Table 2 Hazard ratios of return to work after minimal postal intervention compared to control group, analysed in separate Cox proportional hazards models with length of sick leave as the dependent variable.

Model	n total in analysis (censored)*	−2 log likelihood	p value unadjusted model	p value adjusted model	Hazard ratio (95% CI) unadjusted	Hazard ratio (95% CI) adjusted†
Total	990 (192)	9899.5	0.24	<0.001	1.09 (0.95 to 1.25)	1.07 (0.93 to 1.23)
Low back pain	236 (61)	1688.1	0.38	0.011	0.87 (0.65 to 1.18)	0.79 (0.58 to 1.08)
Rheumatic disorders and arthritis	99 (25)	599.3	0.04	0.42	1.62 (1.02 to 2.57)	1.57 (0.98 to 2.51)
Other musculoskeletal	486 (87)	4345.0	0.99	<0.001	1.00 (0.82 to 1.22)	1.01 (0.83 to 1.24)
Mental disorder	169 (19)	1323.2	0.032	0.046	1.42 (1.03 to 1.96)	1.36 (0.98 to 1.89)
Sick leaves 12 weeks or longer
Total	332 (142)	2023.1	0.024	0.013	1.39 (1.04 to 1.85)	1.42 (1.06 to 1.92)
Low back pain	76 (41)	272.0	0.040	0.135	0.49 (0.25 to 0.98)	0.25 (0.10 to 0.60)
Rheumatic disorders and arthritis	45 (23)	155.1	0.29	0.72	0.61 (0.24 to 1.55)	0.56 (0.21 to 1.49)
Other musculoskeletal	155 (59)	861.0	0.001	<0.001	2.00 (1.30 to 3.08)	2.03 (1.30 to 3.19)
Mental disorder	55 (18)	262.7	0.004	0.047	2.54 (1.32 to 4.87)	3.96 (1.46 to 6.00)

*Sick leaves were censored if benefits changed to maternity benefits or old age pensions, changed by death, and if sick leave reached 365 days.

†Adjusted: gender, age group, educational level, occupation, and the current diagnostic group.