Skip to main content
NCBI home page
Physical Therapy logoLink to Physical Therapy
. 2020 Dec 22;101(3):pzaa219. doi: 10.1093/ptj/pzaa219

A Prospective Observational Study of Emergency Department–Initiated Physical Therapy for Acute Low Back Pain

Howard S Kim 1,2,✉, Jody D Ciolino 3, Nicola Lancki 4, Kyle J Strickland 5, Daniel Pinto 6,7, Christine Stankiewicz 8, D Mark Courtney 9, Bruce L Lambert 10, Danielle M McCarthy 11,12
PMCID: PMC7970627  PMID: 33351942

Abstract

Objective

Low back pain accounts for nearly 4 million emergency department (ED) visits annually and is a significant source of disability. Physical therapy has been suggested as a potentially effective nonopioid treatment for low back pain; however, no studies to our knowledge have yet evaluated the emerging resource of ED-initiated physical therapy. The study objective was to compare patient-reported outcomes in patients receiving ED-initiated physical therapy and patients receiving usual care for acute low back pain.

Methods

This was a prospective observational study of ED patients receiving either physical therapy or usual care for acute low back pain from May 1, 2018, to May 24, 2019, at a single academic ED (>91,000 annual visits). The primary outcome was pain-related functioning, assessed with Oswestry Disability Index (ODI) and Patient-Reported Outcomes Measurement Information System pain interference (PROMIS-PI) scores. The secondary outcome was use of high-risk medications (opioids, benzodiazepines, and skeletal muscle relaxants). Outcomes were compared over 3 months using adjusted linear mixed and generalized estimating equation models.

Results

For 101 participants (43 receiving ED-initiated physical therapy and 58 receiving usual care), the median age was 40.5 years and 59% were women. Baseline outcome scores in the ED-initiated physical therapy group were higher than those in the usual care group (ODI = 51.1 vs 36.0; PROMIS-PI = 67.6 vs 62.7). Patients receiving ED-initiated physical therapy had greater improvements in both ODI and PROMIS-PI scores at the 3-month follow-up (ODI = −14.4 [95% CI = −23.0 to −5.7]; PROMIS-PI = −5.1 [95% CI = −9.9 to −0.4]) and lower use of high-risk medications (odds ratio = 0.05 [95% CI = 0.01 to 0.58]).

Conclusion

In this single-center observational study, ED-initiated physical therapy for acute low back pain was associated with improvements in functioning and lower use of high-risk medications compared with usual care; the causality of these relationships remains to be explored.

Impact

ED-initiated physical therapy is a promising therapy for acute low back pain that may reduce reliance on high-risk medications while improving patient-reported outcomes.

Lay Summary

Emergency department–initiated physical therapy for low back pain was associated with greater improvement in functioning and lower use of high-risk medications over 3 months.

Keywords: Acute Care, Analgesics, Back Pain, Emergency Care, Acute Pain

Introduction

Background

Low back pain afflicts almost one-half of all Americans annually and accounts for nearly 4 million emergency department (ED) visits per year.1,2 A small proportion of these visits leads to an emergent diagnosis (eg, infection, fracture), but the vast majority are due to nonspecific low back pain where symptom relief is the primary goal.3–5 As a result, ED clinicians prescribe more opioids for back pain than for any other diagnosis,6,7 often co-prescribing benzodiazepine and nonbenzodiazepine skeletal muscle relaxants.8,9 Despite substantial prescribing of analgesic medications, outcomes are suboptimal, with 48% of patients reporting persistent functional impairment 3 months after an ED visit for low back pain and 19% reporting continued opioid use.10

ED-initiated physical therapy is an emerging model of care in which physical therapy interventions (eg, exercise, manipulation, mobilization) are delivered directly to the patient within the ED care environment. In this collaborative model of care, the treating ED physician consults the ED physical therapist to perform a bedside diagnostic evaluation, educate the patient on expected symptom trajectory and active strategies to reduce pain, and provide personalized instruction in an evidence-based home exercise regimen.11–16 During a typical 30-minute intervention, the physical therapist utilizes their unique training and expertise to address both the biophysical and psychosocial aspects of acute low back pain.

Although early initiation of physical therapy in the outpatient setting has been demonstrated to improve patient functioning and reduce downstream opioid utilization,17–23 there is little existing research assessing the impact of physical therapy delivered in the ED setting, where outcomes may reasonably differ due to acute pain, the psychosocial stressors of an unplanned health visit, and the relatively resource-constrained care environment (eg, care delivered in hallways). Existing studies of ED-initiated physical therapy are limited to retrospective analyses of electronic health record data,24 and there are no prospective longitudinal studies assessing patient-reported outcomes. Although ED-initiated physical therapy programs are not currently ubiquitous, widespread implementation could be facilitated through the existing infrastructure of inpatient physical therapy services if ED-initiated physical therapy was demonstrated to be an effective intervention.

Objective

The objective of this study was to describe and compare the patient-reported outcomes of pain-related functioning and high-risk medication use over 3 months among ED patients with acute low back pain receiving either ED-initiated physical therapy or usual care. We view this study as a preliminary step to formally testing an ED-initiated physical therapy intervention for acute low back pain in a controlled trial setting.

Methods

Study Design and Setting

This was a prospective observational study of ED patients with acute low back pain who received either ED-initiated physical therapy or usual care, at the discretion of the treating physician. Patients were prospectively enrolled at the index ED visit and provided outcome data over 3 months. The study setting was a single academic ED (>91,000 annual visits) located in Chicago, IL. Study enrollment occurred from May 1, 2018, to May 24, 2019, with data collection concluding on August 31, 2019. This study was approved by the local institutional review board. We adhered to the STROBE Guideline for Cohort Studies in our reporting.25

Study Population

Patients were eligible if they presented to the ED with a complaint of acute low back pain. We defined low back pain utilizing the consensus definition of pain between the twelfth rib and buttocks.26 Pain was considered to be acute if less than 2 weeks in duration10 and if the patient did not have a history of chronic low back pain. We defined chronic as having received a previous lumbar surgery or having previously attended regular visits with a health care provider for episodes of low back pain. We excluded patients who were not adults (<18 years old), were not English speaking, or did not have an email address or telephone number for follow-up data reporting. Additionally, patients with an obvious nonmusculoskeletal etiology for low back pain (eg, shingles, kidney stone), red-flag symptoms that could indicate spinal cord pathology (eg, bladder or bowel incontinence), and a high likelihood of hospital admission were excluded. Patients who met these exclusion criteria after initial study enrollment were later excluded.

