Economic analysis of patient-related effects of an interdisciplinary pain self-management program

Supplemental Digital Content is Available in the Text.

Our economic evaluation indicates that an interdisciplinary intervention such as Active Day Patient Treatment can improve chronic pain patients' labour force participation, reduce healthcare utilization costs, and improve health outcomes.

Abstract

Active Day Patient Treatment (ADAPT) is a well-established 3 week intensive cognitive-behavioural, interdisciplinary pain management program for patients with disabling chronic pain. The aim of this analysis was to conduct an economic analysis of patient-related effects of ADAPT using hospital administrative data, specifically, to compare the costs and health outcomes for patients 1 month after participating in the program, with the preprogram period when they were receiving standard care. This retrospective cohort study included 230 patients who completed ADAPT (including follow-ups) between 2014 and 17 at the Pain Management and Research Centre at the Royal North Shore Hospital in Sydney, Australia. Data on pain-related healthcare utilization and costs before and after the program were assessed. Primary outcome measures were labour force participation for patients' average weekly earnings and cost per clinically meaningful change in Pain Self-efficacy Questionnaire, Brief Pain Inventory (BPI) Severity, and BPI interference scores (n = 224). We estimated patients, on average, earned $59 more each week at 1 month follow-up compared with baseline. The cost per clinically meaningful change in pain severity and interference score based on the BPI severity and BPI interference were AU$9452.32 (95% CI: $7031.76-$12,930.40) and AU$3446.62 (95% CI: $2851.67-$4126.46), respectively. The cost per point improvement and per clinically meaningful change in the Pain Self-efficacy Questionnaire were $483 (95% CI: $411.289-$568.606) and $3381.02, respectively. Our analysis showed a better health outcome, reduced healthcare services' cost, and reduced number of medications taken 1 month after participating in ADAPT.

1. Introduction

Chronic pain is a complex and pervasive health condition which leads to increased healthcare costs and productivity loss among working age adults. Chronic pain commonly affects peoples' ability to perform day-to-day activities and limits their capacity to remain in paid employment.^{7–11,43,49,56} Patients with chronic pain often also experience poor mental health,⁴⁵ such as heightened depression and anxiety symptoms. The costs of chronic pain place a significant public health burden on countries.⁴⁵ For example, in Australia, chronic pain affected up to 3.2 million adults in 2018, costing the economy AU$12.2 billion in healthcare costs and $48.3 billion in productivity losses.¹⁷ In the United States, about 50.2 million adults suffered from chronic pain, resulting in a significantly higher number of missing days at work due to chronic pain (10.3 days vs 2.8) and at a cost to the economy of $US300 billion in 2019.⁵⁹

Previous studies indicate that more structured and intensive interdisciplinary pain management interventions for chronic pain patients with disability were more efficacious and had better functional outcomes than less intensive versions of the same approaches. Many patients gradually resume activities of normal life, despite persisting pain, after participating in such intensive pain management programs.^22,23,35

A systematic literature review of the cost-effectiveness of multidisciplinary pain management programs¹² suggested that although relatively few studies have examined this issue, some have reported favourable cost-effectiveness outcomes.^28,29 The reviewed studies encouraged additional rigorous economic analysis of such interdisciplinary pain management programs.

Active Day Patient Treatment (ADAPT) is a 3-week interdisciplinary intensive cognitive–behavioural, pain management program for patients with disabling chronic pain delivered in a tertiary level hospital setting in Sydney, Australia. Patients are referred by their general practitioner (GP) or medical specialist, but all undergo an interdisciplinary assessment by a pain specialist, clinical psychologist or psychiatrist, and a physiotherapist to determine their suitability for the program. Nicholas et al.^35,37 have reported that pain, disability, and depressive symptoms reduced among patients after completing ADAPT, whereas self-efficacy for functioning despite pain improved.^37,41 However, the economic analysis of patient-related effects of the ADAPT program has not been assessed. A health economic analysis estimating patients' ability to work, work status (ie, working full-time or part-time or unemployed), and cost and health outcomes can help to inform health policy makers when making resource allocation decisions.^57,58

