Predicting three-month functional outcomes after an ED visit for acute low back pain

a b s t r a c t

Background: Nearly 30% of patients who present to an ED with acute, new onset, Low back pain report LBP- related functional impairment three months later. These patients are at risk of chronic LBP, a highly debilitating condition. It has been reported previously that functional impairment, depression, and psychosomatic symptom- atology at the index visit are associated with poor LBP outcomes. We wished to replicate those findings in a co- hort of ED patients, and also to determine if clinical features present at one week follow-up could predict three- month outcomes in individual patients.

Methods: This was a planned analysis of data from a randomized comparative effectiveness study of three anal- gesic combinations conducted in one ED. Patients were followed by telephone one week and three months post- ED visit. The primary outcome was a three-month Roland-Morris Disability Questionnaire (RMDQ) score N 0, in- dicating the presence of LBP-related functional impairment. At the index visit, we measured functional impair- ment (using the RMDQ), depressive symptomatology (using the Patient Health Questionnaire depression module), and psychosomatic features (using the 5-item Cassandra scale). At the one-week follow-up, we ascertained the presence or absence of LBP. We built a logistic regression model in which all the predictors were entered and retained in the model, in addition to socio-demographic variables and dummy variables con- trolling for investigational medication. Results are reported as adjusted odds ratios (adjOR) with 95% CI. To deter- mine if statistically significant associations could be used to predict three-month outcomes in individual patients, we then calculated positive and negative likelihood ratios [LR(+) and LR(-)] with 95% CI for those independent variables associated with the primary outcome.

Results: Of 295 patients who completed the study, 14 (5%) were depressed and 18 (6%) reported psychosomatic symptoms. The median index visit RMDQ score was 19 (IQR: 17, 21) indicating substantial functional impairment. One week after the ED visit, 193 (65%) patients reported presence of LBP. 294 patients provided a three-month RMDQ score, 88 of whom (30%, 95% CI: 25, 35%) reported a score N 0. Neither depression (adjOR 0.7 [95% CI 0.2, 3.1]), psychosomatic symptomatology (adjOR 0.5 [95% CI 0.1, 2.0]), nor index visit functional impairment (adjOR 1.0 [95% CI 1.0, 1.1]) were associated with three-month outcome. Pain at one week was strongly and independently associated with the three-month outcome when examined at the group level (adjOR 4.0 [95% CI 2.1, 7.7]). However, likelihood ratios for pain or its absence at one-week were insufficiently robust to be clinically useful in predicting three-month outcomes in individual patients (LR+: 1.4 [95% CI: 1.3, 1.7]; LR-: 0.4 [95% CI: 0.2, 0.6]).

? We have no conflicts of interest to report.

?? These data were presented at the Society for Academic Emergency Medicine national meeting in New Orleans, LA in May, 2016.

? BWF, PEB, EJG conceived and designed the study. BWF and SG performed the literature review. BWF, AY, RN, and LH supervised the conduct of the trial and data collection. BWF man-

aged the data, including quality control. BWF and SG analyzed the data. BWF drafted the manuscript, and all authors contributed substantially to its revision. BWF takes responsibility for the paper as a whole.

* Corresponding author at: Department of Emergency Medicine, Montefiore Medical Center, Albert Einstein College of Medicine, 111 East 210th Street, Bronx, NY 10467, United States.

E-mail address: bwfriedmanmd@gmail.com (B.W. Friedman).

URL: @benjaminbwf (B.W. Friedman).

http://dx.doi.org/10.1016/j.ajem.2016.11.014

Conclusions: In spite of a strong association at the group level between presence of LBP at one week and functional impairment at three months, when used to predict outcomes in individual patients, presence of pain failed to discriminate with Clinically meaningful utility between acute LBP patients destined to have a favorable versus unfavorable three-month outcome.

Introduction

Low back pain causes more than 2.5 million visits to US emer- gency departments (EDs) annually [1]. Two-thirds of a general popula- tion of ED LBP patients report persistent pain and functional impairment one week after an ED visit for non-traumatic musculoskel- etal LBP [2]. Nearly half of all ED LBP patients will report functional impairment three months later [2]. These patients are at risk for chronic LBP, a debilitating, highly prevalent condition that erodes quality of life and accounts for more than $600 billion in direct and indirect costs annually [3].

