How Effective Is Physical Therapy for Common Low Back Pain Diagnoses?
A Multivariate Analysis of 4597 Patients
Study Design. A retrospective review.
Objective. The aim of this study is to evaluate whether the treatment of low back pain with physical therapy results in clinically significant improvements in patient-reported pain and functional outcomes.
Summary of Background Data. Low back pain is a major cause of morbidity and disability in health care. Previous studies have found poor efficacy for surgery in the absence of specific indications. A variety of nonoperative treatments are available; however, there is scant evidence to guide the practitioner as to the efficacy of these treatments.
Methods. Four thousand five hundred ninety-seven patients who underwent physical therapy for the nonoperative treatment of low back pain were included. The primary outcome measures were pre-and post-treatment scores on the Oswestry Disability Index (ODI), Numeric Pain Rating Scale (NPRS) during activity, and NPRS during rest. Previously published thresholds for minimal clinically important difference (MCID) were used to determine the proportion of patients meeting MCID for each of our outcomes. Patients with starting values below the MCID for each variable were excluded from analysis. Logistic regression analysis was used to determine patient risk factors predictive of treatment failure.
Results. About 28.5% of patients met the MCID for improvement in ODI. Presence of night symptoms, obesity, and smoking were predictors of treatment failure for ODI. Fifty-nine percent of patients met the MCID for improvement in resting NPRS, with a history of venous thromboembolism, night symptoms, psychiatric disease, workers' compensation status, smoking, and obesity predictive of treatment failure. Sixty percent of patients met the MCID for improvement in activity NPRS, with night symptoms, workers' compensation status, and smoking predictive of treatment failure.
Conclusion. We observed that a substantial percentage of the population did not meet MCID for pain and function following treatment of low back pain with physical therapy. Common risk factors for treatment failure included smoking and presence of night symptoms.
Level of Evidence: 4
Low back pain (LBP) continues to be a highly prevalent and challenging condition to manage. In addition, it continues to be a major cause of disability and morbidity in health care.[1–5] The etiology of back pain is multifactorial. Although diagnoses, including deformity, stenosis, instability, neoplasm, and infection are generally thought to be amenable to surgical treatment, the management of axial LBP in the absence of the aforementioned conditions remains challenging. There is no clear consensus on its optimal treatment. Treatment for this type of axial LBP has included physical therapy (PT), acupuncture, chiropractic care, injection, and surgery. Numerous studies have called into question the effectiveness of these treatments, particularly surgery, for LBP.[3,5,6] Furthermore, the utilization and costs of surgery have increased significantly over the past two decades. Extensive discussion in the literature on the effectiveness of surgical treatment has led to increased payor oversight. In comparison, there has been relatively little discussion regarding the effectiveness of nonsurgical treatments of LBP. However, a recent systematic review reported that the although the costs of surgery accounted for 5% of expenditures on chronic LBP, physical therapy accounted for 17% of these expenditures.
The Oswestry Disability Index (ODI) and Numeric Pain Rating Scales (NPRS) are commonly utilized and previously validated measures of patient-reported outcome for spinal conditions.[8–10] The effectiveness for treatment of lumbar spinal conditions has been assessed by using minimal clinically important difference (MCID) thresholds in both NPRS and the ODI.[8,9,11,12] If patients' score improvements with treatment do not exceed these MCID thresholds, the treatment is considered to be not effective for the condition.
We hypothesized that physical therapy would result in improved ODI and NPRS scores in patients with common LBP diagnoses. The purpose of this study was to evaluate the effectiveness of physical therapy treatment for common LBP by assessing whether MCID thresholds for NPRS and ODI changes were met following treatment. This investigation further sought to determine whether any patient factors were associated with lack of improvement in these measures.
Materials and Methods
Data were drawn from a registry maintained by a national commercial physical therapy organization. Consecutive patients undergoing physical therapy for LBP-related diagnoses (lumbar sprain, lumbar disc displacement, lumbago, and lumbosacral spondylosis) from September 9, 2013, to April 22, 2015, were enrolled into a prospective database. Data were available on 40 variables comprising demographic information, comorbidities, and pre- and post-treatment outcome scores. Postsurgical patients were excluded from this dataset. In total, data were available on 4597 patients undergoing physical therapy for LBP.