In this paper, we present an economic analysis of the ADAPT program taking account of healthcare services utilization cost and Brief Pain Inventory (BPI)-pain intensity (severity) and the impact of pain on functioning (interference)^14,54 and Pain Self Efficacy Questionnaire (PSEQ)³⁶ outcomes. The paper also presents the change in labour force participation (LFP) of patients after attending the ADAPT program for average weekly earnings. Assessment measures were completed by patients before and after the program (at 1-month follow-up).^1,2 We compared patients' pretreatment and 1-month follow-up data for this analysis.

2. Methods

2.1. Participants and intervention

The ADAPT program is focused on improving patients' self-reliance in activities of daily living by encouraging the learning and implementation of skills to facilitate more effective pain management at home and work with active family involvement. This is a high-intensity, 120 hours program (15 × 8-hour days over 3 weeks) based on cognitive-behavioural methods implemented by all members of the multidisciplinary team.^39,46 The program offers pain self-management skills training, sleep and stress management, as well as supervised exercise and activity upgrading in a group setting. Each group comprises about 10 patients. The RNSH Pain service offers 12 to 14 such programs per year. The ADAPT team also helps patients seeking to return to work (RTW) with their RTW planning by meeting with rehabilitation providers, where available, during the program.^38,39 The ADAPT team also assists patients to greatly reduce their use of opioid medication while they learn more effective pain self-management strategies.³⁸ Patients are followed up at 1 month and 3 months after completing the program. However, 3 months follow-up data were available only for a 14% of the included patient cohort, and thus, we did not analyse 3 months' data in the current study because of a small sample size. A detailed description of this multimodal program has been published elsewhere.³⁹

2.2. Study design

This is a retrospective cohort study of 230 patients who completed the ADAPT program between 2014 and 2017. We conducted an economic analysis of ADAPT using patient level data.

Ethical approval was obtained from the Northern Sydney Local Health District HREC (Human Resource Ethics Committee), and consent was obtained from all participating patients.

2.3. Data

Routinely collected health-related data are used increasingly for research (amongst other) purposes and are collated within administrative or clinical (disease-based or treatment-based) databases. The hospital administrative data used in this study were collected from the electronic Persistent Pain Outcomes Collaboration (ePPOC) database. electronic Persistent Pain Outcomes Collaboration is an Australian initiative that has become a valuable resource for pain researchers and clinicians because it was established in 2013, and normative data from this data set have been published.⁴⁰ At the time of writing, over 90 adult and paediatric pain management services had either joined or were in the process of implementing ePPOC. ADAPT patients' clinical and healthcare resource use data are administered electronically as part of the ePPOC initiative.

2.3.1. Healthcare utilization and cost

Data related to the cost of the ADAPT program and patients' healthcare services utilization over the prior 3 months (such as the number of visits to a general practitioner, pain specialists, allied health professionals such as physiotherapists, hospital emergency department visits, hospital admissions, and diagnostic tests) were collected from the RNSH patient administrative data. The total average costs of healthcare services utilization included workers' compensation insurance and government expenditure as patients were funded by either Medicare or workers' compensation (or motor-accident) insurance. Costs of Medicare items and workers' compensation items were collected from publicly available Medicare data and data available in the Independent Hospital Pricing Authority and State Insurance Regulatory Authority, New South Wales (SIRA NSW) websites. Costs were calculated as per the 2019 schedule.^4,26,33

2.3.2. Labour force participation

We used patients' labour force status and work hours from the ePPOC database for our labour force participation analysis. Patients' preprogram work status was recorded in the ePPOC database as “full time employment,” “part time employment,” “retired,” “unemployed due to pain,” “unemployed (not due to pain),” “home duties,” “on leave from work,” “student,” “voluntary work,” “retraining” or “limited hours and/or duties,” and the number of hours worked for those in “full-time employment,” “part-time employment,” and “limited hours and/or duties” at preprogram and 1-month follow-up. Average weekly earnings data from the Australian Bureau of Statistics were used to calculate the change in patients' total average weekly income.³