Among the subset of ED patients who present with acute, new-onset LBP, most will recover, though more than 20% of this group report mod- erate or severe LBP three months later and 30% report LBP-related func- tional impairment [4]. Identifying those ED patients at risk for progression from acute, new-onset LBP to Chronic back pain will allow integrated healthcare systems to target these patients with more resources, with the long-term goal of preventing a common and debili- tating chronic disease. In previous work, we have demonstrated that, among all ED patients with LBP, those with chronic or frequent recur- rent LBP, or those with more severe functional impairment at the index visit, were more likely to be impaired at three-month follow-up [5]. In this study, we focus entirely on the transition from acute to chronic LBP.

In this planned analysis of data collected during a randomized comparative effectiveness study [4], we assess whether clinical features available during the index ED visit or one week later can predict poor functional outcome three months later, first at the group level, using an adjusted OR as the measure of association, then in a clinically mean- ingful fashion at the level of the individual patient, using likelihood ratios. We utilized three clinical instruments, each of which has previ- ously demonstrated an association with poor outcomes in LBP: the Roland-Morris Disability Questionnaire, which measures the impact of LBP on daily activities [6]; the five-item Cassandra score, which measures general psychosomatic symptomatology [7]; and the two-question depression screener from the Patient Health Questionnaire [8].

Our hypothesis was that one or more of these three instruments at the index visit would be associated with – or predictive of – functional impairment three months later. We also hypothesized that persistent LBP one week after the ED visit would be associated with the three- month outcome and would improve upon the index visit predictors.

Materials and methods

Overview

This was a planned analysis of data from a randomized comparative effectiveness trial in which patients presenting to one urban ED with acute, new onset, non-traumatic, non-radicular musculoskeletal LBP were randomized to treatment with ten days of: naproxen + placebo or naproxen + cyclobenzaprine or naproxen + oxycodone/acetamino- phen [4]. We then conducted a structured follow-up interview by tele- phone one week and three months after the index ED visit. All patients provided written informed consent. This study was reviewed and approved by the Albert Einstein College of Medicine IRB.

Setting

We conducted this study in the ED of Montefiore Medical Center, the academic medical center for the Albert Einstein College of Medicine in the Bronx, New York. The ED receives 110 000 adult visits annually. Sal- aried, trained, bilingual (English and Spanish) technician-level research associates collect data under the supervision of the principal investiga- tors. During the study period, research associates staffed the ED 18-24 h per day, seven days per week.

Subject selection

Patients were considered for inclusion if they were adults aged 21-64 years who presented to our ED for management of acute LBP, defined as pain originating between the lower border of the scapulae and the upper gluteal folds, and received a diagnosis consistent with non-traumatic, non-radicular, musculoskeletal LBP. Patients were re- quired to have functionally impairing back pain, which we defined as a score greater than five on the Roland-Morris Disability Questionnaire [6]. Patients were excluded for radicular pain, defined as pain shooting below the gluteal folds, direct trauma to the back within the previous month, pain duration greater than two weeks, or history of more than one LBP episode per month. We also excluded pregnant or lactating pa- tients, patients whose follow-up phone numbers could not be con- firmed in the ED, patients with allergy or contra-indication to the investigational medications, or chronic opioid use currently or at any time in the past. Patients could only be enrolled once.

Measures

We collected the following data through a structured, in-person oral interview with the patient during the index ED visit using a standard- ized data collection instrument.

- - level of education completed. This served as a surrogate for socio- economic status and was dichotomized at the level of Bachelor’s de- gree.
  - The 24-item Roland-Morris Disability Questionnaire (RMDQ) was completed verbally. The RMDQ is a valid and reproducible standard- ized instrument used in LBP research, reproduced in the Appendix [5]. It includes 24 questions that assess the extent to which low back pain impairs an individual’s ability to perform activities of daily living, such as putting on socks, climbing stairs, and interacting with others. A score of zero indicates no back pain-related functional impairment, while a score of 24 indicates severe functional impair- ment related to back pain. The severity of the acute attack has been associated with poor outcomes in general LBP populations and the RMDQ itself has previously been shown to be associated with poor three-month outcomes in an unselected ED LBP population [5].
  - We asked two depression screening questions from the Patient Health Questionnaire depression module (PHQ2): 1) “Before your back pain began, were you bothered by little interest or pleasure in doing things?” 2) “Before your back pain began, were you feeling down, depressed, or hopeless?” We considered a positive response to be answers of “more than half the days” or “nearly every day” to either of these questions [8,9]. Depression has been reported to be

associated with poor LBP outcomes in a general LBP population [10].