Participant Screening and Enrollment Process

The lead study investigator trained research assistants on participant screening and enrollment procedures. Research assistants screened the ED tracking board during normal business hours for patients with a chief complaint relating to low back pain (eg, low back pain, back spasms, “threw back out”). In these potentially eligible patients, the research assistant approached the treating ED physician to confirm the inclusion and exclusion criteria noted above (see Suppl. Fig. 1, for eligibility diagram). Research assistants queried the treating ED physician directly regarding nonmusculoskeletal etiologies of low back pain (“Do you think this patient’s back pain could be explained by another diagnosis, such as shingles or a kidney stone?”), absence of red-flag symptoms (“Does this patient have any red-flag symptoms concerning for spinal cord compression?”), and likelihood of hospital admission (“Do you think this patient is likely to be discharged home?”). We have previously found this simplified question to be the most pragmatic screening approach given that visit disposition (admit vs discharge) can change at the last moment, and waiting until the patient is physically ready for discharge results in multiple missed enrollment opportunities. If the treating physician stated that the patient was potentially eligible, the research assistant then approached the patient to confirm study eligibility using the inclusion and exclusion criteria noted above. Patients provided written informed consent for study participation.

Interventions

Patients received either ED-initiated physical therapy or usual care at the discretion of the treating ED physician. At our institution, we maintain 1 full-time employee as a dedicated ED physical therapist (although multiple physical therapists may rotate through this role) who can be consulted to evaluate and treat patients at the request of the treating physician; the physical therapist does not evaluate patients without a consultation order. Thus, usual care consists of any ED testing or treatment not involving an ED physical therapist in accordance with the treating physicians’ usual and customary practice. This could include diagnostic imaging and laboratory tests, administration and prescribing of analgesic medications, and patient education or reassurance.

Patients in the ED-initiated physical therapy group received an evaluation by the ED physical therapist in addition to any usual care by the ED physician. In a typical ED-initiated physical therapy consultation, the physical therapist performs a bedside patient assessment to arrive at a guideline-based diagnosis, provides anticipatory guidance on expected symptom trajectory, counsels the patient on realistic goal-setting, equips the patient with active strategies to reduce pain, and provides detailed instruction in diagnosis-specific home exercise techniques that maximize early mobilization. Our ED physical therapists utilize a treatment-based classification system to select a rehabilitation-matched intervention based on their clinical assessment of the patient’s disability status, symptom acuity, pain intensity, and response to testing.27 Each low back pain classification maps to a specific course of treatment that includes differing forms of exercise and manipulation or mobilization as indicated. Consistent with international guidelines, we do not utilize passive electrical treatment modalities such as ultrasound or transcutaneous electrical nerve stimulation.28,29

Each patient in the ED-initiated physical therapy group was also provided with a customized home exercise regimen consisting of 3 exercises to be performed daily. We additionally arranged for patients in the ED-initiated physical therapy group to return to the hospital 1 week after their index ED visit for a repeat evaluation with the ED physical therapist. The purpose of this evaluation was twofold: first, to standardize a single follow-up visit for all patients in the ED-initiated physical therapy group, and second, to assess the feasibility of a return ED-initiated physical therapy visit and whether patients retained the home exercise skills explained to them at the index ED visit. The results of this demonstrated performance assessment are published separately.30 In brief, each participant was asked to perform all recommended home exercises (eg, supine lower trunk rotation) while being observed by the physical therapist. The physical therapist rated the participant’s performance, then provided additional hands-on instruction in exercise performance if necessary.

Measurements and Key Outcomes

Participants reported the following clinical and demographic characteristics at enrollment: age, sex, race, education level, chief complaint, duration of low back pain prior to the ED visit, and responses to the Keele STarT Back Screening Tool (STarT), a 9-item survey that allocates patients into 1 of 3 risk-defined subgroups for developing chronic back pain.31 We also collected baseline scores for the outcome measures (described below) to be followed longitudinally. Study participants directly input information into a REDCap study database32 using an electronic tablet. The same REDCap interface was sent to the patient via secured e-mail at 1 week, 1 month, 2 months, and 3 months for longitudinal data collection.

After the ED visit concluded, the research assistant collected the following variables from the medical record using a standardized form: diagnostic imaging received, medications prescribed, total ED length of stay, and treating physician. Diagnostic imaging included plain radiography of the lumbar spine and computed tomography of the lumbar spine or abdomen-pelvis. Magnetic resonance imaging was not included, as it is not customary to obtain magnetic resonance images in the ED setting. Analgesic medications included high-risk medications (opioids, benzodiazepines, or skeletal muscle relaxants) and over-the-counter medications (acetaminophen, nonsteroidal antiinflammatory drugs, or lidocaine). High-risk medications included opioids or other high-risk medications with the potential for central nervous system and respiratory depression.33 Additionally, we planned to describe data specific to opioid medications, standardized by morphine milligram equivalents,34 to explore as a potential outcome in future work.

The primary study outcome was pain-related functioning, which we assessed using both a condition-specific instrument (Oswestry Disability Index [ODI]) and a generic instrument (Patient-Reported Outcomes Measurement Information System pain interference [PROMIS-PI]). The ODI is an extensively validated and reliable instrument for quantifying the degree of disability from low back pain.35 The ODI score ranges from 0 (no disability) to 100 (maximum disability).36 The PROMIS-PI measures the self-reported consequences of any pain on relevant aspects of a person’s life, and higher scores indicate worse functioning; we used the computer adaptive testing format of PROMIS-PI. It has excellent face validity, construct validity with other measures of pain interference, and concurrent validity with decreases in overall health.37 The time frame of interest for the ODI is “at the moment,” meaning that patients provide responses based on their symptoms at the time of the survey, while the time frame of interest for the PROMIS-PI is the average symptom burden over the last week.

