Administered orally

WOMAC consists of 24 items divided into 3 subscales:Pain(5 items):during walking,using stairs,in bed,sitting or lying,and standing Stiffness;(2 items):after first waking and later in the day Physical Function;(17 items):stair use,rising from sitting, standing, bending,walking,getting in/out of a car,shopping,putting on/taking off socks,rising from bed,lying in bed,getting in/out of bath,sitting,getting on/off toilet,heavy household duties,light household duties.Each question is answered using a 5-point Likert scale(0 to 4).Pain subscale has a range of scores of 0(none) to 20(extreme).Stiffness subscale has a range of scores of 0(none) to 8(extreme).Physical function subscale has a range of scores of 0(none) to 68(extreme).Total score ranges from 0(none) to 96(extreme).Least squares(LS) mean was calculated using analysis of covariance(ANCOVA) and adjusted for treatment, pooled investigator,and baseline score.Last observation carried forward (LOCF) method was be used for these analyses.

CGI-S measures severity of illness at the time of assessment compared with start of treatment with scores ranging from 1 (normal, not at all ill) to 7 (among the most extremely ill participants). Least squares (LS) mean was calculated using mixed model repeating measures (MMRM) and adjusted for treatment, pooled investigator, visit, and treatment-by-visit interaction, as well as baseline score and baseline score-by-visit interaction.

BPI Severity of Worst Pain is self-reported scale that measures the severity of pain based on the worst pain experienced during the past 24-hours. The severity scores ranged from 0 (no pain) to 10 (pain as severe as you can imagine). BPI Severity of Least Pain is a self-reported scale that measures the severity of pain based on the least pain experienced during the past 24-hours. The severity scores ranged from 0 (no pain) to 10 (pain as severe as you can imagine). BPI Severity of Right Now Pain is a self-reported scale that measures the severity of pain based on the pain right now. The severity scores ranged from 0 (no pain) to 10 (pain as severe as you can imagine). Least squares (LS) mean was calculated using mixed model repeating measures (MMRM) and adjusted for treatment, pooled investigator, visit, and treatment-by-visit interaction, as well as baseline score and baseline score-by-visit interaction.

BPI Interference Average Score is a self-reported scale that measures interference of pain on average of the 7 questions assessing the interference of pain for general activity, mood, walking ability, normal work, relations with other people,sleep, and enjoyment of life.The average Interference scores ranged from 0 to 10. General activity, mood,walking ability, normal work,relations with other people, sleep and enjoyment of life is each is a self-reported scale that measures the interference of pain in the past 24 hours on general activity, mood, walking ability, normal work, relations with other people, sleep and enjoyment of life.The Interference scores ranged from 0 (does not interfere) to 10 (completely interferes).Least squares (LS) mean was calculated using mixed model repeating measures (MMRM) and adjusted for treatment, pooled investigator, visit, and treatment-by-visit interaction, as well as baseline score and baseline score-by-visit interaction.

HADS is a 14-item questionnaire with 2 subscales: anxiety and depression. Each item was rated on a 4-point scale [0 (low level of anxiety or depression) to 3 (high level of anxiety or depression)], giving maximum scores of 21 for anxiety and depression. Scores of 11 or more on either subscale were considered to be a 'significant' case of psychological morbidity, while scores of 8-10 represent 'borderline' and 0-7, 'normal.' Mean was calculated using analysis of covariance (ANCOVA) and adjusted for treatment, pooled investigator, and baseline score. The last observation carried forward (LOCF) method will be used for these analyses.

Evaluation on whether the change in BPI average pain intensity scores is a direct analgesic effect of duloxetine and is independent of treatment effect on mood, as measured by Hospital Anxiety and Depression Scale (HADS) depression subscale (HADS-D), or anxiety as measured by HADS anxiety subscale (HADS-A). Path analysis for the direct analgesic effect was used to test the null hypothesis that the change in BPI average pain severity depends on the improvement of HADS-D or HADS-A, versus the alternative that the improvement in BPI average pain severity is due to a direct analgesic effect of the treatment and not dependent upon the improvement in depression and anxiety symptoms.

