Decision support tools in low back pain




Abstract


Information from individual classification systems or clinical prediction rules that aim to facilitate stratified care in low back pain is important but often not comprehensive enough to be used to support clinical decision-making. The development and implementation of a clinically useful decision support tool (DST) that considering all key features is a challenging enterprise, requiring a multidisciplinary approach. Key features are inclusion of all relevant treatment options, patient characteristics, and benefits and harms and presentation as an accessible and easy to use toolkit. To be of clinical value, a DST should (1) be based on large numbers of high-quality data, allowing robust estimation of benefits and harms; (2) be presented using visually attractive and easy-to-use software; (3) be externally validated with a clinical beneficial impact established; and (4) include a procedure for regular updating and monitoring. As an illustration, we describe the development; presentation; and plans for further validation, implementation, and updating of the Nijmegen Decision Tool for Chronic Low Back Pain (NDT-CLBP).


Introduction


Low back pain


Low back pain (LBP) is considered to have the greatest burden of disease for society globally . This global burden of disease study highlighted that LBP ranked higher than other diseases (e.g., cancer, heart disease, cerebrovascular disease, or diabetes) . Musculoskeletal disorders were a leading cause of years lived with disability, with low back pain specifically highlighted as the number one cause in the industrialized world .


The high prevalence of low back pain and its burden to both patients and society has led to an ever expanding array of diagnostic tests and treatments, including injections, surgical procedures, implantable devices, medications, various types of exercise programs, manual therapy, traction, acupuncture, transcutaneous electrical nerve stimulation, spinal cord stimulators, mattresses, orthotics, back supports, biofeedback, spinal injections, and surgery . The evidence for these different types of interventions varies but the treatments that are generally recommended for LBP as first-line treatments in high-quality international guidelines are education, staying active/exercise, manual therapy, and paracetamol/NSAIDs (although the recommendation of paracetamol for acute LBP is challenged by recent evidence and needs updating) . In addition to being recommended in guidelines, all these treatment options share another important feature: they all have small to moderate average treatment effects. A recurring conclusion of high-quality systematic reviews and randomized controlled trials (RCTs) in the field of low back pain is that most treatments show modest effects compared to natural course and small or no differences between the effectiveness of different interventions, both in a primary care and secondary care settings . Similarly, cost-effectiveness studies do not yield convincing evidence for any particular treatment, making it difficult to choose one treatment over another (see van Dongen et al., this issue).


Stratified care


Several reasons might explain these small to moderate effects. One potential reason is that in the vast majority of patients, a specific etiology is unknown, and therefore, they are labeled to have “non-specific” low back pain. However, this is an umbrella term that lacks diagnostic clarity, resulting in a heterogeneous population. It has been argued that treatments are not specifically tailored to the relevant subgroups of patients but that, instead, many treatments use a (more or less) one-size fits all approach . To achieve more substantial results of treatment and/or a reduction of health care utilization, matching groups of patients with the most appropriate treatment for their profile, referred to as stratified care, has been advocated and is (for many years already) one of the top research priorities in the field of low back pain . Stratified care is hypothesized to provide individual and population-level benefits, including improved patient outcomes and efficiency in resource consumption . Several methods have been proposed to facilitate such stratification, including designing classification systems and developing clinical prediction rules (CPRs) . Although classification systems in LBP may or may not be based on statistical models to classify a heterogeneous group of patients into subgroups, CPRs are typically based on some form of statistical analysis that relates meaningful predictors to the likelihood of an outcome or condition . Patel et al. take a slightly wider view, incorporating in their definition of CPRs also tools that help classify patients into groups more likely to benefit from treatment. This wider view on CPRs is also adopted in this paper.


Decision support tools


In many medical disciplines, the concept of personalized medicine, or stratified care, has become the paradigm. To facilitate tailoring treatment to the individual patient, there is a need for tools that provide predictions of the course of disease without treatment, as well as conditional upon treatment(s) based on a set of unique patient characteristics. Such tools that support medical decision-making are known under the general term “decision support tools” (DSTs) or “decision support systems.” In general, DSTs are computer-assisted technological aids that can support complex decisions by providing personalized predictions and by facilitating dialogue and exchange of information. These tools may, e.g., be used in multidisciplinary meetings as a training tool or in the encounter between the treating physician and his or her patient. They may be available as an online tool or a mobile application.