2.3.3. Pain-related measures

2.3.3.1. Brief Pain Inventory

The BPI (short-form) has been widely used in the clinical pain management setting to measure pain in different conditions such as cancer, musculoskeletal conditions, and depressive disorders and has been shown to have validity and reliability.^6,13,14,51 The BPI short-form uses a 24-hour recall period. A 9-item self-administered questionnaire was used to measure ADAPT patients' pain severity and pain interference scores. Of these 9 items, items 2 to 6 were used to measure intensity or severity of pain over time: pain at its “worst,” “least,” “average,” and “now” (current pain). Seven items (item 9A-G) were used to capture how much pain interfered with various daily activities, including general activity, walking, work, mood, enjoyment of life, relations with others, and sleep.¹³ The 4 pain severity items and 7 pain interference items were rated on 0 to 10 scales. The pain severity scales are 0 = “no pain” and 10 = “pain as bad as you can imagine”. Any improvement ≥ 1 point is clinically significant or improvement of more than 10% from baseline score is considered to be clinically significant. The interference anchors are 0 = “no interference” and 10 = “interferes completely”.⁵¹ Any improvement ≥ 1 point is clinically significant. The BPI has been used specifically for working age adults in clinical and observational research.^5,30,47,51

2.3.3.2. Pain Self-Efficacy Questionnaire

Self-efficacy beliefs have a strong association with individual's capacity to work.^2,32,53 The 10-item PSEQ was used to measure patients' confidence performing day-to-day activities despite pain including household chores, paid and unpaid work, ability to enjoy leisure time, and their ability to cope without medication. At least 9 of 10 items must be complete for the PSEQ total to be valid. These items were rated on 0 to 6 scales where 0 = “not at all confident” and 6 = “completely confident”.³⁴ Any improvement by 7 points is clinically significant. Although ADAPT may help patients to return to work, the main aim of the program is to improve patients' self-efficacy for functioning despite pain as measured by the PSEQ.

2.4. Economic analysis

2.4.1. Labour force participation

We conducted a labour force participation analysis based on patients' participation in paid employment. In this current study, complete data on full-time, part-time, and limited hours were used to calculate labour force participation for weekly income and productivity gain (hours worked per week per patient) before completing ADAPT. Patients who were working part time were, on average, working 20 hours per week before ADAPT. We conducted mean imputation for missing part-time hours' data at 1-month follow-up. The Australian Bureau of Statistics reported that full-time workers were working on average 38 hours per week, and their average weekly income was $1634.7 in May 2019.³ We assumed patients were working 8 hours per week where data were not available (n = 7) for “limited hours of work”, similar to the study by Lambeek et al.²⁹

2.4.2. Health economic analysis

Outcomes were based on patients' work status or LFP and average weekly earnings and healthcare services utilization cost. We calculated an additional cost per clinically meaningful change in pain score based on the BPI severity and the BPI interference scales and a cost per clinically meaningful change in the PSEQ using patient-level hospital administrative data. We also used the average healthcare utilization cost (HCU) to obtain the additional cost per clinically meaningful change in BPI severity, BPI interference, and PSEQ score (n = 224). Mean (SD) scores were calculated for pain severity (0-10), interference (0-10), and pain self-efficacy (0-60). At the same time, we estimated reduction in patients' number of medications taken for pain.² Clinically meaningful changes on these measures have previously been described (Dworkin et al., 2008; Tardif et al., 2016; Dube et al., 2021).^20,21,52 The additional cost per clinically meaningful change in pain score and the cost per clinically meaningful change in PSEQ were calculated as additional cost per improvement in pain score after ADAPT measured as one unit where improvement for the BPI severity, BPI interference, and 7-point improvement gained for the PSEQ.

