Table 2

 Rate ratios (95% CI) for long spells of sickness absence (>7 days) in early follow up (1994–95) by change in psychosocial work characteristics between phase 1 (1985–88) and phase 3 (1991–93)

RateAll participants (n = 3817)Stable employment grade (n = 2746)
Adjusted for age, sex, employment grade+ baseline work characteristics*All adjustments†All adjustments†
Rate of long spells per 100 person years. *Decision latitude, job demands, work social supports. †Adjusted for sex, baseline age, employment grade, baseline work characteristics (decision latitude, job demands, work social supports), change in employment grade between phases 1 and 3, baseline health (longstanding illness, health problems in past year, GHQ caseness), frequency of alcohol consumption in past year, smoking habits, body mass index, baseline sickness absence (rate of short spells per year 1985–89, rate of long spells per year 1985–89, rate of days per year 1985–89). ‡Test for linear trend in continuous work change score.
Decision latitude
    Increase16.11.02 (0.88 to 1.18)0.97 (0.83 to 1.13)1.02 (0.87 to 1.19)1.00 (0.83 to 1.20)
    Stable13.41111
    Decrease20.11.31 (1.10 to 1.55)1.34 (1.13 to 1.59)1.23 (1.03 to 1.46)1.17 (0.96 to 1.42)
Test for linear trend‡p = 0.05p = 0.006p = 0.05p = 0.11
Job demands
    Increase16.71.10 (0.96 to 1.27)1.21 (1.04 to 1.40)1.17 (1.01 to 1.36)1.30 (1.09 to 1.54)
    Stable14.11111
    Decrease14.10.97 (0.82 to 1.15)0.89 (0.75 to 1.06)0.85 (0.71 to 1.01)0.90 (0.73 to 1.11)
Test for linear trend‡p = 0.31p = 0.008p = 0.005p = 0.001
Work social supports
    Increase14.80.87 (0.75 to 1.01)0.80 (0.68 to 0.95)0.79 (0.67 to 0.93)0.78 (0.64 to 0.94)
    Stable15.51111
    Decrease14.30.94 (0.81 to 1.09)0.96 (0.83 to 1.11)0.87 (0.75 to 1.01)0.90 (0.76 to 1.07)
Test for linear trend‡p = 0.10p = 0.003p = 0.10p = 0.06