Abstract
Background Numerous clinical practice guidelines (CPGs) have been developed to assist clinicians in care options for low back pain (LBP). Knowledge of CPGs has been marginal across health-related professions.
Objective The aims of this study were: (1) to measure US-based physical therapists' knowledge of care recommendations associated with multidisciplinary LBP CPGs and (2) to determine which characteristics were associated with more correct responses.
Design A cross-sectional survey was conducted.
Methods Consenting participants attending manual therapy education seminars read a clinical vignette describing a patient with LBP and were asked clinical decision-making questions regarding care, education, and potential referral. Descriptive statistics illustrating response accuracy and binary logistic regression determined adjusted associations between predictor variables and appropriate decisions.
Results A total of 1,144 of 3,932 surveys were eligible for analysis. Correct responses were 55.9% for imaging, 54.7% for appropriate medication, 62.0% for advice to stay active, 92.7% for appropriate referral with failed care, and 16.6% for correctly answering all 4 questions. After adjustment, practicing in an outpatient facility was significantly associated with a correct decision on imaging. Female participants were more likely than male participants to correctly select proper medications, refer the patient to another health care professional when appropriate, and answer all 4 questions correctly. Participants reporting caseloads of greater than 50% of patients with LBP were more likely to select proper medications, give advice to stay active, and answer all 4 questions correctly. Participants attending more continuing education were more likely to give advice to stay active and older, and more experienced participants were more likely to appropriately refer after failed care.
Limitations There was potential selection bias, which limits generalizability.
Conclusions The survey identified varied understanding of CPGs when making decisions that were similar in recommendation to the CPGs. No single predictor for correct responses for LBP CPGs was found.
Worldwide, low back pain (LBP) is a costly and prevalent condition that can lead to substantial morbidity, high levels of disability, and decreased quality of life. Lifetime, annual, and point prevalence rates of 38.9% to 80%, 38.0%, and 18.3%, respectively, have been reported.1–3 Low back pain symptoms peak between the ages of 40 and 69 years, are higher among females than males in all age groups, and are more common in affluent countries with high incomes.1 Acute or chronic LBP can lead to notable functional limitations and disability.3–6
To assist practitioners in providing the most appropriate evidence-based care for LBP, since 1987, more than 15 different clinical practice guidelines (CPGs) have been produced in multiple countries.7 Clinical practice guidelines can assist practitioners in navigating uncertainties about the multiple LBP treatments available8 and may be multidisciplinary or monodisciplinary. Multidisciplinary CPGs are recommended treatments that are identified as appropriate for all kinds of health care professions that treat the dedicated condition (eg, medicine, chiropractic, and physical therapy), whereas monodisciplinary CPGs reflect the care provided by only one specific discipline (ie, physical therapy).9
Physical therapy is a distinct health care profession that has only recently created a set of monodisciplinary CPGs for LBP.10 Physical therapists commonly treat patients with LBP11 and are involved in the dispensation of a number of interventions (eg, exercises, manipulation) that are part of multidisciplinary guidelines. Despite existing evidence that outcomes for LBP are significantly better when CPGs are adhered to,12–14 notable hurdles have been reported in their dispensation and use in clinical practice, including the clinician's trust and compatibility of the guideline recommendations.15–17 Clinicians are unlikely to use CPGs that: (1) are dissimilar to their own practice model, (2) were perceived as irrelevant to their own patient population, and (3) were dissimilar to the clinician's prevailing treatment philosophy.15,16 Other reasons for disuse may include training levels or lack of understanding of the philosophical context of the CPGs.15–17
Another recognized challenge is that particular clinical disciplines are usually only familiar with their own CPGs (monodisciplinary),18 which typically lack the comprehensiveness of multidisciplinary guidelines.9 Nonetheless, during the care of patients with LBP, it is highly likely that patients will consult with their physical therapists on care-related matters often outside a physical therapist's scope of practice, such as imaging, pharmaceutical interventions, and surgical appropriateness. To date, most LBP multidisciplinary CPGs (worldwide) are very consistent in their recommendations in these areas.7 Furthermore, familiarity with CPGs and actual use of the CPGs in clinical practice settings are 2 distinct elements that may or may not be related. For example, many clinicians may be unfamiliar with the actual CPGs on LBP but instinctively follow the appropriate guideline recommendations in clinical practice settings.