The secondary study outcome was high-risk medication use, assessed by binary response to any use of the aforementioned high-risk medications over the last 24 hours via a customized survey (see Suppl. Fig. 2, for medication use survey). This instrument listed common analgesic medications by brand and generic names; a “yes” response to any medication triggered an additional query asking the participant to specify the medication dose (eg, oxycodone, 10 mg) and quantity (eg, 4 pills). We selected a 24-hour period to maximize accuracy in patient recall. We also collected responses to the numeric pain rating scale (NPRS, scored from 0 to 10) at each time point.38 Finally, in participants who received ED-initiated physical therapy, we queried how frequently they performed the recommended home exercises over the last week at each time point.

Data Analysis

Descriptive Statistics

We reported baseline clinical and demographic characteristics using the mean and SD, the median and interquartile range, and the frequency and percentage as appropriate for normal continuous, nonnormal continuous, and categorical variables, respectively. We plotted the observed mean values over time for the primary and secondary outcomes with 95% CIs.

Statistical Modeling

We evaluated the primary outcomes of ODI and PROMIS-PI scores using a linear mixed model with repeated measures on participants—with unstructured covariance—over time, using treatment group as a fixed effect and treating physician as a random effect. This accounts for correlation both within a physician and within a participant. We evaluated the secondary outcome of high-risk medication use employing generalized estimating equations—with participant nested within physician and an exchangeable working correlation structure over time—with a binomial distribution and logit link. We did not perform any imputations for missing data since the frequency was low and the planned modeling techniques were robust when there were small amounts of missing data.

In the models for each outcome evaluated, we utilized the baseline outcome value as a fixed effect and all 4 follow-up time points for analyses. For each model, we used the Wald type III test for fixed effects to evaluate a time-by-group interaction term, which assesses for significant differences in outcome trajectory over time between groups. We then used least squares means and corresponding confidence limits to estimate between-group differences in mean outcome at each given time point and to estimate the change in mean scores over all possible time intervals (relative to 1 week) between groups.

We adjusted all models for the following covariates preselected on clinical rationale: age, sex, race, education, and pain duration at the index ED visit. To evaluate for collinearity of pain-related outcome measures (ODI, PROMIS-PI, NPRS, and STarT), we conducted a Pearson correlation analysis and prespecified the decision to exclude NPRS and STaRT if they correlated with the baseline outcome scores (ODI or PROMIS-PI).39 We evaluated a number of covariance structures for repeated measures (unstructured, autoregressive, Toeplitz, and compound symmetry) and used the Akaike information criterion to evaluate covariance pattern model fit and select the most appropriate structure. For the generalized estimating equations, we evaluated covariance structures for repeated measures using the quasi-likelihood under the independence model criterion. We did not adjust for multiple hypothesis tests given the exploratory nature of this study. All statistical analyses were performed using SAS version 9.4 (SAS Institute Inc, Cary, NC) and 2-tailed testing at the 5% significance level.

Role of the Funding Source

The funders played no role in the design, conduct, or reporting of this study.

Results

Characteristics of Study Participants

A total of 180 patients were approached for enrollment during the study period, and 122 patients (68%) consented. Of the 122 enrolled participants, 21 (17%) were excluded from the final analysis (Fig. 1). Patients excluded from the analysis were older and had fewer over-the-counter medications prescribed at the index ED visit. The final analysis included 101 participants—43 (43%) in the ED-initiated physical therapy group and 58 (57%) in the usual care group.

Figure 1.

Figure 1

Open in a new tab

Participant flowchart.

The median age for the entire sample was 40.5 years; 59% of participants were women, 37% were White, and 36% were Black. Participants were most frequently college educated (41%); 64% presented within 3 days of initial pain onset, and 31% presented within 1 day. The median ODI, PROMIS-PI, and NPRS scores at baseline were 40.0, 64.1, and 7.0, respectively. Table 1 shows the clinical and demographic characteristics of all participants by group.

Table 1.

Baseline Characteristics of Study Participants by Group and Overalla

Characteristic Physical Therapy Group (n = 43) Usual Care Group (n = 58) Total (N = 101)
Age, median (IQR)b 45.0 (35.0–57.0) 38.0 (31.0–53.0) 40.5 (31.5–54.0)
Sex, no. (%) of womenc 27 (62.8) 32 (56.1) 59 (59.0)
Racec
 White 20 (46.5) 17 (29.8) 37 (37.0)
 Black 13 (30.2) 23 (40.4) 36 (36.0)
 Hispanic 06 (14.0) 13 (22.8) 19 (19.0)
 Other 04 (9.3) 04 (7.0) 08 (8.0)
Highest education levelc
 None 01 (2.3) 01 (1.8) 02 (2.0)
 High school/GED 11 (25.6) 17 (29.8) 28 (28.0)
 College 15 (34.9) 26 (45.6) 41 (41.0)
 Graduate or professional school 16 (37.2) 13 (22.8) 29 (29.0)
Pain level, median (IQR) 8.0 (7.0–9.0) 7.0 (6.0–8.0) 7.0 (6.0–8.0)
Initial pain duration
 <1 d 14 (32.6) 17 (29.3) 31 (30.7)
 1–3 d 14 (32.6) 20 (34.5) 34 (33.7)
 3–7 d 09 (20.9) 11 (19.0) 20 (19.8)
 >7 d 06 (13.9) 10 (17.2) 16 (15.8)
Keele STarT score
 Medium risk 11 (25.6) 21 (36.2) 32 (31.7)
 High risk 21 (48.8) 23 (39.7) 44 (43.6)
 Low risk 11 (25.6) 14 (24.1) 25 (24.7)
ODI score, median (IQR) 51.1 (40.0–66.7) 36.0 (20.0–54.0) 40.0 (26.0–58.0)
PROMIS pain interference score, median (IQR)d 67.6 (62.7–71.5) 62.7 (55.2–68.3) 64.1 (56.9–69.6)
Plain radiography of lower back performed 12 (27.9) 16 (27.6) 28 (27.7)
CT of lower back performed 02 (4.7) 09 (15.5) 11 (10.9)
Medications prescribed
at ED visit
 Over-the-counter medication 36 (83.7) 37 (63.8) 73 (72.3)
 Nonsteroidal  anti-inflammatory drug 23 (53.5) 22 (37.9) 45 (44.6)
 Acetaminophen 7 (16.3) 3 (5.2) 10 (9.9)
 Lidocaine patch 27 (62.8) 25 (43.1) 52 (51.5)
 High-risk medication 25 (58.1) 25 (43.1) 50 (49.5)
  Opioid 09 (20.9) 16 (27.6) 25 (24.8)
  Benzodiazepine 08 (18.6) 06 (10.3) 14 (13.9)
  Skeletal muscle relaxant 10 (23.3) 09 (15.5) 19 (18.8)
ED length of stay, min, median (IQR) 223.0 (168.0–307.0) 225.0 (158.0–308.0) 223.0 (167.0–307.0)
Open in a new tab

a Data are presented as number (percentage) of participants unless otherwise indicated. Categorical variables are reported as number (percentage) of participants; continuous variables are reported as median (interquartile range [IQR]). CT = computed tomography; ED = emergency department; GED = general equivalency diploma; ODI = Oswestry Disability Index; PROMIS = Patient-Reported Outcomes Measurement Information System.

b Data for age were missing for 1 participant.

c Data for sex, race, and education were missing for 1 participant.

d Data for baseline PROMIS score were missing for 3 participants.

