What Characterizes People Who Have an Unclear Classification Using a Treatment-Based Classification Algorithm for Low Back Pain? A Cross-Sectional Study
- T.R. Stanton, BScPT, MScRS, PhD, Sansom Institute for Health Research, School of Health Sciences, University of South Australia, GPO Box 2471, Adelaide, South Australia, Australia 5001, and Neuroscience Research Australia, Sydney, New South Wales, Australia.
- M.J. Hancock, BAppSc(Physio), MAppSc, PhD, Faculty of Human Sciences, Macquarie University, Sydney, New South Wales, Australia, and Faculty of Health Sciences, University of Sydney, Sydney, New South Wales, Australia.
- A.T. Apeldoorn, PhD, Department of Epidemiology and Biostatistics and EMGO Institute for Health and Care Research, VU University Medical Centre, Amsterdam, the Netherlands, and Department of Rehabilitation, Medical Centre Alkmaar, Alkmaar, the Netherlands.
- B.M. Wand, PT, PhD, School of Physiotherapy, University of Notre Dame Australia, Fremantle, Western Australia, Australia.
- J.M. Fritz, PT, PhD, Department of Physical Therapy, College of Health, University of Utah, and Physical Therapy Division, Intermountain Healthcare, Salt Lake City, Utah.
- Address all correspondence to Dr Stanton at: tasha.stanton{at}unisa.edu.au.
Abstract
Background A treatment-based classification algorithm for low back pain (LBP) was created to help clinicians select treatments to which people are most likely to respond. To allow the algorithm to classify all people with LBP, additional criteria can help therapists make decisions for people who do not clearly fit into a subgroup (ie, unclear classifications). Recent studies indicated that classifications are unclear for approximately 34% of people with LBP.
Objective To guide improvements in the algorithm, it is imperative to determine whether people with unclear classifications are different from those with clear classifications.
Design This study was a secondary analysis of data from 3 previous studies investigating the algorithm.
Methods Baseline data from 529 people who had LBP were used (3 discrete cohorts). The primary outcome was type of classification, that is, clear or unclear. Univariate logistic regression was used to determine which participant variables were related to having an unclear classification.
Results People with unclear classifications had greater odds of being older (odds ratio [OR]=1.01, 95% confidence interval [CI]=1.003–1.033), having a longer duration of LBP (OR=1.001, 95% CI=1.000–1.001), having had a previous episode(s) of LBP (OR=1.61, 95% CI=1.04–2.49), having fewer fear-avoidance beliefs related to both work (OR=0.98, 95% CI=0.96–0.99) and physical activity (OR=0.98, 95% CI=0.96–0.996), and having less LBP-related disability (OR=0.98, 95% CI=0.96–0.99) than people with clear classifications.
Limitations Studies from which participant data were drawn had different inclusion criteria and clinical settings.
Conclusions People with unclear classifications appeared to be less affected by LBP (less disability and fewer fear avoidance beliefs), despite typically having a longer duration of LBP. Future studies should investigate whether modifying the algorithm to exclude such people or provide them with different interventions improves outcomes.
The treatment-based classification algorithm was developed to help clinicians match patients with acute low back pain (LBP) to the most appropriate intervention.1,2 The broad aim was to improve outcomes for people with LBP because single interventions, when tested in heterogeneous populations with LBP, typically produce small effects.3–5 Early findings from well-conducted randomized controlled trials suggested that using the treatment-based classification approach to match patients with acute LBP to specific interventions results in slightly superior outcomes.6,7 Since those trials, many studies have investigated different aspects of the treatment-based classification approach.2,8–15 These studies may provide insights about how to further improve the classification algorithm.
To provide clear decision-making recommendations and facilitate research, the classification algorithm was developed to be both mutually exclusive (every patient meets 1 and only 1 subgroup) and comprehensive (every patient must meet a subgroup).16 It is the latter feature that we were most interested in for the present study. To allow the algorithm to be comprehensive, additional criteria were added to help guide therapists' decision-making processes for patients who do not clearly fit into a subgroup. Patients for whom the use of these additional criteria are required can be considered to have an unclear classification.