- - - We assessed psychosomatic symptomatology using the five-item Cassandra instrument, a questionnaire that has previously been shown to predict poor outcomes among ED patients with LBP [7]. With this instrument, patients are asked to recall the frequency with which they were bothered by five different symptoms: malaise, getting one’s breath, hot or cold spells, numbness or tingling, or chest pain. Consistent with the methodology utilized by the original authors of the instrument, we used a cut-point of 0.80 (Appendix) to indicate psychosomatic symptomatology [7].
    - One week after the ED visit, we called all patients to determine LBP status, which was assessed on a four point ordinal scale: “severe,” “moderate,” “mild,” or “none.”
    - Three months after the ED visit we again called all patients to deter- mine LBP status, RMDQ score, and medication requirements over the previous 72 h.
Outcomes

The primary outcome measure was functional impairment at 3 months, obtained from the patient’s RMDQ score at final follow-up. We dichotomized this ordinal scale a priori to simplify the outcome and improve its clinical relevance. Patients received either a score great- er than zero on the Roland-Morris questionnaire, indicating the pres- ence of back pain-related functional impairment over the 24 h prior to assessment, or a score of zero, indicating no functional impairment over the same timeframe. We considered a score greater than zero on the RMDQ to constitute an unsatisfactory functional outcome. While mild functional impairment may not be an unreasonable outcome for certain LBP populations, in our study cohort of previously healthy pa- tients, we considered anything less than full functional recovery to be suboptimal.

Secondary outcomes at three months were as follows:

Low back pain itself, assessed on the “severe”, “moderate”, “mild”, or “none” ordinal scale. We considered a patient’s self-report of any pain to constitute an unsatisfactory pain outcome.

Use of any medication for treatment of LBP within 72 h of the three- month phone call, defined by patient response.

Analysis

Index visit characteristics, including age, sex, education level, de- pression, psychosomatic symptoms (Cassandra score), and RMDQ score are reported as mean with standard deviation (SD), median with inter-quartile range (IQR) or frequency with percent and 95% con- fidence interval (95% CI) as appropriate. One week and three-month outcomes are reported as frequency with percent and 95% CI. The bivar- iate associations between our predictors and outcomes of interest are reported as crude odds ratios (OR) with 95% CI.

We then constructed multi-variable logistic models to identify and quantify the independent associations between the hypothesized risk factors at the index visit and our primary outcome variable, the RMDQ score at three months. We repeated this analysis for our two secondary three-month outcome variables, which included any LBP and medica- tion requirements over the preceding 72 h. We included each of our hy- pothesized index visit predictors in the model, as well as age, sex, and educational level, and the medication the patients received in the com- parative effectiveness trial, using SPSS V.21 to conduct all analyses. We present the results of multi-variable logistic regression models as ad- justed odds ratios with 95% confidence intervals. Rather than using arbi- trary bivariate statistical cut-points for inclusion in the multivariable model, we entered and retained all hypothesized predictors and clini- cally relevant index visit covariates in the model. Finally, we re-ran the primary model, this time including the presence of “any LBP at

one week” as a predictor variable, and reported the adjusted OR and 95% CI for this risk factor.

We then used likelihood ratios (reported as LR+ and LR- with 95% CI) to determine the clinical relevance of those predictors that were as- sociated with the primary outcome. Our rationale for the use of LRs was to provide clinicians with clinically actionable data. Aggregate associa- tions calculated using traditional measures such as ORs, although valid at a group or population level, do not readily lend themselves to predic- tion of outcome at the level of the individual patient [11]. For this, the statistical requirements are more rigorous and require not one ratio (such as the OR), but a pair of ratios (LR+/LR- or sensitivity/specifici- ty) that reflect diagnostic test properties applicable to individual pa- tients [12].

