The analyses were performed using the SAS statistical software for Windows (SAS V.9.4; SAS Institute). Associations were modelled using the general linear model (Proc GLM, SAS V.9.4) controlling for confounders and weighted based on information from high-quality national registers at Statistics Denmark, which included gender, age, occupation, highest completed education, family income, family type and origin.25 (link) The residuals of the outcome variable (LBP) were normally distributed through visual inspection. Because a strong correlation exists between a 0–10 NRS and a 0–10 Visual Analogue Scale,36 (link) the 0–10 (11 point) ordinal NRS in this study is treated as a continuous scale. Estimates are reported as least square means pain intensity (NRS) and 95% CIs and between-group least square means differences and 95% CI. An alpha level of <0.05 was chosen as statistically significant differences.