Factors predicting health-related quality of life at baseline and at the eight-year follow up
The association between suggested health factors and baseline SF-36 mean scores, and the predictive value of these health factors with regard to SF-36 development over eight years, were estimated with multivariable logistic regression analyses, controlling for sociodemographic characteristics. Results from the multivariable logistic regressions with odds ratios (OR) and 95% confidence intervals (CI) for these variables are found in Additional files
1 and
2 (Tables 1–4).
For subjects without chronic musculoskeletal pain at baseline, male sex was significantly (P < 0.05) associated with having a health status better than the mean score at baseline in PF, VT, SF, RE, and MH, and a worse score in RP. For those with chronic musculoskeletal pain, male sex was significantly associated with having a health status better than the mean score in PF. In the eight-year follow up, male sex significantly predicted a better score only in SF for subjects with chronic pain.
For subjects without chronic musculoskeletal pain at baseline, younger age was significantly (P < 0.05) associated with having a health status better than the mean score in PF, RP, BP and GH. For those with chronic musculoskeletal pain at baseline, younger age was significantly associated with having a health status better than the mean score for PF, RP, BP and GH. The same pattern could be seen in the eight-year follow up, except that being in the middle age groups significantly predicted better outcome in SF and RE for subjects without chronic pain, and in RE and MH for subjects with chronic pain.
For subjects without chronic musculoskeletal pain at baseline, belonging to a higher socioeconomic status was significantly (P < 0.05) associated with having a health status better than the mean score in PF. For those with chronic musculoskeletal pain, belonging to a higher socioeconomic status was significantly associated with having a health status better than the mean score in PF, RP, and GH. In the eight-year follow up, a higher socioeconomic status significantly predicted a better outcome in PF for subjects with chronic pain.
For subjects without chronic musculoskeletal pain at baseline, regarding immigrant status, being a native Swede was significantly (P < 0.05) associated with having a health status better than the mean score in PF, RP, GH, VT, and MH. For those with chronic musculoskeletal pain, being a native Swede was significantly associated with having a health status better than the mean score in GH and VT. In the eight-year follow up, being a native Swede significantly predicted a better outcome in RE and MH both for subjects with and without chronic pain.
For subjects without chronic musculoskeletal pain at baseline, having emotional support was significantly (P < 0.05) associated with having a health status better than the mean score in RP, GH, VT, SF, RE, and MH. For those with chronic musculoskeletal pain at baseline, emotional support was significantly associated with having a health status better than the mean score in GH, VT, SF, RE, and MH. In the eight-year follow up, emotional support significantly predicted a better outcome in RP, VT, RE, and MH in subjects with chronic pain, but was not significantly predictive in subjects without chronic pain.
For subjects without chronic musculoskeletal pain at baseline, exercise regularly was significantly (P < 0.05) associated with having a health status better than the mean score in PF, BP, and VT. For those with chronic musculoskeletal pain at baseline, exercise regularly was not significantly associated to any of the eight SF-36 health concepts. In the eight-year follow up exercise regularly failed to significantly predict any outcome in SF-36.
For subjects without chronic musculoskeletal pain at baseline, a good sleep structure was significantly (P < 0.05) associated with having a health status better than the mean score in all eight SF-36 health concepts except RP. For those with chronic musculoskeletal pain at baseline, a good sleep structure was significantly associated with having a health status better than the mean score in all eight SF-36 health concepts. In the eight-year follow up a good sleep structure significantly predicted a better outcome in PF, RP, and BP for subjects without chronic pain, and in GH and RE for subjects with chronic pain.
For subjects without chronic musculoskeletal pain at baseline, feeling rested after sleep was significantly (P < 0.05) associated with having a health status better than the mean score in all eight SF-36 health concepts except BP. For those with chronic musculoskeletal pain at baseline, feeling rested was significantly associated with having a health status better than the mean score in all eight SF-36 health concepts except PF. In the eight-year follow up feeling rested significantly predicted a better outcome in BP, GH, VT, SF, RE, and MH for subjects without chronic pain, and in BP, VT, SF, RE, and MH for subjects with chronic pain.
For subjects without chronic musculoskeletal pain at baseline, and regarding smoking habits, never being a smoker or being a former smoker, compared to being a current smoker, was significantly (P < 0.05) associated with having a health status better than the mean score in PF and MH. For those with chronic musculoskeletal pain at baseline, never being a smoker or being a former smoker, was not significantly associated to any outcome in SF-36. In the eight-year follow up, never being a smoker or being a former smoker significantly predicted a better outcome in GH, SF, and RE for subjects without chronic pain, and in RE and MH for subjects with chronic pain.
For subjects without chronic musculoskeletal pain at baseline, and regarding alcohol habits, drinking weekly, compared to never or rarely, was significantly (P < 0.05) associated with having a health status better than the mean score in PF, RP, and GH. For those with chronic musculoskeletal pain at baseline, weekly drinking of alcohol was significantly associated with having a health status better than the mean score in BP and VT. In the eight-year follow up weekly drinking of alcohol significantly predicted a better outcome in RP and MH for subjects without chronic pain, and in PF, RP, BP, and RE for subjects with chronic pain.
Although the multivariable logistic regression analyses were not intended to be complete explanatory models, at baseline 8.2–32.7% (Nagelkerke R2) of the variance in the dependent variables could be explained by the predictor variables for subjects without chronic pain. The figures were between 16.4–30.9% for subjects with chronic pain. In the eight-year follow up it was 14.5–40.9% for subjects without chronic pain and 26.3–49.5% for subjects with chronic pain.