Investigators evaluated the effect of pain on sleep disturbance and vice-versa over time, finding stable differences between patients in sleep and pain.
A recent study found reciprocal effects between sleep disturbance and pain in adults over 50, in that changes in typical sleep disturbance predicted changes in typical pain, and changes in typical pain predicted typical sleep disturbance. Investigators further concluded that pain and sleep disturbance are both trait-like and strongly associated, and while size of effects varied, they were observed to be generally equivalent over time, suggesting bidirectionality.
First author Sarah C. Griffin, PhD, advanced fellow, Mid-Atlantic Mental Illness Research, Education, and Clinical Center, Durham Veterans Affairs Health Care System, and colleagues evaluated data from 7 waves of the Health and Retirement Study (HRS) from 2002 to 2016, focusing on US adults over 50. A total of 17,756 participants were involved in the observational HRS study in 2002, but by 2016, only 43% of the sample was alive and able to participate.
Analyses consisted of 5 models, 1 of which was a traditional cross-lagged panel model (CLPM), and 4 of which were random intercepts cross-lagged panel models (RI-CLPM). Model 1, a CLPM of sleep disturbance and pain, had a poor model fit across fit indices, but consistently concluded that sleep predicted subsequent pain (standardized estimates, 0.09-0.12; P <.001 across all estimates) and pain predicted sleep disturbance (standardized estimates, 0.08-0.14, P <.001).
Model 2, Model 3, and Model 4 were all RI-CLPMs of sleep disturbance and pain. Each model had different constraints, with Model 2 (unconstrained), outperforming both Model 3 (constrained autoregressive parameters, cross-lagged parameters, and covariances to be time invariant), and Model 4 (constrained autoregressive and cross-lagged parameters to be time invariant). Model 2, the primary model, showed significant variance in the traits of sleep disturbance and pain (P <.001). The stable differences between persons in sleep and pain were strongly associated (ß = 0.51; B = 0.12; 95% CI [0.12–0.13]; standardized error [SE] = 0.00; P <.001).
Model 5 was an unconstrained RI-CLPM of sleep disturbance and pain, although, this model also considered and adjusted for baseline age, health, partner status, depressive symptoms, education, and sex. This model found that latent variables of the trait-like natures of sleep disturbance and pain had significant variance (P <.001), and that the characteristics were associated (ß = 0.33; B = 0.06; 95% CI [0.05–0.06]; SE = 0.00; P <.001).
“The results of the present study are in line with and extend prior findings indicating the bidirectional nature of the association between sleep and pain. Specifically, among older adults, sleep difficulties have been found to predict a new onset of pain and vice versa 2–3 years later, suggesting bidirectional causality between sleep and pain in this population,” Griffin et al wrote. “Whereas prior studies have found a stronger association between sleep disturbance and subsequent pain than between pain and subsequent sleep disturbance, the present study's multi-year longitudinal approach suggests that sleep disturbances and pain are approximately equally predictive of each other in adults over 50.”
The study was limited due to its high attrition rate and the measures for sleep disturbance and pain not being validated scales. Additionally, the 2-year waves may be too long to observe, with shorter periods potentially offering the ability to observe bigger effects, and investigators did not take into consideration how chronic disease may affect results. Investigators noted additional studies should be conducted with validated or objective sleep and pain measures, collected over shorter increments of time.