a b s t r a c t

Objectives: Chronic low back pain is an important Public health problem, generating high financial and social costs. While most clinical guidelines stress the importance of managing low back pain in primary care, in practice a disproportionate amount of patients with low back pain present to emergency departments. Patients present- ing to emergency departments may form a specific group with different factors leading to chronicity. This pro- spective cohort study aimed to determine the sociodemographic and psychological predictors of persistent low back pain and the length of sick leave due to pain in patients with acute symptoms visiting an emergency department.

Methods: Patients with a first episode of non-specific acute low back pain in at least three months were qualified for this study. The participants filled a battery of questionnaires, including measures of pain, pain-related disabil- ity, depression, anxiety and pain coping strategies. A structured telephone interview was performed after three months with questions regarding pain and the length of sick leave.

Results: 110 patients participated in the study. 97 patients completed the follow-up, with 70.1% suffering from pain after three months. Lower self-rated health predicted pain after three months. Longer length of sick leave was predicted by lower self-rated health, distraction as a coping strategy and decreased behavioral activity.

Conclusion: Because of its simplicity, a measurement of self-rated health may be included in future clinical prac- tice for assessing the risk of persistent pain.

(C) 2021 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY license (http://

creativecommons.org/licenses/by/4.0/).

1. Introduction

Chronic low back pain is one of the leading causes of disability and a source of substantial Economic burden to individuals, families and com- munities. Worldwide, low back pain has produced 57.6 million Years Lived with Disability (YLD), more than any other disease according to Global Burden of Disease 2016 [31]. Many risk factors for developing an episode of low back pain have been identified [1]. These include hav- ing a previous episode of back pain [11], a chronic condition [30], lifting and heavy physical workload [9] and poor mental health [16]. While most patients suffering from an acute episode of low back pain recover, some develop a chronic, severely debilitating condition that often fails to respond to treatment [10]. Within one year from a previous episode, one third of patients will have a recurrence [25].

Mechanisms for low back pain chronification (the development of persistent low back pain that does not respond to treatment) are not

* Corresponding author at: Department of Quality of Life Research, Medical University of Gdansk, Ul. Tuwima 15, 80-210 Gdansk, Poland.

E-mail address: [email protected] (K. Basinski).

well established. A biopsychosocial model is seen as an appropriate framework to analyze these mechanisms and various factors contribut- ing to pain chronification have been proposed [1]. Implied psychological factors include catastrophizing [7], fear-avoidance beliefs [20], self- efficacy [4] or depression [21]. Societal factors point to the role of low socioeconomic status [12] and poor education [23].

While most clinical guidelines stress the importance of managing low back pain in primary care, in practice a disproportionate amount of patients with low back pain present to emergency departments [19]. A systematic review of 21 studies from 12 countries provided a pooled prevalence estimate of low back pain in emergency settings of 4.39% (95% CI: 3.67-5.18) [5]. Nonetheless, previous research on predic- tors of pain chronification focused on primary care or pain clinics. Pa- tients presenting to emergency departments may form a specific group with different factors leading to chronicity. This disparity may be due to rapid onset of pain, high pain intensity, high anxiety or other factors.

The aim of this prospective cohort study was to determine the psy- chosocial predictors of persistent low back pain after 3 months in pa- tients presenting to the ED with a new episode of acute low back pain.

https://doi.org/10.1016/j.ajem.2021.10.018

0735-6757/(C) 2021 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

We also aimed to determine the predictors of the length of sick leave in this group of patients.

1. Methods
   1. Participants

The study was conducted in an Emergency Department of the Uni- versity Clinical Centre, Medical University of Gdansk, Poland. The De- partment has about 34,000 admissions annually. Patients were qualified for the study if they were aged 18 to 65 years, read and wrote in Polish and had a main diagnosis of non-specific, acute low back pain (with or without sciatica), as determined by an attending physician. Patients were excluded if they had a previous episode of back pain within the last three months or the pain could be attributed to a known condition (such as cancer, trauma, osteoporosis, rheumatoid arthritis). We also excluded patients that were pregnant or had other serious comorbidities (i.e. other unrelated chronic pain conditions, cancer).

• 1. Procedure

Patients who met the inclusion criteria were given an informed con- sent form. Afterwards they completed a battery of questionnaires. After a period of 3 months, patients were contacted for a structured telephone interview. The interview was scripted and included questions about the persistence of pain and the length of sick leave. Three contact attempts were made before the patient was qualified as lost to follow-up. The study protocol was approved by the Independent Bioethics Committee for Research at the Medical University of Gdansk (approval id no. NKBBN/72/2016).