Results

323 patients were enrolled in the comparative effectiveness trial during a 30 month period beginning in April 2012 (Fig. 1). Of these, 295 (91%) had complete outcome data three months after the initial visit. Index visit characteristics of these patients are presented in Table 1. Severe functional impairment, as measured by the RMDQ was common. Depression and psychosomatic symptomatology were uncommon.

One week after the index ED visit, 193 (65%, 95% CI: 60, 71%) patients reported persistence of LBP. By three months after the initial ED visit, 88 of 294 patients (30%, 95% CI: 25, 35%) reported functional impairment,

97 of 295 patients (33%, 95% CI: 28, 38%) reported pain, and 66 of 295

patients (22%, 95% CI: 18, 27%) reported taking analgesics for LBP in the 72 h preceding the follow-up phone call.

Fig. 1. Flow diagram.

Table 1

Index visit characteristics of 295 patients included in the analysis.

Variable Value

(n = 295)

Age in years, mean (SD) 39 (11)

Female, n (%) 151 (51%)

Education, n (%)

Did not graduate high school 80 (27%)

Graduated high school, no Bachelor’s degree 176 (60%)

>=Bachelors degree 35 (12%)

Missing 4

Depression screen positive^a	14 (5%)
Missing	1
Psychosomatic symptoms^b	18 (6%)
RMDQ index visit, median (IQR)^c	19 (17, 21)
Medication allocation
Naproxen + placebo, n (%)	96 (33%)
Naproxen + cyclobenzaprine, n (%)	99 (34%)
Naproxen + oxycodone/acetaminophen, n (%)	100 (34%)

^a Positive depression screen was a response of “often” or “always” to either one of two depression screening questions from the PHQ instrument.

^b A positive psychosomatic symptomatology test was a score >=0.80 on the 5-item Cas- sandra instrument, as detailed in the Appendix.

^c 0 to 24 scale on which 0 = no functional impairment and 24 = severe functional impairment.

LBP-related functional impairment at three months, as reflected by an RMDQ score N 0, was not associated with depression, psychosomatic symptomatology, or LBP-related functional impairment at the index visit (Table 2). After controlling for other index visit characteristics, these associations were unchanged (Table 3). However, presence of LBP at one week was independently associated with three-month out- comes including functional impairment (adjusted OR: 4.0, 95% CI: 2.1, 7.7), pain (adjusted OR: 3.4, 95% CI: 1.9, 6.3), and use of medication (ad- justed OR: 2.6, 95% CI: 1.3, 5.0).

Although the presence of LBP at one week as a dichotomous predic- tor of adverse longer-term group outcomes is strongly associated with poor functional status at three months, it does not accurately discrimi- nate between those individuals destined to experience good longer- term functional status at three months from other patients who will continue to suffer some degree of impairment due to LBP (Table 4).

Limitations

At the three-month follow-up, we asked patients to describe low back symptomatology during the prior 72 h. We did not ask patients to differentiate between persistent LBP (unremitting LBP since the index ED visit) versus recurrent LBP (LBP that initially improved but then re-emerged and was present at three-month follow-up). Because

of this limitation, we cannot determine whether these two patterns of LBP perform differently as predictors of chronicity in individual patients. A second limitation is that in the interest of optimizing homogeneity for the randomized clinical trial, many patients were screened for this study but ultimately not included because they did not meet our rela- tively strict entry criteria. Thus, the study participants analyzed repre- sent only a subset of patients who present to the ED with non- traumatic, non-radicular acute LBP. These results cannot therefore be generalized to patients with other types of LBP, nor do the findings ex-

tend to those suffering from chronic or frequent intermittent LBP.

A third limitation is that this study was conducted in one urban ED serving a socio-economically depressed population. Because back Pain outcomes may be associated with socio-economic variables, our results can most appropriately be generalized to EDs that serve similar disad- vantaged patient populations.

A fourth limitation was our choice of a highly sensitive but less spe- cific primary outcome–an RMDQ score N 0. By choosing any functional impairment at all as our primary outcome, we recognize we may have set the bar for a good outcome quite high. This is a consequence of our a priori decision to classify transition from no functional impairment to any degree of functional impairment as an unsatisfactory clinical outcome.

