IPMPs Versus Usual Care

Primary Outcomes

Pain

Integrated pain management was associated with a statistically significant but clinically unimportant effect on pain compared with usual care postintervention (4 trials, N=1142, pooled difference −0.31, 95% confidence interval [CI] −0.51 to −0.11, I²=0%).^{41,42,46,47,50} IPMPs were associated with a small improvement in pain in the short term (2 trials, N=721, pooled difference −0.59, 95% CI −1.17 to −0.07, I²=0%)^48,50 and the intermediate term (1 trial, N=197, difference −0.70, 95% CI −1.13 to −0.09)⁵⁰ but not in the long term (2 trials, N=688, pooled difference −0.28, 95% CI −0.80 to 0.23, I²=0%),^48,50 (Figure 2). No trials reported the likelihood of experiencing significant improvement in pain.

Figure 2

IPMP versus usual care control: Pain. CI = confidence interval; Combo = combination group and individual sessions; CPG = Chronic Pain Grade Severity subscale; IPMP = integrated pain management program; LBP = low back pain; MSK = musculoskeletal pain; (more...)

Function

Small improvements in function were associated with IPMP versus usual care for continuous measures of function postintervention (4 trials, N=1142, pooled standardized mean difference [SMD] −0.20, 95% CI −0.34 to −0.06, I²=0%) and in the short term (2 trials, N=721, pooled SMD −0.23, 95% CI −0.40 to −0.02, I²=0%),^48,50 but not in the intermediate term (1 trial, N=220, pooled SMD – 0.10, 95% CI −0.38 to 0.17).⁵⁰ Long term, a small improvement tended to favor IPMP but was not statistically significant (2 trials, N=688, pooled SMD −0.19, 95% CI −0.36 to 0.01, I²= 0%)^48,50 (Figure 3). The differences on the original Roland Morris Disability Questionnaire (RMDQ) were −1.0 to −1.3 postintervention, −0.9 to −1.33 in the short term, and −1.0 to −1.11 in the long term. The postintervention difference on the original Western Ontario and McMaster Universities Arthritis Index (WOMAC) Function scale in one trial was −3.3 and the difference in change scores from baseline was −0.2 in another trial.

Figure 3

IPMP versus usual care control: Function. CI = confidence interval; Combo = combination group and individual sessions; IPMP = integrated pain management program; LBP = low back pain; MSK = musculoskeletal pain; RMDQ = Roland Morris Disability Questionnaire; (more...)

More IPMP patients experienced 30 percent or more improvement on the RMDQ (0 to 23 or 24 scale) postintervention compared with those receiving usual care across two trials, in patients with chronic LBP (2 month intervention)⁵⁰ or chronic musculoskeletal pain (12 month intervention)^46,47 (2 trials, N=608, 23% vs. 13%, pooled risk ratio [RR] 1.73, 95% CI 1.14 to 2.80, I²=0). In contrast, two trials in patients with OA found no difference between IPMP and usual care based on 18 percent or more improvement on the WOMAC Function (Scale 0 to 68) postintervention (2 trials, N=399, 18% vs. 21%, pooled RR 1.05, 95% CI 0.69 to 1.65, I²=0).^41,42 In the RCT of patients with LBP (N=207), the 30 percent or greater improvement in RMDQ persisted for IPMP participants to the short-term (RR 1.81, 95% CI 1.20 to 2.72) and intermediate-term (RR 1.97, 95% CI 1.30 to 2.98), but not to long-term (RR 1.35, 95% CI 0.98 to 1.85) followup after the 2-month intervention (Figure 4).

Figure 4

IPMP versus usual care control: Function success. CI = confidence interval; Combo = combination group and individual sessions; IMP = improvement; IPMP = integrated pain management program; LBP = low back pain; MSK = musculoskeletal pain; RMDQ = Roland (more...)

One trial in patients with chronic widespread pain (N=221)^44,49 reported a measure that combined pain and disability that could not be pooled with other studies. It found no differences between IPMP versus usual care postintervention or long term.

Opioid Use

One fair quality trial (N=397) reported no significant differences in opioid prescriptions between IPMP and usual care groups (65% vs. 61%) but IPMP participants were more likely to receive long-acting opioids when prescribed (31% vs. 18%) based on adjusted estimates.^46,47 Prescription of nonsteroidal anti-inflammatory drugs (NSAIDs), antidepressants, and capsaicin were significantly more common in the IPMP group versus usual care. Data are insufficient from one small poor-quality trial, which reported 35 percent attrition.⁴³ The trial found no differences between IPMP delivered 2 hours per week for 8 weeks and waitlist regarding early opioid prescription refill (7.7% vs. 25%) at intermediate term. One fair-quality trial reported that similar proportions of patients receiving the IPMP and usual care received new pain medications or that alternative pain medications were discussed during the 12-month intervention, but specific medication types were not described.⁴¹

IPMPs Versus Active Comparators

One fair-quality RCT conducted in England compared IPMP which incorporated telephone-delivered CBT (TCBT) (N=112) with TCBT (N=112) alone and exercise alone (N=109), as well as usual care as described above, in patients with chronic widespread pain (Results Appendix B, Table B-4; Appendix E, Table E-1).^44,49 Patients were predominately female (71%), with a mean age of 56 years; approximately one third were retired. TCBT consisted of an initial assessment, seven weekly sessions of 30 to 45 minutes, plus one session at 3 months and another at 6 months post-randomization focused on monitoring progress toward goals, problem solving barriers to improvement and relapse prevention. Exercise was geared toward enhancing cardiorespiratory fitness and consisted of six fitness instructor led monthly assessment appointments to guide and monitor home or leisure facility/gym activity. Exercise types and intensity were tailored to the patient with recommendations for 20 to 60 minutes of exercise at least twice per week on gym facility days plus everyday activities (e.g., brisk walking) on other days. Exercise was logged into patient diaries. The IPMP combined the TCBT and exercise protocols and included provider exchange of information regarding treatment; components were individually tailored and delivered.

Key Points

• Comprehensive pain management was associated with small improvements in pain on a 0 to 10 scale compared with usual care or waitlist postintervention (SOE: moderate). Differences were below the threshold for small or not statistically significant (or both) at short-term, intermediate-term and long-term followup (SOE: low).
• Comprehensive pain management was associated with moderate improvement in function compared with usual care or waitlist immediately postintervention and in the short term; there was no difference in the intermediate or long term (SOE: low at all time points).
• Data on harms were only reported by one trial (SOE: insufficient).

