LAUSR

fore were considered to have dropped out of the study. Depression negatively correlated (P = .05) and baseline marijuana use positively correlated (P = .04) with study completion. The Table provides characteristics and results for the sample; we found no sex association with study completion or opioid reduction. Among study completers (n = 51) baseline median MEDD (interquartile range [IQR]) was 288 (153-587) mg, with a median 6-year duration (IQR, 3-9) duration of opioid use. Median pain intensity was moderate (5 out of 10 on a numeric pain rating). After 4 months, the median MEDD was reduced to 150 (IQR, 54-248) mg (P = .002). The likelihood of a greater than 50% opioid dose reduction was not predicted by starting dose, baseline pain intensity, years prescribed opioids, or any psychosocial variable. Neither pain intensity (P = .29) nor pain interference (P = .44) increased with opioid reduction. The Figure shows the relationship between percentage change in MEDD and pain intensity in study completers.