• 1. Outcomes

The presence of persistent low back pain was established in the structured interview with the question “do you still feel pain in your back related to your stay at the Emergency Department three months ago?” The responses were dichotomized as “yes” or “no.” Length of sick leave was determined with the question “were you on a sick leave because of your back pain in the last three months? If so, how long was your sick leave?” If the answer to this question was “no,” the interviewer asked further questions to establish if the sick leave was not needed because the patient was not eligible (i.e. unemployed, re- tired, on pension, etc.) or the pain was not strong enough to warrant a sick leave.

• 1. Measures

Patients were given a set of questionnaires, including measures of:

• • • Self-rated health - an item from the World Health Survey questionnaire (“in general, would you say that your health is (1) excellent, (2) very good, (3) good, (4) fair, or (5) poor?“)
    • social support measure taken from Diagnoza Spoleczna, a Polish large- scale sociological survey (average of two items: “How many people can you count on in case of serious personal problems? (1) None, (2) one,

(3) two-three, (4) four-five, (5) more than five” and “To what extent others are positively interested in how you are doing? (1) Not at all,

(2) somewhat small, (3) neither small nor large, (4) somewhat large,

(5) large.“)

• • • Pain intensity and interference - scales of the Brief Pain Inventory - Short Form (BPI-SF) [6]
    • Disability, measured by the Oswestry Disability Index (ODI) [2]
    • Anxiety and depression symptoms, measured by the Hospital Anxiety and Depression Scale (HADS) [13]
    • Strategies of coping with pain (with subscales for distraction, ignoring

pain sensations, reinterpreting pain sensations, catastrophizing, praying and hoping, and increasing pain behaviour), measured by the Coping Strategies Questionnaire (CSQ) [24]

• • • A survey of socio-demographic data (age, gender, education, Place of residence (urban/rural), BMI)

Participants also completed the Multilevel Health Locus of Control (MHLC) questionnaire. The results were not included in the analysis due to low reliability of MHLC subscales (Cronbach’s ? < 0.70). Two subscales of the CSQ showed moderate reliability (increased behavioral activity, Cronbach’s ? = 0.75; reinterprating pain sensations; Cronbach’s ? = 0.78). Other scales showed high reliability (Cronbach’s ? > 0.80).

• 1. Statistical analyses

Predictors of persistent pain were evaluated using multiple logistic regression. Odds ratios (OR) and their 95% Confidence Intervals (CI) were calculated. Predictors of length of sick leave were evaluated using multiple linear regression. Gender was analyzed as a nominal var- iable. For self-rated health and social support an orthogonal polynomial transform was used (as standard treatment of ordinal predictors in the default linear model implementation in R [17]). To determine the order of polynomials that best fits the underlying structure of responses and avoid overfitting, we performed analyses for linear, quadratic, cubic and fourth-degree polynomials and selected the model that had best fit statistics (Akaike Information Criterion and Bayesian Information Cri- terion for logistic and adjusted R² for linear regression). A linear solution produced best fit for both self-rated health and social support and this transform was used in the final analyses (regression results for higher order polynomials are presented in Appendix 1). The remaining predic- tors were analyzed as continuous variables. Since questionnaire re- sponses were measured using Likert scales and averaged into indices of multiple items, this approach seems to be valid [3]. To evaluate collin- earity in both models, variance inflation factors for all predictors were calculated. Variance inflation factors of the predictors were in the range of 1.39-3.86, indicating low multicollinearity. All data analyses were conducted in R¹⁷ using the packages dplyr [14] and car [8]. Bias due to missing data was minimized using mice (multiple imputations with chained equations) [18].

1. Results

110 patients (44.1% women, 54.1% men, one person did not disclose gender) were recruited for the study. Average age was 39.3 years (SD = 11.1). The patients had mostly high-school (45.9%) or higher (32.4%) education and were living in an urban setting (82.0%). Average BMI was 27.06 (SD = 4.57). Upon admission, the patients reported an aver- age pain intensity of 6.02 (SD = 1.85) on an 11-point numerical rating scale. Demographic characteristics on admission are presented in Table 1.

In the follow-up study, 97 patients (88.1%) were successfully contacted. All patients lost to follow-up were unreachable via telephone and all patients who answered the telephone agreed to the interview. Group comparisons showed no significant differences in baseline vari- ables between the contacted group and patients lost to follow-up. In the follow-up group, 68 patients (70.1%) suffered from back pain after 3 months. Median length of sick leave due to pain was 14 days (inter- quartile range: 0-59). Twenty-eight patients (28.9%) did not require sick leave (included in the analysis as 0 days) while seven patients (7.2%) were unable to receive sick leave due to being retired or on a pen- sion (these were excluded from the regression model).