Detailed Synthesis

Twenty-three RCTs (reported in 30 publications)^{51,52,54–57,60,62,65–67,69,70,73,74,82,83,88,89,91–93,100,102–108} compared CPMPs with usual care or waitlist (Results Appendix B, Table B-5; Appendix E, Table E-2). Randomization occurred at the individual patient-level in all trials. Sample sizes ranged from 39 to 459 (total N=2,961). The mean age of participants ranged from 37 to 50 years in all trials except for two, in which the mean age was 59⁵² and 69¹⁰⁶ years. The trial in which mean age was 69 years enrolled U.S. veterans 60 years of age or older with LBP. Three trials restricted enrollment to female patients;^52,54,88 in the other trials, the proportion female ranged from 4 percent¹⁰⁶ to 96 percent.^102,103 Race or ethnicity was reported by seven studies.^{62,82,83,106–108} Only five studies reported the proportion of non-White patients (range, 4% to 30%; two U.S. trials^106,107 had 30% African Americans).^{62,82,83,106–108} The pain condition was LBP in nine trials,^{51,55–57,60,66,67,74,91,92,100,106,107} FM in five trials,^{54,73,88,102–105} rheumatoid arthritis in two trials,^52,89 mixed chronic pain in six trials,^{65,69,70,82,83,93,108} and mixed traumatic injury in one trial.⁶² One trial⁶² enrolled patients with acute (<4 weeks) pain following trauma; the other trials enrolled patients with chronic pain (mean duration 3 months⁷⁴ to 168 months^66,67 in trials that reported this information). Baseline pain intensity ranged from 4.4⁷⁴ to 8.2⁸⁸ on a 0 to 10 scale. Eleven trials ^{54,62,65,73,74,88,89,93,100,102,107} reported baseline depression and patients with psychological comorbidities were excluded in three trials.^51,54,70 The proportion of patients who used opioids at baseline ranged from 10 percent⁶² to 61 percent¹⁰⁸ in six trials that reported this information.^{55,62,82,83,93,106,108} Information regarding Medicare-qualifying criteria other than age was not reported. Five trials focused on employed patients or vocational rehabilitation.^{56,57,60,65–67,70,107}

Four trials evaluated high intensity (≥20 hours per week or >80 hours total) CPMPs^{56,57,60,70,83,102} and 17 trials evaluated low intensity (<20 hours per week)^{51,52,54,55,62,66,67,69,73,74,88,89,91–93,100,104,105,107,108} programs (one trial evaluated both high and lower intensity programs¹⁰⁸); in two trials the intensity was unclear.^65,107 In addition to psychological, educational, and exercise components, additional interventions in the CPMP included drug management and medication in 14 trials,^{51,52,55–57,60,62,65,70,73,82,83,88,89,93,106–108} two trials included massage^66,67,70 and one trial included massage, acupuncture and chiropractic therapy.¹⁰⁶ The number of sessions varied from one session a week to 5 days of inpatient sessions a week⁷⁰ and duration of treatment ranged from 2 weeks⁸⁹ to 12 months.⁵² Programs were provided individually in four trials,^{62,93,106,107} in group sessions in fourteen trials,^{51,54,56,57,60,66,67,69,70,73,74,82,83,89,102–105,108} and in a combination of group and individual sessions in four trials;^{52,88,91,92,100} one trial did not report delivery format.⁶⁵ Sixteen trials^{51,52,55–57,60,62,65–67,69,73,74,82,83,88,93,102,103,106,107} compared CPMP to usual care and seven trials^{54,70,89,91,92,100,104,105,108} to waitlist controls. The majority of CPMPs were conducted in outpatient settings, with the exception of three trials with separate inpatient and outpatient programs,^{66,67,82,83,108} and one that was online.⁹³

Trials were conducted in fourteen different countries including three studies in the United States,^100,106,107 sixteen studies in Europe,^{52,54–57,60,65–67,69,70,74,82,83,88,89,91,92,102–105,108} two in Australia,^62,93 and one each in Canada,⁷³ Iran,⁵¹ New Zealand,^82,83 and Turkey.⁸⁸ Four trials^{52,54,56,57,60,89} were conducted in rheumatology clinics and the remainder were in pain management or rehabilitation clinic settings with the exception of one trial conducted at the patients home or a local Young Men’s Christian Association⁷³ and one trial of an online program.⁹³ Six trials had followup of 1 year or greater (longest 60 months).^56,57,60

Fifteen trials were fair quality^{52,54–57,60,65,69,70,73,74,88,91–93,102–106} and eight were poor quality.^{51,62,82,83,89,100,107,108,66,67} The trials were unable to blind care providers or patients; in addition, only two trials blinded outcome assessors.^56,57,60,106 Other common limitations included unclear randomization and allocation concealment methods and high attrition (Appendix F, Table F-2).

Primary Outcomes

Pain

CPMPs were associated with a small improvement in pain compared with usual care or waitlist at postintervention followup (11 trials, N=764, pooled difference −0.53 on a 0 to 10 scale, 95% CI −0.80 to −0.25, I²=0%).^{51,55,62,70,73,83,92,93,100,103,106} At other time points the difference was below the threshold for small effects, was not statistically significant, or both (short term: 6 trials, N=943, pooled difference −0.39, 95% CI −0.83 to 0.04, I²=36.6%;^{54,60,66,70,93,108} intermediate term: 4 trials, N=690, pooled difference −0.85, 95% CI −2.01 to 0.21, I²=83.5%;^65,67,88,107 long term: 6 trials, N=906, pooled difference −0.13, 95% CI −0.71 to 0.22, I²=19.5%^{51,56,65,67,74,103}) (Figure 5). Findings were similar in sensitivity analyses excluding poor-quality trials,^{51,62,66,67,82,83,100,107,108} excluding the trial in patients with acute (<4 weeks) trauma,⁶² using the most common duration of followup for the long-term analysis,^57,65–67 and excluding trials that used the McGill Pain Questionnaire (MPQ) (which is based on pain descriptors, rather than a visual analog scale [VAS] or numerical rating scale [NRS] for pain intensity)¹⁰⁰ (Appendix I, Figures I-4 to I-7). The effect sizes and associated variability were similar for trials of higher intensity programs (>20 hours/week or >80 total hours) and lower intensity programs. Meta-regression confirmed no significant differences in pain effect estimates between higher and lower intensity programs immediately postintervention (p=0.67), in the short-term (p=0.74) or the long-term (p=0.50). Effect sizes and variability were also similar across programs that were delivered individually, as group sessions, or a combination of these.

All trials enrolled patients with chronic pain except for one small (n=67) trial of patients with acute (<4 weeks) trauma. In this trial, which only reported results at postintervention, there was no difference between CPMP (lower intensity) versus usual care, but the estimate was imprecise.⁶² One small (N=55), fair quality trial evaluated effects of CPMP (intensity unclear) in patients 60 and older with chronic LBP.¹⁰⁶ Postintervention effect on pain (difference −1.22, 95% CI −2.28 to −0.16) was somewhat greater compared to the other trials, which enrolled younger populations. However, evidence was too limited to determine how older age impacts effectiveness.

Figure 5

CPMP versus usual care or waitlist control: Pain. BPI = Brief Pain Inventory; CBNP = chronic back/neck pain; CI = confidence interval; CLBP = chronic low back pain; Combo = combination group and individual sessions; CP = chronic pain; CPMP = comprehensive (more...)

Function

CPMPs were associated with moderate improvements in function compared with usual care or waitlist at postintervention (13 trials, N=981, pooled SMD −0.52, 95% CI −0.88 to −0.16, I²=83.0%)^{51,52,55,62,73,83,89,92,93,100,103,105,106} and short-term followup (7 trials, N=1,097, pooled SMD −0.62, 95% CI −1.02 to −0.24, I²=83.7%).^{54,60,66,89,93,105,108} There was no difference at intermediate-term followup (4 trials, N=588, pooled SMD −0.33, 95% CI −0.81 to 0.05, I²=66.9%%),^65,67,88,89 or at long-term followup (6 trials, N=974, pooled SMD −0.21, 95% CI −0.47 to −0.00, I²=42.1%).^{51,56,65,67,74,89,103} The measures used to assess function varied (Figure 6). The difference on the original RMDQ scale ranged from −1.40 to 2.80 postintervention and was −1.9 in the short term in one trial. Differences on the original Fibromyalgia Impact Questionnaire (FIQ) ranged from −11.50 to −3.40 postintervention and from −12.1 to 0.10 in the short term. For the Sickness Impact Profile (SIP), differences on the original scale ranged from −68.80⁸³ to −3.50 postintervention and was −11.50 in the short term in a single trial.