Our objective was to measure the accuracy of treatment selections of US-based physical therapists compared with currently advocated LBP multidisciplinary CPGs. By providing a clinical vignette, decision making was evaluated in lieu of a direct query of familiarity. Secondarily, our objective was to determine whether there are experiences or personal features associated with accurate responses. We hypothesized that, proportionally, more than 75% of physical therapists would appropriately select treatment options that are “in line” with CPGs and that increased experience and dedicated training should be associated with more accurate responses.
Method
Sample
During a 16-month period (January 13, 2012–May 19, 2013), 3,932 rehabilitation clinicians were solicited for participation in this cross-sectional survey. Surveys were administered in 24 states within the contiguous United States and in the state of Hawaii in the following geographic regions: Southwest, Northwest, Northeast, Southeast, and Midwest. Clinicians who were participants of a continuing education course with a focus on manual therapy were the targeted population. These individuals were targeted because it was assumed that the majority of the participants would have contextual knowledge and experience in treatment of LBP and because the course is frequently attended and there was good potential for recruiting higher numbers of participants.
The Survey
The pen-and-paper survey was administered during the registration period of the manual therapy continuing education course, prior to beginning any lecture and following the reading of the human subject's review board–approved informed consent statement. Early solicitation was performed to eliminate biasing the participants regarding LBP treatment options. In addition, it allowed participants to make selections based on their current knowledge of the treatment of LBP.
The survey involved the presentation of a patient case scenario (a clinical vignette). Simple survey tools using clinical vignettes to elicit information about clinical decision making can be valuable in drawing inferences about quality and applicability of care.19,20 The patient case was derived from one of this study's coauthor's case files. It depicted a patient with nonspecific LBP that suggested central canal stenosis–like features. The clinical vignette was shared with 5 academic and clinical physical therapists for input, and modifications were made for clarity. Through feedback, modifications were made to further suggest that the patient did not exhibit “red flag”–oriented symptoms and to support that the condition was mechanical LBP.
Participants were instructed to make decisions about the clinical vignette as a physical therapist who practiced alone in an outpatient setting in an environment of direct access (Appendix). The clinical vignette described a 56-year-old man who reported having bilateral low back and buttock pain. A pain drawing was supplied to identify pain location. The survey participants were instructed that the pain had been present for 3 weeks and that he had a recurrent, long-term history (8 years) of similar LBP and buttock pain. The patient history was designed to reflect lumbar central canal stenosis–like features21 and included consistent clinical identifiers of this condition such as pain during walking and standing and pain relief upon sitting; however, the survey respondents were not provided a diagnosis. The instructions indicated that he was not currently taking medication, his symptoms declined in the evening when he reduced his activity, and he had no history of cancer, rheumatoid arthritis, ankylosing spondylitis, other forms of arthropathy, radiculopathy, or any other red flags. The patient self-treated with rest and diminished movement. Within the clinical vignette, the participants were told they could reproduce the patient's symptoms during dedicated mechanical movements (ie, repeated side-flexion) and that range of motion was limited in all directions.
The survey had 2 distinct components. The first component queried the participants on appropriate treatment options for the patient. The questions included treatment options associated with imaging, medications, self-care choices, and referral of the patient upon 1 month of treatment with no improvement. Additional distracter questions were created to increase the complexity of the vignette. To improve the content validity of the questions, multiple-choice options for each question were designed to provide a wide cadre of potential selections, including a single proper choice advocated by the majority of LBP guidelines7 or by the recommendations outlined by the American College of Physicians22 and subsequently reinforced by the Orthopaedic Section of the American Physical Therapy Association (APTA).10 Each multiple-choice question also included 3 improper choices that are not advocated by CPGs. To ensure clarity and understanding of each item, physical therapy students initially examined the questions, and feedback was sought.