A multiple logistic regression model of predictors of persistent pain at three months included: age, gender, BMI, self-rated health, pain in- tensity, pain interference, disability, social support, anxiety, depression, and seven CSQ pain coping strategies. The Akaike Information Criterion for the model was 132.32, while the Bayesian Information

Table 1

Demographic characteristics of the studied group (on admission)

Baseline variable	Value
Age; M (SD)	39.3 (11.1)
Gender; N (%)
Women	49 (44.1%)
Men	60 (54.1%)
Education; N (%) Elementary/vocational	18 (16.2%)
High school	51 (45.9%)
Higher	36 (32.4%)
Place of residence; N (%)

Urban 91 (82.0%)

Rural 15 (13.5%)

BMI; M (SD) 27.06 (4.57)

Self-rated health; N (%)

Very good	8 (7.2%)
Good	46 (41.4%)
Moderate	38 (34.2%)
Bad	11 (9.9%)
Very bad	2 (1.8%)
Pain intensity; M (SD)	6.02 (1.85)
Pain interference; M (SD)	48.98 (15.0)
ODI; M (SD)	22.61 (11.85)
Social support; M (SD)	4.15 (0.98)
HADS Anxiety; M (SD)	7.91 (4.24)
HADS Depression; M (SD) CSQ; M (SD):	5.36 (4.3)
Distraction	12.69 (8.01)
Reinterpreting sensations	9.94 (6.95)
Catastrophizing	11.18 (7.73)
Ignoring sensations	14.31 (7.59)
Praying and hoping	14.59 (8.77)
Statements	19.03 (7.47)
Behavioral activity	14.79 (6.88)

Scale ranges: self-rated health 1-5 (1 - best, 5 - worst); pain intensity 0-10; pain interference 0-100; ODI 0-50; Social support 1-5; anxiety and depres- sion 0-21; CSQ subscales 0-36.

Criterion was 178.66. Self-rated health emerged the only statistically significant predictor of persistent pain (OR = 2.34; 95% CI 1.15-5.41; P = .028). Odds ratios and their 95% confidence intervals of the predic- tors are presented in Table 2. Due to overfitting considerations, a stepwise regression using backward selection and AIC has been per- formed to evaluate if a model with less predictors would yield similar results (Appendix 2). Self-rated health remained statistically significant (P = .049).

Table 2

Odds ratios (OR) and their 95% confidence intervals of the predictors of persistent low back pain at three months

Predictor	OR	95% C.I.	P
Age	0.98	0.93-1.03	0.359
Gender (male)	0.52	0.15-1.70	0.287
BMI	1.10	1.00-1.26	0.130
Self-rated health	2.34	1.15-5.41	0.028?
Pain intensity	1.39	0.98-2.06	0.081
Pain interference	0.99	0.95-1.04	0.815
ODI	1.01	0.95-1.07	0.738
Social support	0.66	0.30-1.25	0.238
HADS Anxiety	1.05	0.86-1.30	0.612
HADS Depression	0.96	0.77-1.18	0.673
CSQ:

Distraction	1.07	0.95-1.22	0.274
Reinterpreting sensations	0.97	0.87-1.07	0.516
Catastrophizing	0.97	0.87-1.07	0.509
Ignoring sensations	1.08	0.97-1.22	0.179
Praying and hoping	0.93	0.85-1.02	0.145
Statements	1.05	0.94-1.17	0.419
Behavioral activity	0.89	0.77-1.01	0.082

Scale ranges: self-rated health 1-5 (1 - best, 5 - worst); pain intensity 0-10; pain interfer- ence 0-100; ODI 0-50; Social support 1-5; anxiety and depression 0-21; CSQ subscales

0-36.

* p < .05.

A linear regression model was constructed with length of sick leave as a dependent variable. This model included all of the predictors from the model described above. The model was statistically significant (F (17,72) = 2.313; P = .007) and explained 35.3% of the variance (R² = 0.353; adjusted R² = 0.20). Statistically significant predictors in- cluded self-rated health (? = 13.13; 95% CI 4.54-21.73; P = .003), dis- traction (? = 2.02; 95% CI 0.53-3.51; P = .008) and increased

behavioral activity (? = -1,61; 95% CI -3.07 - -0.14; P < .032).

Regression coefficients and p-values are presented in Table 3. As before, stepwise regression was also conducted and self-rated health, distrac- tion and behavioral activity remained statistically significant (P =

.001, P = .002, P = .018 respectively).