We argue that to facilitate medical decision making, multifaceted DSTs are required. We use the term multifaceted to stress the fact that for decision-making, generally, multiple facets of the decision problem need to be addressed. More specifically, a multifaceted DST includes the following:



  • •

    all relevant treatments for the patient population at hand;


  • •

    all relevant patient characteristics;


  • •

    all relevant benefits and harms;


  • •

    an easy to use interface with information presented in an accessible manner.



All relevant benefits refers to the effectiveness of treatments on the most important outcomes such as pain, physical functioning, and health-related quality of life, while harms refer to treatment side effects. However, one may also consider including costs and cost-effectiveness in a DST. Presentation of all relevant outcomes will allow patients to participate in the decision-making process as patients may differ substantially in their preferences and attitude toward risks . Finally, a DST may include explicit elicitation of patient preferences as this may further facilitate the use of the DST for shared decision-making. In LBP, a DST that can facilitate stratified care may consist of several CPRs or classification models that address different aspects of the decision problem. The construction of multifaceted DSTs requires the use of rigorous procedures for development and validation based on dedicated prospective studies and necessitates a procedure for regular updating of the tool.


The aim of this paper is to summarize the state-of-affairs regarding stratified care in LBP and discuss various aspects of developing DSTs that are important to produce clinically useful tools in this field.




Overview of stratified care in LBP


As stated above, various methods (or approaches) have been proposed to optimize clinical decision-making in LBP, such as the use of classification systems or CPRs. Foster et al. define stratified care for low back pain as targeting treatment to subgroups of patients. In their paper, they describe three (partly overlapping) approaches: (1) based on patients’ prognosis, (2) based on underlying causal mechanisms, or (3) based on treatment responsiveness. For each of these approaches, one key example with at least one high-quality RCT is described. First, they conclude that stratified care for low back pain based on patients’ prognosis (i.e., The STarT Back Tool) has demonstrated changes in clinicians’ behavior, benefit for patients, and cost savings in primary care. Second, stratified care based on underlying mechanisms (i.e., Classification Based Cognitive Functional Therapy) is potentially useful, but it was concluded that further clinical evaluation is needed. Third, they conclude that stratified care based on identifying treatment responders (i.e., responding to spinal manipulation) has reached the stage of broad validation but no high-quality studies were available.


A second review, by Fairbank et al. , focused on classification systems including diagnostic, prognostic, and treatment-based systems. They describe a total of 28 classification systems. Five classification systems were termed “treatment based,” i.e., treatments were targeted to different subgroups. Of these five, two (the Canadian Back Institute Classification and the McKenzie classification) were assessed in comparative studies. This means that patients who were treated according to the classification system were compared to patients who received (standard or usual) care not using the classification system. The review concluded that the McKenzie classification did not show any significant reduction in pain when compared to unclassified control patients. For the Canadian Back Institute Classification, it was concluded that using this classification system decreased VAS pain rating, improved physical functioning, and decreased medication use compared to the generic, traditional, treatment strategy. It is important to acknowledge that the researchers used an observational study design. One interesting and important finding is that all the classification systems focused on non-operative management in primary care, more specifically on physiotherapy practice. Fairbank et al. suggest that perhaps the greatest potential for a classification system would be one that can predict which patients should be considered for surgical intervention .


Taking a slightly different approach to evaluate the evidence regarding stratified care for low back pain, Patel et al. focused on randomized clinical trials that aimed to validate the added value of using CPRs to match physical therapy to an individual patient. CPRs were defined as clinical tools with various components drawn from the history, examination, and laboratory tests used to inform treatment choices. They identified 3 RCTs, each of them assessing a different CPR (where treatment was matched to the CPR) against a different comparator. Overall, the conclusion was that there were various methodological shortcomings in these RCTs and that the evidence for the use of CPRs was weak. One of their recommendations for testing whether targeting the treatment according to the CPR is effective and cost-effective is to design a study that randomizes patients to receive the “tool” versus “no tool.” The study by Apeldoorn et al. , (published after the publication of this systematic review) is illustrative of such a design. The study evaluated the effectiveness and cost-effectiveness of the CPR as developed and tested by Brennan (one of the CPRs included in the systematic review by Patel et al. ). The overall conclusion by Apeldoorn et al. was that this classification-based treatment approach was neither effective nor cost-effective in comparison with usual physical therapy care in a population of patients with sub-acute and chronic LBP .