Pre-treatment demographic and risk factors collected included age, gender, body mass index, tobacco use, payor status, duration of treatment, comorbid diagnoses (history of diabetes, hypertension, osteoporosis, stroke, venous thromboembolism, malignancy, respiratory, renal, cardiac, or psychiatric conditions), and symptomatology (sexual dysfunction, nighttime symptoms, or groin numbness). Clinical outcomes included pre- and post-treatment scores on the ODI and NPRS. NPRS scores were recorded both at rest and during activity.
For each of the three outcome scores (ODI, resting NPRS, and activity NPRS), success of treatment was defined as achievement of a MCID between pre- and post-treatment scores. Previously published LBP-specific thresholds for MCID were used. For ODI, the threshold was set at 10 points improvement from pre- to post-treatment score (on a scale of 0 to 100); for NPRS, the threshold was set at 2 points improvement (on a scale of 0 to 10).[8,9]
Microsoft Excel (Microsoft Corp., Redmond, WA) was used for data analysis. Each of the three clinical outcome variables was analyzed separately. For our analysis of ODI, patients with starting scores below the 10 were excluded, as by definition, they could not meet the MCID for improvement. For the remaining patients, the average pre- and post-treatment scores were calculated as well as the percentage of patients meeting or exceeding the MCID for improvement. Paired two-sample t test was utilized to assess the means pre and post-treatment. Bivariate analysis of pre-treatment demographic and risk factors was performed to identify factors associated with failure to meet MCID. Chi-square test was used with a threshold of P < 0.1 for significance. Pre-treatment risk factors meeting this threshold of significance were then included in a logistic regression model to identify independent predictors of treatment failure. A similar analysis was performed for activity and resting NPRS.
Four thousand five hundred ninety-seven patients were included for analysis. The average number of visits was 14. Average duration of therapy was 44.6 days (Table 1).
Oswestry Disability Index
Of 4597 total patients, 986 were excluded due to a starting ODI less than the MCID of 10. This left 3611 patients for analysis. For the sample as a whole, the average ODI score improved from 20.9 to 15.3, for an average improvement of 5.6 points (SEM 0.13, P < 0.001). About 28.5% of patients achieved the MCID of 10 points improvement (Table 2). In logistic regression analysis, night symptoms, obesity, and smoking were independent predictors of failure to meet MCID for ODI (Table 3).
Resting Visual Analog Scale
One thousand five hundred three patients were excluded due to a starting resting NPRS less than the MCID of 2. Thus, 3094 patients were included for analysis. The average score pre-treatment was 4.4 and post-treatment was 2.3, for an average improvement of 2.1 points (SEM 0.04, P < 0.001). Fifty-nine percent of patients met the MCID for improvement (Table 2). In logistic regression analysis, history of venous thromboembolism, night symptoms, psychiatric history, workers' compensation status, smoking, and obesity were all predictive of failure to meet MCID. Osteoporosis was associated with an increased rate of meeting MCID for improvement (Table 4).
Activity Visual Analog Scale
Four hundred twenty-seven patients were excluded due to a starting activity NPRS less than the MCID of 2, leaving 4170 patients for analysis. The average score pre-treatment was 7.2 and post-treatment was 4.5, for an average improvement of 2.7 points (SEM 0.04, P < 0.001). Sixty percent of patients met the MCID for improvement (Table 2). In logistic regression analysis, night symptoms, workers' compensation status, and smoking were independent predictors of failure to meet MCID for improvement. History of cancer was associated with increased odds of meeting MCID for improvement (Table 5).
