Table 1 Included studies: n = 6 studies (representing 2 randomized controlled trials)

Vitiello et al. 2013 [31]

USA

Lifestyles RCT (9-mo FU)

Osteoarthritis

Grade II to IV pain on Graded Chronic Pain Scale (GCPS)

N = 367

78% female

Mean age = 73y (SD 8.2)*

DSM-IV-TR for insomnia: self-reported sleep difficulties ≥3 nights/week during past month with ≥1 daytime sleep related problem

CBT pain and insomnia (CBT-PI)

(n = 122)

CBT pain (CBT-P)

(n = 122)

Education only control (EOC)

(n = 123)

CBT pain as below and standard CBT for insomnia (sleep hygiene education, stimulus control, sleep restriction, daily sleep monitoring).

Pain education, physical education, goal setting, relaxation, activity pacing, guided imagery, cognitive restructuring).

Educational content related to pain and sleep management. Classes facilitated in nondirective, self-help format.

All interventions delivered as group interventions by mental health professionals; 90-min group sessions (5–12 individuals) 1x/week, for 6 weeks

Sleep

Insomnia severity:

Insomnia Severity Index (ISI): 0–28 (lower is better)

MCID: 3.45 (defined as 30% reduction from baseline)

Sleep efficiency (SE): wrist actigraphy (over 1 week): 0–100% (higher is better)

MCID: 5%*

Pain

Pain severity:

Graded Chronic Pain Scale (GCPS): 0–10 (lower is better)

MCID: 1.3 (defined as 30% reduction from baseline)

Arthritis symptoms:

Arthritis Impact Measurement Scale V2 (AIMS): 1–10 (lower is better)

MCID: 1.8* (defined as 30% reduction from baseline)

9-mo FU

Treatment effect estimate [95%CI]:

CBT-P vs. EOC

ISI: 0.13 [− 0.89,1.16] (no statistically significant difference)

GCPS: 0.08 [− 0.21, 0.38] (no statistically significant difference)

SE: 2.91% [0.85, 4.97] (non-clinically important difference favouring CBT-P)

AIMS: − 0.06 [− 0.39, 0.28] (no statistically significant difference)

CBT-PI vs. EOC

ISI: − 1.89 [− 2.83, − 0.96] (non-clinically important difference favouring CBT-PI)

GCPS: − 0.09 [− 0.37, 0.18] (no statistically significant difference)

SE: 2.64% [0.44, 4.84] (non-clinically important difference favouring CBT-PI)

AIMS: 0.20 [− 0.26, 0.66] (no statistically significant difference)

CBT-PI vs. CBT-P

ISI: − 2.03 [− 3.01, − 1.04] (non-clinically important difference favouring CBT-PI)

SE: − 0.26% [− 2.82, 2.29] (no statistically significant difference)

Clinically significant treatment effect (OR [95% CI]):

CBT-P vs. EOC

ISI: 0.81 (0.48, 1.36) (no statistically significant difference)

GCPS: 0.79 (0.39, 1.60) (no statistically significant difference)

CBT-PI vs. EOC

ISI: 2.20 (1.25, 3.90) (clinically important difference favouring CBT-PI)

GCPS: 0.96 (0.55, 1.68) (no statistically significant difference)

CBT-PI vs. CBT-P

ISI: 2.72 (1.59, 4.64) (clinically important difference favouring CBT-PI)

Treatment effect estimate [95%CI] for participants with severe pain at baseline:

CBT-P vs. EOC

ISI: − 0.29 (− 2.36, 1.77) (no statistically significant difference)

GCPS: − 0.16 (− 0.68, 0.37) (no statistically significant difference)

SE: 5.45% (1.56, 9.33) (clinically important difference favouring CBT-P)

AIMS: 0.05 (− 0.60, 0.69) (no statistically significant difference)

CBT-PI vs. EOC

ISI: − 2.71 (− 4.91, − 0.51) (non-clinically important difference favouring CBT-PI)