A recent systematic review makes a distinction between prognostic CPRs and prescriptive CPRs for CPRs related to non-surgical management in low back pain. Prognostic CPRs aim to identify variables that predict the course of complaints or prognosis. These variables then could be used to prioritize a patient for an intervention and/or inform the patient about the anticipated course of complaints. Prescriptive CPRs are a special type of prognostic CPR including treatment effect modifiers (at baseline) that inform which patient profiles are more (or less) likely to benefit from a certain treatment. In other words, these prescriptive CPRs aim to inform treatment selection. Despite this distinction, they include all tools that consist of variables related to diagnosis, prognosis, or treatment response. Tools were only included as long as a formal derivation process was followed, in which a larger pool of candidate predictors was refined using multivariate statistical procedures. Of the 30 CPRs identified, validity was evaluated for only three. Their main conclusion was that validation studies and studies evaluating the clinical impact of these CPRs are needed. However, their operationalization of CPRs related to non-surgical treatment led to the exclusion of, among others, the CPR developed by Brennan and the STarT Back Tool, two CPRs for which impact studies are available .


Our summary of the literature regarding stratified care in low back pain highlights some important issues. First of all, the identified reviews describe a few distinct but partly overlapping methods and approaches to optimize clinical decision-making in LBP. There are differences in the operationalization of CPRs and in the included treatment options in each review. Moreover, the majority of the CPRs focus on one treatment option, mainly physiotherapy, often compared to usual care. CPRs that include additional relevant options, including surgical interventions or medication, are missing.


It is clear that developing and implementing a DST that incorporates all relevant treatment options, patient characteristics, and benefits and harms and that is presented in an accessible manner is a challenging enterprise. In the next section of this paper, we will focus on important features of the development and testing of multifaceted DSTs, which require (ideally) a focused and multidisciplinary research program (and not single studies). As an illustration, we will describe a tool that is currently under development, namely the Nijmegen Decision Tool for Chronic Low Back Pain (NDT-CLBP) . The purpose of the NDT-CLBP is to supports secondary or tertiary back care specialists to decide which patients should be considered for a surgical intervention and which patients for specific non-surgical interventions. In the first step, the DST is set up to address the triage of patients to either a surgical specialist or a conservative treatment.




Overview of stratified care in LBP


As stated above, various methods (or approaches) have been proposed to optimize clinical decision-making in LBP, such as the use of classification systems or CPRs. Foster et al. define stratified care for low back pain as targeting treatment to subgroups of patients. In their paper, they describe three (partly overlapping) approaches: (1) based on patients’ prognosis, (2) based on underlying causal mechanisms, or (3) based on treatment responsiveness. For each of these approaches, one key example with at least one high-quality RCT is described. First, they conclude that stratified care for low back pain based on patients’ prognosis (i.e., The STarT Back Tool) has demonstrated changes in clinicians’ behavior, benefit for patients, and cost savings in primary care. Second, stratified care based on underlying mechanisms (i.e., Classification Based Cognitive Functional Therapy) is potentially useful, but it was concluded that further clinical evaluation is needed. Third, they conclude that stratified care based on identifying treatment responders (i.e., responding to spinal manipulation) has reached the stage of broad validation but no high-quality studies were available.


A second review, by Fairbank et al. , focused on classification systems including diagnostic, prognostic, and treatment-based systems. They describe a total of 28 classification systems. Five classification systems were termed “treatment based,” i.e., treatments were targeted to different subgroups. Of these five, two (the Canadian Back Institute Classification and the McKenzie classification) were assessed in comparative studies. This means that patients who were treated according to the classification system were compared to patients who received (standard or usual) care not using the classification system. The review concluded that the McKenzie classification did not show any significant reduction in pain when compared to unclassified control patients. For the Canadian Back Institute Classification, it was concluded that using this classification system decreased VAS pain rating, improved physical functioning, and decreased medication use compared to the generic, traditional, treatment strategy. It is important to acknowledge that the researchers used an observational study design. One interesting and important finding is that all the classification systems focused on non-operative management in primary care, more specifically on physiotherapy practice. Fairbank et al. suggest that perhaps the greatest potential for a classification system would be one that can predict which patients should be considered for surgical intervention .


