Miranda Van Hooff

The ultimate aim of the research presented in this thesis is to contribute to the body of knowledge on outcomes of interventions for chronic low back pain (CLBP) and to identify outcome-based subgroups of patients having different profiles. Based on work included in this thesis, an article was published in a provincial newspaper in 2014 highlighting the development of a tool to triage the right patient to the right medical specialist, based on subgroups with different patient profiles, i.e. the Nijmegen Decision Tool for CLBP. The headline stated ‘A questionnaire helps in low back pain’ [1]. Whilst we think we have taken an important step forward in terms of the knowledge gained and identifying pre-treatment profiles related to outcomes of interventions, a questionnaire as such is only a small piece of the puzzle towards solving the tremendous worldwide burden of low back pain.

Despite decades of research, honed expertise, and improved quality of clinical trials, the evidence regarding the effectiveness of the treatments offered to patients with CLBP is still inconsistent [2-8], rarely shows more than a small to moderate overall benefit [2,7,9-12], and demonstrates a lack of long-term efficacy in changing the prognostic paths [10]. Most patients recover, but in around 20% the complaints persist for more than three months, resulting in disabling CLBP. These patients bear the greatest proportion of the disease burden. One reason for the overall disappointing outcomes is that the CLBP population is heterogeneous because the condition lacks diagnostic clarity; as a consequence, there exists a plethora of invasive and non-invasive interventions in secondary or tertiary healthcare for the same symptom. To reduce the global societal burden of CLBP it is crucial to improve treatment outcomes and to reduce the related costs, to improve the value of delivered healthcare. To achieve this it is essential to know which outcomes are relevant to both patients and medical specialists and to know which patients will benefit from spine surgery or from non-surgical treatments.

In the General introduction of this thesis the aims are described using three separate, though related themes. In the first theme a non-surgical combined physical and psychological (CPP) programme was evaluated and insight is given as to who could benefit from this programme. In the second theme the focus is on the methodology used for outcomes assessment in the evaluation of clinical practice and research of degenerative lumbar spine disorders. In the third theme the development of a clinical decision tool for patient triage to a surgical or a non-surgical medical specialist is described. In this General discussion the overall research questions are answered by summarising and discussing the main findings per theme. Considerations are described that warrant further exploration in relationship to the methodology used, and some implications and recommendations for clinical practice and for future research are presented. The discussion ends with concluding remarks.

Theme A: Introduction of a combined physical and psychological programme

Research questions
1. Does the novel CPP programme for CLBP improve patient outcomes and reduce healthcare consumption?
2. Is it possible to identify a subgroup of patients that benefits most from the novel CPP programme so that selection criteria can be optimised?

Summary of main findings
In the one- and two-year follow-up cohort studies presented in Chapter 2 and 3, selected and motivated patients with longstanding CLBP participated in an intensive multidisciplinary CPP programme. Participating patients were moderately to severely disabled, which is comparable to patients being treated in secondary spine care. They learned to manage their CLBP, and their daily functioning and quality of life improved meaningfully. The magnitude of the improvement is comparable to that achieved with spinal surgery (Standardised Morbidity Ratio [SMR] 98%) and better than that achieved with less intensive rehabilitation programmes (SMR 136%) (Chapter 2). At the two-year follow-up of this cohort, the significant and clinically relevant improvements in functional ability, pain and health-related quality of life achieved at the one-year follow-up assessment were maintained. Above all, most of the participants were employed and the results indicate that the use of both pain medication and healthcare decreased substantially (Chapter 3). In Chapter 4, pre-treatment indicators of a successful treatment outcome were identified. Successful treatment outcome was defined as a one-year follow-up functional disability (disability) score falling to values seen in healthy populations (Oswestry Disability Index, ODI ≤22). Patients who are employed at pre-treatment [OR 3.61 (95 % CI 1.80–7.26) and who are mild to moderately disabled at the start of a CPP programme [OR 0.94 (95% CI 0.92–0.97)] are most likely to benefit from this programme (R2 =22%; 67% correctly classified). No interaction effects between pre-treatment characteristics were found and, to our surprise, no predictive value was found for psychological distress.

As continuous outcomes monitoring is part of the CPP programme, this gave us the opportunity to further substantiate the ‘pilot’ results presented in Chapter 2 with one-year follow-up results of a large cohort (n=848). The results of this recent study showed that these patients had similarly good results: patients improved during the programme, showed further improvement at the one-year follow-up, and half of the patients (51%, n=433) improved such that their functional status was comparable to that of the healthy population [13].

Discussion
In a recent systematic review, moderate-quality evidence for moderate effects of multidisciplinary bio-psychosocial treatments was found compared to usual care [12]. These treatments are recommended for CLBP patients [14-16], but not often implemented.

In this thesis the clinical relevance of the CPP programme has been demonstrated: patients improved meaningfully, healthcare consumption was reduced, and a relevant treatment effect was found. We found large effect sizes for functional ability at the one-year follow-up (Chapter 2) and the two-year follow-up (Chapter 3), meaning that the programme shows relevant effects. The two-year follow-up findings presented in Chapter 3 were further substantiated by the results of a recently performed long-term follow-up post-marketing surveillance study involving 277 ex-participants (mean follow-up of 6.5 years [range: 5.5–7.5]; response 85%; no baseline differences between responders and non-responders to the survey); positive results were maintained after 6.5 years on average, and 80% of the participants were satisfied with the treatment results [17].

