Nationwide Evaluation of Multimodal Rehabilitation in Patients With Chronic Musculoskeletal Pain

Chronic musculoskeletal pain is a public health problem, which frequently cause long-term sickness absence (>180 days). Societal costs of chronic pain are immense, with direct and indirect costs amounting to 90 billion Swedish crowns per year according to Swedish Council on Technology Assessment (SBU). Despite this, few policies have aimed specifically at chronic pain as a public health problem, also acknowledged in the later literature. Multimodal rehabilitation (MMR) is an increasingly used method in this patient group and denotes a combination of psychological measures, physical activity/exercise and manual or physical methods, administered by multidisciplinary teams. Its approach considers the patient's needs from a bio-psycho-social perspective, including work or employment status. At date, MMR is commonly offered by a number of specialist clinics throughout Sweden, of which more than 95% (currently 40) are connected to the Swedish Quality Registry for Pain Rehabilitation (SQRP) with the purpose to collect data for work with quality improvement and research.

According to the latest report from SBU, medium to high evidence from available research supports that MMR is effective for patients with chronic pain, with a few studies suggesting that MMR is superior to traditional and single-treatment programs. However, there is a paucity in the scientific literature with respect to how MMR should be designed to optimize results, and comparisons between different MMR designs or durations are limited. This void is a cause for uncertainty which delays improvement of existing MMR programs treating patients with chronic pain. Further, it is necessary to investigate whether reported results from RCTs and systematic reviews truly apply to naturalistic practice settings with a consecutive non-selective flow of patients, which may be done using a research methodology known as practice-based evidence through prospective observational cohort study designs. Our intention with the present research program is therefore to clarify effectiveness of different MMR designs. Furthermore, knowledge on risk predictors of poor treatment outcomes, and how long-term pain can be prevented and treated is greatly limited. Here, it is currently unclear what factor, patient related and/or treatment component, are more or less effective in different subgroups of patients with chronic pain. This has been discussed in the national SQRP-network and is in demand in the scientific literature. Knowledge of this will enable MMR providers to customize treatment according to patient profile in order to maximize treatment outcomes in relation to costs.

The overall aim of this research program is to prospectively evaluate multimodal rehabilitation (MMR) in patients with chronic pain conditions. The long-term goal is to develop and implement updated knowledge into Swedish MMR programs. The following two strategic aims will be targeted: 1) effectiveness of MMR programs with different durations (short, moderate and long-duration) , 2) prediction and risk of positive and negative outcomes, respectively.

Methods Study design This research project is carried out with longitudinal cohort-study designs based on data extracted from the Swedish Quality Registry for Pain Rehabilitation (SQRP; www.ucr.uu.se/nrs). Data covering several years will be used to analyse the effects of MMR. We expect a large sample size, it entail high test-power for the analyses of overall effects. Therefore, in addition to tests of statistical significance, effect sizes will be calculated in order to avoid type I error, i.e. avoid to falsely claiming that effect exists with support the support of p-values in such large sample. Included subjects are patients aged between 16 and 67 with a referral to a MMR clinic (specialist clinic) for chronic musculoskeletal pain problems.

Data from SQRP allows for longitudinal evaluation of treatment effects, prediction analyses. SQRP is a registry mainly covering patient-reported questionnaire data intended for health care research and quality work for healthcare providers that offer MMR. SQRP is connected to the Swedish Association of Local Authorities and Regions (Sveriges Kommuner och Landsting) and part of a system of national quality registries in Sweden (www.kvalitetsregister.se). Currently, 40 Swedish MMR specialist-treatment units (representing over 95% of the pain rehabilitation centres in the country) are associated with the registry and continuously transmit their patient data to SQRP (www.ucr.uu.se/nrs/index.php/om-registret). This means that nationwide data from all parts of Sweden is included in SQRP which constitutes a major strength of this database. Based on this data, an annual report is compiled by the SQRP steering group and co-workers describing and comparing the performance of the associated MMR treatment units on the overall scale. Since 2007 these accounts have been available to the public (http://www.ucr.uu.se/nrs/index.php).

Data from established questionnaires are registered in SQRP to measure health-related quality of life (SF-36), the consequences of prolonged pain problems on the level of activity and function (Multidimensional Pain Inventory, MPI), perceived health (EQ-5D), pain intensity (numeric rating scale), and anxiety and depression (The Hospital Anxiety and Depression Scale, HAD). Furthermore, data on socio-demographic details, information on employment and sick leave status as well as pain duration and future prospects is collected. In addition to the patient-reported data, an attending MMR practitioner complements the registry with the patient's diagnosis, information on type of pain and expected form of employment (post-rehabilitation).

Data analysis Health-related quality of life, pain intensity, function, activity level, as well as anxiety will serve as dependent outcome measures in the evaluation of effects. The use of a single outcome measure is rarely sufficient to describe the patient's experience, symptoms or treatment outcome in terms of understanding potential effects. However, multiple outcome variables can be inter-correlated and therefore it is reasonable to apply multivariate analyses to control for such dependent relationships. Therefore, multivariate repeated measures analyses will be used to approach effect variables prospectively, while logistic regressions (or similar) with a multiple approach will be applied to analyse prediction models.