The second component included a series of questions designed to describe the background of the survey participant. Background information included age, sex, years of practice, level of physical therapy degree, additional credentialing, practice setting, continuing education courses attended, journal articles read per month, and percentage and prominence of patient population that are low back oriented.
Data Coding and Variables
Within the survey, multiple variables were collected that were associated with formal and informal training, demographics, practice setting, and years of experience. In selected cases, recoding was used to enhance the ease of interpretation of the data analysis. For example, “possessing additional credentials” was categorized as having APTA-recognized credentials, non–APTA-recognized credentials, or physical therapy–specific credentials. Physical therapy degree was dichotomized as a graduate degree (master's [MPT] or doctorate [DPT]) or non-graduate degree (bachelor's degree or certificate, or both). Transitional DPT graduates were coded as graduate degrees. Practice setting was dichotomized into primary outpatient and primary inpatient. Years of clinical experience was categorized as <10 years, 10–20 years, and 20+ years. The quartile ranking of the percentage of total patients treated as having lumbar spine conditions was dichotomized as over and under the 50% threshold. The clinical questions based on the clinical vignette provided were recoded as having been answered correctly or incorrectly (incorrect answers were coded as any 1 of the 3 wrong answers).
Data Analysis
The survey data were organized and coded using a Microsoft Excel (Microsoft Corp, Redmond, Washington) spreadsheet. Descriptive statistics were reported, including means and standard deviations for the continuous demographic variables (age and years in practice) and frequencies and percentages for categorical variables. Percentages of those participants answering each of the clinical decision-making questions correctly and those participants who answered all questions correctly or incorrectly were separately analyzed. A multicollinearity analysis was conducted using correlation coefficients, variance inflation factors (>3.0 for each covariate), and tolerance values (<0.40 for each covariate). Age and years of experience were found to have a substantial correlation (Pearson r=.83), variance inflation factor values >5.0, and tolerance values <0.40. Re-centering and categorizing these variables resulted in only small changes to reduce collinearity. As such, age was not included in the multivariable models, whereas the variable representing years of experience was included as a covariate because this variable represented a meaningful relationship to guideline responses.
Binary logistic regression was used to determine the association between demographic and clinical practice characteristics and the outcomes of: (1) each the 4 guideline areas separately and (2) whether participants answered all 4 guideline questions correctly. Very few participants answered all 4 guideline questions incorrectly, making analysis of this variable inconsequential. Each model was adjusted for demographic and clinical practice characteristics of the participants. All data analyses were performed using IBM SPSS 20.0 (IBM Corp, Armonk, New York). A P value of ≤.05 was considered significant for all inferential analyses.
Results
A total of 1,226 (31.3%) of the 3,932 surveys were completed, of which 82 were removed for various data-related reasons. Forty-five respondents were not physical therapists, 5 did not answer at least half the clinical scenario questions, and 32 did not provide any demographic data, leaving 1,144 surveys for this analysis.
Table 1 outlines the descriptive statistics of the sample. The median and 25th and 75th quartiles for age and years of experience were 32 (28, 40) and 6 years (2, 14), respectively, suggesting a strong positive skew secondary to a large number of participants being younger and having less clinical experience. A majority of participants possessed a graduate degree. Nearly 91% of the participants worked in a primary outpatient orthopedic practice, and the majority of participants read a relatively small number of journal articles per month (0–3). A high percentage of participants reported treating LBP frequently, with 25% to 50% representing the largest quartile selected.
Descriptive Statistics of the Survey Respondents (N=1,144)a
Table 2 outlines the frequencies and percentages of participants who correctly answered the clinical decision-making questions based on LBP treatment guideline suggestions. Survey respondents were most accurate when given the opportunity to refer the patient to another health care practitioner after failure of improvement with conservative care (92.7% correct). Sixty-two percent of the survey respondents correctly suggested that the patient in the case should stay active. Fewer than 56% of the respondents correctly identified the appropriate pharmaceutical and imaging options for this patient case. Furthermore, only 16.6% of the participants answered all 4 clinical decision-making questions correctly, and a small proportion (n=10, 0.87%) answered all 4 questions incorrectly.