Pain severity was measured using an 11 point BPI scale from 0 (no pain) to 10 (worst pain) to determine average pain in the past 24 hours (average pain). A 30% (or 50%) improvement was defined as a ≥30% (or ≥50%) reduction in BPI pain severity from baseline to endpoint. Percentage of participants = (number of participants with ≥30% or ≥50% pain reduction / total number of participants in treatment group) * 100.The last observation carried forward (LOCF) method will be used for these analyses.

PGI-I measures the participant's perception of improvement at the time of assessment compared with the start of treatment. Scores ranged from 1 (very much better) to 7 (very much worse). Response to treatment is defined by endpoint PGI rating of either "much better" or "very much better".The last observation carried forward (LOCF) method will be used for these analyses.

Double Blind Treatment Phase:

Placebo administered by mouth once a day (QD) for 13 weeks.

Extension Treatment Phase:

60 mg duloxetine administered by mouth QD for 13 weeks. Started on duloxetine 30 mg QD for 1 week and then titrated up to 60 mg duloxetine QD.

Taper Phase: 1-week taper - Placebo administered for 1 week if the participant discontinued from double blind treatment phase, or duloxetine 30 mg QD administered for 1 week if the participant discontinued from extension treatment phase or completed treatment. 1 week taper is to minimize discontinuation-emergent adverse events (DEAEs).

Total of all reporting groups

Double Blind Treatment Phase:

Placebo administered by mouth once a day (QD) for 13 weeks.

Extension Treatment Phase:

60 mg duloxetine administered by mouth QD for 13 weeks. Started on duloxetine 30 mg QD for 1 week and then titrated up to 60 mg duloxetine QD.

Taper Phase: 1-week taper - Placebo administered for 1 week if the participant discontinued from double blind treatment phase, or duloxetine 30 mg QD administered for 1 week if the participant discontinued from extension treatment phase or completed treatment. 1 week taper is to minimize discontinuation-emergent adverse events (DEAEs).

PGI-I measures a participant's perception of improvement at the time of assessment compared with the start of treatment. Score ranges from 1 (very much better) to 7 (very much worse). Least squares (LS) mean was calculated using mixed model repeating measures (MMRM) and adjusted for treatment, pooled investigator, visit, and treatment-by-visit interaction, as well as baseline score and baseline score-by-visit interaction.

All participants who were randomized and had a baseline and at least 1 post-baseline observation.

WOMAC consists of 24 items divided into 3 subscales:Pain(5 items):during walking,using stairs,in bed,sitting or lying,and standing Stiffness;(2 items):after first waking and later in the day Physical Function;(17 items):stair use,rising from sitting, standing, bending,walking,getting in/out of a car,shopping,putting on/taking off socks,rising from bed,lying in bed,getting in/out of bath,sitting,getting on/off toilet,heavy household duties,light household duties.Each question is answered using a 5-point Likert scale(0 to 4).Pain subscale has a range of scores of 0(none) to 20(extreme).Stiffness subscale has a range of scores of 0(none) to 8(extreme).Physical function subscale has a range of scores of 0(none) to 68(extreme).Total score ranges from 0(none) to 96(extreme).Least squares(LS) mean was calculated using analysis of covariance(ANCOVA) and adjusted for treatment, pooled investigator,and baseline score.Last observation carried forward (LOCF) method was be used for these analyses.

All participants who were randomized and had a baseline and at least 1 post-baseline observation.

CGI-S measures severity of illness at the time of assessment compared with start of treatment with scores ranging from 1 (normal, not at all ill) to 7 (among the most extremely ill participants). Least squares (LS) mean was calculated using mixed model repeating measures (MMRM) and adjusted for treatment, pooled investigator, visit, and treatment-by-visit interaction, as well as baseline score and baseline score-by-visit interaction.

All participants who were randomized and had a baseline and at least 1 post-baseline observation.

BPI Severity of Worst Pain is self-reported scale that measures the severity of pain based on the worst pain experienced during the past 24-hours. The severity scores ranged from 0 (no pain) to 10 (pain as severe as you can imagine). BPI Severity of Least Pain is a self-reported scale that measures the severity of pain based on the least pain experienced during the past 24-hours. The severity scores ranged from 0 (no pain) to 10 (pain as severe as you can imagine). BPI Severity of Right Now Pain is a self-reported scale that measures the severity of pain based on the pain right now. The severity scores ranged from 0 (no pain) to 10 (pain as severe as you can imagine). Least squares (LS) mean was calculated using mixed model repeating measures (MMRM) and adjusted for treatment, pooled investigator, visit, and treatment-by-visit interaction, as well as baseline score and baseline score-by-visit interaction.

