Work stress, worries, and pain interact synergistically with modelled traffic noise on cross-sectional associations with self-reported sleep problems

Abstract

Objectives

To examine the risk of sleep problems associated with work stress (job strain, job demands, and decision authority), worries and pain and to investigate the synergistic interaction between these factors and traffic noise.

Methods

Sleep problems and predictor variables were assessed in a cross-sectional public health survey with 12,093 respondents. Traffic noise levels were assessed using modelled A-weighted energy equivalent traffic sound levels at the residence. The risk of sleep problems was modelled using multiple logistic regression analysis.

Results

With regard to sleep problems not attributed to any external source (general sleep problems), independent main effects were found for traffic noise (women), decision authority (women), job strain, job demands, suffering from pain or other afflictions, worries about losing the job, experiencing bullying at work, having troubles paying the bills, and having a sick, disabled, or old relative to take care of (women). Significant synergistic effects were found for traffic noise and experiencing bullying at work in women. With regard to sleep problems attributed to traffic noise, strong synergistic interactions were found between traffic noise and, respectively, job demands (men), having pain or other afflictions, taking care of a sick, old, or disabled relative, and having troubles paying the bills. Main effects were found for worries about losing the job, experiencing bullying at work, job strain (men), and decision authority (men). Synergistic interactions could potentially contribute with 10–20% of the sleep problems attributed to traffic noise in the population.

Conclusions

Work stress, pain, and different worries were independently associated with general sleep problems and showed in general no synergistic interaction with traffic noise. In contrast, synergistic effects between traffic noise and psychological factors were found with regard to sleep problems attributed to traffic noise. The synergy may contribute significantly to sleep problems attributed to traffic noise in the population.

Appendix

Calculation of excess attributable fraction

Consider two interacting variables as a single variable with level low (both variables at low level) and high (both high), and RR is the relative risk of sleep problems at the high level. The attributable fraction of sleep problem cases among the exposed, AFE, is then:

$$ {\text{AFE}} = \left( {{\text{RR}} - 1} \right)/{\text{RR }}\left( {{\text{Greenland 2}}00 8} \right). $$

Likewise, assume that RR* would be the relative risk if there was no interaction between the variables, and the corresponding attributable fraction is AFE* = (RR*− 1)/RR*. We define the excess attributable fraction of sleep problem cases among the exposed, EAFE, as the difference between these two attributable fractions:

$$ {\text{EAFE}} = {\text{AFE}} - {\text{AFE}}^{*} = \left( {{\text{RR}} - 1} \right)/{\text{RR}} - \left( {{\text{RR}}^{*} - 1} \right)/{\text{RR}}^{*} . $$

RR and RR* are related through AP. Thus

$$ {\text{AP}} = \left( {{\text{RR}}_{ 1 1} - {\text{RR}}_{ 10} - {\text{RR}}_{0 1} + 1} \right)/{\text{RR}}_{ 1 1} = \left( {{\text{RR}}_{ 1 1} - {\text{RR}}^{*}_{ 1 1} } \right)/{\text{RR}}_{ 1 1} $$

where RR*₁₁ = RR₁₀ + RR₀₁ − 1. Rearranging this expression for AP one reaches the relation RR*₁₁ = RR₁₁ × (1 − AP). Inserting this in the expression for EAFE (and making use of the identities RR = RR₁₁ and RR* = RR*₁₁), we arrive to the expression for EAFE used in this study:

$$ {\text{EAFE}} = {\text{AP}}/\left( { 1- {\text{AP}}} \right) \times 1/{\text{RR}}_{ 1 1} . $$