A recent study evaluating the treatment-based classification algorithm for acute or subacute LBP found that 34% of patients had unclear classifications.14 This finding was generally supported by 2 additional studies that used modified versions of the algorithm; in those studies, 26% of patients with subacute or chronic pain8 and 50% of patients with acute or subacute pain17 had unclear classifications. Because the duration of symptoms is a criterion in the classification algorithm, it is possible that the prevalence of unclear classifications is related to the duration of symptoms in the population investigated. Furthermore, the reliability of assigning patients with unclear classifications to a treatment group has been found to be poor and lower than the reliability of assigning patients with clear classifications.14 That is, it seems that the use of additional criteria in the algorithm for patients with unclear classifications (which requires weighing the relative evidence for and against a certain treatment subgroup) results in more between-rater variability when assigning a patient to a treatment subgroup. In contrast, for patients with clear classifications, assignment to a treatment subgroup is more straightforward, and no weighing of relative evidence is necessary.
In addition to poor reliability findings for unclear classifications, a recent study of people with subacute and chronic LBP found that clinical outcomes obtained with the classification-based approach were slightly superior to those obtained with usual physical therapy care only when patients with unclear classifications were excluded from the analysis (using a per-protocol analysis).8 These studies suggested that future improvements to the classification algorithm may involve patients with unclear classifications.
It is currently not known whether patients with unclear classifications are different in some consistent way from patients with clear classifications when the algorithm is used. It is possible that patients with unclear classifications have a presentation suggesting that a treatment approach not included in the current algorithm is more appropriate. In this situation, adding such an intervention to the algorithm could be considered. Alternatively, such patients may represent a group for whom selection of an intervention is not important (eg, highly favorable or very poor prognosis regardless of intervention), and the classification algorithm may not be suitable for use in these patients. What is certain is that a better understanding of patients with unclear classifications has the potential to improve the treatment-based classification algorithm.
Therefore, the aims of this study were to determine the characteristics of people who receive unclear classifications when the classification algorithm is used and to determine whether these characteristics are significantly different from those of people who receive clear classifications when the algorithm is used.
Method
Study Design
This study is a secondary analysis of baseline data from 3 previously published studies investigating the treatment-based classification algorithm: Brennan et al,6 Stanton et al,14 and Apeldoorn et al.8
Setting
Participants were recruited from US private physical therapy clinics from June 2000 through July 20036 and from June 2008 through June 2010,14 from Australian private physical therapy clinics and public hospital physical therapy outpatient departments from June 2008 through June 2010,14 and from private physical therapy clinics in the Netherlands from June 2008 through October 2009.8
Participants
A total of 529 participants were included in this secondary analysis. Participants from all studies were consecutively recruited when seeking care for their back pain. All participants were required to have LBP as their primary complaint (with or without associated leg pain) and to be between 18 and 65 years of age. Two of the data sets consisted of participants with acute or subacute LBP6,14 (defined as a current episode of LBP lasting <90 days) and required a minimal level of disability to be present (modified Oswestry Disability Questionnaire scores of ≥25%6 and ≥20%14). The third data set consisted of participants with subacute or chronic LBP8 (defined as a current LBP episode lasting at least 6 weeks).
For all data sets, patients were excluded if they had any known or suspected serious spinal pathology (eg, fracture, cauda equina compression, or tumor), if they did not have LBP at the time of assessment (ie, inability to reproduce symptoms during a physical examination), or if they were currently pregnant. All studies excluded patients who had lumbar surgery, although how recent the surgery needed to be to cause exclusion differed among the studies (ie, in the past 6 months,14 in the past year,8 or ever6). Additionally, in the study of Stanton et al,14 patients were excluded if they had spinal fusion, scoliosis rods or screws, spinal injections within the last month, sclerotic injections, botulinum toxin injections, or denervation procedures. In the study by Apeldoorn et al,8 patients were excluded if they had severe radiculopathy, spondylolisthesis (grade 2 or more), serious comorbidity, psychopathology, more than 1 low back surgery, more than 1 year of absence from work because of LBP, given birth in the past 3 months, an inability to attend 6 or more regular physical therapy appointments, or moderate complaints about 1 or more items on the Urogenital Distress Inventory (version 6, short form)18 or if they could not speak or read Dutch. In the study of Brennan et al,6 exclusion criteria for patients included a visible lateral shift, acute kyphotic deformity, or signs of nerve root compression (positive results in the straight leg raise test and reflex or strength deficits).