The findings for function were similar in sensitivity analyses excluding an outlier trial,⁸⁹ excluding poor-quality trials,^{51,62,66,83,89,100,108} and analysis excluding the trial in acute trauma (<4 weeks).⁶² Using the most common duration of followup (primarily 12 to 18 months) for the long-term analysis^65,67 the effect size was decreased and again there was no difference between CPMP and usual care or waitlist (6 trials pooled SMD −0.04, 95%CI −0.21 to 0.09, I²=0%)^{51,57,65,67,74,103} (Appendix I, Figures I-8 to I-11). Meta-regression confirmed that there were no significant differences in function estimates between higher and lower intensity programs immediately postintervention (p= 0.813) and in the long-term (p=0.154). The trial of patients with acute trauma found no difference between CPMP versus usual care at postintervention, but the estimate was imprecise.⁶² The trial of patients 60 years or older reported results for postintervention function that were consistent with the overall pooled results.¹⁰⁶

Figure 6

CPMP versus usual care or waitlist control: Function. CBNP = chronic back/neck pain; CI = confidence interval; CLBP = chronic low back pain; Combo = combination group and individual sessions; CP = chronic pain; CPMP = comprehensive pain management program; (more...)

Opioid Use

Evidence on CPMP and changes in opioid prescribing was very limited. One trial found no difference between CPMP versus usual care in likelihood of opioid use at postintervention (63.3% vs. 50.0%, RR 1.27, 95% CI 0.82 to 1.95) or 3 months (60.0% vs. 51.0%, RR 1.16, 95% CI 0.75 to 1.81).⁹³ One other trial reported that among patients randomized to CPMP, the proportion not using opioids at followup increased compared to baseline (at 12 months, the proportion not using opioids increased from 47% to 80% for inpatient CPMP and from 33% to 55% for outpatient CPMP), but did not report opioid use in the waitlist control arm.¹⁰⁸

Key Points

• There were no differences on a 0 to 10 scale comparing CPMP with physical activity postintervention, at short term, intermediate term or long term (SOE: moderate for postintervention and long term, low for short term and intermediate term).
• CPMP was associated with a small improvement in short-term function compared with physical activity, but not postintervention, intermediate term or long term (SOE: moderate for all timepoints).
• Only two trials reported on harms. One noted no adverse events related to the intervention, and the other reported less pain in new locations in the CPMP group compared to physical activity, RR 0.38, 95% CI 0.08 to 1.7 (SOE: insufficient).

Detailed Synthesis

Fifteen RCTs (reported in 21 publications) compared 17 CPMPs with physical activity (Results Appendix B, Table B-6; Appendix E, Table E-2).^{53,56–59,61,68,69,71,72,75,79,84–86,90–92,100,102,103} One was a cluster RCT that randomized patients in clusters of four consecutive participants;⁹² the remainder randomized individual patients. Sample sizes ranged from 33 to 409 (total sample=2,344). The diagnosis was FM,¹⁰³ chronic musculoskeletal pain,⁷⁹ and a mix of subacute and chronic LBP pain⁶⁸ in one trial each and chronic LBP with or without leg pain in the remaining 12 trials. Mean symptom duration ranged from 2.2 to 12.9 years across six trials that reported it,^{69,72,79,92,100,103} and in seven trials, pain 3 months or more,^71,84–86 6 months or more,^53,59,90 or longer than 6 weeks (for the trial including subacute pain),⁶⁸ was required for inclusion. The remaining two trials^61,75 only stated that pain was chronic. Baseline pain intensity was moderate in most trials. None of the studies specifically included Medicare patients and mean age in all studies ranged from 40 to 49 years. One trial in FM was comprised of primarily females (95%)¹⁰³ and another (chronic LBP) enrolled only females.⁷² The proportion of male participants ranged from 21 to 67 percent across the other trials. Only one trial reported on race (all patients were White).¹⁰⁰ Comorbidities were poorly reported across the trials and included smoking (range, 31% to 66%) in three trials,^59,71,72 prior back surgery (15% to 25%) in three trials,^59,71,92 previous depression diagnosis in one trial (30%),⁷¹ and one trial stated that 39 percent of patients had one or more comorbidity.⁷⁹ In addition, most trials excluded patients with severe psychiatric disorders and contraindications to exercise. Other exclusion criteria related to comorbidities included substance abuse (3 trials),^68,69,103 severe comorbidities (not further defined) (2 trials),^69,90 cardiovascular disease (1 trial)¹⁰⁰ and current depression (1 trial).⁷⁹ Studies inconsistently reported disability. One trial did report that 38 and 24 percent of its population, respectively, was receiving full and partial sick leave/disability pensions.⁹² Nine trials included patients who were either currently sick-listed, at risk for sick leave/work disability, or who had numerous episodes of sick leave prior to study entry.^{59,61,68,69,71,79,84,86,90} One trial¹⁰⁰ was conducted in the United States and the remainder were conducted in Europe.

Regarding the physical activity components of both groups, eight trials included only active physical activity such as aerobics, strength training, flexibility and stretching exercises, and proprioception and coordination exercises^{59,61,71,84,86,92,100,103} and seven included both active and passive modalities (e.g., massage, traction, spinal mobilization, ultrasound, heat therapy, and electrotherapies).^{53,68,69,72,75,90} Ten trials also included an occupational therapy or work therapy component.^{59,61,68,71,72,75,79,84,86,103} The intensity of the programs varied substantially, ranging from 24 to 150 hours. Ten programs met our criteria for high intensity (≥20 hours/week or >80 hours total),^{56–59,61,68,69,71,75,79,86,90,102,103} six for low intensity,^{53,56–59,72,84–86,100} and one unclear.⁹² The duration of the programs ranged from 3 to 12 weeks. Program components were delivered in group sessions in five trials^{56–59,61,69,84,85,102,103} and in a combination of group and individual sessions in ten trials.^{53,68,71,72,75,79,86,90–92,100}

Eleven trials were rated fair quality^{53,56–59,61,68,69,71,72,84,85,90–92,102,103} and two poor quality^86,100 (Appendix F, Table F-2). The major methodological limitation in the fair-quality trials was the inability to effectively blind patients and caregivers to the CPMP. Other methodological shortcomings included unclear randomization, unclear allocation concealment methods, and high attrition.

Primary Outcomes

Pain

There were no differences in low back or musculoskeletal pain on a 0 to 10 scale comparing CPMP with physical activity alone postintervention (8 trials, N=1,312, pooled difference −0.05, 95% CI −0.32 to 0.19, I²=0%),^{72,75,79,84,90,91,100,103} at short term (1 trial, N=106, difference −0.35, 95% CI −1.49 to 0.79),⁵⁹ intermediate term (4 trials, N=341, pooled difference −0.15, 95% CI −0.73 to 0.38, I²=0%)^71,72,91,100 or long term (9 trials, N=2,492, pooled difference 0.05, 95% CI −0.30 to 0.42, I²=0%),^{56,61,72,75,85,86,91,100,103} (Figure 7). The findings for pain were similar in sensitivity analyses excluding two poor-quality trials postintervention, and in the intermediate and long term^86,100 (Appendix I, Figure I-21) and using the most common duration of followup for the long-term analysis^56,58,72 (Appendix I, Figure I-22). There were three fair-quality trials that assessed pain radiating into the leg on a 0 to 10 scale, which found no difference comparing CPMP with physical activity alone postintervention (1 trial, N=120, difference 0.20, 95% CI −0.75 to 1.15),⁷² short term (1 trial, N=106, difference −0.94, 95% CI −2.30 to 0.42),⁵⁹ intermediate term (1 trial, N=115, difference 0.50, 95% CI −0.44 to 1.44),⁷² and long term (3 trials, N=263, pooled difference −0.61, 95% CI −1.59 to 0.37, I²=0%)^56,61,72 (Appendix I, Figure I-23). There were no significant differences between higher and lower intensity programs in effects on pain immediately postintervention (p=0.14), at intermediate term (p=0.30) or at long term (p=0.55), based on meta-regression.