1. Discussion

Over 70% of patients in this study had persistent back pain after three months from admission to the emergency department. This figure is higher than most chronification rates reported in primary care stud- ies, which are in the range between 11% and 42% [28]. Prospective co- hort studies on emergency patients report a higher percentage after three months, between 39% and 53% [27]. The higher rate may be a re- sult of group specificity, with patients presenting to the emergency de- partment having a more serious low back pain condition in comparison to patients in primary care. The result can also be attributed to a very sharp criterion that was applied in the current study, namely that the pain was classified as persistent if the patient was not pain-free. This cri- terion was chosen as non-arbitrary, clear and precise, yet it may limit- ing. One way to overcome this problem in future studies is to use disability measures along with pain scales to further differentiate be- tween patients with functional disabilities.

Lower self-rated health predicted persistent pain after three months as well as the length of sick leave. A one point (on a five-point scale) de- crease in self-rated health is associated with a 2.34-time increase in the chance of persistent pain after three months. Similarily, a one point de- crease in self-rated health was associated with 13.13 more days on sick leave due to persistent pain. Self-rated health has been established as a good mortality predictor in epidemiological studies [29]. To our knowl- edge, no previous studies explored this predictor in acute low back pain patients. Poor self-rated health may be indicative of other comorbidities that contribute to pain persistence by increasing the allostatic load [22]. It may also be interpreted as a set of negative beliefs about the patients

Table 3

Linear regression coefficients of the predictors of the length of sick leave (in days) associated with low back pain after 3 months

Predictor	Beta	95% C.I.	P
Age	-0.12	-0.82 - 0.58	0.737
Gender (male)	-3.66	-18.90 - 11.58	0.634
BMI	-0.19	-0.50 - 0.12	0.220
Self-rated health	13.13	4.54-21.73	0.003??
Pain intensity	-0.77	-5.48 - 3.94	0.746
Pain interference	0.12	-0.52 - 0.75	0.713
ODI	0.60	-0.14 - 1.34	0.111
Social support	-1.93	-10.48 - 6.62	0.654
Anxiety	0.85	-1.88 - 3.59	0.535
Depression	-0.85	-3.69 - 2.00	0.553
CSQ:

Distraction	2.02	0.53-3.51	0.008??
Reinterpreting sensations	-0.49	-1.80 - 0.81	0.451
Catastrophizing	-0.33	-1.64 - 0.97	0.613
Ignoring sensations	-0.29	-1.74 - 1.16	0.692
Praying and hoping	-0.02	-1.23 - 1.19	0.973
Statements	-0.09	-1.35 - 1.18	0.892
Behavioral activity	-1.61	-3.07 - -0.14	0.032?

Scale ranges: self-rated health 1-5 (1 - best, 5 - worst); pain intensity 0-10; pain interfer- ence 0-100; ODI 0-50; Social support 1-5; anxiety and depression 0-21; CSQ subscales

0-36.

* p < .05.

?? p < .01.

well being, contributing to the vicious circle of fear-avoidance [26]. As a single-item measure, self-rated health is potentially a useful clinical tool for selecting patients that could develop persistent pain. Future studies could investigate the predictive value of this measure in comparison to an established screening method, such as the STaRT Back questionnaire. Two pain coping strategies were predictive of the length of sick leave at three months. Patients who were prone to use distraction as a coping strategy spend more time on sick leave. This finding is surprising, as pre- vious research suggest that distraction lowers pain intensity and should contribute to better outcomes [15]. One way to explain this is that the length of sick leave may not necessarily be directly related to the amount of pain in some patients. Some psychological characteristics un- related to pain may contribute to the overall length of stay on sick leave. The measure of distraction as a coping strategy may be indicative of these characteristics. Further research is needed to investigate this ef- fect. Patients who increased their behavioral activity as a coping strat- egy for pain spent less time on sick leave. This is in line with previous research showing that remaining active despite back pain leads to bet-

ter outcomes [1].