Therapists
Registered physical therapists performed the baseline assessments in each study. All physical therapists received training in the assessment procedures and in the use of a treatment-based classification algorithm. In the study of Brennan et al,6 5 physical therapists considered to be expert in the use of the algorithm (ie, they had been trained with written instructions and had previous research experience with the algorithm) performed the assessments. In the study by Stanton et al,14 the initial assessments were performed by 10 physical therapists from the United States (who had a mean of 7.6 years of experience with the algorithm and who received no additional training) and 16 physical therapists from Australia (who had a mean of 1.2 years of experience with the algorithm and who received an initial 45-minute training session and 1 follow-up session). In the study by Apeldoorn et al,8 4 research physical therapists who were from the Netherlands and had manual therapy qualifications performed the initial assessments (1 physical therapist had 1.5 years of experience with the algorithm, and the remaining 3 had no previous experience with the algorithm; 2 physical therapists had Mechanical Diagnosis and Therapy [MDT] qualifications, and the remaining 2 took a 1-day MDT course; all 4 therapists received 10 hours of formal training in the use of the algorithm).
Baseline Assessments
Participants in all studies completed an 11-point numerical rating scale (0–10) to assess the intensity of current LBP,19 the modified Oswestry Disability Questionnaire to determine LBP-specific disability,20 and the Fear-Avoidance Beliefs Questionnaire (FABQ)21 to measure fear-avoidance beliefs related to work (FABQ work subscale) and physical activity (FABQ physical activity subscale). Participants also completed a pain drawing, on which their symptoms in the last 24 hours were drawn in detail on a body map.22 Data for demographic variables (age, weight, height, and sex) about the participants were also collected.
After completion of the questionnaires, all participants underwent standardized history taking and physical examination by trained physical therapists. The components of a participant's history included duration of LBP, number of previous episodes of LBP, frequency of previous episodes of LBP (increasing, decreasing, no change, or not applicable), best position for symptoms, and worst position for symptoms. The physical examination items that were similar in the 3 studies included assessment of bilateral passive straight leg raise range of motion,23 aberrant movements occurring during active range of motion of the trunk (including instability catch,24 painful arc of movement,25 and reversal of lumbopelvic rhythm or thigh climbing26), lumbar mobility with the lumbar posterior-anterior pressure test (hypomobile, hypermobile, or normal),27,28 pain response to the lumbar posterior-anterior pressure test (yes or no),28 and the prone instability test (positive, negative, or not applicable).29,30
In all 3 studies, a repeated movement examination was used to determine direction-specific responses to movement; however, the methods (eg, repetitions and directions) differed among the studies. Direction-specific responses to movement in the sagittal plane were assessed in all studies (trunk flexion and extension), and status changes were monitored (centralization, peripheralization, or no change). Stanton et al14 assessed immediate changes in both pain intensity (increase, decrease, or no change) and pain location (centralization, peripheralization, or no change) in response to 10 repetitions of various trunk movements2 (flexion, extension, repeated extension and flexion with the participant standing, repeated extension with the participant lying down, sustained prone extension, and repeated flexion with the participant seated). In addition to the lumbar movements listed above, Brennan et al6 assessed status changes (centralization or peripheralization) in response to 10 repetitions2 of side bending and pelvic translation. Apeldoorn et al8 assessed directional preference (defined as centralization or decreased pain intensity for 1 minute in a neutral standing position) and performed a complete MDT assessment with a maximum of 40 movement repetitions, including repeated side gliding and sustained flexion or rotation, if indicated. Only results for direction-specific responses to movement in the sagittal plane were considered in the present study.