Overall, 29 unique physicians provided care for the 101 study participants (median = 3/physician; interquartile range = 2–5). Intraclass correlation coefficient (ICC) estimates (ie, physician effect) for the outcomes of ODI and PROMIS-PI were negligible (the ICC was not estimable for ODI; for PROMIS-PI, ICC < 0.01 [P = .266]), while the unstructured covariance parameters estimating covariance between time points were highly significant overall (P < .001 in most cases). Three unique physical therapists provided care for the 43 patients in the ED-initiated physical therapy group, with 1 physical therapist providing the majority of assessments (n = 36; 83.7%).

Evaluation of Collinearity and Model Selection

All baseline measures of pain-related outcomes were correlated (for PROMIS-PI and ODI, r = 0.58; for PROMIS-PI and STarT, r = 0.55; for ODI and STarT, r = 0.55; for ODI and NPRS, r = 0.30).39 For each linear mixed model, we therefore included only the baseline value of the specific outcome of interest as a fixed effect (ie, we did not include additional pain-related outcomes in the model due to collinearity). We selected the unstructured covariance model for use in all linear mixed models because it had the lowest Akaike information criterion of all covariance structures assessed. For the generalized estimating equation models, we used an exchangeable correlation structure.

Main Results

Observed data for ODI, PROMIS-PI, and high-risk medication use are presented in Figure 2. Both groups demonstrated a reduction in the primary outcomes of ODI and PROMIS-PI over time (Fig. 2A–B); however, participants in the ED-initiated physical therapy group had higher baseline scores (ODI = 51.1 vs 36.0; PROMIS-PI = 67.6 vs 62.7). For the secondary outcome of high-risk medication use, both groups demonstrated a reduction over time (Fig. 2C) and for opioid use specifically (Fig. 3A). The proportions of participants reporting opioid use for the total cohort were 25.8%, 11.1%, 11.9%, and 8.8% at 1 week, 1 month, 2 months, and 3 months, respectively. Participants in the ED-initiated physical therapy group reported performing the recommended home exercises a median of 3, 4, 3, and 3 times weekly, respectively, at the follow-up time points.

Figure 2.

Figure 2

Figure 2

Open in a new tab

Observed primary and secondary outcomes over time by treatment group. (A) Mean Oswestry Disability Index (ODI) and 95% CI. Follow-up data were missing for 0, 4, 11, 18, and 21 participants at the baseline, 1-week, 1-month, 2-month, and 3-month time points, respectively. (B) Mean Patient-Reported Outcomes Measurement Information System pain interference (PROMIS-PI) and 95% CI. Follow-up data were missing for 3, 5, 11, 19, and 20 participants at the baseline, 1-week, 1-month, 2-month, and 3-month time points, respectively. (C) Percentage with high-risk medication use and 95% CI. High-risk medication use includes opioids, benzodiazepines, and skeletal muscle relaxants. Follow-up data were missing for 4, 11, 17, and 21 participants at the 1-week, 1-month, 2-month, and 3-month time points, respectively.

Figure 3.

Figure 3

Open in a new tab

Observed opioid use over time by treatment group. (A) Proportion of participants (95% CI) reporting opioid use within the last day at each time point by group. (B) Box and whisker plot of daily morphine milligram equivalents (MME) of opioid use among those reporting opioid use at each time point by group.

Primary Outcome

Adjusted model results are presented in Table 2. For both ODI and PROMIS-PI, the time-by-group interaction term was significant overall (P < .001 and P = .04, respectively). In the between-group comparisons for each time point (Tab. 2), ODI scores in the ED-initiated physical therapy group relative to usual care were higher (ie, worse functioning) in the early follow-up time points and lower (ie, better functioning) at the later time points, although only the comparison at 2 months was statistically significant. Similarly, PROMIS-PI scores in the ED-initiated physical therapy group were higher in the early time points and lower or similar at later time points, although only the comparisons at 1 week and 1 month were significant.

Table 2.

Adjusted Model Results for the Outcomes of ODI, PROMIS-PI, and High-Risk Medication Usea

Outcome Follow-up Time Physical Therapy
Group (Raw)		 Usual Care Group
(Raw)		 Between-Group Difference at Time
Pointb,c 		Change in Mean Score Relative to Score
at Week 1b,d
Mean
(SD)		 No, (%) of Participants Mean
(SD)		 No. (%) of Participants Mean
(95% CI)		 P OR
(95% CI)		 P Mean
(95% CI)		 P OR
(95% CI)		 P
ODI 1 wk 39.38 (19.91) 25.49 (21.52) 7.06 (−0.22 to 14.33) .0572
1 mo 24.82 (17.83) 16.64 (18.74) 0.66 (−6.89 to 8.21) .8622 −6.40 (−13.24 to 0.45) .0669
2 mo 16.62 (17.75) 17.64 (22.59) −9.93 (−18.90 to −0.95) .0306 −16.98 (−25.83 to −8.14) .0003
3 mo 14.39 (17.33) 13.74 (20.72) −7.33 (−16.38 to 1.71) .1107 −14.39 (−23.04 to −5.74) .0014
PROMIS-PI 1 wk 65.95 (19.91) 60.37 (21.52) 5.44 (2.28 to 8.60) .0010
1 mo 57.27 (17.83) 52.98 (18.74) 4.69 (0.57 to 8.81) .0263 −0.75 (−5.10 to 3.60) .7325
2 mo 51.49 (17.75) 51.47 (22.59) −0.29 (−5.03 to 4.46) .9040 −5.73 (−10.67 to −0.78) .0237
3 mo 49.61 (17.33) 49.89 (20.72) 0.31 (−4.23 to 4.85) .8927 −5.13 (−9.90 to −0.36) .0353
High-risk
medication
use		 1 wk 24 (57.14) 19 (34.55) 2.170 (0.83 to 5.64) .1123
1 mo 3 (7.32) 13 (26.53) 0.113 (0.02 to 0.58) .0086 0.05 (0.01 to 0.35) .0024
2 mo 3 (8.57) 7 (14.29) 0.300 (0.06 to 1.56) .1524 0.14 (0.02 to 0.97) .0468
3 mo 1 (3.03) 7 (14.89) 0.099 (0.01 to 1.11) .0607 0.05 (0.004 to 0.58) .0175
Open in a new tab