GCPS: − 0.44 (− 1.00, 0.11) (no statistically significant difference)

SE: 3.69% (0.72, 6.66%) (non-clinically important difference favouring CBT-PI)

AIMS: 0.28 (− 0.55, 1.10) (no statistically significant difference)

CBT-PI vs. CBT-P

ISI: − 2.42 (− 4.15, 0.68) (no statistically significant difference)

SE: − 1.76% (− 6.10, 2.58%) (no statistically significant difference)

Clinically significant treatment effect (OR [95% CI]) for participants with severe pain at baseline:

CBT-P vs. EOC

ISI: 0.75 (0.27, 2.07) (no statistically significant difference)

GCPS: 1.18 (0.49, 2.85) (no statistically significant difference)

CBT-PI vs. EOC

ISI: 2.41 (0.93, 6.21) (no statistically significant difference)

GCPS: 1.36 (0.57, 3.24) (no statistically significant difference)

CBT-PI vs. CBT-P

ISI: 3.21 (1.22, 8.43) (clinically important difference favouring CBT-PI)

McCurry et al. 2014 [42]

USA

Lifestyles RCT (18-mo FU)

18-mo FU

Treatment effect estimate [95%CI]:

CBT-P vs. EOC

ISI: 0.32 [− 0.97, 1.61] (no statistically significant difference)

GCPS: − 0.04 [− 0.53, 0.45] (no statistically significant difference)

SE: 0.91% [− 2.10, 3.91%] (no statistically significant difference)

AIMS: − 0.09 [− 0.63, 0.45] (no statistically significant difference)

CBT-PI vs. EOC

ISI: − 0.86 [− 2.13, 0.40] (no statistically significant difference)

GCPS: − 0.36 [− 0.82, 0.10] (no statistically significant difference)

SE: 2.10% [− 1.39, 5.59%] (no statistically significant difference)

AIMS: 0.07 [− 0.55, 0.68] (no statistically significant difference)

CBT-P vs. CBT-PI*

ISI: − 0.53 [− 3.08, 2.02] (no statistically significant difference)

GCPS: − 0.54 [− 1.63, 0.55] (no statistically significant difference)

SE: 2.59% [− 4.63, 9.81%] (no statistically significant difference)

AIMS: − 0.01 [− 1.16, 1.15] (no statistically significant difference)

Clinically significant treatment effect (OR [95% CI]):

CBT-P vs. EOC

ISI: 1.06 (0.59, 1.90) (no statistically significant difference)

GCPS: 1.05 (0.52, 2.13) (no statistically significant difference)

CBT-PI vs. EOC

ISI: 1.51 (0.72, 3.12) (no statistically significant difference)

GCPS: 1.21 (0.52, 2.82) (no statistically significant difference)

Treatment effect estimate [95%CI]) for participants with severe pain at baseline:

CBT-P vs. EOC

ISI: 0.45 (− 1.84, 2.74) (no statistically significant difference)

GCPS: 0.09 (− 0.78, 0.96) (no statistically significant difference)

SE: 1.75% (− 2.36, 5.86) (no statistically significant difference)

AIMS: 0.14 (− 0.88, 1.16) (no statistically significant difference)

CBT-PI vs. EOC

ISI: − 1.63 (− 3.77, 0.50) (no statistically significant difference)

GCPS: − 0.55 (− 1.48, 0.39) (no statistically significant difference)

SE: 2.53% (− 3.29, 8.35) (no statistically significant difference)

AIMS: 0.42 (− 0.69, 1.52) (no statistically significant difference)

Clinically significant treatment effect (OR [95% CI]) for participants with severe pain at baseline:

CBT-P vs. EOC

ISI: 0.59 (0.17, 2.11) (no statistically significant difference)

GCPS: 1.01 (0.35, 2.92) (no statistically significant difference)

CBT-PI vs. EOC

ISI: 2.06 (0.51, 8.41) (no statistically significant difference)