The additional cost for a clinically meaningful improvement following ADAPT was derived as follows.

2.4.2.1. BPI Severity (or BPI Interference)

Additional cost per clinically meaningful change in BPI severity (or BPI interference) =

([C_ADAPT + HCU_post-ADAPT] − HCU_pre-ADAPT)/(BPI_post-ADAPT − BPI_pre-ADAPT)

Where,

C_ADAPT = intervention cost of ADAPT.

HCU_post-ADAPT = healthcare utilisation cost (1 month after ADAPT);

HCU_pre-ADAPT = healthcare utilisation cost (before ADAPT)

BPI_post = BPI severity (or BPI interference) score (1 month after ADAPT)

BPI_pre = BPI severity (or BPI interference) score (before ADAPT)

2.4.2.2. Pain Self-Efficacy Questionnaire

Additional cost per clinically meaningful change in PSEQ =

([C_ADAPT + HCU_post-ADAPT] − HCU_pre-ADAPT)/([PSEQ_post-ADAPT − PSEQ_pre-ADAPT]/7)

Where,

C_ADAPT = intervention cost of ADAPT

HCU_post-ADAPT = healthcare utilisation cost (1 month after ADAPT);

HCU_pre-ADAPT = healthcare utilisation cost (before ADAPT)

PSEQ_post = PSEQ score (1 month after ADAPT)

PSEQ_pre = PSEQ score (before ADAPT)

From the additional cost per clinically meaningful change in pain scores (and PSEQ), we estimated the additional costs for clinically significant improvement in pain (BPI severity and interference) and self-efficacy (PSEQ). 95% confidence intervals (CI) for the additional cost per clinically meaningful change in pain scores (and PSEQ) analysis were estimated using bootstrapping techniques by generating 1000 replicated data sets.

We used bootstrapped data to estimate cost effectiveness planes (CE planes) and cost-effectiveness acceptability curves (CEAC).^28,29,57 To assess the variability of this estimate, we plotted the changes in costs and changes in pain scores on the CE planes (Figs. 1A-C).

Figure 1.

Cost-effectiveness plane based on BPI severity, BPI interference, and PSEQ scores, before and 1 month after participating the ADAPT-healthcare perspective. (A) BPI severity (0-10). (B) BPI interference (0-10). (C) PSEQ (0-60).

2.4.3. Statistical analysis

We used a paired sample t test to analyse if there was an improvement in pain outcomes 1 month after participating in the ADAPT program. Paired sample t tests were also undertaken to compare changes in the number of hours worked per week, reduction in all medications taken for pain, and healthcare utilization at preprogram and 1 month after completing ADAPT. Data processing was undertaken in Microsoft Excel with bootstrapping performed in SAS version 9.4.

3. Results

3.1. Baseline characteristics

Of the 230 patients' data assessed as potentially suitable for our retrospective analysis, 6 patients did not have complete data for either LFP or healthcare utilization, leaving a total sample of 224 for the final analysis. We compared patients' pretreatment and 1-month follow-up data for this analysis. Participants were adults aged 16 to 91 years (56% female and 44% male) with pain lasting more than 3 months. We included all patients as some of them were still working after the retirement age and covered by workers' compensation.

Table 1 shows characteristics of the participants.

Table 1.

Baseline (pre-ADAPT) characteristics and pain score.

Age (n = 227), mean (SD) = 46 (13) (min 16, max 91)
Age group (y)	n (%)
16-20	5 (2)
21-35	45 (20)
36-50	93 (40)
51-65	75 (32.5)
66-80	11 (5)
81-91	1 (0.5)
Sex (n = 227)	n (%)
Female	126 (56)
Male	101 (44)
Employment (n = 227)	n (%)
Full time (FT)	42 (18.5)
Home duty (HD)	10 (4.4)
Limited hours (LH)	24 (10.6)
Leave due to pain (LP)	21 (9.3)
Part time (PT)	23 (10.1)
Retired (ret)	20 (8.8)
Retraining (retrain)	5 (2.2)
Student (ST)	4 (1.7)
Unemployed due to pain (UP)	82 (36.1)
Volunteer (vol)	1 (0.4)
Number of analgesics (n = 224), mean (SD) = 3 (1.48)
BPI severity (scale 0-10) (n = 227), mean (SD) = 5.99 (1.5) min 2.25, max = 10)
BPI interference (scale 0-10) (n = 227), mean (SD) = 7.12 (1.8) (min 1.28, max = 10)
PSEQ (scale 0-60) (n = 227), mean (SD) = 19.81 (11.6) (min 0, max 60)

