Clinical prediction rules are tools designed to aid clinicians in making daily clinical decisions. They provide physical therapists with an evidence-based tool to aid patient management in determining a particular diagnosis or prognosis, or in predicting a response to a particular intervention. They are most commonly used in musculoskeletal practice, but can also be used in other areas such as the airway. They are generally prescriptive and have grown in popularity over the past few years, but are they actually logical and factual?
According to Beattie and Nelson (2006), a clinical prediction rule reads: “ is a combination of clinical findings that have shown statistically meaningful predictability in determining a selected condition or prognosis of a patient given a particular treatment ". Steyerberg et al. (2013) state that a CPR is a statistical combination of several predictors, which can be used to calculate the risks of a particular endpoint for individual patients. Basically, it is the process of selecting a treatment based on evidence that the treatment has worked for other people with the same / similar presentation.
CPRs are more complex than diagnostic or prognostic rules because they predict differences in outcomes under different conditions and their development requires experimental studies to investigate effectiveness, followed by validation of the tool itself 2013 study titled PROGnosis RESearch Strategy raised concerns about the CPR release rate without thorough validation from experimental studies.
The PROGRES study was started shortly after a systematic review by Stanton et al. Published in 2010 which found that 13 commonly used CPRs were not validated and therefore concluded that there is little evidence of the use of CPRs for many musculoskeletal practices. This also applies to CPRs, which have been used for back pain, spinal manipulation, and many rehabilitation practices in the middle of the last decade. Essentially, too much emphasis has been placed on creating new CPR instead of validating, updating, and refining existing ones.
A new systematic review published in the Journal of Clinical Epidimiology aims to build on the work of Stanton et al. To be built by reviewing CPRs that have undergone validation tests to predict and critically assess response to physiotherapy-related interventions in ambulatory musculoskeletal disorders associated with lead studies.
Discover the treatment-based classification for LBP
Methods
This systematic review followed PRISMA guidelines and a search strategy and fowchart are available in the article. The Steyerberg definition of CPR was used. Studies were included that aimed to validate a published CPR in which patients with musculoskeletal disorders were stratified to different interventions relevant to outpatient physical therapy. Only studies with a comparator or control arm were included. Detailed inclusion and exclusion criteria are available.
Descriptive features of clinical prediction rule validation studies
Pubmed, EMBASE, CINAHL and Cochrane Library were the databases used for the review. The search was mimicked from 2007 to September 2020, and the justification for this was based on the work of Stanton et al. Build up.
Two review authors checked the search results and excluded irrelevant studies. To review the remaining articles, the two review authors were accompanied by a third. Disagreements about inclusion or exclusion were resolved amicably.
A total of 9 articles were included in the review (reduced from 6,058 studies) and CPR characteristics, reference of the original lead study, study design, setting, population, condition of the musculoskeletal system, proposed treatment, definition of successful outcome, follow-up duration, locked in. Sample size and demographics were all extracted.
The risk of bias was a major focus for the authors, so a two-stage process was used to determine the degree of bias of the included studies. This was done with the PROBAST tool and then the Cochran risk bias tool.
Results & clinical significance
The overview contains seven derivative studies, 6 of which were developed in the USA and one in Spain. Of these three, back pain, two were neck pain, one was carpal tunnel syndrome, and one was shoulder pain. Treatment options within these CPRs included manipulation, traction, movement, or a combination of these treatments. The duration of follow-up ranged from immediately to 8 weeks.
Brief summary of the CPRs in the MSK practice
Results consistent with previous systematic reviews by Stanton et al
Layered screening tools were not the subject of this review
An increasing number of CPRs are published, with v very few CPRs reaching the validation phase
This systematic review does not support the use of CPRs to aid treatment selection in adults with LBP, neck or shoulder pain, or carpal tunnel syndrome
Many validation studies are methodologically flawed and should consider blinding in the future and report results using the same definitions as derivation studies
If using CPR in clinical practice, verify the validity of the tool before using it
A controlled design was not used in any derivative study, although the aim was to develop prediction rules. All had sample sizes of 54-95 participants. The ratio of the number of patients with positive results to the number of variables in the multivariate analysis was between 2 and 5 when PROBAST recommends including at least 20 for a valid analysis.
Nine validation studies are included in the review. Of the CPRs included in a CPR, validation tests were performed in three different studies to predict the response to spinal manipulation in LBP. Each of the remaining CPRs was subjected to a single validation study. Six validation studies had longer follow-up times than the original and six used different primary outcome definitions than the original derivation studies .
Based on the results of this systematic review and because of their quality and accuracy, it is difficult to support the use of clinical prediction tools in musculoskeletal practice. When using CPR, it is important to evaluate the validation of the tool first.
All validation studies were RCTs and overall of good quality. However, it was unclear whether many were blinded, and several validation studies reported outcomes definitions other than derivative studies and the reasons for this were not fully explained.
Looking at the statistics, it seems that all the rules were designed to identify patients who were most likely to respond to treatment, rather than excluding those who would not. Of the validation studies, only one recommended the use of CPR. This was that of Flynn et al. Derived CPR for back pain and spinal manipulations for short-term pain relief. However, two subsequent validation studies failed to replicate the results. All other CPRs did not reach statistical significance and therefore remain invalid.
Based on this systematic review, it is difficult to support the use of clinical predictive tools in musculoskeletal practice. When using CPRs it is important to evaluate the validation of the tool first. For an explanation of what tools are best for back pain and to learn more about the concepts discussed in this article, take the course below.