Hip medial (internal) rotation passive range of motion31 was assessed in 2 studies.8,14 A neurological baseline assessment (strength, sensation, ankle and knee jerk, and crossed straight leg raise)29 was completed by Stanton et al14 only; in the other 2 studies, a neurological screening examination was used to exclude patients.
Recoding of Baseline Assessment Data
To achieve consistency in baseline findings among data sets, it was necessary to recode numerous variables. These recoding changes included dichotomizing pain drawing findings into symptoms distal to the buttock (yes or no), previous episodes of LBP (yes or no), and frequency of previous episodes of LBP (increasing versus decreasing, no change, or not applicable). For direction-specific responses to movement, assessment findings related to movements in the transverse and frontal planes were excluded because they were not present in all data sets. Centralization and peripheralization findings were recoded from the data sets in the studies by Brennan et al6 and Apeldoorn et al8 to a “yes” or “no” answer to the following question: “Does the participant centralize with 2 or more movements in the same direction or centralize with movement in 1 direction and peripheralize with an opposite movement?”
Version of the Classification Algorithm
Different versions of the classification algorithm were used in all 3 studies. These different versions of the algorithm exist because of the incorporation of new information, specifics of the research protocol, local constraints in the application of the algorithm, or a combination of these factors. Brennan et al6 excluded people who would fit into the traction subgroup and, therefore, used only 3 subgroups in their algorithm (ie, manipulation, stabilization, and direction-specific exercise subgroups). Stanton et al14 used a 4-treatment-subgroup version of the algorithm (ie, similar to the algorithm used by Brennan et al6 but with the addition of the traction subgroup), and Apeldoorn et al8 used a modified, updated version of the 3-treatment-subgroup algorithm. The main modifications in the algorithm used by Apeldoorn et al8 were more inclusive direction-specific exercise criteria, inclusion of postpartum factors in the stabilization subgroup, and—because they also studied chronic LBP—removal of the LBP duration criterion from the manipulation subgroup.
Outcome Measures
On the basis of the original data (questionnaires, history, and findings from physical examination), all participants were classified with the 4-treatment-subgroup algorithm14 (Figure). This required the participants from the studies of Apeldoorn et al8 and Brennan et al6 to be reclassified with the 4-treatment-subgroup algorithm. This version of the algorithm was chosen because it contains information from all currently used subgroups.
Treatment-based classification algorithm for low back pain (4-treatment-subgroup version). FABQ=Fear-Avoidance Beliefs Questionnaire, FABQ-PA=physical activity subscale of the FABQ, FABQ-W=work subscale of the FABQ, ROM=range of motion, SLR=straight leg raise.
The primary outcome of the present study was the type of classification in the 4-treatment-subgroup algorithm: “clear classification” or “unclear classification.” Clear classifications were defined as participants who fit into a subgroup directly (ie, they could be classified using the top part of the algorithm only). Unclear classifications were defined as participants who did not fit into a subgroup directly (ie, the bottom of the algorithm, containing additional subgroup information, had to be used to classify them). Comparisons of clear classifications and unclear classifications for the specific subgroups (ie, manipulation, stabilization, specific exercise, and traction) were not made; comparisons were made for all participants with clear classifications and all participants with unclear classifications.
Data Analysis
All demographic, questionnaire, history, and physical examination variables that were similar in the data sets were considered for inclusion in the analyses. Decisions regarding which variables could plausibly be related to clear classifications versus unclear classifications were made a priori through discussion and consensus of the authors by considering both the clinical value of the variables and recommendations in the literature. The following variables were selected for inclusion in the analyses: age, sex, duration of LBP, presence of leg pain (ie, symptoms distal to the buttock), previous episodes of LBP, frequency of previous episodes of LBP, pain intensity, LBP-related disability, and fear-avoidance beliefs (physical activity and work subscales of the FABQ). There were no missing data for the main outcome (unclear classification or clear classification) or the variables of interest, with the exception of fear-avoidance beliefs (data for physical activity were missing for 11 participants, and data for work were missing for 37 participants), initial pain score (missing data for 7 participants), and frequency of previous episodes of LBP (missing data for 146 participants). No data were imputed for missing values.