Accuracy Data for the 4 Clinical Decision-Making Questions
Table 3 provides the adjusted associations between demographic and clinical practice characteristics and responding correctly to delay requesting imaging referral. Practice setting and having 20 years or more of clinical experience were significantly associated with the correct response to imaging. Participants practicing in a primary outpatient facility were 1.63 (95% confidence interval [95% CI]=1.06, 2.51) times more likely to correctly delay imaging compared with those not practicing in nonoutpatient facilities. Participants with 20 years or more of clinical experience were 1.92 (95% CI=1.14, 3.24) times as likely to respond correctly compared with clinicians with less than 10 years of clinical experience. Table 3 provides the adjusted associations between demographic and clinical practice characteristics and responding correctly to medication use. Two variables were significant for appropriate selection of medication use: (1) participant's sex and (2) high percentage of patients seen with LBP. Female therapists were 1.63 (95% CI=1.27, 2.08) times more likely than male therapists to correctly identify appropriate medication use. In addition, those treating a high percentage of patients with LBP were 1.36 (95% CI=1.03, 1.81) times more likely than those who do not treat a high percentage of patients with LBP to correctly identify the appropriate medication for use.
Binary Logistic Regression Determining the Association Between Demographic and Clinical Practice Characteristicsa
Adjusted associations between demographic and clinical practice characteristics and responding correctly to advising the patient to stay active are provided in Table 3. Both “high percentage of patients with LBP seen” and “continuing education courses attended” were significantly associated with advising the patient to stay active. Years of clinical experience were significantly associated with correctly answering to advise the patient to stay active. Participants with 10 to 20 years and 20 years or more of clinical experience were strongly associated (1.50 [95% CI=1.07, 2.10] and 2.06 [95% CI=1.20, 3.57], respectively) with correct responses compared with participants with less than 10 years of clinical experience. Participants who see a high percentage of patients with LBP were 1.36 (95% CI=1.02, 1.81) times more likely than participants with a lower percentage of providing LBP care to correctly answer to advise the patient to stay active. Those with a lower amount of attendance (0–2 courses per year) at continuing education were 1.49 (95% CI=1.05, 2.11) times more likely to appropriately advise patients to stay active. Participants' sex and years of experience were significantly associated with appropriately deciding to refer the patient after 30 days of conservative care failing to result in a meaningful change in the patient's symptoms. Female therapists were 1.78 (95% CI=1.09, 2.89) times more likely than men to refer the patient when conservative care did not improve the patient's condition. Furthermore, participants with 20 years or more of clinical experience were 3.98 (95% CI=1.03, 15.4) times more likely to refer the patient after conservative care did not result in a meaningful change in the patient's symptoms (Tab. 3).
One variable was significant for identifying individuals who answered all 4 of the clinical questions appropriately. Female therapists were 1.54 (95% CI=1.09, 2.16) times more likely than men to answer all 4 clinical decision-making questions correctly (Tab. 3).
Discussion
This survey evaluated the treatment selection options of physical therapists when provided with a clinical vignette designed to represent lumbar stenosis. Specifically, the study evaluated accuracy of selected options in comparison with those advocated by most LBP CPGs. Secondarily, we analyzed numerous practice experience characteristics and general demographic components that were related to the accuracy of the CPG questions. Overall, most physical therapists appropriately referred a patient for additional help when there was little progression, but answers associated with imaging, medications, and advice to stay active did not meet the 75% a priori level of accuracy that we hypothesized and were dissimilar to what has been recommended by LBP CPGs. Furthermore, we found that few participant characteristics predicted answering the clinical vignette questions correctly, and there were no universal predictor variables for clinical knowledge, although 20 years or more of experience was related to correctly answering 3 of the 4 questions, and sex and treating a high percentage of patients with LBP were associated with 2 of the 4 CPG questions and answering all 4 questions correctly. It is worth noting that although our sample size was larger than that needed for precision and power estimates, the study group was a sample of convenience and does not reflect the US-based physical therapy population as a whole.