a ODI = Oswestry Disability Index; OR = odds ratio; PROMIS-PI = Patient-Reported Outcomes Measurement Information System Pain-Interference.

b From the model adjusting for the value at the index emergency department visit, age, sex, race, education, pain duration (and index ODI for high-risk medication use). ODI and PROMIS-PI were evaluated using linear mixed models; high-risk medication use was evaluated using a generalized estimating equation.

c The between-group difference at time point was the statistical model–based estimate of the difference between the mean scores for the emergency department–initiated physical therapy group and those for the usual care group at the given time point.

d The change in mean score relative to score at week 1 was the statistical model–based estimate comparing the change in mean scores over the given time interval (relative to week 1) in the emergency department–initiated physical therapy group compared with the usual care group.

When assessing the change in mean score relative to week 1 (Tab. 2), there was a significant negative estimate for the change in mean ODI and PROMIS-PI scores for the ED-initiated physical therapy group compared with the usual care group (ie, greater decrease in score signifying a greater improvement in functioning) at the 2- and 3-month time points relative to the 1-week time point.

Secondary Outcome

In the model for high-risk medication use, the time-by-group interaction term was significant overall (P = .006). In the between-group comparisons for each time point, the odds of high-risk medication use were higher in the ED-initiated physical therapy group at the 1-week time point and lower at all remaining time points (Tab. 2), although only the 1-month time point was statistically significant. When assessing the change in mean score relative to week 1 (Tab. 2), there was a significant negative estimate for the change in mean scores for the ED-initiated physical therapy group compared with the usual care group (ie, greater decrease in odds of medication use) at the 1-, 2-, and 3-month time points relative to the 1-week time point.

Discussion

In this prospective observational study of ED patients with acute low back pain, we compared 3-month outcomes among patients receiving either ED physical therapy or usual care. ED physical therapy consisted of a physical therapist evaluation during the ED visit as well as a follow-up assessment at 1 week. We accounted for potential confounding through incorporation of baseline outcome values and other pre-specified clinically important covariates in our statistical models, which assessed the association of ED-initiated physical therapy with the outcomes in 2 ways. First, at individual time points, we estimated the difference in mean outcome scores between groups. Patients in the ED-initiated physical therapy group had higher mean scores across all outcomes in the early stages of follow-up and similar to lower scores in the later stages of follow-up, likely reflecting higher pain-related disability at enrollment. Second, we evaluated the effect of the intervention over time using an overall interaction term and specific time interval estimates relative to the 1-week time point. The overall interaction term across all outcomes was significant, indicating a significant intervention effect over time, and the specific time interval estimates at 2 and 3 months demonstrated greater reduction in ODI score, PROMIS-PI score, and high-risk medication use. In sum, these data suggest greater improvements in pain-related functioning and larger reductions in analgesic medication usage among patients receiving ED-initiated physical therapy. It is important to note that while these data provide evidence of association between ED-initiated physical therapy and improved outcomes, we cannot claim this is a causal relationship.

To our knowledge, this is the first prospective study of patient-reported outcomes following a physical therapy intervention delivered in the ED setting. Our findings on the primary outcome of pain-related functioning are consistent with 2 randomized trials of physical therapy in the primary care setting.19,40 Fritz et al found that patients randomized to early vs delayed physical therapy reported greater improvements in ODI at 4 weeks and 3 months. Rhon et al reported similar findings among active duty military personnel. These 2 trials suggest that earlier intervention in the natural history of acute low back pain can mitigate the development of chronic disability and pain.

Nearly two-thirds of patients in this study presented to the ED within 3 days of initial pain onset, and one-third presented on the same day. The very early presentation of ED patients with acute low back pain therefore represents a critical opportunity to intervene into the progression of low back pain symptoms when initial expectations of recovery and pain-related disability are nascent. In administrative claims studies, delay of physical therapy even 2 weeks after injury is associated with higher utilization of downstream care, including advanced imaging, doctor visits, and invasive procedures.17,18,20–22,41–44

Our findings on the secondary outcome of reduced analgesic medication use among patients receiving ED-initiated physical therapy are consistent with prior research indicating that outpatient physical therapy reduces downstream opioid utilization.18,20–22 These findings further support the recent #ChoosePT Campaign sponsored by the American Physical Therapy Association to reduce reliance on opioid medications. Uniquely, we defined analgesic medication usage as a composite outcome of all high-risk medications given that we predicted relatively low absolute numbers of opioid use based on a previously published estimate of 20 percent at 3 months,10 which we expected to be lower given recent secular trends.45,46 Additionally, a singular focus on opioid utilization ignores the high frequency of benzodiazepine and nonbenzodiazepine skeletal muscle relaxant use in treating low back pain symptoms,8,9,24 and the accompanying risks of abuse, dependence, and overdose with these medications.47–51

These study findings must be interpreted in the context of our current understanding of the generally favorable natural history of low back pain, in which the majority of patients experience symptomatic recovery within 3 months.52 Despite this predictable course, a small proportion of patients go on to develop chronic low back pain and long-term opioid use.10 It is within this context that our findings on high-risk medication use are notable, as differences in medication use between patients receiving ED-initiated physical therapy and usual care signify that ED-initiated physical therapy may be a meaningful intervention to reduce opioid use in the long term.