Prevalence rates were calculated for clear classifications and unclear classifications. The primary analysis included all participants, regardless of LBP duration. To determine which participant variables were related to having an unclear classification, univariate logistic regression was first completed (outcome: clear classification versus unclear classification) for each of the variables of interest. Next, multivariate logistic regression was completed, placing all variables in the model. Because of the large number of missing data for frequency of previous episodes of LBP, all multivariate analyses were repeated without this variable.
Two sensitivity analyses, identical in method to the above-described analyses, also were completed. The first sensitivity analysis restricted participants to only those with acute or subacute LBP. This analysis was done because the treatment-based classification algorithm was created for use in people with acute or subacute LBP, so it is relevant to determine which unique predictors are present for this participant group. Furthermore, because participants with chronic LBP were included in the main analysis, it is possible that participants with chronic LBP were overrepresented in the unclear classification group. Because of the reclassification of participants with chronic LBP for the 4-treatment-subgroup analysis, participants with chronic LBP had a higher prevalence of unclear classifications (ie, they could not receive a clear classification for the manipulation subgroup because of the symptom duration criterion of <16 days). The second sensitivity analysis restricted participants to only those with chronic LBP. We used IBM SPSS (version 19.0, SPSS Inc, IBM Corporation, Armonk, New York) for all analyses.
Role of the Funding Source
This project was supported by a Physiotherapy Research Foundation tagged Physiotherapists' Registration Board of Western Australia research grant. Dr Stanton was supported by a Canadian Institutes of Health Research Postdoctoral Training Fellowship (ID 223354).
Results
In the primary analysis of all participants (ie, regardless of LBP duration, N=529), 57.3% of the participants received a clear classification when the classification algorithm was used, and 42.7% received an unclear classification. Of participants with acute or subacute LBP (n=446), 61.0% had a clear classification when the algorithm was used, and 39.0% had an unclear classification. Only 37.3% of participants with chronic LBP (31/83) had a clear classification when the algorithm was used; the majority (62.7%) had an unclear classification. Table 1 shows the prevalence of clear classifications for the treatment subgroups, separated by duration of LBP (acute or subacute LBP and chronic LBP).
Prevalence of Clear Classifications Determined With the Algorithm for People With Acute or Subacute Low Back Pain (LBP) and for People With Chronic LBP
The primary univariate analysis results suggested that participants who received unclear classifications had greater odds of being older, having a longer duration of LBP symptoms, having had a previous episode of LBP, having fewer fear-avoidance beliefs related to both work and physical activity, and having less LBP-related disability than participants who received clear classifications (Tab. 2). The multivariate analysis indicated that only LBP-related disability (ie, less disability) was significantly related to having an unclear classification (odds ratio [OR]=0.98, 95% confidence interval [CI]=0.97–0.99) (Tab. 2). That is, for every point increase in the Oswestry Disability Questionnaire score, a 2% decrease in the odds of that participant having an unclear classification would be expected. The findings of the multivariate analysis were unchanged when the variable “frequency of previous episodes of LBP” was removed.
Results of Analyses for All Participants With Low Back Pain (LBP)a
When only participants with acute or subacute LBP were considered, the results of the univariate analysis were similar to those of the primary univariate analysis. Participants who received unclear classifications had greater odds of having a longer duration of LBP symptoms, having fewer fear-avoidance beliefs related to physical activity, and having less LBP-related disability (Tab. 3). However, in the multivariate analysis, only the duration of LBP symptoms remained significant (OR=1.04, 95% CI=1.02–1.05) and sex became significant (OR=0.55, 95% CI=0.32–0.96), with men being less likely than women to receive unclear classifications (Tab. 3). When the variable “frequency of previous episodes of LBP” was removed from the multivariate analysis, only the duration of LBP symptoms remained significant (OR=1.03, 95% CI=1.02–1.04).
