How to sustain and improve client centred (matched) care in chronic musculoskeletal pain? Start by changing the way policy makers select and judge the large amount of available data, and get rid of the dogma of stepped care

When I was invited to write an editorial for this issue of EJPT, I was faced with the large dilemma how to deal with the notification of a Dutch governmental agency that they will evaluate whether interdisciplinary multimodal pain treatment (IMPT) programs should still be funded. This type of action, as in many other countries, seems to be an almost automated response to shrinking resources and often exponentially growing expenses. Nevertheless, as clinicians and researchers, we all have an obligation to improve the efficiency of care and treat as many as possible patients within the limits of the available resources. We therefore should display the utmost transparency regarding the way we select patients, the content of treatment and the process and outcome data [1]. However, being around in this field for almost 30 years, I see a large gap between the way policy makers approach this topic and how they rate the evidence we have provided versus the way I, and many others think this should be done.

First, in the discussion regarding growing expenses, the global increase of prevalence of patients experiencing disability due to chronic musculoskeletal pain caused by population growth and ageing is totally being ignored [2]. But also, other aspects are neglected. For example, in my practice as physiatrist specialised in chronic musculoskeletal pain (secondary care facility), I see a rapid increase of young patients (18-30 years) who are not able to cope with their pain due the challenges of modern society and the malleability of health. This development definitely does not deserve the reflex of controlling the budget or restricting the availability of proven effective care, but a revaluation of primary and early secondary prevention. We know that for example mass media campaigns for low back pain can change beliefs and behaviour and have the potential to reduce personal as well as societal burden [3].

Second, the doctrine of stepped care should be abandoned. There are several great examples of cost-effective risk stratified care [4,5] and emerging evidence for matched care [6]. Stepped care is defined as a more conservative (and cheaper) intervention that should be provided first, and a more complex (and expensive) intervention should be progressed to only when the simpler ones fail. Stratified care, categorises patients based on risk factors (e.g. low, medium and high) and provides an accordingly level of treatment comprehensiveness, and matched care assesses key risk factors and individualises the intervention on the patient’s needs [7]. Needless to say, that be holding on to the stepped care, in the long run this will increase costs, but also will inflict iatrogenic, and hence preventable damage to patients. We do have high levels of evidence for the prognostic value of modifiable, mostly psychosocial factors in musculoskeletal pain [8], so why not implement the appropriate screening and early treatment? In my opinion, matched care is the way forward as in this model, next to the risk factors, information on social factors, comorbidity, previous and personal experiences with other treatments can be taken into account [9], and real client centred care in which shared decision making is also integrated can be delivered. For us as clinicians there is the obligation to collect data and to start using promising techniques like machine learning to build up clinically based evidence and to improve the effectiveness of this type of care.

Third, in this abovementioned governmental evaluation process but also often during the development of evidence-based guidelines, three specific errors can be identified. Often, only studies using a RCT design are still being regarded as the highest and only level of evidence that should count in the decision whether or not a treatment should be reimbursed. This despite the fact that their clinical applicability can often be debated (e.g. inclusion of often highly homogeneous population, hardly any room for client centred approach (one size fits them all), and the use of often not personally relevant outcome measure (e.g. Pain Disability Index which addresses activities that often are irrelevant for the patient being treated in IMPT [10]. In 2013, I already pleaded for the incorporation of Single Case Replicated Designs (SCED) [1], especially as nowadays better and highly relevant and powerful statistical techniques and e.g. randomisation procedures are available [11], and the possibility to preregister such studies [12]. This makes them highly relevant and powerful to incorporate in the weighing process of the available evidence. But not only SCEDs should be taken into account. Due to availability of routinely collected data and new statistical methods and designs for establishing causality (e.g. propensity scores, etc.; for an overview, see [13]), high quality gathered observational data could become a very important and powerful source of new information to compare and assess effectiveness of care. Another important development is the use of living systematic reviews (LSR) in which new data can be quickly uploaded and synthesised to update the existing evidence [14]. Such a LSR has to possibility for additions of for example the specific content of treatment while using appropriate checklists like the TIDieR [15]. This enables clinicians to design the proper treatment for their patients. The other stakeholders and especially policy makers are herewith invited to join researchers and clinicians in these exciting developments and to adapt their weighing processes based on new knowledge form this rapidly developing field [16].

The second error is that only studies in which IMPT is compared to a not-IMPT, are incorporated in the weighing process. So RCTs in which two types of IMPT are compared in specific patient populations are totally disregarded. This is highly problematic as especially these studies, due to the provision of treatment specifically targeted at the personal and modifiable factors show larger effect sizes than the more general, one size fits them all IMPT [17,18].

The third error is the use of the Minimally Important Clinical Change (MCIC). Often, and definitely in the process of the abovementioned evaluation this error causes me great concerns and resulted in this editorial. In the weighing process, a treatment is only regarded effective in case the change in primary outcome after providing the treatment (i.e. IMPT) at group level exceeds the MCIC level compared to the change at group level of the control treatment (often a monodisciplinary or less integrated treatment). This is a huge methodological flaw [19], as the MCIC is developed to evaluate whether the change of an individual patient is a clinically relevant change. In order to overcome this problem, we could express effectiveness based on the MCID as the % of responders in each treatment group. However, in order to have sufficient power, a larger number of participants is necessary and the individual change data should also be provided. For this, free accessible datasets which could also be uploaded in LSR, could help us to pool individual data of different studies, unless the content of treatment is similar. Another way forward is to develop consensus on what we all, patients, care-providers, policymakers and other stakeholders, define as a clinically worthwhile change when comparing treatments.