3.2. Health outcomes

Health outcome data including BPI severity and interference and PSEQ scores were available for 227. Of these, cost data were available for 224 patients. Table 2 shows changes in BPI severity and interference and PSEQ scores between pre-ADAPT and 1-month follow-up. Patients showed improvements on all the pain measures, and these changes reached both clinical and statistical significance.

Table 2.

Effectiveness of ADAPT at 1-mo follow-up (n = 227).

Pain scale	Mean (SD)	Change in pain scale	Clinical effectiveness (Y/N)	Statistical significance (Y/N)
BPI* severity (0-10)	5.15 (1.9)	0.83	>10% improvement on average/>1 point improvement (Y)	P < 0.05 (Y)
BPI* interference (0-10)	4.9 (2.5)	2.2	>1-point improvement (Y)	P < 0.05 (Y)
PSEQ† (0-60)	35.64 (14.67)	−15.86	>7-point improvement (Y)	P < 0.05 (Y)

^*Brief Pain Inventory.

^†Pain Self-efficacy Questionnaire: higher score indicates better outcome; therefore, negative change in pain scale indicates lower score at baseline and improvement at 1-month follow-up.

There was a reduction in mean BPI severity scores at 1-month follow-up (mean difference = 0.83; [P < 0.05]). More than 1-point improvement in pain score or improvement of more than 10% from baseline score is considered to be clinically significant. Based on our calculation, on average, there was a 13.93% improvement in the mean pain score, ie, a clinically significant improvement. Of the 227 patients, 92 patients (40.5%) had clinically significant improvements in the BPI severity score. Of these 92 patients, 65 patients (28.6% of the 227) had ≥10% improvement in the BPI severity score, indicating a minimally important change; 20 patients had ≥30% improvement in their score, indicating a moderate clinically important change, and 7 patients had ≥50% improvement in BPI Severity score, indicating a substantial clinically important change. By contrast, 72 patients (31.7% of the 227) had less than 10% improvement, whereas 63 patients (27.8% of the 227) did not have an improved BPI severity score 1 month after participating in ADAPT (ie, no change).

The mean BPI interference was also lower at 1-month follow-up (mean difference = 2.2; [P < 0.05]). Any improvement ≥ 1 point is clinically significant; therefore, the reduction in the pain interference score was clinically significant. Mean PSEQ scores also improved on average at 1-month follow-up (mean difference= −15.87; (P < 0.05). Any improvement by 7 points is clinically significant. On the PSEQ, higher scores indicate better outcomes. At the individual patient level, a PSEQ score of >40 indicates patients should be able to RTW. Available data indicated that 8 patients had PSEQ scores >40 before participating in ADAPT. Our calculation showed the number of patients with a PSEQ > 40 increased to 94 (41.4% of the 227) 1 month after participating in ADAPT. The PSEQ score remained unchanged for 52 patients (22.9% of the 227).

3.3. Labour force participation

Of 224 patients, 33 patients reported that they were on sick leave pre-ADAPT. Thirty seven patients continued to maintain same work status (full-time or part-time or limited hours) 1 month after participating in ADAPT. Table 3 shows number of patients in paid (full-time or part-time or limited hours) and unpaid work before ADAPT and at 1-month follow-up. On average, patients (n = 224) were estimated to earn $59 more per week at 1-month follow-up compared with baseline. On average, there was 1.4 hours productivity gain each week after participating in ADAPT.