Naturally, the delivery of any additional consultation or intervention in the course of ED care may have negative implications on operational efficiency and length of stay. In a previous qualitative study, some ED physicians expressed concern about decreased patient throughput due to physical therapist consultation. However, in that same study, other physicians cited gains in productivity as ED-initiated physical therapy consultation allowed physicians to focus on other tasks that would better benefit overall department flow, and patients were highly satisfied with the expert evaluation and discharge recommendation of the physical therapist.14 In the present study, overall ED lengths of stay were similar between patients in the ED-initiated physical therapy group and those in the usual care group at a median of 223 and 225 minutes, respectively, despite patients in the ED-initiated physical therapy group reporting greater pain-related disability at baseline. This indicates that an ED-initiated physical therapy intervention can be delivered without detriment to departmental goals of operational efficiency.

Limitations

This was an observational study, and participants were not randomized to the intervention. The decision to refer patients to ED-initiated physical therapy was based on the discretion of the treating ED physician outside of the study protocol, which may have biased treatment allocation. Although we attempted to control for baseline outcome differences using robust models, it is possible that unmeasured confounders were consciously or unconsciously perceived by the treating physician and affected the decision to refer a patient to ED-initiated physical therapy. Additionally, we excluded a total of 21 participants, most commonly due to being hospitalized after initial study enrollment or providing zero follow-up data. Although our intent in excluding hospitalized patients was to focus on the effect of the discrete intervention provided within the ED environment, it is possible that hospitalized patients or those lost to follow-up might have reported different findings.

This study was conducted at a single center with access to a full-time ED physical therapist. These results may not be generalizable to other hospitals with different patient populations or clinical staffing patterns. However, given the lack of any existing longitudinal outcome data on an ED-initiated physical therapy intervention, we view this single-center study as a necessary first step prior to conducting a larger study with a more heterogeneous patient population.

We did not isolate the effect of individual components of the ED-initiated physical therapy intervention, such as anticipatory guidance and reassurance, realistic goal setting, active pain management strategies, or home exercise. It is therefore unclear to what extent specific components of the ED-initiated physical therapy intervention are necessary, sufficient, or efficacious. Additionally, we conducted a 1-week follow-up evaluation with the ED physical therapist for patients in the ED-initiated physical therapy group. This additional “dose” of physical therapy may not be generalizable outside of the study setting, as patients would not return to the ED for repeat physical therapy visits but rather follow-up with an outpatient physical therapist of their choosing.

In this observational study, we present evidence of an association between ED physical therapy for acute low back pain and improvements in pain-related functioning and reductions in analgesic medication usage over 3 months compared with usual care. These findings suggest a potential benefit for physical therapy delivered in the ED setting; however, baseline outcome differences between groups point to potential biases in treatment allocation that could be addressed in a randomized study.

Supplementary Material

Supplementary_Figure1_pzaa219
Click here for additional data file. (120.6KB, pdf)
Supplementary_Figure2_pzaa219
Click here for additional data file. (90.3KB, pdf)

Contributor Information

Howard S Kim, Department of Emergency Medicine, Northwestern University Feinberg School of Medicine, Chicago, Illinois, USA; Center for Health Services and Outcomes Research, Northwestern University Feinberg School of Medicine, Chicago, Illinois, USA.

Jody D Ciolino, Division of Biostatistics, Department of Preventive Medicine, Northwestern University Feinberg School of Medicine, Chicago, Illinois, USA.

Nicola Lancki, Division of Biostatistics, Department of Preventive Medicine, Northwestern University Feinberg School of Medicine, Chicago, Illinois, USA.

Kyle J Strickland, Department of Rehabilitation Services, Northwestern Memorial Hospital, Chicago, Illinois, USA.

Daniel Pinto, Department of Physical Therapy, Marquette University College of Health Sciences, Milwaukee, Wisconsin, USA; Department of Medical Social Sciences, Northwestern University Feinberg School of Medicine, Chicago, Illinois, USA.

Christine Stankiewicz, Department of Rehabilitation Services, Northwestern Memorial Hospital, Chicago, Illinois, USA.

D Mark Courtney, Department of Emergency Medicine, University of Texas Southwestern Medical School, Dallas, Texas, USA.

Bruce L Lambert, Department of Communication Studies, Center for Communication and Health, Northwestern University School of Communication, Chicago, Illinois, USA.

Danielle M McCarthy, Department of Emergency Medicine, Northwestern University Feinberg School of Medicine, Chicago, Illinois, USA; Center for Health Services and Outcomes Research, Northwestern University Feinberg School of Medicine, Chicago, Illinois, USA.

Author Contributions

Concept/idea/research design: H.S. Kim, D.M. Courtney, B.L. Lambert, D.M. McCarthy

Writing: H.S. Kim, J.D. Ciolino, K.J. Strickland, D. Pinto, D.M. Courtney, B.L. Lambert, D.M. McCarthy

Data collection: H.S. Kim, K.J. Strickland

Data analysis: H.S. Kim, J.D. Ciolino, N. Lancki, D.M. Courtney, B.L. Lambert, D.M. McCarthy

Project management: H.S. Kim

Fund procurement: H.S. Kim, B.L. Lambert

Providing participants: H.S. Kim, D.M. Courtney

Providing facilities/equipment: H.S. Kim

Providing institutional liaisons: H.S. Kim, C. Stankiewicz

Consultation (including review of manuscript before submitting): J.D. Ciolino, D. Pinto, C. Stankiewicz, B.L. Lambert, D.M. McCarthy

Ethics Approval

This study was approved by the Northwestern University Institutional Review Board.

Funding

This research was supported by grants from the following sources: US Department of Health and Human Services, Agency for Healthcare Research and Quality (K12HS023011); the Davee Foundation; and US Department of Health and Human Services, National Institutes of Health, National Institute on Drug Abuse (P50DA044121).

Disclosures

The authors completed the ICMJE Form for Disclosure of Potential Conflicts of Interest and reported no conflicts of interest.