The findings in this study are limited by several issues. A second follow-up after a longer period would be needed to establish the predic- tors of long-term chronicity. Some patients in this study reported still being on sick leave at the time of follow-up. These patients were in- cluded in the analysis with their sick leave length set at 90 days. A longer follow-up would also allow for a more reliable measurement of the length of sick leave. Furthermore, as is common with questionnaire studies, a’priori decisions about the choice of methods can influence the outcomes to some extent. The choice of methods was also limited by the availability of reliable questionnaires that were properly trans- lated and culturally adapted for use in a Polish population. There is

thus a possibility that some significant psychological variables were left out of the study. Similarily, some of the variance explained in our models may be due to overfitting with many predictors. This problem is reflected in the discrepancy between R² and adjusted R² measures. Many predictors included in our models have low estimates and are not statistically significant. However, we did not remove these predic- tors from the final models as we pursued a confirmatory approach, selecting the model before the data was collected. Furthermore, simpler models, as presented in Appendix 2, provide similar results. A larger sample size would increase the statistical Power of the study and pro- vide grounds for using cross-validation based approaches to model se- lection. It would also be beneficial for future research efforts to further explore the role of self-rated health in pain persistence and the transi- tion from acute to chronic pain in context of LBP as well as other pain conditions.

Financial support

This work did not receive any financial support.

Author contributions

KB and MM designed the study. KB and AZR performed the study. KB analyzed the data and drafted the manuscript while AZR and MM pro- vided revisions.

Declaration of Competing Interest

The authors report no conflicts of interest.

Appendix 1 Logistic and Linear regression models for persistent low back pain and sick leave length at three months with different orders of orthogonal polyno- mial transformation for ordinal variables (self-rated health and social support). Models were constructed assuming linear (reported in the paper, see Tables 2 and 3), quadratic (A, D), cubic (B, E) and fourth-degree polynomial (C, F) structure underlying the ordinal scale responses. Model estimates and fit measures are given for each polynomial degree.

Model A. Predictors of persistent low back pain at three months, orthogonal polynomial for ordinal scales degree = 2. AIC = 135.67; BIC = 187.17.

Predictor	Estimate	Standard error	Statistic	P
(Intercept)	-1.72	2.53	-0.68	0.496
Age	-0.02	0.03	-0.92	0.356
Gender (male)	-0.59	0.63	-0.93	0.35
BMI	0.1	0.07	1.45	0.148
Self-rated health (linear)	7.03	3.22	2.18	0.029
Self-rated health (quadratic)	-2.24	3.06	-0.73	0.465
Pain intensity	0.34	0.19	1.77	0.076
Pain interference	-0.01	0.03	-0.28	0.781
ODI	0.01	0.03	0.45	0.652
Social support (linear)	-4.18	3.8	-1.1	0.271
Social support (quadratic)	1.52	3.58	0.42	0.672
HADS Anxiety	0.05	0.1	0.46	0.644
HADS Depression	-0.04	0.11	-0.36	0.716
CSQ Distraction	0.08	0.07	1.19	0.233
CSQ Reinterpreting sensations	-0.04	0.05	-0.81	0.418
CSQ Catastrophizing	-0.04	0.05	-0.73	0.464
CSQ Ignoring sensations	0.08	0.06	1.35	0.178
CSQ Praying and hoping	-0.07	0.05	-1.47	0.141
CSQ Statements	0.04	0.06	0.8	0.425
CSQ Behavioral activity	-0.12	0.07	-1.73	0.084

Model B. Predictors of persistent low back pain at three months, orthogonal polynomial for ordinal scales degree = 3. AIC = 138.69; BIC = 195.33.

Predictor	Estimate	Standard error	Statistic	P
(Intercept)	-1.82	2.62	-0.69	0.487
Age	-0.03	0.03	-1.06	0.29
Gender (male)	-0.56	0.63	-0.89	0.374
BMI	0.1	0.07	1.42	0.156
Self-rated health (linear)	7.2	3.51	2.05	0.04

(continued)
Predictor	Estimate	Standard error	Statistic	P
Self-rated health (quadratic)	-1.8	3.66	-0.49	0.623
Self-rated health (cubic)	2.96	3.22	0.92	0.357
Pain intensity	0.39	0.21	1.89	0.059
Pain interference	-0.01	0.03	-0.43	0.667
ODI	0.02	0.03	0.62	0.536
Social support (linear)	-3.95	4.4	-0.9	0.369
Social support (quadratic)	2.11	4.46	0.47	0.636
Social support (cubic)	-0.64	4.14	-0.15	0.878
HADS Anxiety	0.06	0.1	0.53	0.597
HADS Depression	-0.03	0.11	-0.27	0.788
CSQ Distraction	0.07	0.07	0.97	0.331
CSQ Reinterpreting sensations	-0.04	0.05	-0.78	0.436
CSQ Catastrophizing	-0.04	0.05	-0.75	0.451
CSQ Ignoring sensations	0.08	0.06	1.38	0.168
CSQ Praying and hoping	-0.07	0.05	-1.52	0.129
CSQ Statements	0.04	0.06	0.78	0.437
CSQ Behavioral activity	-0.11	0.07	-1.46	0.145

Model C. Predictors of persistent low back pain at three months, orthogonal polynomial for ordinal scales degree = 4. AIC = 141.25; BIC = 203.04.