Table 3.

Number of patients in paid (full-time/part-time/limited hours) and unpaid work before ADAPT and at 1-mo follow-up (n = 224).

Work status	Before ADAPT	At 1-month follow-up	Average number of hours worked	Average weekly earnings
Full-time (FT)	36	45	38	$1658.70*
Part-time (PT)	27	29	20	$873.00
Limited hours (LH)	27	22	8	$349.20
Not in paid work (HD/LP/Ret/Retrain/ST/UN/UP/Vol)†	134	128	0	0

^*Average Weekly Earnings, Australia (Reference period May 2019) Australia 2019 (Australian Bureau of Statistics).

^†HD, house duty; LP, leave pain; Ret, retired; Retrain, retraining; ST, student; UN, unemployed; UP, unemployed due to pain; Vol, voluntary work.

We further performed 2 subgroup analysis, one focusing on the working age population and another on those patients who were unemployed or on leave due to pain before ADAPT. When the calculation was restricted to only the working age population of 21 to 65 years (n = 207), it was estimated that patients earned approximately $51 more per week and on average, and there was 1.2 hours productivity gain each week at 1 month after participating in ADAPT. When the analysis was focused on those patients who were unemployed or on leave due to pain before ADAPT (n = 100), 20% of them returned to paid employment 1 month after ADAPT. It was estimated that these patients, on average, earned approximately $194 more per week, with an estimated productivity gain of 4.5 hours per week after participating in ADAPT.

3.4. Economic analysis

3.4.1. Costs

Healthcare utilization data were available for 224 patients (42% male and 58% female). Of the 224 patients, 118 patients (52.7%) were covered by Medicare (male= 47, female = 71) and 106 patients were covered by NSW workers compensation insurance (male = 46, female = 60). Table 4 shows changes in the average number of healthcare utilization before ADAPT and at 1 month follow-up. The average HCU cost, before the cost of the ADAPT program, was significantly lower at 1 month after completing the ADAPT program, resulting in an average cost saving of approximately $1420 per patient over 3 months (Table 5). The reduced costs 1 month after participating in the program comprised savings from GP costs ($127.43), specialist costs ($138.04), allied health costs ($172.22), emergency department (ED) costs ($202.61), hospital costs ($385.81), and diagnostic costs ($394.65). The cost of ADAPT at that time was $9150.00 (2019, administrative data). Therefore, the additional cost per patient 1 month after participating in ADAPT was $7729.24 (ie, $9150.00-$1420.00).

Table 4.

Average number of healthcare utilization before ADAPT and at 1-mo follow-up (n = 224).

Healthcare utilization	Before ADAPT (number of visits) [mean (SD)]	At 1-mo Follow-up (number of visits) [mean (SD)]	Difference (average number of visits)
General practitioner	5 (3.5)	3 (2.9)	2
Specialist	2 (2.1)	1 (1.2)	1
Allied health	7 (7.8)	3 (5.5)	4
Emergency department	0.4 (1.8)	0.1 (0.61)	0.3
Hospital admission	0.18 (0.7)	0.04 (0.19)	0.14
Diagnostic test	1 (1.6)	0.3 (0.74)	0.7

Table 5.

Cost of ADAPT per patient (2019 Australian Dollar) (n = 224).

Cost (2019)	Before ADAPT	At 1 month follow-up	Difference in cost
Intervention cost of ADAPT*	$9150.00		$9150.00
Healthcare utilization cost
GP visit	$291.48	$164.05	$127.43
Specialist visit	$224.54	$86.50	$138.04
Allied health visit	$265.11	$92.90	$172.22
Emergency department visit	$287.57	$84.96	$202.61
Hospital admission	$613.10	$227.29	$385.81
Diagnostic test	$585.43	$190.78	$394.65
Total healthcare utilization cost	$2267.23	$846.47	$1420.76
Change in cost per patient = intervention cost of ADAPT − (difference in healthcare utilization cost)			$7729.24

^*Based on hospital administrative data.