The treatment of back pain continues to be extremely challenging. Numerous causes of back pain such as spondylolisthesis, scoliosis, neoplasm, infection, and neuro-compressive pathology are generally thought to be responsive to surgical and nonsurgical treatment. However, other causes of axial back pain (such as lumbago, lumbar strain, disc degeneration, disc displacement, and spondylosis) have been suggested to be less effectively treated, particularly by surgery.[3,5] To our knowledge, there is scant quality literature examining the effectiveness of nonsurgical treatment for these common LBP diagnoses using patient-reported outcomes. We sought to examine the effectiveness of physical therapy for these LBP diagnoses.
MCID thresholds for patient-reported outcome measures should be regarded cautiously. There are numerous ways of creating and calculating these thresholds utilizing absolute or percentage improvement.[11–14] The outcome of the analysis and subsequent interpretation of the data may be affected by the choice of threshold. For ODI, MCID thresholds have ranged from 8.2 to 19.9.[11–14] For NPRS, MCID thresholds have ranged from 1.2 to 4.5.[11–14] On the basis of our review of the literature, we selected MCID for ODI to be 10 and for NPRS to be 2.0.
Interestingly, we observed that a substantial percentage of this population did not achieve MCID after physical therapy. For ODI, 71.5% of patients did not achieve a MCID threshold of 10 points improvement. For resting NPRS, 41% of patients did not achieve the MCID of 2 points improvement. For activity NPRS, 40% of patients did not achieve MCID of 2 points improvement. This may be a function of co-occurring comorbidities. Using multiple logistic regression analysis in a large prospective data registry, we identified statistically significant risk factors for not achieving MCID after PT for common LBP causes (Table 3, Table 4 and Table 5). Common risk factors across resting NPRS, activity NPRS, and ODI included smoking and night pain. Our data suggest that patients with these risk factors are less likely to have their LBP successfully treated with physical therapy.
Rodeghero et al. performed a similar analysis to the present study. In their study, they evaluated 6379 patients using a data registry compiled by Focus on Therapeutic Outcomes Inc. (FOTO) (Knoxville, TN). Whereas our study examines previously validated patient reported outcomes such as NPRS and ODI, they utilized Patient Inquiry, which is a computer program developed by FOTO. It is unclear how the Patient Inquiry differs from other patient-reported outcome measures. In their analysis, they report age, duration of symptoms, insurance, and medication use as significant risk factors for poor outcome.
The findings in our study seemingly call into question the effectiveness of PT for common LBP diagnoses, particularly as a high percentage of patients did not achieve MCID in NPRS or ODI. However, the decision to prescribe physical therapy for LBP should be assessed in the context of each patient's unique situation.
From a safety standpoint, physical therapy is vastly superior to interventional treatments such as injections or surgery. Although the likelihood of achieving MCID may be considered low, the likelihood for harm from complication is also low, at least in comparison to other treatments. Thus, from a risk-benefit analysis of various treatments for common LBP diagnoses, physical therapy is still advisable in the majority of patients, particularly as a first line of treatment.
The cost-effectiveness of physical therapy for common LBP diagnoses, however, is unclear. Prior literature has suggested that PT for LBP may account for more than three times the cost of surgical treatment. Cost-effectiveness analysis was outside the scope of the present study and future study is required.
There are some weaknesses of our study that should be noted. The entry of covariate data such as comorbidities was dependent upon the infrastructure of a nationwide physical therapy group. These data were entered by patient self-report as categorical and not continuous and thus have limited granularity for analysis. Furthermore, all potential confounding covariates may not have been recorded. For example, duration of symptoms before PT was not prospectively recorded. Thus, our analysis of risk factors for not achieving MCID should be interpreted cautiously.
Secondly, the exact physical therapy provided may not be uniform across the patient population. Although there have been several randomized controlled trials that have compared the efficacy of various PT modalities for this diagnosis, there is still a lack of consensus on a specific treatment protocol, and decisions as to which modality to employ are often made by the individual therapist. Our intention with this study was not to conduct a well-controlled trial comparing the efficacy of various PT modalities, but rather to document the real-world effectiveness of PT as currently practiced, as well as identifying patient factors predictive of outcome. We believe that this information will be useful to primary care physicians and spine specialists who must choose where to initially refer patients presenting with nonspecific LBP (PT, pain clinic, advanced imaging, etc.).
