PurposeTo assess whether patients with knee osteoarthritis pain who have early pain reduction or treatment-related adverse events of special interest (TR-AESIs; constipation, decreased appetite, malaise, nausea, somnolence, thirst) with duloxetine treatment are more likely to have later improvements in pain and quality of life (QOL) relative to placebo than patients without these early indicators.Patients and methodsThis was a post hoc analysis of 14-week randomized trial of Japanese patients with knee osteoarthritis pain (Brief Pain Inventory [BPI]-Severity average pain score ≥4) receiving duloxetine 60 mg/day (n=177 analyzed) or placebo (n=176). Primary trial outcome was change from baseline in BPI-Severity average pain at Week 14. Subgroups included early pain reduction (≥30%, 10%–30%, or <10% decrease in BPI-Severity average pain at Week 4) and early TR-AESIs (with/without TR-AESIs by Week 2). Measures included changes from baseline in BPI-Severity average pain, QOL (BPI-Interference, Western Ontario and McMaster Universities Osteoarthritis Index), Patient Global Impression of Improvement (PGI-I), and response rate (proportion achieving ≥30% or ≥50% pain reduction at Week 14).ResultsThe ≥30% early pain reduction subgroup (n=93) had significantly greater improvements in pain, QOL, and PGI-I and higher ≥30% and ≥50% response rates than placebo; the 10%–30% (n=45) and the <10% (n=33) pain reduction subgroups did not show the same (except 10%–30% group: PGI-I at Week 10 and some QOL at Weeks 10 and/or 14). Both TR-AESI subgroups (with, n=52; without, n=125) had significantly greater improvements in pain, PGI-I, and most QOL measures and higher response rates than placebo.ConclusionEarly efficacy responses to duloxetine treatment, but not early TR-AESIs, may predict later pain reduction and QOL improvements in Japanese patients with knee osteoarthritis pain.ClinicalTrials.govNCT02248480.