GCPS: 1.64 (0.40, 6.80) (no statistically significant difference)

Vitiello et al. 2014 [41]

USA

Secondary analysis of Lifestyles RCT

Sleep

Insomnia severity:

Insomnia Severity Index (ISI): 0–28 (lower is better)

MCID: 3.47 (defined as 30% reduction from baseline)

Sleep efficiency (SE): wrist actigraphy (over 1 week): 0–100% (higher is better)

MCID: 5%*

Sleep quality: Pittsburgh Sleep Quality Index (PSQI): 0–21 (lower is better)

MCID: 3*

Sleep beliefs and attitudes: Dysfunctional Beliefs and Attitudes About Sleep Scale (DBAS-10): 0–100 (lower is better)

MCID: 14.6* (defined as 30% reduction from baseline)

Fatigue: Flinders Fatigue Scale (FFS): 0–31 (lower is better)

MCID: 3.4* (defined as 30% reduction from baseline)

Daytime sleepiness: 8-item Epworth Sleepiness Scale (ESS): 0–24 (lower is better)

MCID: 2*

Daytime function: 10-item Functional Outcomes of Sleep Questionnaire (FOSQ-10): 5–20 (higher is better)

MCID: 5.22* (defined as 30% higher from baseline)

Pain

Pain severity:

Graded Chronic Pain Scale (GCPS): 0–10 (lower is better)

MCID: 1.3 (defined as 30% reduction from baseline)

Arthritis symptoms:

Arthritis Impact Measurement Scale V2 (AIMS): 1–10 (lower is better)

MCID: 1.8* (defined as 30% reduction from baseline)

Catastrophizing:

Pain Catastrophizing Scale (PCS): 0–52 (lower is better)

MCID: 3.36* (defined as 30% reduction from baseline)

Fear avoidance:

Tampa Scale for Kinesiophobia (TSK): 17–68 (lower is better)

MCID: 4*

Depression:

Geriatric Depression Scale (GDS): 0–30 (lower is better)

MCID: 2.1* (defined as 30% reduction from baseline)

Improvers: ≥30% reduction on Insomnia Severity Index (ISI) from baseline to 2 months

9-mo & 18-mo FU

Mean difference improvers vs. non-improvers [95% CI]:

9mo:

ISI: 3.33 (1.35, 5.31) (non-clinically important difference favouring non-improvers)

SE: − 2.49% (− 8.15, 3.17%) (no statistically significant difference)

PSQI: 1.05 (− 0.40, 2.50) (no statistically significant difference)

PSQI-1: 0.17 (− 0.16, 0.50) (no statistically significant difference)

DBAS: 1.24 (− 7.91, 10.39) (no statistically significant difference)

FOSQ: − 0.2 (− 0.92, 0.52) (no statistically significant difference)

FFS: 2.3 (− 1.08, 5.68) (no statistically significant difference)

ESS: − 0.05 (− 1.85, 1.75) (no statistically significant difference)

GCPS: 0.7 (− 0.22, 1.62) (no statistically significant difference)

AIMS: − 0.83 (− 1.80, 0.14) (no statistically significant difference)

PCS: 2.73 (− 0.62, 6.08) (no statistically significant difference)

Tampa Scale for Kinesiophobia: 2.97 (− 1.01, 6.95) (no statistically significant difference)

GDS: 1.12 (− 1.21, 3.45) (no statistically significant difference)

Mean difference improvers vs. non-improvers [95% CI]:

18mo:

ISI: 3.33 (1.26, 5.42) (non-clinically important difference favouring non-improvers)

SE: − 2.49% (− 8.71, 3.73%) (no statistically significant difference)

PSQI: 1.05 (− 0.47, 2.57) (no statistically significant difference)

PSQI-1: 0.18 (− 0.17, 0.53) (no statistically significant difference)

DBAS: 1.24 (− 7.91, 10.39) (no statistically significant difference)