Results of Analyses for Participants With Acute or Subacute Low Back Pain (LBP)a
Finally, when only participants with chronic LBP were considered, a longer duration of LBP symptoms and fewer fear-avoidance beliefs related to work characterized participants with unclear classifications (Tab. 4). In the multivariate analysis, low levels of fear-avoidance beliefs related to work remained significantly associated with unclear classifications (OR=0.92, 95% CI=0.85–0.98) (Tab. 4). The findings of the multivariate analysis were unchanged when the variable “frequency of previous episodes of LBP” was removed.
Results of Analyses for Participants With Chronic Low Back Pain (LBP)a
Discussion
Principal Findings
Overall, participants who had unclear classifications tended to be less affected by LBP (ie, lower levels of disability and fewer fear-avoidance beliefs related to work and physical activity). They also had a longer duration of symptoms than participants who had clear classifications. These findings are not particularly surprising. Duration of LBP is an explicit part of the decision-making algorithm; a shorter duration of LBP favors clear classification into the manipulation subgroup. Furthermore, earlier versions of treatment-based classification were not considered to be appropriate for patients with lower levels of disability (ie, stage II).16
Strengths and Weaknesses
By combining the individual participant data from 3 separate studies,6,8,14 we were able to create a reasonably large data set and investigate participant-level factors associated with clear and unclear classifications. The limitation of this approach was that the studies had somewhat different inclusion criteria and were performed in different settings. It is possible that participant characteristics associated with an unclear presentation were different across the studies. For example, the study of Apeldoorn et al8 included participants with LBP lasting longer than 6 weeks, whereas the other 2 studies6,14 included participants with pain lasting less than 3 months. For this reason, we performed secondary analyses for participants with acute or subacute LBP and participants with chronic LBP, yet the findings were similar. The results of the primary analysis included participants from 3 countries, assessed by approximately 30 therapists with different levels of experience in using the treatment-based classification algorithm. Therefore, it is likely that our findings can be generalized to a range of settings.
We chose to classify all participants with the 4-treatment-subgroup algorithm because this version includes all of the subgroups currently used in treatment-based classification and because there is more evidence for the use of the 4-treatment-subgroup algorithm. We believed that analyzing this specific algorithm would be most informative. To do so, we had to reclassify participants from 2 studies.6,8 Because of the reclassification, no participants with subacute or chronic LBP could receive a clear classification for the manipulation subgroup because none could meet the criterion of LBP lasting less than 16 days.
Indeed, when participants with subacute and chronic pain from the study of Apeldoorn et al8 were evaluated with a modified version of the algorithm, only 26% had unclear classifications (versus 62.7% of participants with chronic LBP in the present study). These percentage differences may be explained by modifications of the algorithm itself or by differences in assessment methods (ie, a more thorough direction-specific assessment was completed by Apeldoorn et al,8 whereas in the present study we utilized only the sagittal-plane, direction-specific movement results, possibly resulting in more participants receiving a clear classification for the specific exercise subgroup in the original study8). Regardless, our finding that a large proportion of participants with chronic LBP received unclear classifications when the 4-treatment-subgroup algorithm was used suggests that it is not well suited for people with a longer duration of symptoms. This finding is contingent on the assessment method used in the present study (ie, direction-specific movement assessed only in the sagittal plane). However, this finding does support the original intent of the developers of the subgroup algorithm in targeting its use toward people with acute (ie, stage I) LBP.16
Meaning of the Study
The results of the present study provide some insight into the group of people who receive unclear classifications when the treatment-based algorithm is used. These people appear to be less disabled by LBP, have fewer fear-avoidance beliefs, and have a longer duration of symptoms. This interpretation is consistent with the description of patients for whom the algorithm, in its original presentation, is considered to be appropriate.16 Patients with low levels of disability and a longer duration of symptoms were described as having a different stage of LBP, so that the application of this decision-making algorithm was not considered to be appropriate. The findings of the present study may support the relevance of this perspective of considering such patients to be a separate subgroup.