A fifth limitation was that very few of our patients reported psycho- somatic or depressive symptoms. Therefore, our results are most appro- priately generalized to populations with similarly low frequencies of depression and psychosomatic symptomatology. We do not know if the depression and psychosomatic instruments would have performed better in a population with a higher prevalence of these findings.

Finally, we did not follow our patients beyond three months. While

we may have obtained a more accurate assessment of their long-term prognosis with follow-up of longer duration, we believed that the three-month follow-up provides a reasonable approximation of long- term outcomes and is feasible in an ED population.

Discussion

One-third of patients will report functional impairment three months after an ED visit for acute, non-traumatic, non-radicular low back pain (LBP). These patients are at risk of chronic LBP, a devastating illness responsible for more than $600 billion in direct and indirect costs annually [3]. We failed to identify clinically useful predictors of poor outcome in individual patients. In spite of a strong group association be- tween LBP at one week and functional impairment at three months, when LBP at one week was used as a diagnostic outcome predictor for individual patients, it failed to discriminate with clinically meaningful utility between those individuals destined to have a favorable versus unfavorable three-month functional outcome.

Table 2

Bivariate associations between risk factors (index visit predictors) obtained at the index ED visit and three-month outcomes, expressed as crude odds ratios (ORs) with 95% CI’s.

Predictors Outcomes

Any functional impairment^a Any low back pain^b Required analgesics^c

	Crude OR	95% CI		Crude OR	95% CI		Crude OR	95% CI
Depression^d
Positive screen	0.6	0.2, 2.3		0.8	0.2, 2.6		1.4	0.4, 4.6
Negative screen Psychosomatic symptomatology^e	1.0			1.0			1.0
Positive test	0.5	0.1, 1.6	0.6		0.2, 1.8	1.0		0.3, 3.1
Negative test	1.0		1.0			1.0
RMDQ^f, per 1 point increment	1.0	1.0, 1.1	1.0		1.0, 1.1	1.0		0.9, 1.1

Variables whose 95% CI does not cross 1.0 are considered statistically significant. None of these variables satisfied this criterion.

^a Functional impairment defined as Roland-Morris Disability Questionnaire score at 3 months greater than 0, indicating some functional impairment.

^b “Severe,” “moderate,” or “mild” pain.

^c Use of any analgesic medication for LBP within previous 72 h.

^d Positive depression screen was a response of “often” or “always” to either one of two depression screening questions from the PHQ instrument.

^e A positive psychosomatic test was a score >=0.80 on the 5-item Cassandra instrument, as detailed in the Appendix.

^f 0 to 24 scale. Higher scores indicate worse functional impairment.

Table 3

Multivariable associations among independent risk factors recorded at index ED visit and three-month outcomes.

Predictor	Any functional	impairment^a		Low back pain^b		Analgesic requir	ement^c
	Adj OR	95% CI		Adj OR	95% CI	Adj OR	95% CI
Depression^d
Positive screen	0.9	0.2, 3.6		1.0	0.3, 3.7	1.5	0.4, 5.6
Negative screen Psychosomatic symptomatology^e	1.0			1.0		1.0
Positive test	0.5	0.1, 1.9	0.7		0.2, 2.2 1.0		0.3, 3.4
Negative test	1.0		1.0		1.0
RMDQ^f, per 1 point increment	1.0	1.0, 1.1	1.0		1.0, 1.1 1.0		0.9, 1.1
Age, per year	1.0	1.0, 1.0	1.0		1.0, 1.0 1.0		1.0, 1.0
Sex
Female	1.4	0.8, 2.3	1.4		0.8, 2.2 1.5		0.8, 2.6
Male	1.0		1.0		1.0
Education
>=Bachelor’s degree	1.2	0.6, 2.5	1.1		0.5, 2.2 1.4		0.6, 3.2
b Bachelor’s degree	1.0		1.0		1.0
Study meds
Cyclobenzaprine	1.1	0.6, 2.0	1.3		0.7, 2.4 1.5		0.7, 3.0
Oxycodone/APAP	1.1	0.6, 2.1	1.1		0.6, 2.0 1.6		0.7, 3.1
Placebo	1.0		1.0		1.0

Adj OR = adjusted multivariable OR, reflecting independent association after adjustment for covariates in logistic model. Variables whose 95% CI does not cross 1.0 are considered statistically significant. None of these variables satisfied this criterion.