Figure 7

CPMP versus physical activity: Pain. CI = confidence interval; CLBP = chronic low back pain; C. MSK = chronic musculoskeletal disease; Combo = combination group and individual sessions; CPMP = comprehensive pain management program; FIQ = Fibromyalgia (more...)

Function

CPMP was associated with a small improvement in short-term function compared with physical activity alone (3 trials, N=459, pooled SMD −0.37, 95% CI −0.61 to −0.16, I²=0%),^53,59,68 but not postintervention (9 trials, N=1,379, pooled SMD −0.05, 95% CI −0.16 to 0.05, I²=0%),^{68,72,75,79,84,90,91,100,103} or in the intermediate (6 trials, N=695, pooled SMD −0.11, 95% CI −0.36 to 0.13, I²=38.3%)^{53,68,71,72,91,100} or long term (10 trials, N=1,214, pooled SMD −0.12, 95% CI −0.31 to 0.06, I²=43.3%),^{58,61,68,72,75,85,86,91,100,103} (Figure 8). The findings for function were similar in sensitivity analyses excluding two poor-quality trials^86,100 (Appendix I, Figure I-24) and using the most common duration of followup for the long-term analysis^56,58,72 (Appendix I, Figure I-25). There were no significant differences between higher and lower intensity programs in effects on function immediately postintervention (p=0.50) or at long-term (p=0.91), based on meta-regression.

Figure 8

CPMP versus physical activity: Function. CI = confidence interval; CLBP = chronic low back pain; C. MSK = chronic musculoskeletal disease; Combo = combination group and individual sessions; CPMP = comprehensive pain management program; DPQ = Dallas Pain (more...)

Key Points

• Comprehensive pain management programs were associated with moderate improvement in average pain (VAS or NRS 0 to 10 scale) versus pharmacologic treatments alone postintervention and small improvement in the intermediate term, but not in the short or long term (SOE: low at all time points).
• More CPMP participants with FM experienced 30 percent or more improvement on 0 to 10 NRS pain scale in one fair-quality trial compared with pharmacologic therapy alone.
• Comprehensive pain management programs were associated with moderate improvement in function based on FIQ Total Score (0 to 100 scale) postintervention and small improvements at short, intermediate, and long term versus pharmacologic therapy alone (SOE: moderate for intermediate term; low for postintervention, short term and long term). At all timepoints, more CPMP patients experienced 14 percent or more improvement in function.
• Comprehensive pain management programs were associated with moderate pain improvement postintervention, and in the intermediate and long term compared with the combination of pharmacologic therapy (antidepressants) and physical activity. There were no differences in function between groups at any time point (SOE: low).
• Evidence regarding opioid use from one trial comparing CPMP versus pharmacologic therapy plus physical activity is insufficient.
• No trial reported on harms.

Detailed Synthesis

Six RCTs (reported in 14 publications) (Appendix E, Table E-2) compared CPMP with pharmacologic therapy alone (5 RCTs, 13 publications)^{63,64,76–78,80,81,87,94–98} (Results Appendix B, Table B-7) or plus physical activity (2 RCTs, 3 publications)^80,81,99 (Results Appendix B, Table B-8). Randomization occurred at the individual patient-level in all trials. Sample sizes ranged from 63 to 197 (total sample=702). Pain diagnoses included chronic LBP^{80,81,94–98} and FM^{63,64,76–78,87,99} in three trials each. Mean pooled pain duration in the FM trials was 13.4 years (range, 11.7 to 16.3 years) while in the two LBP trials that reported it, pain duration ranged from 9 months⁹⁷ to 6.9 years.^94–96,98 Mean ages ranged from 44 to 50 years. Across the FM trials, populations were almost entirely female (pooled 97%, range 91% to 100%) while there was more variability across the LBP trials (pooled 75% female, range 43% to 100%). Comorbidities were reported by three trials, and included hypothyroidism (12%), high blood pressure (10%), chronic obstructive pulmonary disease (12%), diabetes mellitus (3%), rheumatoid arthritis (3%), and others (42%) in one trial,^76–78 and smoking (4% to 6%) in two others.^94–98 In addition, exclusion criteria included major psychiatric disorders in two trials^80,81,99 and substance abuse disorders in one trial.^80,81 None of the studies specifically included Medicare patients. Fifteen percent of patients in one trial^76–78 were on disability and 6 percent^63,64,87 and 20 percent⁹⁹ were receiving workers’ compensation in two other trials, respectively. Four trials^{63,64,76–78,80,81,87,99} were conducted in Europe and two^94–98 in Iran. Two trials received government funding^{76–78,80,81} and two trials received university funding.^94–98 Funding was not reported in the remaining two trials.^63,64,87,99

Intervention durations ranged from 1 to 12 weeks. All but one CPMP^80,81 were considered low intensity (<20 hours per week) though the total hours varied widely across the trials. Program components were delivered individually (2 RCTs),^80,81,97 in groups (2 RCTs),^{63,64,87,94–96,98} and via a combination of the two (1 RCT);⁹⁹ the format could not be determined in the sixth trial.^76–78 Four trials^{63,64,76–78,87,94–98} were conducted in an outpatient setting and two^80,81,99 in an inpatient setting (3 rheumatology, 1 rehabilitation, and 1 pain clinic; unclear location in the last trial).

The content and delivery of program components varied substantially across the trials (Appendix E, Table E-2). Within each trial, patients in both groups received the same medications. Across the five trials that compared CPMP to pharmacologic therapy only, medication regimens included analgesics and/or NSAIDs in all trials as well as combinations of antidepressants in three,^{63,64,76–78,87,94–96,98} sedatives (e.g., benzodiazepine, chlordiazepoxide) in two,^63,64,87,97 and muscle relaxants,^94–96,98 omeprazole,^80,81 and tramadol^76–78 in one trial each. Three trials did not provide information regarding medication dosages.^{63,64,87,94–98} In the two trials evaluating CPMP versus a combination of pharmacologic therapy and physical activity medications included diclofenac, omeprazole, and acetaminophen in one^80,81 and antidepressants only in the other (type and dose not reported).⁹⁹ Both trials employed primarily passive physical therapy approaches (e.g., massage, transcutaneous electrical nerve stimulation [TENS]) with more active components not well described.^80,81,99

Four trials were rated fair quality^{63,64,87,94–99} and two poor quality^{76–78,80,81} (Appendix F, Table F-2). The major methodological limitation in the fair-quality trials was the inability to effectively blind patients and caregivers to the CPMP. Other methodological shortcomings included unclear randomization, unclear allocation concealment methods and high attrition.