Of particular interest was the accuracy (55.9% and 54.7%) associated with the selection of the appropriate imaging modality and appropriate medication use, respectively. Both of these areas of care are not traditionally addressed within the practice act of a US-based physical therapist; however, both areas were principal content focused on during the advancement of the DPT degree. Physical therapy graduates are required to demonstrate reasonable understanding of imaging recommendations and utility, and all should have a working knowledge of appropriate medication interactions, effectiveness, and recommendations. Within the LBP CPG guidelines, paracetamol (acetaminophen) is the first line of recommendation, followed by nonsteroidal anti-inflammatory drugs (NSAIDs), for short-term LBP.7 The patient represented in this case had only 3 weeks of LBP during this current bout, which qualified him as having short-term LBP, and the other listed options of opioids, tricyclic antidepressants, and corticosteroids are not advocated early options.7 Furthermore, because the case demonstrated symptoms that were of less than 1 month's duration, no imaging would be recommended. This finding is worth noting because some authors have expressed concern regarding the “over-imaging” of LBP, specifically by caregivers within the United States.23–26
Advice to stay active is a basic tenet associated with the care of individuals with LBP and is advocated by nearly all LBP CPGs, regardless of level of acuity.7 We found that both participants with 10 to less than 20 years of experience and those with 20 years or more of experience were inclined to recommend keeping active for the patient in the vignette. For patients with acute pain, moderate-quality evidence supports small benefits in pain relief and functional improvement from advice to stay active compared with advice to rest in bed.27 Sixty-two percent of the respondents selected advice to stay active as the appropriate answer in this survey. In the clinical vignette, the patient reported reduction of symptoms during rest—a finding that was designed to be consistent with most patients with lumbar stenosis. This finding may have been why many physical therapists indicated they would advise the individual to decrease activity because the patient in the vignette reported it diminished his current symptoms. A recent summary review28 for treatment of stenosis suggests that patients with lumbar stenosis should be encouraged to identify activities and situations that cause discomfort and problem solve with the therapist to determine appropriate movement strategies and easing positions. Perhaps this recent publication is associated with our significant finding that those respondents with fewer continuing education courses were more inclined to recommend staying active because they may have had less likelihood of being exposed to the idea of being more cautious with activity.
The accuracy of our findings is similar to that reported in other studies of both physical therapists and other health care professionals. The accuracy of our findings is slightly better than that reported by Simmonds and Derghazarian,16 who investigated the CPG knowledge of Quebec-based physical therapists. Only 12% of Canadian participants were able to identify CPGs for LBP. Hendrick and associates29 reported higher percentages of behaviors “in line” with CPGs, including work recommendations (60% in line), activity recommendations (87.6% in line), and bed rest recommendations (63% in line) for New Zealand-based physical therapists. Worth noting is the difference in the study designs. In our study, we were interested in whether clinicians picked the advocated LBP CPG interventions when provided with a clinical vignette. Previous studies measured only knowledge of the CPG guidelines. We hypothesized that individuals may be unfamiliar with guidelines but may make the proper selections based on a number of factors, including experience and training.
Primary care physicians also are nonadherent to established CPGs for the treatment of LBP, with 4.3% and 26.9% being fully adherent to guidelines for a case with and a case without sciatica, respectively.30 In particular, discordant recommendations with respect to referral for imaging31,32 and passive modalities32 were commonly advocated. Furthermore, orthopedic surgeons are more likely to order more powerful medications than NSAIDs or paracetamol despite CPGs.31 Knowledge barriers to specific LBP CPGs appear to be specialist specific. Dahan and colleagues33 found that general practitioners exhibited significantly lower levels of knowledge of LBP CPGs compared with family physicians. These studies represent those that are both knowledge-based and practice-based with use of clinical vignettes. In both forms of designs, there are notable deficiencies in accuracy of selections with advocated CPGs.
