Study Design

The present study, funded by the Federal Joint Committee under the acronym PAIN2020 (“patient-oriented.graduated.interdisciplinary.network”) (Innovation Fund 01NVF17049), is a nationwide multicentre randomised controlled trial (RCT). The evaluation design envisaged 6000 patients at baseline (t1) and a calculated drop-out rate of 20% at each of the two follow-ups after 3 and 6 months (t2 and t3). Participants were included in 28 pain centres across Germany between February 2019 and August 2021. After inclusion, participants were assigned to the intervention group (IG) or the control group (CG) with a probability of 70% (IG) and 30% (CG), respectively. The unequal assignment to the two groups was based on the ethical consideration that study participants should have  a higher probability of being assigned to the intervention group. Computer-generated block randomization with variable block length was used. The IG received an IMA; the CG was assigned to a unimodal (physician-only) MPA. Participants were informed about the treatment they received in the two groups. Blinding was therefore not possible.

PAIN2020 is registered at the German Clinical Trials Register, which contains the most important key points of the study [21]. A published study protocol for detailed information is available [22]. The study was conducted in accordance with the 1964 Declaration of Helsinki and its subsequent amendments. It was approved by all participating institutions, including the Ethics Committee of the University Hospital Carl Gustav Carus in Dresden (EK 216062018) as the main ethics committee. Written informed consent was obtained from all participants.

Participants

The study included people in the region of one of the participating centres, aged ≥ 18 years, who were insured by BARMER statutory health insurance (the second largest in the German health insurance system with around 8.7 million insured persons). The project was later opened up to the Kaufmännische Krankenkasse (KKH) and finally patients insured by all statutory health insurance funds to facilitate recruitment.

BARMER played a key role in patient recruitment. In all regions with participating centres, insured persons were informed by post about the PAIN2020 project if the care data showed a continuous intensive use of services in association with pain diagnoses [e.g. visits to general practitioners (GP) and specialists in several consecutive quarters]. Interested insured persons could contact BARMER's telemedical advisory service, which, after a preliminary check of the inclusion and exclusion criteria, referred the insured persons to the participating centres for an information appointment. Further recruitment paths were based on informing GP practices and medical specialists. These efforts were supplemented by contacting specialist associations, selective press releases and social media activities. Each interested person who attended the information appointment and met the inclusion criteria was consecutively included in the study.

The inclusion criteria for the study were:

Persistent pain for at least 6 weeks and/or recurrent pain during the last 2 years,

Pain-related limitations relevant to the patient (e.g. ≥ 4 weeks of previous sick leave or cumulative sick leave of ≥ 6 weeks in the past year, interference with daily activities, family, leisure, work and homework);

Risk factors for chronic pain [including spreading pain, signs of stress in family/partnership/working life, depressive symptoms in experience and/or behaviour, feelings of frustration/anger, maladaptive behaviour (fear avoidance, task persistence or endurance), signs of somatisation, high health care utilisation including seeking further diagnostics].

Patients had to be ≥ 18 years old, have sufficient written and spoken German language skills, live in the vicinity of the participating healthcare facility and give verbal and written consent to participate.

The following criteria led to exclusion:

Clinical signs of a serious illness requiring urgent acute therapy or other serious illnesses (red flags, e.g. severe cardiac insufficiency, tumour disease) that make activation treatment impossible,

A manifest chronic pain condition that had already occurred (e.g. sick leave due to pain for > 6 months, pain-relevant diagnosis for > 4 quarters, previous treatment with strong opioids for > 3 months, previous interdisciplinary multimodal pain therapy in the last 2 years),

A severe and active psychiatric disorder (personality disorder, severe depression or anxiety disorder, signs of suicidal tendencies),

An ongoing application for retirement or a rehabilitation programme planned for the near future,

Linguistic and/or cognitive impairments.

The listed inclusion and exclusion criteria were checked by the physician as part of the anamnesis during the information appointment. For this purpose, the physician also had the information provided by the patient in the German Pain Questionnaire (“Deutscher Schmerzfragebogen”, DSF) [23, 24].

Participating Centres

Healthcare institutions (n = 28) specialised in pain care from all over Germany (11 out of 16 federal states) were involved in the implementation of the new healthcare diagnostic procedure IMA.

20 centres were located in large cities, four centres in mid-sized cities and four centres in a rural region. Six centres could be classified as outpatient pain practices, 15 were located at smaller or larger hospitals with pain units and seven were located at university pain clinics. The study centres already offered interdisciplinary multimodal pain therapy (IMPT) or had the prerequisites for cooperation between the professions required for IMPT according to the consensus recommendations of the German pain societies [25, 26]. They received information about the study from the project team, including data management, obligations related to study procedures and training of professionals and teams to implement a standardised framework for the IMA. The centres were monitored and supervised by the project team during the complete course of the study.

Interventions

The centres implemented the IMA based on Casser et al. [15] and adapted the procedures according to the study protocol and the target patients with risk factors for chronic pain. Each discipline carried out the anamnesis and diagnostics and documented the respective results of the patient examination and evaluation (60 min allocated per profession) as well as potential treatment options. In a team session (20 min), the findings were discussed and the joint evaluation and treatment recommendation were formulated. Recommendations included the entire spectrum of established treatment options within the German healthcare system. In addition, two special multimodal treatment group programmes were offered for IG patients that were not part of standard care. Finally, these interdisciplinary findings were explained to the patient (including diagnosis, disease model and potential treatment options) followed by a mutual decision on the final treatment recommendation with the patient (20 min). All steps of the IMA were documented in standardised form.