^a Functional impairment defined as Roland-Morris Disability Questionnaire score at 3 months greater than 0, indicating some functional impairment.

^b “Severe,” “moderate,” or “mild” pain.

^c Use of any analgesic medication for LBP within previous 72 h.

^d Positive depression screen was a response of “often” or “always” to either one of two depression screening questions from the PHQ instrument.

^e A positive psychosomatic test was a score >=0.80 on the 5-item Cassandra instrument, as detailed in the Appendix.

^f 0 to 24 scale. Higher scores indicate worse functional impairment.

The five-item Cassandra psychosomatic score was not associated with three-month outcomes. In another ED setting, higher scores on the five item Cassandra instrument were associated with worse LBP outcomes [7]. We were unable to replicate those prior results. The most plausible explanation for this discordance is that our population and entry criteria were substantially different from the original ED pop- ulation studied [6]. The previous ED study was conducted among French Canadians with acute, subacute, or chronic low or upper back pain or neck pain. By contrast, our study consisted of English and Spanish speakers in the US with acute, new onset LBP. We also excluded patients with radicular symptoms, a subset known to have a worse prognosis, comprising 11% of the Canadian study population. Dionne et al.’s prima- ry outcome also differed from ours: a score in the upper half of the RMDQ assessed two years after the ED visit, which was appropriate for a LBP population with mixed chronicity, but is inappropriate for our previously healthy population, in whom we regarded anything less than complete functional recovery at 3-months as suboptimal. The differing features of the two populations may also explain another unexpected finding: only 6% of our study cohort had a positive score on the 5-item Cassandra instrument compared to 49% in the Canadian study [7].

Similarly, a positive result on the depression screen was not associ- ated with poor LBP outcomes in our analysis, even though previously published data have demonstrated an association between depression and poor LBP prognosis [10]. Depression is an appealing predictor be- cause it is a modifiable risk factor for chronicity. As with the five-item Cassandra instrument, few of our subjects screened positive for depres- sion (5%), indicating that it does not appear to be an important co- morbid condition in our ED population with recent-onset of LBP. It is surprising that so few of our patients screened positive for depression given that others have reported that one-third of a cross-section of gen- eral ED patients screen positive for depression [13]. The lower rates in our population may reflect the fact that we excluded many patients with chronic pain syndromes and selected for previously healthy patients.

Also surprising was that index visit RMDQ, a measure of LBP-related functional impairment, was not associated with LBP outcomes. In previ- ous work, we demonstrated an association between index visit RMDQ and unfavorable outcomes in a sample that included patients with sub-acute and chronic LBP as well as acute LBP [5]. We speculate that, in the sub-acute and chronic population, the RMDQ may have had an opportunity to stabilize over time and is therefore more likely to remain

Table 4

Despite a strong overall group association, presence or absence of low back pain at one week is not a clinically important predictor of three-month outcome in individual patients.

One week predictor	Three-month outcome
	RMDQ N 0	RMDQ = 0	N
LBP at 1 week	74	119	193
No LBP at 1 week	14	87	101
N	88	206	294

Prediction in individual patients (using presence or absence of LBP as a diagnostic test predicting 3-month outcome) Likelihood ratio -: 0.4 (95% CI: 0.2, 0.6)

Likelihood ratio +: 1.4 (95% CI: 1.3, 1.7) OR

Sensitivity: 84% (95% CI: 75, 90%)

Specificity 42% (95% CI: 36, 49%)

Aggregate association within study cohort (using presence of absence of LBP as a risk factor associated with population 3-month outcome) Odds ratio (unadjusted) = 3.9 (95% CI: 2.0, 7.3)

constant over the subsequent three months, while in patients with acute LBP, the RMDQ spikes sharply, causing the patient to come to the ED. In many of these acute patients, the RMDQ then reverts to a nor- mal baseline well before the three-month follow-up. This occurred in the majority of patients in the present study, in spite of the fact that

Over the last 24 h, my appetite was not very good because of my back pain:

Over the last 24 h, I have had trouble putting on my socks ( or stockings) because of the pain in my back or leg:
Over the last 24 h, I could only walk short distances because of my back pain:
No⁰ Yes¹

No⁰ Yes¹

No⁰ Yes¹

our index visit RMDQ scores were clustered near the severe end of the

Over the last 24 h, I slept less well because of my back: No⁰ Yes¹

scale, a phenomenon we have reported previously in ED-based work [2].