It is interesting to consider whether the characteristics of patients with unclear classifications indicate that such patients are likely, given current knowledge, to benefit from an approach different from those included in the classification algorithm (direction-specific exercise,9 manipulation,10,11 stabilization exercise,13 and traction12). Practice guidelines for chronic or persistent LBP recommend supervised exercise as the primary intervention.32 Although 2 of the groups in the treatment-based classification algorithm involve some form of exercise (direction-specific exercise and stabilization exercise), neither of these is typically high-intensity, long-duration, supervised exercise, which is widely recommended for persistent LBP.33,34 It seems reasonable that patients who have persistent LBP and low levels of disability and fear avoidance and who do not clearly meet the criteria for more targeted exercise approaches may be good candidates for a more general exercise program that meets the criteria (ie, supervised, high intensity, and long duration) previously shown to be associated with better outcomes from exercise programs.33 Future research could assess the impact of adding such a group to the classification algorithm.
An alternative way forward, based on the results of the present study, is to suggest that people with unclear classifications are typically less affected by LBP (lower levels of disability and fear avoidance) and, therefore, should be initially treated with minimal intervention (good advice and reassurance). This approach was successfully used in the recent STarT Back trial, in which minimal intervention was found to be as effective as (and less expensive than) more intense intervention in people with a more favorable prognosis.35 It would be interesting to assess the prognosis, regardless of treatment, for people with unclear classifications. Indeed, particularly for people with acute or subacute LBP and low disability and fear-avoidance levels, the combination of reassurance, general advice to stay active, and simple analgesics may be an appropriate addition to the algorithm because of the consistent recommendations of this treatment as first-line care for acute or subacute LBP.32
Unanswered Questions and Future Research
The current study describes people who were assigned unclear classifications by the treatment-based classification algorithm. This information can be used to suggest reasonable changes that may improve the classification algorithm. However, the current study provides no evidence that any changes will result in better outcomes. Controlled trails are needed to compare the outcomes of any future modifications of the treatment-based classification algorithm with the outcomes of previous versions or other recommended interventions.
The Bottom Line
What do we already know about this topic?
A treatment-based classification algorithm for low back pain has been developed that aims to match treatments to specific patient subgroups. Recent studies suggest that classifications using the algorithm are unclear for approximately one third of people with low back pain.
What new information does this study offer?
People with low back pain who have unclear classifications tend to be less affected by their back pain (ie, lower levels of disability and fewer fear-avoidance beliefs), despite having a longer duration of low back pain. Outcomes may be improved by modifying the algorithm to exclude such people or provide them with different interventions.
If you're a patient or a caregiver, what might these findings mean to you?
If you have low back pain that has been present for more than 3 months and you are able to do most of your normal daily activities, use of the classification algorithm by your therapist may not be indicated.
Footnotes
-
All authors provided concept/idea/research design and data collection. Dr Stanton, Dr Hancock, Dr Apeldoorn, and Dr Wand provided writing. Dr Stanton and Dr Wand provided data analysis. Dr Stanton provided project management, facilities/equipment, and institutional liaisons. Dr Stanton, Dr Hancock, Dr Wand, and Dr Fritz provided fund procurement. Dr Stanton, Dr Wand, and Dr Fritz provided study participants. Dr Wand and Dr Fritz provided consultation (including review of manuscript before submission). The authors thank Nate Savage, Anne Thackeray, Faris al-Odaibi, Mohammed Qasheesh, and Jake Magel for help in recoding data.
-
A poster presentation of the results of this study was given at the Low Back Pain Forum; October 16–19, 2012; Odense, Denmark.
-
This project was supported by a Physiotherapy Research Foundation tagged Physiotherapists' Registration Board of Western Australia research grant. Dr Stanton was supported by a Canadian Institutes of Health Research Postdoctoral Training Fellowship (ID 223354).
- Received June 26, 2012.
- Accepted November 2, 2012.
- © 2013 American Physical Therapy Association