The MPA in the CG consisted of a visit to a specialist pain physician, located in either a centre providing IMA or an external practice setting. As with the IMA, the MPA was fundamentally open-ended regarding treatment recommendations. This MPA was not a standard treatment within the German healthcare system but rather an early referral to a specialist. The physicians who conducted the MPA were only trained in the required documentation; the MPA itself was carried out in the respective clinical routine.

For both MPA and IMA, the medical documentation was based on the standard pain quality assurance protocol of the German Pain Association [27], including the Mainz Pain Staging System (MPSS) [28].

Power Calculations

The original case number planning with 6000 patients and three primary outcome measures had to be revised, especially because of the COVID-19 pandemic and its impact on patient recruitment. The revised sample size calculation performed with MLPowSim [29] was based on two primary outcome variables (treatment satisfaction was excluded but remained as a secondary outcome measure for analysis) and two measurement time points, each with a dropout rate of 30%. Based on a probability of error for the first type of error in a two-tailed test with a significance level of α = 0.05, the α error was adjusted to α = 0.05/4 = 0.0125 according to Bonferroni because of the correction for multiple testing. A total sample size of 4500 participants at baseline (net sample 2175 at 6-month follow-up) was sufficient to detect small intervention effects with a power of 0.80.

Outcome Variables

Outcome variables and additional patient data were collected using the German Pain Questionnaire (“Deutscher Schmerzfragebogen”, DSF) and its follow-up version (“Schmerz-Verlaufsfragebogen”, VFB) via paper and pencil [23, 24]. At a later stage of the survey, study participants were also given the opportunity to use an electronic version of the VFB.

DSF and VFB comprise the two primary outcome measures Characteristic Pain Intensity (PI) and Disability Score (DS), which are also components of the Graded Chronic Pain Scale (GCPS) [30, 31] (Table 1). Mean scores were calculated for each of the three numeric rating scales belonging to PI (pain right now/average pain/worst pain) and DS (pain interference in daily activities/recreational, social and family activities/ability to work). The endpoints of these rating scales were labelled "no pain" or "no interference" (0) and "pain as bad could be" or "unable to carry on any activities" (10).

Table 1 Schedule of enrolment, interventions and assessments

Additionally, DSF and VFB also regularly include the Depression Anxiety Stress Scale (DASS) [32, 33], the Veterans Rand 12 Item Health Survey (VR-12) [34, 35], the Pain Description List (PDL) [24] and the Marburg Questionnaire for Habitual Well-Being (MFHW) [36], assessing the secondary outcome measures. The Pain Catastrophizing Scale (PCS) [37, 38], not integrated as standard in the DSF or VFB, was added. At both follow-ups, to assess treatment satisfaction from the patient's perspective, participants were additionally surveyed with a global change item (“When you look at it all together, how would you rate the success of your treatment so far?"; 1 = very good, 5 = very poor) and the Client Satisfaction Questionnaire (CSQ-8/ZUF-8) [39, 40]. DSF and PCS were assessed at baseline (assessment, t1), the VFB and PCS as well as CSQ-8/ZUF-8 at the two follow-ups after 3 (t2) and 6 months (t3).

Data Management

All data were entered into a study database. The principles of data management and data protection were described in a comprehensive data management plan and approved by the responsible supervisory authorities.

Monitoring

A extensive monitoring concept was developed to ensure that the intervention was carried out as well as possible in accordance with the study protocol. This included two personal on-site visits to the centres. Among other things, the project team checked whether the medical documentation was correct and whether the practitioners had received sufficient training. A weekly telephone conference was offered by the project team for any questions from the centres. Furthermore, the data in the study database were continuously monitored by the evaluator to provide ongoing feedback on data quality and the development of case numbers.

Statistical Analysis

For the statistical analysis, the intention-to-treat principle (ITT) was applied: All participants with valid consent who had been randomised and who had received an assessment were included in the analysis.

The comparison of the outcome measures of IG and CG over time was based on a mixed model for repeated measures (MMRM) [41]. The value expressions of the primary and secondary outcome measures at t2 and t3 were predicted from (1) the baseline values of the outcome measures at t1 (yt1), (2) the factors of group membership (group) and the relevant time point (time) and (3) the interaction effect of group membership and time point. A random effect was included in the model for the regression constant.

$$\begin yt = & a + b1*yt1 + b2*group \\ & + b3* time + b4* group \times time \\ \end$$

The regression models were parameterised so that the baseline values (t1) were centered and the CG and t3 were defined as the reference category. The regression constant therefore describes the value expected under the model for the respective dependent variable in the CG at t3.

The time effect defines to what extent the value at t2 deviates from that at t3. The group effect describes the extent to which the value of the IG deviates from that of the CG. The same applies to the interaction effect. The analysis model thus corresponds to an analysis of covariance as recommended for the analysis of controlled studies with longitudinal data as an alternative to using difference values to control for baseline values [42]. Missing values at measurement time points were not imputed. Analyses were performed using IBM SPSS Statistics 28.

To investigate possible centre effects, the sample composition of the respective patient collective was analysed in more detail, including characteristics such as age, sex, education, pain diagnoses or other structural parameters at the centre level (e.g. case numbers, response rates, patient satisfaction or whether standard care was provided in-house or externally in a specialist practice).