Predictor	Estimate	Standard error	Statistic	P
(Intercept)	-1.43	84.63	-0.02	0.986
Age	-0.03	0.03	-1.14	0.253
Gender (male)	-0.6	0.64	-0.93	0.351
BMI	0.11	0.07	1.61	0.107
Self-rated health (linear)	19.07	1723.31	0.01	0.991
Self-rated health (quadratic)	13.39	2215.19	0.01	0.995
Self-rated health (cubic)	15.68	1792.43	0.01	0.993
Self-rated health (fourth)	10.38	1094.07	0.01	0.992
Pain intensity	0.41	0.21	1.95	0.051
Pain interference	-0.01	0.03	-0.36	0.719
ODI	0.02	0.03	0.49	0.621
Social support (linear)	-12.72	1773.25	-0.01	0.994
Social support (quadratic)	13.55	2345.5	0.01	0.995
Social support (cubic)	-11.76	2163.85	-0.01	0.996
Social support (fourth)	5.97	1005.97	0.01	0.995
HADS Anxiety	0.07	0.11	0.61	0.54
HADS Depression	-0.04	0.11	-0.31	0.755
CSQ Distraction	0.08	0.07	1.12	0.262
CSQ Reinterpreting sensations	-0.04	0.05	-0.75	0.452
CSQ Catastrophizing	-0.04	0.05	-0.84	0.404
CSQ Ignoring sensations	0.08	0.06	1.35	0.177
CSQ Praying and hoping	-0.08	0.05	-1.57	0.117
CSQ Statements	0.04	0.06	0.69	0.488
CSQ Behavioral activity	-0.11	0.07	-1.53	0.126

Model D. Predictors of sick leave length at three months, orthogonal polynomial for ordinal scales degree = 1. R² = 0.359; adjusted R² = 0.186.

Predictor	Estimate	Standard error	Statistic	P
(Intercept)	38.25	23.99	1.59	0.115
Age	-0.11	0.35	-0.3	0.763
Gender (male)	-4.68	7.82	-0.6	0.551
BMI	-0.19	0.16	-1.18	0.242
Self-rated health (linear)	111.82	38.03	2.94	0.004
Self-rated health (quadratic)	27.99	36.57	0.77	0.447
Pain intensity	-0.98	2.4	-0.41	0.685
Pain interference	0.13	0.32	0.4	0.688
ODI	0.51	0.39	1.32	0.192
Social support (linear)	-16.13	41.79	-0.39	0.701
Social support (quadratic)	-17.14	39.21	-0.44	0.663
HADS Anxiety	0.99	1.41	0.71	0.482
HADS Depression	-0.94	1.45	-0.65	0.518
CSQ Distraction	1.96	0.76	2.57	0.012
CSQ Reinterpreting sensations	-0.33	0.69	-0.48	0.632
CSQ Catastrophizing	-0.27	0.67	-0.4	0.691
CSQ Ignoring sensations	-0.37	0.74	-0.5	0.62
CSQ Praying and hoping	-0.04	0.61	-0.07	0.946
CSQ Statements	-0.08	0.65	-0.13	0.898
CSQ Behavioral activity	-1.62	0.74	-2.18	0.033

Model E. Predictors of sick leave length at three months, orthogonal polynomial for ordinal scales degree = 1. R² = 0.373; adjusted R² = 0.180.

Predictor	Estimate	Standard error	Statistic	P
(Intercept)	38.82	24.17	1.61	0.113
Age	-0.04	0.36	-0.11	0.916
Gender (male)	-4.93	7.97	-0.62	0.538
BMI	-0.15	0.16	-0.96	0.339
Self-rated health (linear)	115.28	38.39	3.0	0.004
Self-rated health (quadratic)	30.55	37.08	0.82	0.413
Self-rated health (cubic)	-42.64	34.83	-1.22	0.225
Pain intensity	-1.61	2.47	-0.65	0.516
Pain interference	0.19	0.33	0.57	0.571
ODI	0.44	0.4	1.1	0.276
Social support (linear)	-26.66	43.07	-0.62	0.538
Self-rated health (quadratic)	-20.21	39.96	-0.51	0.615
Self-rated health (cubic)	2.41	41.03	0.06	0.953
HADS Anxiety	0.84	1.42	0.59	0.556
HADS Depression	-1.04	1.46	-0.71	0.478
CSQ Distraction	2.05	0.82	2.49	0.015
CSQ Reinterpreting sensations	-0.34	0.69	-0.49	0.624
CSQ Catastrophizing	-0.24	0.67	-0.36	0.722
CSQ Ignoring sensations	-0.38	0.75	-0.51	0.612
CSQ Praying and hoping	0.05	0.63	0.08	0.938
CSQ Statements	-0.06	0.65	-0.09	0.928
CSQ Behavioral activity	-1.78	0.79	-2.25	0.028

Model F. Predictors of sick leave length at three months, orthogonal polynomial for ordinal scales degree = 1. R² = 0.403; adjusted R² = 0.195.