In the outpatient setting, a nine-item instrument called the STarT Back tool, which incorporates items from the RMDQ as well as questions about depression, catastrophizing, and radicular symptoms has been shown to result in better allocation of resources than Clinical impression alone [14]. Further work is needed to determine whether these findings can be replicated among ED patients with new-onset symptoms. The Orebro Musculoskeletal pain Questionnaire and the acute low back pain Screening Questionnaire, while similar in scope, are more cumber- some and therefore less appropriate for efficient use in an acute care set- ting [15,16].

It is not self-evident that factors related to psychiatric and emotional distress would be better predictors of poor pain outcomes than factors related to the injury itself, such as mechanism of injury or findings on physical exam. However the association between psychiatric or emo- tional distress and poor pain outcomes has been observed across a vari- ety of pain syndromes, including widespread pain after Motor vehicle collisions [17], chronic migraine [18], and persistent post-operative pain [19]. Although an integrated pain management program that in- corporates psychosocial screening and treatment is intuitively appeal- ing, further work is needed in acute LBP before it is clear that such interventions limit the development of chronic LBP and its accompany- ing morbidity.

In conclusion, nearly 30% of ED patients will report functional im- pairment and pain three months after an ED visit for acute, non- traumatic, non-radicular LBP. In spite of a strong aggregate association at the cohort level between presence of LBP at one week and functional impairment at three months, when presence of LBP at one week is used to predict outcomes in individual patients, it fails to identify with satis- factory clinical accuracy those individuals likely to have an unfavorable longer-term outcome.

Over the last 24 h, I got dressed with the help of someone else because of my back pain:
Over the last 24 h, I sat down for most of the day because of my back:
Over the last 24 h, I avoided heavy jobs around the house be- cause of my back pain:
Over the last 24 h, I was more irritable and bad tempered with people than usual because of my back pain:
Over the last 24 h, I went upstairs more slowly than usual be- cause of my back pain:
Over the last 24 h, I stayed in bed most of the time because of my back pain:
Appendix B. Cassandra 5-item instrument

No⁰ Yes¹

No⁰ Yes¹

No⁰ Yes¹

No⁰ Yes¹

No⁰ Yes¹

No⁰ Yes¹

Appendix A. Roland-Morris Disability Questionnaire

Cassandra score is obtained by calculating the mean of all items. The cut-off value for this questionnaire is a mean >=0.80.

Over the last 24 h, I have stayed home most of the time because of my back pain:
Over the last 24 h, I changed position frequently to try to get my back comfortable:
Over the last 24 h, I walked more slowly than usual because of my back:
Over the last 24 h, I have not been doing any jobs that I usually do around the house because of my back pain:
Over the last 24 h, I used a handrail to get upstairs because of my back pain:
Over the last 24 h, I lay down to rest more often because of my back pain:
Over the last 24 h, I have had to hold on to something to get out of an easy chair because of my back pain:
Over the last 24 h, I have tried to get other people to do things for me because of my back pain:
Over the last 24 h, I got dressed more slowly than usual because of my back pain:
Over the last 24 h, I only stood up for short periods of time because of my back pain:
Over the last 24 h, I tried not to bend or kneel down because of my back pain:
Over the last 24 h, I found it difficult to get out of a chair because of my back pain:
No⁰ Yes¹ No⁰ Yes¹ No⁰ Yes¹ No⁰ Yes¹ No⁰ Yes¹ No⁰ Yes¹ No⁰ Yes¹ No⁰ Yes¹ No⁰ Yes¹ No⁰ Yes¹ No⁰ Yes¹ No⁰ Yes¹

