Methods

308 adult primary care patients seen at either a university integrated primary care clinic (n= 147; 47.7%) or federally qualified health center (n= 161; 52.3%) from August 2019 to March 2020 completed intake paperwork including questions assessing psychiatric symptoms and sleep.

Results

A structural equation model was used. To test for statistical mediation, we performed 500 bootstraps and examined bias-corrected estimates. The model evidenced good fit, X2 (16)= 22.53, p=. 127, CFI=. 98, TLI=. 95, SRMR=. 03, RMSEA=. 04 [90% CI:. 00,. 07]. All manifest variables loaded significantly onto their latent factors (standardized loadings>. 40, p values<. 05). The model explained 14.7% of the variance in total sleep time (TST), 14.8% in sleep procrastination, and 54.0% in sleep problems. Anxiety was negatively associated with TST (β=-. 36, p<. 001) and positively associated with bedtime procrastination (β=. 35, p<. 001) and greater sleep problems (β=. 63, p=. 006). The association between anxiety and sleep problems was not statistically mediated by TST. There was a significant indirect effect of anxiety on sleep problems through bedtime procrastination. The total effect of anxiety on sleep problems was significant (β=. 70, p=. 003). We found a significant direct association between age and TST, and age and bedtime procrastination. However, there was no significant direct association between age and sleep problems (β=. 00, p=. 984). Sleep attitudes were unrelated to any of the primary study variables. None of the exogenous variable covariances were significant. The covariance between TST and bedtime procrastination was significant.

Conclusion

Sleep problems are pervasive and complex. This study highlights factors related to sleep problems and support anxiety and pre-bedtime behaviors as treatment targets.