Predictor	Estimate	Standard error	Statistic	P
(Intercept)	38.12	24.23	1.57	0.12
Age	0.01	0.36	0.03	0.972
Gender (male)	-5.26	7.9	-0.67	0.508
BMI	-0.19	0.16	-1.18	0.243
Self-rated health (linear)	105.95	38.7	2.74	0.008
Self-rated health (quadratic)	29.64	36.88	0.8	0.425
Self-rated health (cubic)	-41.87	34.53	-1.21	0.23
Self-rated health (fourth)	-45.5	34.64	-1.31	0.194
Pain intensity	-1.45	2.45	-0.59	0.555
Pain interference	0.14	0.33	0.43	0.669
ODI	0.43	0.39	1.09	0.281
Social support (linear)	-24.3	42.98	-0.57	0.574
Social support (quadratic)	-23.39	39.91	-0.59	0.56
Social support (cubic)	11.12	41.05	0.27	0.787
Social support (fourth)	-48.84	36.36	-1.34	0.184
HADS Anxiety	0.8	1.41	0.57	0.571
HADS Depression	-1.15	1.45	-0.79	0.432
CSQ Distraction	1.99	0.82	2.43	0.018
CSQ Reinterpreting sensations	-0.34	0.69	-0.49	0.627
CSQ Catastrophizing	-0.1	0.67	-0.14	0.886
CSQ Ignoring sensations	-0.45	0.74	-0.61	0.542
CSQ Praying and hoping	0.03	0.64	0.04	0.965
CSQ Statements	0.12	0.66	0.18	0.861
CSQ Behavioral activity	-1.83	0.8	-2.3	0.025

Appendix 2

Predictors of persistent low back pain at three months. Stepwise logistic regression with backwards selection using Akaike Information Criterion (AIC).

Predictor	Estimate	Standard error	Statistic	P
(Intercept)	-0.16	1.08	-0.15	0.884
Self-rated health (linear)	5.65	2.88	1.96	0.049?
Pain intensity	0.24	0.15	1.67	0.096
Social support (linear)	-4.16	2.84	-1.47	0.142
CSQ Distraction	0.09	0.05	1.75	0.079
CSQ Praying and hoping	-0.08	0.04	-2.08	0.037?
CSQ Statements	0.08	0.04	1.87	0.062
CSQ Behavioral activity	-0.11	0.06	-1.93	0.054
AIC = 118.01.

* p < .05;

Predictors of sick leave length at three months. Stepwise linear regression with backwards selection using Akaike Information Criterion (AIC).

Predictor	Estimate	Standard error	Statistic	P
(Intercept)	27.48	10.06	2.73	0.008??
BMI	-0.21	0.13	-1.59	0.115
Self-rated health (linear)	107.32	30.59	3.51	0.001???
ODI	0.61	0.27	2.22	0.029?
CSQ Distraction	1.9	0.58	3.28	0.002??
CSQ Reinterpreting sensations	-0.7	0.51	-1.37	0.175
CSQ Behavioral activity	-1.62	0.67	-2.42	0.018?

AIC = 613.5; R² = 0.332; adjusted R² = 0.284; F(6, 83) = 6.881, p < .0001.

* p < .05.

?? p < .01.