References
1. Friedman BW, Chilstrom M, Bijur PE, Gallagher EJ. Diagnostic testing and treatment of low back pain in United States emergency departments: a national perspective. Spine 2010;35(24):E1406-11.
2. Friedman BW, O’Mahony S, Mulvey L, Davitt M, Choi H, Xia S, et al. One-week and 3- month outcomes after an emergency department visit for undifferentiated musculo- skeletal low back pain. Ann Emerg Med 2012;59(2):128-33 [e3].
3. Deyo RA, Dworkin SF, Amtmann D, Andersson G, Borenstein D, Carragee E, et al. Re- port of the NIH Task Force on research standards for Chronic low back pain. Pain Med 2014;15(8):1249-67.
4. Friedman BW, Dym AA, Davitt M, Holden L, Solorzano C, Esses D, et al. Naproxen with cyclobenzaprine, oxycodone/acetaminophen, or placebo for treating acute low back pain: a randomized clinical trial. JAMA 2015;314(15):1572-80.
5. Friedman BW, Mulvey L, Davitt M, Choi H, Esses D, Bijur PE, et al. Predicting 7-day and 3-month functional outcomes after an ED visit for acute nontraumatic low back pain. Am J Emerg Med 2012;30(9):1852-9.
6. Roland M, Fairbank J. The Roland-Morris Disability Questionnaire and the Oswestry Disability Questionnaire. Spine 2000;25(24):3115-24.
7. Dionne CE, Le Sage N, Franche RL, Dorval M, Bombardier C, Deyo RA. Five questions predicted long-term, severe, back-related Functional limitations: evidence from three large prospective studies. J Clin Epidemiol 2011;64(1):54-66.
8. Haughey MT, Calderon Y, Torres S, Nazario S, Bijur P. Identification of depression in an inner-city population using a simple screen. Acad Emerg Med 2005;12(12): 1221-6.
9. Whooley MA, Avins AL, Miranda J, Browner WS. Case-finding instruments for de-
10. Over the last 24 h, my back was painful almost all of the time: No⁰ Yes¹
11. Over the last 24 h, I found it difficult to turn over in bed because No⁰ Yes¹ of my back pain:
  pression. Two questions are as good as many. J Gen Intern Med 1997;12(7):439-45.
  
  Pinheiro MB, Ferreira ML, Refshauge K, Maher CG, Ordonana JR, Andrade TB, et al. Symptoms of depression as a prognostic factor for low back pain: a systematic re- view. Spine J 2016;16(1):105-16.
12. Pepe MS, Janes H, Longton G, Leisenring W, Newcomb P. Limitations of the odds ratio in gauging the performance of a diagnostic, prognostic, or screening marker. Am J Epidemiol 2004;159(9):882-90.
13. Gallagher EJ. Clinical utility of likelihood ratios. Ann Emerg Med 1998;31(3):391-7.
14. Boudreaux ED, Cagande C, Kilgannon H, Kumar A, Camargo CA. A prospective study of depression among adult patients in an urban emergency department. Prim Care Companion J Clin Psychiatry 2006;8(2):66-70.
15. Hill JC, Whitehurst DG, Lewis M, Bryan S, Dunn KM, Foster NE, et al. Comparison of stratified primary care management for low back pain with current best practice (STarT Back): a randomised controlled trial. Lancet 2011;378(9802):1560-71.
16. Hockings RL, McAuley JH, Maher CG. A systematic review of the Predictive ability of the Orebro Musculoskeletal Pain Questionnaire. Spine (Phila Pa 1976) 2008;33(15): E494-500.
17. Linton SJ, Hallden K. Can we screen for problematic back pain? A screening question- naire for predicting outcome in acute and sub Acute back pain. Clin J Pain 1998; 14(3):209-15.
18. Hu J, Bortsov AV, Ballina L, Orrey DC, Swor RA, Peak D, et al. Chronic widespread pain after motor vehicle collision typically occurs through immediate development and nonrecovery: results of an emergency department-based cohort study. Pain 2016; 157(2):438-44.
19. Ashina S, Serrano D, Lipton RB, Maizels M, Manack AN, Turkel CC, et al. Depression and risk of transformation of episodic to chronic migraine. J headache pain 2012; 13(8):615-24.
20. Lewis GN, Rice DA, McNair PJ, Kluger M. Predictors of persistent pain after total knee arthroplasty: a systematic review and meta-analysis. Br J Anaesth 2015;114(4): 551-61.