FOSQ: − 0.2 (− 0.96, 0.56) (no statistically significant difference)

FFS: 2.3 (− 1.42, 6.02) (no statistically significant difference)

ESS: − 0.04 (− 2.2, 1.94) (no statistically significant difference)

GCPS: − 0.3 (− 1.30, 0.70) (no statistically significant difference)

AIMS: − 0.83 (− 1.96, 0.30) (no statistically significant difference)

PCS: 2.73 (− 0.86, 6.32) (no statistically significant difference)

TSK: 2.97 (− 1.28, 7.22) (no statistically significant difference)

GDS: 1.12 (− 1.42, 3.66) (no statistically significant difference)

Mean difference improvers vs. non-improvers [95% CI]:

Sleep and fatigue outcomes:

ISI: − 3.03 (− 3.74, − 2.32) (non-clinically important difference favouring improvers)

SE: 1.29% (− 0.18, 2.76) (no statistically significant difference)

PSQI: − 1.45 (− 1.97, − 0.93) (non-clinically important difference favouring improvers)

DBAS: − 2.44 (− 4.74, − 0.15) (non-clinically important difference favouring improvers)

FOSQ: 0.20 (− 0.03, 0.43) (no statistically significant difference)

FFS: − 1.99 (− 3.01, − 0.98) (non-clinically important difference favouring improvers)

ESS: − 0.35 (− 1.00, 0.29) (no statistically significant difference)

Pain and depression outcomes:

GCPS: − 0.51 (− 0.80, − 0.21) (non-clinically important difference favouring improvers)

AIMS: 0.63 (0.26, 1.00) (non-clinically important difference favouring non-improvers)

PCS: 1.33 (− 2.94, 0.29) (no statistically significant difference)

TSK: − 2.27 (− 3.95, − 0.58) (non-clinically important difference favouring improvers)

GDS: − 0.52 (− 1.36, 0.32) (no statistically significant difference)

Smith et al. 2015 [32]

USA

RCT

Knee OA

American College of Rheumatology criteria for classification of knee OA Kellgren/Lawrence Grade ≥ 1

Typical knee pain ratings ≥2 of 10 experienced > 5 days/week for > 6 months

N = 100

79% female

Mean age = 59.4y (SD 9.5)

DSM-IV-TR for insomnia

Level of sleep difficulty: moderate-severe insomnia

CBT insomnia (CBT-I)

Behavioral desensitization (BD) (placebo)

Sleep restriction therapy, stimulus control therapy, cognitive therapy for insomnia, sleep hygiene education

Presented as means of eliminating the conditioned arousal through imagery

Individual 45 min sessions 1 x/week, for 8 weeks delivered by mental health professionals

Sleep

Sleep continuity: all measures recorded by diary, actigraphy, or PSG (polysomnography)

Wake after sleep onset (WASO):

(lower is better)

MCID: ≥0.2*

Total Sleep Time (TST): (higher is better)

MCID: ≥0.2*

Sleep-onset latency (SOL): (lower is better)

MCID: ≥0.2*

Sleep efficiency (SE): 0–100% (higher is better)

MCID: ≥0.2*

Insomnia Severity Index (ISI): 0–28 (lower is better)

MCID: ≥0.2*

Pain

Pain intensity:

Visual Analogue Scale (VAS): 0–100 (lower is better)

MCID: ≥0.2*

Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC pain): 0–10 (lower is better)

MCID: ≥0.2*

Conditioned Pain Modulation (CPM): (> 1 better)

Temporal summation: (lower is better)

MCID: ≥0.2*

Post-treatment, 3 and 6 months FU

Between-group treatment effect sizes CBT-I vs. BD (cohen’s d [95%CI*])

WASO:

Diary

Posttreatment: − 0.28 (− 0.93, − 0.09) (clinically important difference favouring CBT-I)

3 mo: − 0.38 (− 1.16, − 0.21) (clinically important difference favouring CBT-I)