Finally, our follow-up ranged from one week to 35 weeks. Long-term follow-up data are not available. It is unclear whether patients continued to perform learned exercises on their own afterwards and how well they fared.
However, there are strengths to this study. First, this registry prospectively recorded pre and post-treatment validated patient-reported outcomes data (NPRS, ODI), which allows for a direct assessment of effectiveness. Second, the sheer size of this sample population allows for the identification and quantification of significant risk factors using multivariate analysis. Although there is likely some heterogeneity in how some risk factor data were collected, other risk factors such as payor status and obesity are likely to be collected more homogenously. A data set of this size can allow for analysis of some of these risk factors.
In conclusion, physical therapy for common LBP can be effective for patients; however, it would be prudent to consider risk factors for poor outcome. Although generally considered safe when compared with other treatments of common LBP diagnoses, future analyses in cost are needed to determine the true value of physical therapy in this patient population.
- Chapman JR, Norvell DC, Hermsmeyer JT, et al. Evaluating common outcomes for measuring treatment success for chronic low back pain. Spine 2011;36 (21 Suppl):S54–68.
- Maetzel A, Li L. The economic burden of low back pain: a review of studies published between 1996 and 2001.Best Pract Res Clin Rheumatol 2002;16:23–30.
- Mirza SK, Deyo RA. Systematic review of randomized trials comparing lumbar fusion surgery to nonoperative care for treatment of chronic back pain. Spine 2007;32:816–23.
- Standaert CJ, Friedly J, Erwin MW, et al. Comparative effectiveness of exercise, acupuncture, and spinal manipulation for low back pain. Spine 2011;36 (21 Suppl):S120–30.
- Turner JA, Ersek M, Herron L, et al. Patient outcomes after lumbar spinal fusions. J Am Med Assoc1992;268:907–11.
- Deyo RA, Gray DT, Kreuter W, et al. United States trends in lumbar fusion surgery for degenerative conditions.Spine 2005;30:1441–5.
- Dagenais S, Caro J, Haldeman S. A systematic review of low back pain cost of illness studies in the United States and internationally. Spine J 2008;8:8–20.
- Childs JD, Piva SR, Fritz JM. Responsiveness of the numeric pain rating scale in patients with low back pain.Spine 2005;30:1331–4.
- Davidson M, Keating JL. A comparison of five low back disability questionnaires: reliability and responsiveness.Phys Ther 2002;82:8–24.
- Devine J, Norvell DC, Ecker E, et al. Evaluating the correlation and responsiveness of patient-reported pain with function and quality-of-life outcomes after spine surgery. Spine 2011;36 (21 Suppl):S69–74.
- Parker SL, Mendenhall SK, Shau DN, et al. Minimum clinically important difference in pain, disability, and quality of life after neural decompression and fusion for same-level recurrent lumbar stenosis: understanding clinical versus statistical significance. J Neurosurg Spine 2012;16:471–8.
- Paul AR, Kumar V, Roth SG, et al. 126 Determining the minimally clinical important difference in pain, quality of life, and disability for failed neck and failed back syndromes. Neurosurgery 2015;62(Suppl 1):205–6.
- Copay AG, Glassman SD, Subach BR, et al. Minimum clinically important difference in lumbar spine surgery patients: a choice of methods using the Oswestry Disability Index, Medical Outcomes Study questionnaire Short Form 36, and pain scales. Spine J 2008;8:968–74.
- Van der Roer N, Ostelo RW, Bekkering GE, et al. Minimal clinically important change for pain intensity, functional status, and general health status in patients with nonspecific low back pain. Spine 2006;31:578–82.
- Rodeghero JR, Cook CE, Cleland JA, et al. Risk stratification of patients with low back pain seen in physical therapy practice. Man Ther 2015;20:855–60.
- Dagenais S, Caro J, Haldeman S. A systematic review of low back pain cost of illness studies in the United States and internationally. Spine J 2008;8:8–20.