Taking a slightly different approach to evaluate the evidence regarding stratified care for low back pain, Patel et al. focused on randomized clinical trials that aimed to validate the added value of using CPRs to match physical therapy to an individual patient. CPRs were defined as clinical tools with various components drawn from the history, examination, and laboratory tests used to inform treatment choices. They identified 3 RCTs, each of them assessing a different CPR (where treatment was matched to the CPR) against a different comparator. Overall, the conclusion was that there were various methodological shortcomings in these RCTs and that the evidence for the use of CPRs was weak. One of their recommendations for testing whether targeting the treatment according to the CPR is effective and cost-effective is to design a study that randomizes patients to receive the “tool” versus “no tool.” The study by Apeldoorn et al. , (published after the publication of this systematic review) is illustrative of such a design. The study evaluated the effectiveness and cost-effectiveness of the CPR as developed and tested by Brennan (one of the CPRs included in the systematic review by Patel et al. ). The overall conclusion by Apeldoorn et al. was that this classification-based treatment approach was neither effective nor cost-effective in comparison with usual physical therapy care in a population of patients with sub-acute and chronic LBP .


A recent systematic review makes a distinction between prognostic CPRs and prescriptive CPRs for CPRs related to non-surgical management in low back pain. Prognostic CPRs aim to identify variables that predict the course of complaints or prognosis. These variables then could be used to prioritize a patient for an intervention and/or inform the patient about the anticipated course of complaints. Prescriptive CPRs are a special type of prognostic CPR including treatment effect modifiers (at baseline) that inform which patient profiles are more (or less) likely to benefit from a certain treatment. In other words, these prescriptive CPRs aim to inform treatment selection. Despite this distinction, they include all tools that consist of variables related to diagnosis, prognosis, or treatment response. Tools were only included as long as a formal derivation process was followed, in which a larger pool of candidate predictors was refined using multivariate statistical procedures. Of the 30 CPRs identified, validity was evaluated for only three. Their main conclusion was that validation studies and studies evaluating the clinical impact of these CPRs are needed. However, their operationalization of CPRs related to non-surgical treatment led to the exclusion of, among others, the CPR developed by Brennan and the STarT Back Tool, two CPRs for which impact studies are available .


Our summary of the literature regarding stratified care in low back pain highlights some important issues. First of all, the identified reviews describe a few distinct but partly overlapping methods and approaches to optimize clinical decision-making in LBP. There are differences in the operationalization of CPRs and in the included treatment options in each review. Moreover, the majority of the CPRs focus on one treatment option, mainly physiotherapy, often compared to usual care. CPRs that include additional relevant options, including surgical interventions or medication, are missing.


It is clear that developing and implementing a DST that incorporates all relevant treatment options, patient characteristics, and benefits and harms and that is presented in an accessible manner is a challenging enterprise. In the next section of this paper, we will focus on important features of the development and testing of multifaceted DSTs, which require (ideally) a focused and multidisciplinary research program (and not single studies). As an illustration, we will describe a tool that is currently under development, namely the Nijmegen Decision Tool for Chronic Low Back Pain (NDT-CLBP) . The purpose of the NDT-CLBP is to supports secondary or tertiary back care specialists to decide which patients should be considered for a surgical intervention and which patients for specific non-surgical interventions. In the first step, the DST is set up to address the triage of patients to either a surgical specialist or a conservative treatment.




Toward multifaceted DSTs in LBP


The main goal of a DST in LBP is to predict all important benefits and harms for an individual patient considering the relevant treatment options. To be of clinical value, a DST should be based on high-quality data, allowing robust and methodologically sound estimation of the benefits and harms. Moreover, predictions should be validated in samples other than the one used for development (external validation). Before implementation of the DST in daily practice, an accessible online or mobile software application should be developed. Accessibility of the DST refers to a number of aspects, such as visual attractiveness, understandable risk communication in words or graphs, ease of access (e.g., free access with/without registration), and ease of navigation. All these aspects influence user acceptance and thus ultimately the potential of the DST to contribute to decision-making. Therefore, both during the development and testing phase of the application, input from treating physicians, therapists, and patients is required.


To ensure that the DST remains up-to-date, a procedure for regular updating and monitoring is important . That is, ideally at regular intervals, the predictions of the DST are validated in ongoing cohorts or registry data, and where necessary, the tool’s underlying models are recalibrated. It may be necessary not only to recalibrate the DST in time but also to include new treatments that have been developed. In addition, requirements for the software application may change over time.


Fig. 1 highlights the key steps in the development, presentation, validation, and updating of a multifaceted DST for personalized and shared decision-making in low back pain. For each step, some guidance is provided as to how to proceed, without the intention to be exhaustive.


Tags:
Nov 10, 2017 | Posted by in RHEUMATOLOGY | Comments Off on Decision support tools in low back pain

Full access? Get Clinical Tree
Get Clinical Tree app for offline access