6 mo: − 0.15 (− 0.67, 0.27) (no statistically significant difference)

Actigraphy

Posttreatment: − 0.42 (− 0.83, 0.04) (no statistically significant difference)

3 mo: − 0.27 (− 0.75, 0.19) (no statistically significant difference)

6 mo: − 0.21 (− 0.57, 0.45) (no statistically significant difference)

PSG

Posttreatment: − 0.31 (− 1.09, − 0.21) (clinically important difference favouring CBT-I)

3 mo: 0.45 (− 0.32, 0.63) (no statistically significant difference)

6 mo: 0.00 (− 0.79, 0.16) (no statistically significant difference)

TST:

Diary

Posttreatment: − 0.49 (− 1.03, − 0.18) (clinically important difference favouring BD)

3 mo: 0.01 (− 0.46, 0.46) (no statistically significant difference)

6 mo: 0.03 (− 0.56, 0.38) (no statistically significant difference)

Actigraphy

Posttreatment: − 0.44 (− 1.08, − 0.19) (clinically important difference favouring BD)

3 mo: 0.09 (− 0.53, 0.39) (no statistically significant difference)

6 mo: − 0.25 (− 0.99, 0.05) (no statistically significant difference)

PSG

Posttreatment: − 0.40 (− 0.78, 0.08) (no statistically significant difference)

3 mo: − 0.04 (− 0.47, 0.48) (no statistically significant difference)

6 mo: − 0.22 (− 0.74, 0.22) (no statistically significant difference)

SOL:

Diary

Posttreatment: 0.07 (− 0.67, 0.16) (no statistically significant difference)

3 mo: − 0.15 (− 1.17, − 0.22) (non-clinically significant difference favouring CBT-I)

6 mo: 0.01 (− 0.79, 0.16) (no statistically significant difference)

Actigraphy

Posttreatment: 0.20 (− 0.19, 0.69) (no statistically significant difference)

3 mo: 0.24 (− 0.28, 0.65) (no statistically significant difference)

6 mo: 0.06 (− 0.52, 0.50) (no statistically significant difference)

PSG

Posttreatment: 0.42 (− 0.16, 0.70) (no statistically significant difference)

3 mo: − 0.09 (− 0.89, 0.07) (no statistically significant difference)

6 mo: 0.06 (− 0.53, 0.42) (no statistically significant difference)

SE:

Diary

Posttreatment: 0.22 (− 0.14, 0.69) (no statistically significant difference)

3 mo: 0.39 (0.24, 1.18) (clinically important difference favouring CBT-I)

6 mo: 0.20 (− 0.12, 0.82) (no statistically significant difference)

Actigraphy

Posttreatment: − 0.06 (− 0.43, 0.43) (no statistically significant difference)

3 mo: 0.11 (− 0.33, 0.60) (no statistically significant difference)

6 mo: − 0.09 (− 0.65, 0.37) (no statistically significant difference)

PSG

Posttreatment: 0 (− 0.14, 0.72) (no statistically significant difference)

3 mo: − 0.12 (− 0.40%, 0.56) (no statistically significant difference)

6 mo: 0.01 (− 0.23, 0.73) (no statistically significant difference)

ISI:

Posttreatment: − 0.44 (− 0.86, − 0.03) (clinically important difference favouring CBT-I)

3 mo: − 0.24 (− 0.58, 0.35) (no statistically significant difference)

6 mo: − 0.62 (− 1.01, − 0.07) (clinically important difference favouring CBT-I)

VAS:

Posttreatment: 0.04 (1.45, 2.45) (non-clinically important difference favouring BD)

3 mo: − 0.17 (− 0.64, 0.25) (no statistically significant difference)

6 mo: 0.09 (− 0.43, 0.51) (no statistically significant difference)

WOMAC pain:

Posttreatment: 0.12 (− 0.44, 0.38) (no statistically significant difference)

3 mo: 0.01 (− 0.48, 0.42) (no statistically significant difference)