There are several possible explanations for why clinicians with greater clinical experience were more likely to incorporate the recommendations of the CPGs. The current graduate-level curricula may not incorporate these relatively recent CPGs, and the physical therapy monodisciplinary guidelines were published during data collection of this survey. Physical therapy curricula may not adequately emphasize the clinical implications of following these CPGs in the reasoning process, which could lead to a failure to retain what is taught. Younger clinicians may simply lack the necessary relevant experience that would develop and reinforce the intuitive knowledge that early imaging and failure to stay active do not improve clinical outcomes, and failure to improve with 1 month of conservative care warrants physician referral.
Slightly more than 16% of the respondents were correct on all 4 LBP CPG questions. A significantly higher number (53.6%) selected 3 of 4 guideline recommendations correctly; thus, 70% of the sample selected at least 3 of the 4 LBP CPG recommendations correctly. There were very few factors associated with correctly selecting the 4 LBP CPG items. Unlike our initial hypothesis, physical therapy degree (training) and possessing advanced credentialing did not appear to significantly contribute to the accuracy of observed decision making. It is possible that the generally low number of individuals possessing board certification or fellowship training may have played a role in these results. Interestingly, the respondent's sex influenced decision making for 2 questions and the likelihood of getting all 4 decision-making questions correct. Perhaps female therapists are more conservative than male therapists in their approach to examination and treatment of people with LBP. Additionally, there appeared to be some relationship with how frequently patients with LBP are treated. To our knowledge, these findings are novel and warrant further exploration.
Limitations
The majority of respondents had limited years of experience, and most had a DPT degree. As stated, the clinical scenario was designed to reflect acute, recurrent stenosis-related LBP. Although clinical vignettes are commonly used to ascertain knowledge of clinical scenarios,19,20 there is a chance that a different case could have produced different accuracy findings. Furthermore, previous studies demonstrated that knowledge and use of CPGs for LBP is influenced with the perception that the guidelines are dissimilar to their own practice model, are perceived as irrelevant to their own patient population, and are dissimilar to the clinician's prevailing treatment philosophy.15,16 We did not identify any variables associated with these concepts in this study (this was not an objective), and these variables may have more accurately identified explanations toward correctness of responses. The sample represented in this study was a convenience sample that might limit the generalizability of the results. Nevertheless, the sample size was large and provided tight CI values around the point estimates, implying precise findings.
In conclusion, deficiencies in adherence to LBP CPGs were most pronounced in the areas of imaging and medications and were similar to those deficits identified by other health care professionals. There were no universal predictors for American College of Physicians guideline adherence during selection of interventions from a vignette, but the best predictors happened to be female sex and having a high percentage of your caseload the treatment of LBP. The survey study may suggest that a stronger educational strategy for treatment guidelines is warranted.
Appendix.
Survey of Treatment Options for a Clinical Vignette of a Patient With Low Back Paina
a The appendix may not be used or reproduced without written permission from the authors. APTA=American Physical Therapy Association, AAOMPT=American Academy of Orthopaedic Manual Physical Therapists.
Footnotes
Dr Learman, Dr Goode, Mr Showalter, and Dr Cook provided concept/idea/research design and writing. Dr Learman, Ms Ellis, and Mr Showalter provided data collection. Dr Learman, Ms Ellis, Dr Goode, and Dr Cook provided data analysis. Dr Learman and Mr Showalter provided project management and study participants. Dr Learman provided facilities/equipment. Dr Cook provided institutional liaisons. Dr Learman, Ms Ellis, Dr Goode, and Mr Showalter provided consultation (including review of manuscript before submission).
The study was approved and overseen by the Walsh University Human Subject's Review Board.
- Received November 21, 2013.
- Accepted February 21, 2014.
- © 2014 American Physical Therapy Association