Primary Outcomes

Pain

More CPMP patients achieved 30 percent or more improvement in pain (NRS, 0 to 10 scale) compared with those receiving conventional pharmacologic therapy alone in one fair-quality trial of patients with FM postintervention (22.2% vs. 6.7%, RR 3.3, 95% CI 1.3 to 8.4) and in the intermediate term (16.0% vs. 5.4%, RR 3.0, 95% CI 1.0 to 8.7).⁶⁴ Results at short term (13.6% vs. 10.8%, RR 1.3, 95% CI 0.53 to 3.0) and long term (8.6% vs. 0%, RR not calculable) were not statistically significant. CPMPs were associated with a moderate improvement in average pain (VAS or NRS 0 to 10 scale) postintervention across one fair-quality trial in FM and one poor-quality trial in LBP (2 trials, N=204, pooled difference −1.28, 95% CI, −2.14 to −0.63, I²=0%)^64,81 and a small improvement at intermediate term across one fair- and one-poor quality trial in FM (2 trials, N= 265, pooled difference −0.84, 95% CI −1.64 to −0.15, I²=0%),^64,76 (Figure 9). Medications in the fair-quality trial in patients with FM⁶⁴ included analgesics, antidepressants, benzodiazepines, and nonbenzodiazepine hypnotics and in the poor-quality trial in FM, amitriptyline, paracetamol, and tramadol;^76–78 the third poor-quality trial of LBP patients⁸¹ prescribed diclofenac, acetaminophen and omeprazole. The fair-quality trial, in patients with FM (N=155), reported moderate pain improvement postintervention (difference −1.20, 95% CI −1.78 to −0.62) but no clear difference was found between CPMP and pharmacologic therapy in the short (difference −0.40, 95% CI −0.98 to 0.18), intermediate (difference −0.60, 95% CI −1.20 to 0.0), or long term (difference −0.40, 95% CI −0.94 to 0.14)⁶⁴ (Figure 9).

For the comparison between CPMP and the combination of pharmacologic treatment and physical therapy, CPMP was associated with moderate improvements in Multidimensional Pain Inventory (MPI) pain intensity (differences −1.2 to −2.1 on a 0 to 6 scale) and MPI pain interference (differences −1.9 to −2.5 on 0 to 6 scale) at postintervention, intermediate, and long term in the fair-quality trial in patients with FM; only antidepressants were prescribed in this trial.⁹⁹ In contrast, the poor-quality trial in patients with LBP reported no difference between groups on VAS pain postintervention (difference 0.93, 95% CI −0.19 to 2.1, on a 0 to 10 scale);⁸¹ medications in this trial included diclofenac, acetaminophen, and omeprazole.

Figure 9

CPMP versus pharmacologic therapy alone: Pain. CI = confidence interval; CPMP = comprehensive pain management program; FM = fibromyalgia; LBP = low back pain; NR = not reported; NRS = numerical rating scale; Pharma = pharmacologic therapy alone; SD = (more...)

Function

More CPMP patients experienced 14 percent or greater improvement in function (FIQ Total 0 to 100 scale) compared with those receiving conventional pharmacologic therapy alone in one fair-quality trial (N=155) of patients with FM postintervention (64.2% vs. 24.3%, RR 2.6, 95% CI 1.7 to 4.1).⁶⁴ Improvement persisted to the short (48.1% vs. 23.0%, RR 2.1, 95% CI 1.3 to 3.4), intermediate (42.0% vs. 18.9%, RR 2.2, 95% CI 1.3 to 3.8) and long term (27.2% vs. 4.0%, RR 6.7, 95% CI 2.1 to 21.5). No improvement in function was seen with CPMP versus pharmacologic therapy postintervention on continuous outcomes (2 trials, N=204, pooled SMD −0.57, 95% CI −1.66 to 0.62, I²= 74.5%),^64,81 however heterogeneity was substantial across studies with the higher-quality trial in patients with FM having moderate improvement (difference −18.2, 95% CI −24.0 to −12.4, on the 0 to 100 FIQ).⁶⁴ The small poor-quality trial in patients with LBP found no difference (difference −0.24, 95% CI −4.2 to 3.8, on the 0 to 24 RMDQ).⁸¹ Heterogeneity may be due to differences in study quality, functional measures used, conditions (FM vs. LBP), program duration and delivery, and/or medications used. Analgesics (acetaminophen), antidepressants, benzodiazepines, and nonbenzodiazepine hypnotics were used in one trial;^63,64,87 diclofenac, acetaminophen, and omeprazole were used in the other.^80,81 CPMPs were associated with small improvements in function versus pharmacologic treatment in the short term (2 trials, N=342, pooled SMD −0.37, 95% CI −0.67 to −0.08, I²=0%),^64,94 intermediate term (3 trials, N= 453, pooled SMD −0.44, 95% CI −0.67 to −0.22, I²=0%),^64,76,94 and long term (2 trials, N=301, pooled SMD −0.46, 95% CI −0.76 to −0.16, I²=0%)^64,96 (Figure 10). Sensitivity analyses removing one poor-quality trial⁷⁶ at intermediate term did not impact the effect estimates. Similarly, sensitivity analyses using 12 months versus 30 months followup from one trial⁹⁶ in the long-term estimates lead to similar conclusions (see Appendix I, Figures I-34 to I-35 for sensitivity analyses). Mean differences in RMDQ (0 to 24 scale) were −1.6, −1.8, −3.1 in the short, intermediate, and long term (>12 months), respectively.⁹⁶ Mean differences in total FIQ (0 to 100 scale) were −9.1, −12.0 and −10.8 for these time frames, respectively.⁶⁴ Differences based on these scales suggest small to moderate improvement. Two fair-quality trials also reported on additional functional measures.^{63,64,87,94–96,98} CPMPs were consistently associated with at least small functional improvement at all time frames compared with pharmacologic therapy.

For the comparison between CPMP versus the combination of pharmacologic treatment and physical therapy, no differences in function between groups were reported in either study based on MPI total activity at any time frame (1 trial, difference ranges −0.27 to −0.22, on a 0 to 6 scale)⁹⁹ or on postintervention RMDQ (1 trial, difference 1.50 on a 0 to 24 scale).⁸¹

Figure 10

CPMP versus pharmacologic therapy alone: Function. CI = confidence interval; CPMP = comprehensive pain management program; FIQ = Fibromyalgia Impact Questionnaire; FM = fibromyalgia; LBP = low back pain; NR = not reported; Pharma = pharmacologic therapy (more...)

Opioid Use

Changes in opioid use were not reported in any of the trials comparing CPMP with pharmacologic therapy alone. One trial reported greater overall decrease in medication use for CPMP versus combined medication/physical therapy, indicating that overall opioid use was also reduced with CPMP but did not provide sufficient data for comparisons on use.⁹⁹

Key Points

• There were no differences in pain (on a 0 to 10 scale) or function comparing CPMP versus psychological therapy alone postintervention, intermediate term, and long term (SOE: low for all).
• One trial reported increased low back or radiating leg pain leading to withdrawal in three (5.5%) CPMP patients compared with no patient who received psychological therapy (SOE: insufficient).