All participants who were randomized and had a baseline and at least 1 post-baseline observation.

BPI Interference Average Score is a self-reported scale that measures interference of pain on average of the 7 questions assessing the interference of pain for general activity, mood, walking ability, normal work, relations with other people,sleep, and enjoyment of life.The average Interference scores ranged from 0 to 10. General activity, mood,walking ability, normal work,relations with other people, sleep and enjoyment of life is each is a self-reported scale that measures the interference of pain in the past 24 hours on general activity, mood, walking ability, normal work, relations with other people, sleep and enjoyment of life.The Interference scores ranged from 0 (does not interfere) to 10 (completely interferes).Least squares (LS) mean was calculated using mixed model repeating measures (MMRM) and adjusted for treatment, pooled investigator, visit, and treatment-by-visit interaction, as well as baseline score and baseline score-by-visit interaction.

All participants who were randomized and had a baseline and at least 1 post-baseline observation.

HADS is a 14-item questionnaire with 2 subscales: anxiety and depression. Each item was rated on a 4-point scale [0 (low level of anxiety or depression) to 3 (high level of anxiety or depression)], giving maximum scores of 21 for anxiety and depression. Scores of 11 or more on either subscale were considered to be a 'significant' case of psychological morbidity, while scores of 8-10 represent 'borderline' and 0-7, 'normal.' Mean was calculated using analysis of covariance (ANCOVA) and adjusted for treatment, pooled investigator, and baseline score. The last observation carried forward (LOCF) method will be used for these analyses.

All participants who were randomized and had a baseline and at least 1 post-baseline observation.

Evaluation on whether the change in BPI average pain intensity scores is a direct analgesic effect of duloxetine and is independent of treatment effect on mood, as measured by Hospital Anxiety and Depression Scale (HADS) depression subscale (HADS-D), or anxiety as measured by HADS anxiety subscale (HADS-A). Path analysis for the direct analgesic effect was used to test the null hypothesis that the change in BPI average pain severity depends on the improvement of HADS-D or HADS-A, versus the alternative that the improvement in BPI average pain severity is due to a direct analgesic effect of the treatment and not dependent upon the improvement in depression and anxiety symptoms.

All participants who were randomized and had a baseline and at least 1 post-baseline observation.

Pain severity was measured using an 11 point BPI scale from 0 (no pain) to 10 (worst pain) to determine average pain in the past 24 hours (average pain). A 30% (or 50%) improvement was defined as a ≥30% (or ≥50%) reduction in BPI pain severity from baseline to endpoint. Percentage of participants = (number of participants with ≥30% or ≥50% pain reduction / total number of participants in treatment group) * 100.The last observation carried forward (LOCF) method will be used for these analyses.

All participants who were randomized and had a baseline and at least 1 post-baseline observation.

PGI-I measures the participant's perception of improvement at the time of assessment compared with the start of treatment. Scores ranged from 1 (very much better) to 7 (very much worse). Response to treatment is defined by endpoint PGI rating of either "much better" or "very much better".The last observation carried forward (LOCF) method will be used for these analyses.

All participants who were randomized and had a baseline and at least 1 post-baseline observation.

Placebo administered by mouth once a day (QD) for 13 weeks.

60 mg duloxetine administered by mouth QD for 13 weeks.

60 mg duloxetine administered by mouth QD for 13 weeks. Started on duloxetine 30 mg QD for 1 week and then titrated up to 60 mg duloxetine QD.

1-week taper - participants taking 60 mg QD duloxetine during the study had their dosage reduced to 30 mg to minimize discontinuation-emergent adverse events (DEAEs).

Taper Phase: 1-week taper - Placebo administered for 1 week if the participant discontinued from double blind treatment phase, or duloxetine 30 mg QD administered for 1 week if the participant discontinued from extension treatment phase or completed treatment. 1 week taper is to minimize discontinuation-emergent adverse events (DEAEs).