3.4.2. Cost and health effects of Active Day Patient Treatment

We combined 224 patients' pain scores and healthcare utilization cost data to calculate the additional cost per clinically meaningful change in BPI severity, BPI interference, and PSEQ scores. There were estimated additional costs of $9452.32 (95% CI: $7031.76- $12,930.40) for a 1-point improvement in pain severity, $3446.62 (95% CI: $2851.67-$4126.46) for a 1-point improvement in pain interference, $483 (95% CI: $411.289-$568.606) for a 1-point improvement in the PSEQ, and $3381.02 for a 7-point improvement (clinically meaningful improvement) in the PSEQ score.

In summary, an additional $9452.32 would need to be invested in ADAPT for a 1-point improvement in the BPI pain severity score. Similarly, an additional $3446.62 would need to be invested for a 1-point improvement in BPI interference score, and an additional $3381.02 would need to be invested for a 7-point improvement in the PSEQ score.

Figure 1A-C show the cost-effectiveness (CE) plane with the X-axis showing the additional cost per clinically meaningful change in health outcomes (eg, improvement in BPI severity, BPI interference, and PSEQ) at 1 month after ADAPT, and the Y-axis showing increased cost at 1 month after ADAPT (including the cost of the ADAPT intervention). Each point in the plots represents the estimate of a replicated data set. This demonstrates although there is an additional cost to provide the ADAPT intervention, it results in improved health outcomes for a clinically meaningful change in BPI severity, BPI interference, and PSEQ.

Cost-effectiveness acceptability curves show the probability that an intervention is cost effective if the decision makers are willing to pay (WTP) to gain one extra unit of effect. An intervention is cost-effective if the decision makers' WTP is greater than the calculated cost per improvement in pain scores. Figure 2A indicates that for a WTP of $11,000 for a clinically meaningful change (1-point improvement in BPI interference (1-10) score), the probability that the ADAPT is cost-effective at 1-month follow-up was 82%. Figure 2B shows that for a WTP of $4000 for a clinically meaningful change (1-point improvement in the BPI interference score), the probability of the ADAPT being cost-effective at 1-month follow-up was 95%. In addition, Figure 2C indicates that for a WTP of $3600 for a clinically meaningful change (7 point improvement in the PSEQ score), the probability of the ADAPT being cost-effective at 1 month follow-up was 81%.

Figure 2.

Cost-effectiveness acceptability curve based on BPI severity, BPI interference, and PSEQ scores. (A) BPI severity (0-10). (B) BPI interference (0-10). (C) PSEQ (0-60).

3.5. Reduction in analgesics

The mean number of analgesics per patients reduced to 1.6 per day at 1 month after participating ADAPT, down from 2.7 daily at pre-ADAPT (n = 205). This result was statistically significant (P < 0.05) and was consistent with what clinicians have seen in other studies.^38,39,55

4. Discussion

At 1-month follow-up, the 227 patients reported an average 15.86 point improvement in their PSEQ scores compared with their pre-ADAPT scores. Previous studies suggest that pain self-efficacy is positively associated with adjustment to a persistent pain condition.^27,53 Improvements in pain self-efficacy after an intervention are expected to be reflected in reductions of pain-related disability.^1,15,18,53 A study by Thomas et al. indicated that the effects of a pain management program on occupational performance were influenced by gains in skills and confidence, including self-efficacy.⁵³ Pain self-efficacy beliefs (ie, confidence to function despite pain) have a stronger association with disability, pain behavior, and confidence to perform at work than with pain intensity.^2,32,53 This is consistent with the previous findings by Nicholas et al.³⁹ Although this study does not include longer term data, previous studies show that improvement in health outcomes were consistent over 1 year.^38,39 However, economic data were not collected by the authors for those studies.