Detailed Synthesis

Five RCTs compared CPMPs with psychological therapy alone (Results Appendix B, Table B-9; Appendix E, Table E-2).^{69,74,91,92,100,101} Randomization occurred at the individual patient-level in all trials. Sample sizes ranged from 36 to 138 (total sample=578). Pain diagnoses included chronic LBP (2 trials),^91,92,100 chronic nonspecific spinal pain (2 trials),^69,74 and nonspecific chronic pain (1 trials).¹⁰¹ Comorbidities were not reported and none of the studies specifically included Medicare patients. However, four trials reported on patients’ disability status with 38 percent and 24 percent receiving a full or partial sick leave/disability pension in one trial^91,92 and 21 percent currently on sick leave in a second trial;¹⁰¹ in the other two trials, inclusion criteria included current and continuous sick leave for at least 1 month (maximum 6 months) before study entry⁶⁹ and “at risk” for long-term disability.⁷⁴ Mean patient age ranged from 42 to 49 years. Intervention durations ranged from 4 to 10 weeks. Program components were delivered via a combination of individual and group formats in two trials^91,92,100 and in a group format only in the remaining three trials.^69,74,101 All CPMPs but one⁶⁹ were considered low intensity (<20 hours per week). Four trials had psychological components based on CBT and one¹⁰⁰ on behavioral modification.

Three trials were considered fair quality^69,74,91,92 and two poor quality^100,101 (Appendix F, Table F-2). The major methodological limitation in the fair-quality trials was the inability to effectively blind patients and caregivers to the CPMP. Other methodological shortcomings in the poor-quality trials included unclear allocation concealment methods and high attrition.

Primary Outcomes

Pain

There were no differences in pain improvement for CPMPs compared with psychological therapy alone postintervention (3 trials, N=259, pooled difference 0.03 on a 0 to 10 scale, 95% CI −0.30 to 0.31, I²=0.0%),^91,100,101 and at intermediate-term (3 trials, N=228, pooled difference −0.09, 95% CI −0.50 to 0.21, I²=0.0%)^91,100,101 and long-term (3 trials, N=256, pooled difference 0.05, 95% CI −0.35 to 0.47, I²=26.1%)^74,91,100 followup (Figure 11). Results from sensitivity analyses excluding the two poor-quality trials^100,101 at postintervention and intermediate term changed the effect estimates somewhat but did not change the conclusions (Appendix I, Figure I-36). Likewise, no differences were seen at any timepoint for other measures of pain (MPQ Pain Rating Index, VAS worst pain, number of pain free days in the past week) as reported by two trials^74,91 (Appendix E, Table E-2).

Figure 11

CPMP versus psychological therapy alone: Pain. CBNP = chronic back and neck pain; CI = confidence interval; CLBP = chronic low back pain; Combo = combination group and individual sessions; CPMP = comprehensive pain management program; CSP = chronic spinal (more...)

Function

There were no differences in function for CPMPs compared with psychological therapy alone at any time point: postintervention (3 trials, N=262, pooled SMD 0.10, 95% CI −0.23 to 0.36, I²=0%),^91,100,101 intermediate term (3 trials, N=231, pooled SMD 0.11, 95% CI −0.32 to 0.41, I²=0%),^91,100,101 and long term (3 trials, N=259, pooled SMD 0.16, 95% CI −0.18 to 0.48, I²=0%),^74,91,100 (Figure 12). Results from sensitivity analyses excluding the two poor-quality trials^100,101 at postintervention and intermediate term changed the effect estimates somewhat but did not change the conclusions (Appendix I, Figure I-37). There were also no difference between groups on a modified activities of daily living questionnaire in the long term in one fair-quality trial⁷⁴ and pain interference at postintervention and in the intermediate term in one poor-quality trial (Appendix B, Table B-9).¹⁰¹

Figure 12

CPMP versus psychological therapy alone: Function. CBNP = chronic back and neck pain; CI = confidence interval; CLBP = chronic low back pain; Combo = combination group and individual sessions; CPMP = comprehensive pain management program; CSP = chronic (more...)

Detailed Synthesis

One cluster RCT^41,110 conducted in the United States randomized 10 community-based primary care clinics to electronically receive patient-specific OA treatment recommendations made by the study team based on patient assessment and treatment guideline care algorithms (5 clinics) or to continue usual care (5 clinics). Although the primary care team was not multidisciplinary, patients could be referred for or receive care from providers from multiple disciplines. Patients with OA within each of the 10 clinics’ arms were randomized to a 12-month phone-based intervention focused on physical activity, cognitive behavioral strategies for pain management, weight management, or usual care resulting in 4 arms (N=537). This care was delivered by individuals with training in counseling and/or health education and behavioral change and was overseen by the co-investigators from multiple disciplines. Patient intervention participants also received educational materials, an exercise video, and a compact disc of relaxation exercises in addition to the phone-based intervention. Patients in the intervention arms (N=408) were primarily female (75%), White (59%), had knee OA (85%) with a mean symptom duration of 124.8 months, and mean age of 63 years; 20 percent rated their health as fair or poor. Patients with active psychosis, personality disorder or uncontrolled substance abuse disorder were excluded. The trial was considered fair quality due to lack of blinding and unclear concealment of treatment allocation (Appendix F, Table F-1).

In the cluster RCT, results comparing the provider plus patient intervention group with the provider intervention group alone and provider plus patient intervention versus patient intervention alone showed no difference in pain scores. The difference in change scores from baseline to postintervention (12 months) between the provider plus patient intervention group versus provider intervention only group for pain did not meet the threshold for a small effect (difference −0.3, 95% CI −0.8 to −0.2, on a 0 to 10 scale). Further for this comparison, there were no differences in changes scores for WOMAC function (0 to 68 scale, difference −2.5, 95% CI −5.0 to 0.0), or in the proportion of patients achieving 18 percent or more improvement from baseline (44% versus 35% based on author imputation). PHQ depression scores were also not different. Similarly, there were no differences in change scores for the comparison of provider plus patient intervention versus patient intervention only for pain (difference −0.05, 95% CI −0.45 to 0.55, on a 0 to 10 scale), WOMAC function (difference 0.80, 95% CI −1.8 to 3.4) or in the proportion of patients achieving 18 percent or more improvement from baseline (44% versus 49% based on author imputation). PHQ depression scores were also not different between groups (Appendix E, Table E-1). Authors reported that no study-related adverse events occurred.

One trial (N=109) conducted in Sweden compared the impact of adding Web-based behavioral change support to IPMP versus usual IPMP delivery in patients with persistent musculoskeletal pain.¹⁰⁹ Randomization in this trial occurred at the individual patient-level. For IPMP, a multidisciplinary team conferred with the patient to individualize treatment. In addition to the primary IPMP components (i.e., medication adaptation, psychological care including CBT, and physical activity), treatment could include acupuncture, TENS, manual therapy and others. IPMP in both groups involved a minimum of two or three treatment sessions a week for at least 6 weeks. The 8-module Web-based support program was available for 16 weeks and focused on enhancing patient physical and cognitive activity in their rehabilitation. Only 36 percent of participants opened all modules. Total program intensity was not reported but was considered low based on information provided. Patients were predominantly male (57%) with a mean age of 43 years. Information on race/ethnicity and comorbidities was not provided. The trial was rated fair due to lack of blinding and unacceptable attrition (27% at 12 months) (Appendix F, Table F-1).

There were no differences between those who did and those who did not have Web-based support in the proportions of patients meeting 30 percent or more improvement on either VAS pain (0 to 10 scale) in the short (22% vs. 23%; RR 1.00, 95% CI 0.45 to 2.23) or long term (28% vs. 22%; RR 1.23, 95% CI 0.56 to 2.67) or function based on 30 percent or more improvement in Pain Disability Index (PDI, 0 to 70 scale) in the short (20% vs. 24%, RR 0.91, 95% CI 0.40 to 2.07) or long term (31% vs. 30%; RR 1.04, 95% CI 0.54 to 2.01). No differences between groups in average effects for either pain or functional measures were seen. Similarly, there were no differences between groups on mean differences in these measures at either time or on individual SF-36 subscales. Authors did not report on harms (Results Appendix B, Table B-10).