References

  • 1. Rui  P, Kang  K, Ashman  JJ.  National Hospital Ambulatory Medical Care Survey: 2016 Emergency Department Summary Tables. 2016. Accessed December 16, 2020. https://www.cdc.gov/nchs/data/nhamcs/web_tables/2016_ed_web_tables.pdf.
  • 2. Blackwell  DL, Lucas  JW, Clarke  TC. Summary health statistics for U.S. adults: National Health Interview Survey, 2012. Vital Health Stat. 2014;10:1–161. [PubMed] [Google Scholar]
  • 3. Waterman  BR, Belmont  PJ  Jr, Schoenfeld  AJ. Low back pain in the United States: incidence and risk factors for presentation in the emergency setting. Spine J.  2012;12:63–70. [DOI] [PubMed] [Google Scholar]
  • 4. Maher  C, Underwood  M, Buchbinder  R. Non-specific low back pain. Lancet. 2017;389:736–747. [DOI] [PubMed] [Google Scholar]
  • 5. Dubosh  NM, Edlow  JA, Goto  T, Camargo  CA  Jr, Hasegawa  K. Missed serious neurologic conditions in emergency department patients discharged with nonspecific diagnoses of headache or back pain. Ann Emerg Med.  2019;74:549–561. [DOI] [PubMed] [Google Scholar]
  • 6. Hoppe  JA, Nelson  LS, Perrone  J, Weiner  SG. Prescribing opioids safely in the emergency department study i, prescribing opioids safely in the emergency department PSI. Opioid prescribing in a cross section of us emergency departments. Ann Emerg Med.  2015;66:253–259.e1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7. Kea  B, Fu  R, Lowe  RA, Sun  BC. Interpreting the national hospital ambulatory medical care survey: United States emergency department opioid prescribing, 2006-2010. Acad Emerg Med.  2016;23:159–165. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8. Kim  HS, McCarthy  DM, Hoppe  JA, Courtney  DM, Lambert  BL. Emergency department provider perspectives on benzodiazepine-opioid co-prescribing: a qualitative study. Acad Emerg Med. 2018;25:15–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9. Kim  HS, McCarthy  DM, Mark Courtney  D, Lank  PM, Lambert  BL. Benzodiazepine-opioid co-prescribing in a national probability sample of ED encounters. Am J Emerg Med.  2017;35:458–464. [DOI] [PubMed] [Google Scholar]
  • 10. Friedman  BW, O'Mahony  S, Mulvey  L, et al.  One-week and 3-month outcomes after an emergency department visit for undifferentiated musculoskeletal low back pain. Ann Emerg Med.  2012;59:128–33 e3. [DOI] [PubMed] [Google Scholar]
  • 11. Fleming-McDonnell  D, Czuppon  S, Deusinger  SS, Deusinger  RH. Physical therapy in the emergency department: development of a novel practice venue. Phys Ther.  2010;90:420–426. [DOI] [PubMed] [Google Scholar]
  • 12. Fruth  SJ, Wiley  S. Physician impressions of physical therapist practice in the emergency department: descriptive, comparative analysis over time. Phys Ther.  2016;96:1333–1341. [DOI] [PubMed] [Google Scholar]
  • 13. Kim  HS, Strickland  KJ, Mullen  KA, Lebec  MT. Physical therapy in the emergency department: a new opportunity for collaborative care. Am J Emerg Med.  2018;36:1492–1496. [DOI] [PubMed] [Google Scholar]
  • 14. Lebec  MT, Cernohous  S, Tenbarge  L, Gest  C, Severson  K, Howard  S. Emergency department physical therapist service: a pilot study examining physician perceptions. IJAHSP. 2010;8:1–12. [Google Scholar]
  • 15. Lebec  MT, Jogodka  CE. The physical therapist as a musculoskeletal specialist in the emergency department. J Orthop Sports Phys Ther.  2009;39:221–229. [DOI] [PubMed] [Google Scholar]
  • 16. Pugh  A, Roper  K, Magel  J, et al.  Dedicated emergency department physical therapy is associated with reduced imaging, opioid administration, and length of stay: a prospective observational study. PLoS One.  2020;15:e0231476. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17. Fritz  JM, Childs  JD, Wainner  RS, Flynn  TW. Primary care referral of patients with low back pain to physical therapy: impact on future health care utilization and costs. Spine (Phila Pa 1976).  2012;37:2114–2121. [DOI] [PubMed] [Google Scholar]
  • 18. Fritz  JM, King  JB, McAdams-Marx  C. Associations between early care decisions and the risk for long-term opioid use for patients with low back pain with a new physician consultation and initiation of opioid therapy. Clin J Pain.  2018;34:552–558. [DOI] [PubMed] [Google Scholar]
  • 19. Fritz  JM, Magel  JS, McFadden  M, et al.  Early physical therapy vs usual care in patients with recent-onset low back pain: a randomized clinical trial. JAMA.  2015;314:1459–1467. [DOI] [PubMed] [Google Scholar]
  • 20. Frogner  BK, Harwood  K, Andrilla  CHA, Schwartz  M, Pines  JM. Physical therapy as the first point of care to treat low back pain: an instrumental variables approach to estimate impact on opioid prescription, health care utilization, and costs. Health Serv Res.  2018;53:4629–4646. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21. Sun  E, Moshfegh  J, Rishel  CA, Cook  CE, Goode  AP, George  SZ. Association of early physical therapy with long-term opioid use among opioid-naive patients with musculoskeletal pain. JAMA Netw Open.  2018;1:e185909. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22. Thackeray  A, Hess  R, Dorius  J, Brodke  D, Fritz  J. Relationship of opioid prescriptions to physical therapy referral and participation for Medicaid patients with new-onset low back pain. J Am Board Fam Med.  2017;30:784–794. [DOI] [PubMed] [Google Scholar]
  • 23. Kazis  LE, Ameli  O, Rothendler  J, et al.  Observational retrospective study of the association of initial healthcare provider for new-onset low back pain with early and long-term opioid use. BMJ Open.  2019;9:e028633. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24. Kim  HS, Kaplan  SH, McCarthy  DM, et al.  A comparison of analgesic prescribing among ED back and neck pain visits receiving physical therapy versus usual care. Am J Emerg Med.  2019;37:1322–1326. [DOI] [PubMed] [Google Scholar]
  • 25. von  Elm  E, Altman  DG, Egger  M, et al.  The strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. Ann Intern Med.  2007;147:573–577. [DOI] [PubMed] [Google Scholar]
  • 26. Dionne  CE, Dunn  KM, Croft  PR, et al.  A consensus approach toward the standardization of back pain definitions for use in prevalence studies. Spine (Phila Pa 1976). 2008;33:95–103. [DOI] [PubMed] [Google Scholar]