The ADAPT program runs within a public hospital. Although some patients are supported by worker's compensation, patients from all other circumstances are also included in this program. For the patients who are supported by workers' compensation, ADAPT may help them with return to work (RTW). However, other participants may not have any formal relationship with work. They might nevertheless improve their work capacity. Our analysis indicates that patients' labour force participation improved and their average weekly earnings increased at 1-month follow-up, with average weekly income increasing by $51 for working age patients (21-65 years of age). Other studies have also suggested that workers with chronic pain may have reduced productivity, as well as higher healthcare costs.^{16,24,25,42,44} Our findings also indicate that mean BPI severity, BPI interference, and PSEQ scores improve for patients with chronic disabling pain within 1 month after participating in ADAPT. In addition, at 1-month follow-up, there was a significant reduction in average healthcare utilization cost. The economic analysis also indicated that ADAPT was more effective compared with the standard care provided before ADAPT, but these improvements came at a cost.

There was also reduced number of medications taken by patients at 1-month follow-up which was consistent with previous findings.³⁹

Before this study, there was limited economic analysis related to interdisciplinary pain management programs. A small number of previous studies have reported increased labour force participation for subacute and acute pain management interventions that were also found to be cost-effective and cost-beneficial.^25,29,31

Schofield et al. (2012) suggested that it is important to maintain the labour force participation of patients with chronic pain to maintain their living standards at levels comparable with others who do not suffer from this condition where possible.^48,50 Patients who work limited hours due to pain or retire early due to pain have lower income available which also reduces their long-term capacity for wealth accumulation. This increases their dependence on their family members and welfare payments. Thus, any intervention that can improve a person's capacity to work should be of interest to both the patients, their families, and to those responsible for funding treatments.

4.1. Strengths and limitations

Our study has a number of strengths. In this economic analysis, we have used data from a standardised clinical database, ePPOC. The ePPOC database provides a means of using the data at an individual patient level with a software called epiCentre (ePPOC Patient Information Centre), which facilitates consistent collection, terminology, process, and protocol among participating pain management services and promotes research into areas of importance in pain management. Nicholas et al.¹² reports that the low proportion of missing responses in the data sets available in ePPOC provides robust data for an economic analysis. In this study, we have included labour force participation outcomes for patients' average weekly earnings as an outcome measure. Labor force participation is an important indicator beyond improved health outcomes of tax paid after RTW, less dependence on welfare payments, and less dependence on family members for financial support.¹²

Our study also has a number of limitations. First, although our analyses controlled for important known baseline differences for ADAPT patients, other unknown or unmeasured confounders could have contributed to our findings. Second, to participate in the ADAPT program, patients need to be referred by GP or medical specialists. The ADAPT is an intensive program run over 3 weeks and patients unable to take leave from work, study, or home duties may not be able to attend the program, and therefore, we cannot rule out the possibility of selection bias. Third, this study analysed short-term cost-effectiveness of the ADAPT program at 1 month compared with pre-ADAPT. As we had to estimate many of the costs, it is possible we may have underestimated the cost-effectiveness of ADAPT that could have been captured if we analysed long-term cost-effectiveness beyond the follow-up period.¹⁹ Fourth, although the number of analgesics was reported, the types of medication were not recorded in the available data set. Accordingly, it was not possible to estimate their cost, but assuming the medication types did not change, the reduction in number would represent a reduced cost. Finally, this study did not use a specific QALY measure, but it did include measures of mood and physical disability, as well as confidence in functioning, all of which are related to quality of life and sensitive to change. Despite these limitations, we believe, our study adds to the limited body of evidence related to health and labour outcomes of interdisciplinary pain management programs.

4.2. Conclusion

Our analysis indicates that an interdisciplinary intervention such as ADAPT can improve chronic pain patients' labour force participation, reduce healthcare utilization costs, and improve health outcomes.

Conflict of interest statement

The authors have no conflicts of interest to declare.

Appendix A. Supplemental digital content

Supplemental digital content associated with this article can be found online at http://links.lww.com/PAIN/B856.