Key Points

• In one fair-quality trial (N=75), CPMP with greater total hours was associated with moderate improvement in pain and function over the short and long term compared with CPMP with fewer hours. Applicability to the Medicare population might be limited due to high program intensity, inclusion of work-related therapy, and young age.
• Across three lower-intensity trials (total N=307), two fair and one poor quality, no differences in pain or function were seen for longer versus shorter CPMPs postintervention and in the intermediate term across different pain diagnoses.
• No trial-related adverse events were reported across two fair-quality trials, although patients in both groups experienced occasional mild increases in pain after exercise sessions.

Detailed Synthesis

Four RCTs (reported in 7 publications)^{56–59,88,111,112} directly compared CPMPs with greater versus fewer total program hours (Results Appendix B, Table B-11; Appendix E, Table E-2). Sample sizes ranged from 44 to 153 (total sample=328). The weighted mean age was 42.3 years and 42 percent of the patients were male. Pain diagnoses included chronic LBP (with or without referred leg pain) in two trials,^56–59,112 FM in one trial,⁸⁸ and mixed musculoskeletal (mostly neck/shoulder, low back, and lower extremity) pain in the fourth trial.¹¹¹ The duration of symptoms was 6 months or longer in one trial,^56–59 greater than 1 year in the majority of patients (74%) in another,¹¹¹ and a mean 7.5 and 8.1 years in two trials.^88,112 None of the studies specifically included Medicare patients. However, 64 percent of patients were receiving a disability or sickness benefit in one trial,¹¹² 77 percent were incapable of working due to their condition in a second,^56–59 and 49 percent were on full or partial sick leave in another trial.¹¹¹ None of the trials reported comorbidities with the exception of smoking (64%) in one trial.^56–59 In one trial, 13 percent had previous spinal surgery.¹¹² No trial reported opioid use at baseline. One trial received no funding,⁸⁸ one received hospital funds,¹¹¹ one received nonprofit funding,^56–59 and one trial¹¹² did not report their source of funding. Three trials were conducted in Europe^{56–59,111,112} and one⁸⁸ in Turkey.

Treatment components were delivered to patients in group sessions in two trials,^56–59,112 individually in one trial,¹¹¹ and via a combination of both in one trial.⁸⁸ All programs were delivered on an outpatient basis, regardless of the group. Only one trial was considered high intensity with 135 total hours (39 hours/week over 3 weeks) in the intervention group (compared with 24 hours in the CPMP of shorter duration). This trial demanded a high level of physical activity, included work-related therapy, and enrolled younger participants (mean 42 years); given the structure and demands of the program this trial will be reported separately below. Across the remaining three lower-intensity programs, the total number of hours ranged from 60 to 70 (over a range of 1 to 10 weeks) and from 10 to 15 (over a range of 2 days to 1 week) across programs with greater versus fewer total hours, respectively, in two trials. The third trial intended to compare two programs of different “dosages” (a conventional program and a “short form”) but they ultimately ended up with comparable mean durations (11.7 vs. 10.8 weeks) and mean number of patient contact hours with providers (30.7 vs. 29.8).¹¹¹ Two trials included a work rehabilitation component in their program.^56–59,111

Of the four trials, three were rated fair quality^{56–59,88,111} and one poor quality¹¹² (Appendix F, Table F-2). The major methodological limitation in the fair-quality trials was the inability to effectively blind patients and caregivers to the interventions; in addition, two of the trials had unacceptable levels of attrition. Other methodological shortcomings in the poor-quality trial included unclear randomization, unclear allocation concealment methods and lack of an intention-to-treat analysis.

Key Points

• Evidence comparing CPMPs conducted in an inpatient versus outpatient setting from four poor-quality trials is insufficient to draw conclusions about benefits or safety.

Detailed Synthesis

Four RCTs (reported in 6 publications) compared CPMPs conducted in an inpatient versus an outpatient setting (Results Appendix B, Table B-11; Appendix E, Table E-2).^{66,67,82,83,86,108} Pain diagnoses included mixed chronic pain (primarily back, neck, head, arms and legs) in two trials^82,83,108 and chronic LBP in two trials.^66,67,86 Mean duration of symptoms was 14.4 years in one trial^66,67 and 8.8 years in another;¹⁰⁸ in a third trial symptom duration was 3 months or longer⁸⁶ and in the fourth, ranged from 6 months to over 20 years.^82,83 The weighted mean age of participants was 44.6 years (oldest population mean 50 years)¹⁰⁸ and 56.9 percent were male; across the two trials that reported race, only 12 percent of participants were non-White.^82,83,108 While none of the trials specifically included Medicare patients, 60 percent were receiving disability income in one trial.¹⁰⁸ In two other trials, patients had numerous incidences of sick leave due to their pain condition prior to study enrollment.^66,67,86 None of the trials reported comorbidities but one reported that 65 percent of patients used opioids, 58 percent reported excess drug use, and 40 percent had undergone at least one prior surgery.¹⁰⁸ Trials were conducted in New Zealand,^82,83 France,⁸⁶ Finland,¹⁰⁸ and the United Kingdom.^66,67 Two trials^82,83,108 received government funding and two trials^66,67,86 received funding from nonprofit organizations.

Sample sizes ranged from 52 to 306 (total sample size=551). In the inpatient groups, intervention durations ranged from 3 to 5 weeks and were considered higher intensity (≥20 hours per week; range, 30 to 40 hours per week). In the outpatient groups, intervention durations ranged from 5 to 9 weeks and were considered lower intensity (<20 hours per week; range, 2 to 11 hours per week) and content differed for the inpatient versus outpatient programs. Two programs^66,67,86 took place in a rehabilitation clinic, one^82,83 in a pain clinic, and one¹⁰⁸ in a general hospital ward. Program components were delivered in groups sessions in three trials^{66,67,82,83,108} and in a combination of individual and group sessions in one.⁸⁶ A work rehabilitation component was included in both the inpatient and outpatient programs in one trial,^66,67 and only in the inpatient group for another trial.^82,83 The other two trials did not include such a component. In one trial, both groups received multiple reinforcement followup sessions 1.5 years after the initial program.^66,67

All four trials were rated poor quality (Appendix F, Table F-2). The major methodological limitations included the inability to effectively blind patients and caregivers to the interventions, unclear allocation concealment methods, unacceptable levels of attrition, and lack of an intention-to-treat analysis.

Key Points

• There were no differences in pain or function at postintervention, intermediate- and long-term followup across two trials (N=63) comparing CPMP with and without additional psychological components.
• There were no differences in function at postintervention, short-, intermediate- and long-term followup in one trial (N=78) comparing CPMP with and without additional physical components.
• None of the trials provided data on harms.