  • 27. Alrwaily  M, Timko  M, Schneider  M, et al.  Treatment-based classification system for low back pain: revision and update. Phys Ther.  2016;96:1057–1066. [DOI] [PubMed] [Google Scholar]
  • 28. Qaseem  A, Wilt  TJ, McLean  RM, Forciea  M, for the Clinical Guidelines Committee of the American College of Physicians . Noninvasive treatments for acute, subacute, and chronic low back pain: a clinical practice guideline from the American College of Physicians. Ann Intern Med.  2017;166:514–530. [DOI] [PubMed] [Google Scholar]
  • 29. Stochkendahl  MJ, Kjaer  P, Hartvigsen  J, et al.  National Clinical Guidelines for non-surgical treatment of patients with recent onset low back pain or lumbar radiculopathy. Eur Spine J.  2018;27:60–75. [DOI] [PubMed] [Google Scholar]
  • 30. Kim  HS, Strickland  KJ, Pinto  D, et al.  Evaluating the feasibility of an emergency department physical therapy intervention for acute low back pain. J Acute Care Phys Ther.  2020;11:113–119. [Google Scholar]
  • 31. Hill  JC, Dunn  KM, Lewis  M, et al.  A primary care back pain screening tool: identifying patient subgroups for initial treatment. Arthritis Rheum.  2008;59:632–641. [DOI] [PubMed] [Google Scholar]
  • 32. Harris  PA, Taylor  R, Thielke  R, Payne  J, Gonzalez  N, Conde  JG. Research electronic data capture (REDCap)--a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform.  2009;42:377–381. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33. Dowell  D, Haegerich  TM, Chou  RCDC. Guideline for prescribing opioids for chronic pain--United States, 2016. JAMA. 2016;315:1624–1645. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34. Centers for Disease Control and Prevention . Calculating total daily dose of opioids for safer dosage. 2016. Accessed November 30, 2020. https://www.cdc.gov/drugoverdose/pdf/calculating_total_daily_dose-a.pdf.
  • 35. Fairbank  JC, Pynsent  PB. The Oswestry Disability Index. Spine (Phila Pa 1976). 2000;25:2940–2952  discussion 2952. [DOI] [PubMed] [Google Scholar]
  • 36. Kosloff  TM, Elton  D, Shulman  SA, Clarke  JL, Skoufalos  A, Solis  A. Conservative spine care: opportunities to improve the quality and value of care. Popul Health Manag.  2013;16:390–396. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37. Amtmann  D, Cook  KF, Jensen  MP, et al.  Development of a PROMIS item bank to measure pain interference. Pain.  2010;150:173–182. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38. Childs  JD, Piva  SR, Fritz  JM. Responsiveness of the numeric pain rating scale in patients with low back pain. Spine (Phila Pa 1976).  2005;30:1331–1334. [DOI] [PubMed] [Google Scholar]
  • 39. Evans  JD. Straightforward Statistics For The Behavioral Sciences. Pacific Grove, CA: Brooks/Cole Publishing Co; 1996. [Google Scholar]
  • 40. Rhon  DI, Miller  RB, Fritz  JM. Effectiveness and downstream healthcare utilization for patients that received early physical therapy versus usual care for low back pain: a randomized clinical trial. Spine (Phila Pa 1976).  2018;43:1313–1321. [DOI] [PubMed] [Google Scholar]
  • 41. Gellhorn  AC, Chan  L, Martin  B, Friedly  J. Management patterns in acute low back pain: the role of physical therapy. Spine (Phila Pa 1976).  2012;37:775–782. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42. Horn  ME, Fritz  JM. Timing of physical therapy consultation on 1-year healthcare utilization and costs in patients seeking care for neck pain: a retrospective cohort. BMC Health Serv Res.  2018;18:887. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43. Fritz  JM, Brennan  GP, Hunter  SJ, Magel  JS. Initial management decisions after a new consultation for low back pain: implications of the usage of physical therapy for subsequent health care costs and utilization. Arch Phys Med Rehabil.  2013;94:808–816. [DOI] [PubMed] [Google Scholar]
  • 44. Arnold  E, La Barrie  J, DaSilva  L, Patti  M, Goode  A, Clewley  D. The effect of timing of physical therapy for acute low back pain on health services utilization: a systematic review. Arch Phys Med Rehabil.  2019;100:1324–1338. [DOI] [PubMed] [Google Scholar]
  • 45. Guy  GP  Jr, Zhang  K, Bohm  MK, et al.  Vital signs: changes in opioid prescribing in the United States, 2006-2015. MMWR Morb Mortal Wkly Rep.  2017;66:697–704. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46. Guy  GP  Jr, Zhang  K, Schieber  LZ, Young  R, Dowell  D. County-level opioid prescribing in the United States, 2015 and 2017. JAMA Intern Med.  2019;179:574–576. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47. Jones  CM, McAninch  JK. Emergency department visits and overdose deaths from combined use of opioids and benzodiazepines. Am J Prev Med.  2015;49:493–501. [DOI] [PubMed] [Google Scholar]
  • 48. Overdyk  FJ, Dowling  O, Marino  J, et al.  Association of opioids and sedatives with increased risk of in-hospital cardiopulmonary arrest from an administrative database. PLoS One.  2016;11:e0150214. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49. Park  TW, Saitz  R, Ganoczy  D, Ilgen  MA, Bohnert  AS. Benzodiazepine prescribing patterns and deaths from drug overdose among US veterans receiving opioid analgesics: case-cohort study. BMJ.  2015;350:h2698. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50. Sun  EC, Dixit  A, Humphreys  K, Darnall  BD, Baker  LC, Mackey  S. Association between concurrent use of prescription opioids and benzodiazepines and overdose: retrospective analysis. BMJ.  2017;356:j760. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51. Votaw  VR, Geyer  R, Rieselbach  MM, McHugh  RK. The epidemiology of benzodiazepine misuse: a systematic review. Drug Alcohol Depend.  2019;200:95–114. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52. Friedman  BW, Conway  J, Campbell  C, Bijur  PE, John Gallagher  E. Pain one week after an emergency department visit for acute low back pain is associated with poor three-month outcomes. Acad Emerg Med.  2018;25:1138–1145. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary_Figure1_pzaa219
Click here for additional data file. (120.6KB, pdf)
Supplementary_Figure2_pzaa219
Click here for additional data file. (90.3KB, pdf)

Articles from Physical Therapy are provided here courtesy of Oxford University Press

RESOURCES