6 mo: 0.25 (− 0.26, 0.64) (no statistically significant difference)

CPM:

Posttreatment: − 0.19 (− 0.52, 0.33) (no statistically significant difference)

3 mo: 0.31 (− 0.04, 0.91) (no statistically significant difference)

6 mo: − 0.03 (− 0.40, 0.60) (no statistically significant difference)

Temporal summation:

Posttreatment: − 0.20 (− 0.52, 0.34) (no statistically significant difference)

3 mo: − 0.13 (− 0.45, 0.50) (no statistically significant difference)

6 mo: − 0.17 (− 0.52, 0.43) (no statistically significant difference)

Lerman et al. 2017 [43]

USA

Secondary analysis of RCT by Smith et al. 2015 [32]

Pain

Catastrophizing:

Pain Catastrophizing Scale (PCS): 0–52 (lower is better)

MCID: 4.5* (defined as 30% reduction from baseline)

Daytime catastrophizing: 0–100 (lower is better)

MCID: 8.9* (defined as 30% reduction from baseline)

Nocturnal catastrophizing: 0–100 (lower is better)

MCID: 7.1* (defined as 30% reduction from baseline)

Post-treatment, 3 and 6 months FU

Between-group mean differences (CBT-I vs. BD) [95% CI]*

PCS:

Posttreatment: 1.04 (− 3.07, 5.15) (no statistically significant difference)

3 mo: − 1.28 (− 5.43, 2.87) (no statistically significant difference)

6 mo: 0.54 (− 3.64, 4.72) (no statistically significant difference)

Daytime catastrophizing:

Posttreatment: 2.09 (− 4.85, 9.03) (no statistically significant difference)

3 mo: 2.75 (− 4.58, 10.08) (no statistically significant difference)

6 mo: 0.37 (− 7.11, 7.85) (no statistically significant difference)

Nocturnal catastrophizing:

Posttreatment: − 0.32 (− 6.82, 6.18) (no statistically significant difference)

3 mo: 0.65 (− 5.82, 7.12) (no statistically significant difference)

6 mo: − 2.73 (− 9.63, 4.17) (no statistically significant difference)

Salwen et al. 2017 [36]

USA

Secondary analysis of RCT by Smith et al. 2015 [32]

Knee OA

American College of Rheumatology criteria for classification of knee OA. Kellgren/Lawrence Grade ≥ 1

Typical knee pain ratings ≥2 of 10 experienced > 5 days/week for > 6 months

N = 74

77% female

Mean age = 59.5y (SD 9.9)

Sleep

Sleep continuity:

Total sleep time (TST): self-reported, actigraphy (higher better)

MCID: 40 min*

Sleep efficiency (SE): 0–100% (higher better)

MCID: 5%*

Sleep onset latency (SOL): (lower better)

MCID: > 30 min

Wake after sleep onset (WASO):

(lower better)

MCID: 30 min

Pain

Western Ontario and McMaster Universities Osteoarthritis Index

(WOMAC pain): 0–10 (lower is better)

MCID: 1.4* (defined as 30% reduction from baseline)

Responders: reported ≥30% improvement in self-reported pain from baseline to 6 m follow-up

6 months FU

Pain responders: 31/74 (42%)

Nonresponders: 43/74 (58%)

Mean difference mid-treatment (pain responders – nonresponders) [95% CI]*)

SE: − 0.02 (− 0.06, 0.02) (no statistically significant difference)

SOL: − 4.24 (− 15.13, 6.65) (no statistically significant difference)

TST: − 14.67 (− 43.04, 13.70) (no statistically significant difference)

WASO: 8.89 (− 6.14, 23.92) (no statistically significant difference)

Patients who achieved 382 min (6 h) TST per night by mid-treatment (4 weeks) were more likely to report clinically significant pain reduction (≥30% reduction) at FU regardless of treatment group (sensitivity 54.8%, specificity 81.4%)