Detailed Synthesis

Three RCTs^51,113,114 assessed the effectiveness of adding additional program components to a standard CPMP (Results Appendix B, Table B-11; Appendix E, Table E-2). Sample sizes ranged from 19 to 94 (total sample=158). Two trials assessed the addition of psychological components. One added operant conditioning, relaxation, and biofeedback¹¹³ and the other⁵¹ included spouses to assist in the participants’ training. The third trial assessed the addition of psychomotor therapy.¹¹⁴ Two trials^51,113 included patients with chronic low back pain and one trial¹¹⁴ included patients with any musculoskeletal pain. Trials delivered the program in the outpatient setting in two trials^51,114 and the inpatient setting in one trial.¹¹³ Program lengths ranged from 3 to 12 weeks. Program components were delivered in group sessions in two trials^51,114 and via a combination of group and individual session in the third trial.¹¹³ Two trials^113,114 administered a high-intensity program (≥20 hours/week or >80 hours total) and one trial⁵¹ administered a low-intensity program (<20 hours/week or ≤80 hours total). The weighted mean age was 41 years and 18 percent of patients were male (across 2 trials).^113,114 Symptom duration was reported by two trials; one reported median of 74 months⁵¹ while in the other, 75 percent of patients had had symptoms for two or more years.¹¹⁴ None of the trials specifically included Medicare patients. However, one trial only included patients who were disabled and not working due to pain (for at least 3 but no more than 30 months),¹¹³ while the other two did not provide information on disability status. No trial reported on comorbidities or opioid use at baseline, but two trials^51,113 did exclude patients with coexisting psychiatric morbidity and one¹¹³ stated that no patient entered the program using narcotics. Two trials reported receiving funding from government sources;^51,113 the third did not report a funding source. Trials were conducted in the United States,¹¹³ Iran,⁵¹ and the Netherlands.¹¹⁴

Two trials were rated fair quality^113,114 and one poor quality⁵¹ (Appendix F, Table F-2). The major methodological limitation in the fair-quality trials was the inability to effectively blind patients and caregivers to the interventions. Other methodological shortcomings in the poor-quality trial included inadequate randomization methods, unclear allocation concealment methods and lack of an intention-to-treat analysis.

Key Points

• CPMP delivered based on a functional capacity preassessment tool was associated with a small improvement in function in one trial at long term, but there was no difference in pain or function in a second trial that employed a biopsychological preintervention assessment.
• None of the trials provided data on harms.

Detailed Synthesis

Two RCTs compared CPMPs conducted with versus without a pretreatment assessment to help inform subsequent therapy (Results Appendix B, Table B-10; Appendix E, Table E-2).^115,116 Sample sizes ranged from 207 to 222 (total sample=429). The mean age in one trial was 46 years and the median age in the other trial was 40 years. Across the two trials, 54 percent of patients were male. Neither trial reported on race/ethnicity, comorbidities, or opioid use at baseline. Both trials included patients with various chronic musculoskeletal pain conditions (primarily back pain in one¹¹⁶). Only one trial reported duration of symptoms (median of 18 months).¹¹⁵ Neither trial specifically included Medicare patients. In one trial, 81 percent of patients were on sick leave at the start of the trial.¹¹⁶ One trial¹¹⁶ utilized a preintervention assessment in the intervention group designed to evaluate patients’ functional capacity in relation to their workplace (or intended workplace) based on a kinesiophysical approach. The CPMPs in this trial were 3-week, lower intensity (50-60 hours total, 3-4 hours/day) inpatient programs and treatment was delivered on an individual basis. The second trial¹¹⁵ employed a preintervention assessment in the intervention group based on a multidisciplinary, biopsychosocial approach to guide subsequent therapies. The CPMPs in this trial were outpatient programs delivered to groups of patients; program intensity was unclear in this trial but considered to be lower intensity. One trial received nonprofit funds¹¹⁶ and the other did not report funding information. Both trials were conducted in Europe.

Both trials were rated fair quality (Appendix F, Table F-2). The major methodological limitation was the inability to effectively blind patients and caregivers to the interventions. In addition, one trial had a high attrition rate.¹¹⁶

Key Points

• CPMP using a function-centered approach was associated with small improvements in pain and function postintervention, but not pain in the short term, compared with CPMP using a pain-centered approach in one trial.
• There were no differences in pain or function postintervention or at intermediate-term followup comparing CPMP using an “exposure in vivo” versus a graded activity approach in one trial.
• One patient (2%) in the graded activity group (vs. none in the “exposure in vivo” group) deteriorated during treatment (i.e., treatment counterproductive).

Detailed Synthesis

Two RCTs (reported in 3 publications) assessed different approaches to CPMPs (Results Appendix B, Table B-11; Appendix E, Table E-2).^117–119 One trial compared a function-centered versus a pain-centered approach to therapy^117,118 while the other compared an “exposure in vivo” approach (i.e., systematically reducing pain-related fear using Pavlovian conditioning and CBT) versus a graded activity approach (i.e., increase healthy behavior via operant learning principles).¹¹⁹ Sample sizes ranged from 85 to 174 (total sample=259). Both trials included patients with chronic LBP, with pain radiating to the legs in 83 percent of patients in one trial^117,118 and 98 percent in the other trial.¹¹⁹ Weighted mean age was 43 years and 70 percent of patients were male. Neither trial specifically included Medicare patients. In one trial, 54 percent of patients were on sick leave or in receipt of a disability pension¹¹⁹ and in the other, patients were required to have a minimum of 6 weeks of sick leave in the 6 months prior to enrollment.^117,118 Neither trial provided information on comorbidities, however, one trial excluded patients with substance abuse, medical disorders or cardiovascular disease preventing physical exercise, and serious psychopathology.¹¹⁹ Across the two trials, 73 percent of patients were taking pain medication (not otherwise specified) at baseline and in one trial 31 percent of patients had previous back surgery.¹¹⁹ One trial delivered a high-intensity program (≥20 hours/week or >80 hours total) in an inpatient setting for 3 weeks.^117,118 The program in the other trial was considered low intensity (<20 hours/week or ≤80 hours total) and was delivered over 8 to 12 weeks in an outpatient setting.¹¹⁹ In both trials, program components were delivered to patients individually. One trial received government funding¹¹⁹ and the other did not report funding information. Both trials were conducted in Europe.

Both trials were rated fair quality (Appendix F, Table F-2). The major methodological limitations were the inability to effectively blind patients and caregivers to the interventions and high attrition rates.

Key Points

• Evidence comparing CPMPs delivered in a group versus an individual format from one small poor-quality trial did not permit conclusions about effectiveness. This trial did not report data on harms.

Detailed Synthesis

One RCT (N=50) conducted in the United Kingdom compared outpatient CPMPs delivered in a group versus an individual format (Results Appendix B, Table B-11; Appendix E, Table E-2).¹¹² Minimal information on the program characteristics were provided. All patients had chronic LBP (mean duration 8.1 years). Population demographics include data on patients included in another study conducted by the same authors. Including those patients, mean age was 42 years and 41 percent were male. Most patients (64%) were in receipt of sickness or disability benefit and 13 percent had undergone spinal surgery. Funding information for the trial was not reported.

This trial was rated poor quality due to major methodological limitations: unclear randomization, unclear allocation concealment methods, the inability to effectively blind patients and caregivers to the interventions and an unacceptable attrition rate (Appendix F, Table F-2).

Key Points

• There were no differences in pain or function postintervention or at long-term followup comparing a CPMP with and without additional booster sessions in one fair-quality trial. This trial did not report data on harms.

Detailed Synthesis

One RCT (N=232) conducted in Germany compared a 4-week CPMP with and without the addition of seven, 20-minute booster sessions over the course of a year⁷⁵ (Results Appendix B, Table B-11; Appendix E, Table E-2). The program took place in an inpatient setting and was delivered via a combination of group and individual sessions. All patients had chronic back pain (duration not reported). The mean patient age was 49 years and 23 percent were male. The trial did not report on race, comorbidities or opioid use at baseline. Funding was provided, in part, by government.

This trial was rated fair quality due to the inability to effectively blind patients and caregivers to the interventions (Appendix F, Table F-2).