097 | (0.057, 0.14) | <0.005 | … that time to leaving role increases for employees who spend more time outside JCT tasks | |
| Incident category (per shift worked) | Calls from people with life-threatening illnesses or injuries (Category 1) | −0.33 | (−0.64, −0.018) | <0.05 | … that time to leaving role decreases for employees who respond to more ‘Category 1’ incidents |
| Emergency calls (Category 2) | −0.32 | (−0.63, −0.00) | <0.05 | … that time to leaving role decreases for employees who respond to more ‘Category 2’ incidents | |
| Urgent calls (Category 3) | −0.068 | (−0.42, 0.28) | – | – | |
*The accelerated failure time model is operating as a survival model in the implementation reported and hence a significant positive coefficient is indicative of an increased average survival time as the covariate increases.
†P value limits have been drawn from ‘An Introduction to Medical Statistics (Bland 2015)’.29
‡Coefficients with p value>0.1 are represented by ‘-’.
§IMDs are ordered from IMD:1 (highest levels of deprivation) to IMD: 10 (lowest levels of deprivations).
¶IMD: 3 was removed from the analysis feature space to avoid over-specification of the model and was selected for removal as the most frequently attended IMD, and hence giving the most power as a reference category.
IMD, Index of Multiple Deprivation; JCT, job cycle time.