We studied the intervention as an integral programme. The programme uses a wide range of techniques based on cognitive behavioural principles. As yet, the working mechanisms are unknown, so it is unclear which techniques or parts of the intervention are the effective elements. In Chapter 4 we speculate that several aspects contribute to the success of this programme, such as the programme’s structure (a. the intensity or dose of the programme), the programme entrance criteria (b. motivation to change behaviour), and the content of the programme (c. improvement of dysfunctional cognitive behavioural factors).

a) Intensity of the programme. Although conflicting evidence exists concerning stability over time [18,19], it is assumed that dysfunctional behavioural cognitions in patients with persistent pain of long duration are resistant to change [20]. In a systematic review, Guzman et al. recommend 100 hours or more of intensive multidisciplinary rehabilitation including cognitive behavioural interventions to improve functionality [21]. The CPP programme studied in this thesis follows this recommendation in a ‘pressure-cooker’ structure. This ‘pressure-cooker’ structure might moderate the dysfunctional behavioural cognitions. Patients with longstanding CLBP (mean 12 years, SD 11) benefit from this programme (Chapter 2-3), but the influence of duration and intensity of the programme as moderating factors remains elusive and needs to be further explored.

b) Motivation of participants to change behaviour. Although motivation is a selection criterion for the programme, we neither assessed this factor in a clearly valid and reproducible way at pre-treatment nor assessed it systematically over time. As ‘motivation’ is viewed as a state that is amenable to change rather than a trait that is constant [22], it is possible that those patients who were unsuccessful after having followed this programme were actually not ready or motivated to change their pain-related behaviour. Motivation or readiness to change pain-related behaviour is important for treatment compliance; it plays an important role in accomplishing and achieving treatment goals and it positively influences quality of life and may predict healthcare costs [23,24], completion of a treatment programme [25], and treatment outcome [23,26,27]. The pain Readiness to Change (RtC) model is used for conceptualising the process of adopting a self-management approach to chronic pain [22]. According to the RtC model, individuals vary in their degree of readiness to adopt a self-management approach. In Chapter 2 it is shown that behavioural change is possible and that participants adopt self-management strategies to cope with back pain complaints. The Multidimensional Pain Readiness to Change Questionnaire (MPRCQ) [28,29] is designed to measure RtC related to specific components of self-management targeted in multidisciplinary treatment programmes [30], completion of treatment, and to predict treatment outcomes [29]. However, a recent study shows that, in patients with CLBP at pre-treatment, coping behaviours instead of readiness to engage those behaviours are associated with pain-related functioning [31]. Further research is needed to assess this indicator in a clearly valid and reproducible way and to evaluate its contribution to the outcome over time.

c) Improvement of dysfunctional cognitive behavioural factors. It has been suggested that improvement of such factors as catastrophising cognitions (i.e. exaggeration of the threat value of pain sensations) and fear of movement behaviour might contribute to the development of CLBP [32], to persistence of CLBP [33,34], and jeopardize successful treatment outcomes [12,35-38]. This phenomenon is described in the fear avoidance model (FAM) [32]. The model postulates a causal relationship between pain catastrophising (a sign of serious injury or pathology [39]), fear of movement, disability, and experienced pain severity [32,39]. Some studies have concluded that the impact of these dysfunctional cognitive behavioural factors on outcome is diminished [40,41] or is even absent [42], which is consistent with the results presented in Chapter 4. These findings corroborate the suggestion that the sequence and relationships between pain catastrophising, fear of movement, and disability postulated in the FAM maybe different for the development of CLBP than for the recovery of disability as a result of an intervention [43]. The CPP programme has a beneficial impact on cognitive behavioural variables [44], but a closer exploration of these cognitive behavioural factors and their impact on functional ability is needed.

Methodological considerations
The studies presented in Chapter 2 and 3 followed an observational study design. To enhance the internal validity of the studies, several precautions in the study methodology were taken to minimise the potential influence of confounding on the outcomes. In evidence-based medicine the randomised controlled trial (RCT) is regarded as the gold standard to secure the internal validity of the study, because only the RCT is thought to resemble a true, pure experiment from which causal inference can be concluded. However, in the last decade the value of the RCT has been questioned for decisions about the use of interventions due to drawbacks of the design and available evidence from RCTs [45]. In certain situations, as in an RCT, ‘experimentation’ may be unnecessary, inappropriate, impossible, or inadequate [46]. Some of these are relevant in both surgical and conservative treatments. For example, an RCT may be inappropriate because of the random allocation and blinding procedures used. When the clinician and patient agreed upon the treatment, the clinician or patient (or both) have their preferences. This arises when the effectiveness of the intervention depends on patients’ active participation in conservative programmes, which, in turn, depends on patients’ beliefs and preferences. In (spine) surgery the results and effects also depend on surgical skills as well as surgeons’ beliefs and preferences. As a consequence, the lack of any subsequent difference in outcome between comparison groups may underestimate the benefits of the intervention. To overcome these problems, alternatives have been suggested for which the key is to use all available study designs, depending on the research question, and to perform every study with scientific rigour [47]. One of the alternatives to the RCT is routine outcome monitoring with an outcome registry, set up with an observational study design (Chapter 5). Recent spine-related studies comparing RCTs with observational study designs show comparable results [48-52], which suggests that observational study designs are complementary to RCTs.

The external validity of the study results presented in the Chapters 2-4 might be limited because we studied the prospective cohort with carefully selected patients in a secondary care setting. The conclusions of the studies performed in Chapter 2 to 4 are based on data gathered through routine outcome monitoring rather than an RCT. The outcome monitoring is performed through the web-based outcome registry of the CPP programme. The outcomes registry follows the guidelines of observational studies [53] and the recommendations listed in Chapter 5. To further substantiate the effects found in Chapter 2, an historical controlled trial was performed to compare the magnitude of treatment effects to those of other published studies for similar populations. For this comparison an SMR was used, which is a rate ratio to compare estimates of relative treatment improvement. We were limited to the external