??? p < .001

References

1. Hartvigsen J, Hancock MJ, Kongsted A, et al. What low back pain is and why we need to pay attention. Lancet. 2018;391(10137):2356-67.
2. Fairbank JC, Pynsent PB. The Oswestry disability index. Spine. 2000;25(22):2940-52 discussion 2952. Available from http://www.ncbi.nlm.nih.gov/pubmed/11074683.
3. Norman G. Likert scales, levels of measurement and the “laws” of statistics. Adv Health Sci Educ. 2010;15:625-32.
4. Lee H, Hubscher M, Moseley GL, et al. How does pain lead to disability? A systematic review and meta-analysis of mediation studies in people with back and neck pain. Pain. 2015;156(6):988-97.
5. Edwards J, Hayden J, Asbridge M, Gregoire B, Magee K. Prevalence of low back pain in emergency settings: a systematic review and meta-analysis. BMC Musculoskelet Disord. 2017;18(1):1-2.
6. Cleeland CS, Ryan KM. pain assessment: global use of the Brief Pain Inventory. Ann Acad Med Singap. 1994;23(2):129-38 Available from: http://europepmc.org/ abstract/med/8080219.
7. Wertli MM, Eugster R, Held U, Steurer J, Kofmehl R, Weiser S. Catastrophizing - a prognostic factor for outcome in patients with low back pain: a systematic review. Spine J. 2014;14(11):2639-57 Available from: https://doi.org/10.1016/j. spinee.2014.03.003.
8. Fox J, Weisberg S. An R Companion to Applied Regression [Internet]. Third. Thousand Oaks {CA}: Sage; 2019 Available from: https://socialsciences.mcmaster.ca/jfox/ Books/Companion/.
9. Heneweer H, Staes F, Aufdemkampe G, van Rijn M, Vanhees L. physical activity and low back pain: a systematic review of recent literature. Eur Spine J. 2011;20(6): 826-45.
10. Apkarian AV, Baliki MN, Farmer MA. Predicting transition to chronic pain. Curr Opin Neurol. 2013;26(4):360-7 Available from: http://content.wkhealth.com/linkback/ openurl?sid=WKPTLP:landingpage&an=00019052-201308000-00006.
11. Taylor JB, Goode AP, George SZ, Cook CE. Incidence and risk factors for first-time in- cident low back pain: a systematic review and meta-analysis. Spine J. 2014;14(10): 2299-319.
12. Lacey RJ, Belcher J, Croft PR. Does life course socio-economic position influence chronic disabling pain in older adults? A general population study. The Eur J Public Health. 2013;23(4):534-40.
13. Zigmond AS, Snaith RP. The hospital anxiety and depression scale. Acta Psychiatr Scand. 1983;67(6):361-70.
14. Wickham H, Francois R, Henry L, Muller K. dplyr: A Grammar of Data Manipulation [Internet]. Available from https://cran.r-project.org/package=dplyr; 2018.
15. Bushnell MC, Ceko M, Low LA. Cognitive and emotional control of pain and its dis- ruption in chronic pain. Nat Rev Neurosci. 2013;14(7):502-11.
16. Currie SR, Wang J. More data on major depression as an antecedent risk factor for

first onset of Chronic back pain. Psychol Med. 2005;35(9):1275.

1. R Core Team. R: A Language and Environment for Statistical Computing [Internet]. Available from https://www.r-project.org/; 2018.
2. Azur MJ, Stuart EA, Frangakis C, Leaf PJ. Multiple imputation by chained equations: what is it and how does it work? Int J Methods Psychiatr Res. 2012;20(1):40-9.
3. Foster NE, Anema JR, Cherkin D, et al. Prevention and treatment of low back pain: evidence, challenges, and promising directions. Lancet. 2018;391(10137):2368-83.
4. Wertli MM, Rasmussen-Barr E, Weiser S, Bachmann LM, Brunner F. The role of fear avoidance beliefs as a prognostic factor for outcome in patients with nonspecific low back pain: a systematic review. Spine J. 2014;14(5) 816-36.e4. Available from http://www.ncbi.nlm.nih.gov/pubmed/24412032.
5. Pinheiro MB, Ferreira ML, Refshauge K, et al. Symptoms of depression as a prognostic factor for low back pain: a systematic review. Spine J. 2015;16(1):105-16 Available from http://www.ncbi.nlm.nih.gov/pubmed/26523965.
6. Simons LE, Elman I, Borsook D. Psychological processing in chronic pain: a neural systems approach. Neurosc Biobehav Rev. 2014;39:61-78 Available from: http:// linkinghub.elsevier.com/retrieve/pii/S0149763413003035.
7. Shmagel A, Foley R, Ibrahim H. Epidemiology of chronic low back pain in US adults: data from the 2009-2010 National Health and nutrition examination survey. Arthri- tis Care Res. 2016;68(11):1688-94.
8. Rosenstiel AK, Keefe FJ. The use of coping strategies in chronic low back pain pa- tients: relationship to patient characteristics and current adjustment. Pain. 1983; 17(1):33-44.
9. Da Silva T, Mills K, Brown BT, Herbert RD, Maher CG, Hancock MJ. Risk of recurrence of low back pain: a systematic review. J Orthop Sports Phys Ther. 2017;47(5): 305-13.
10. Crombez G, Eccleston C, Van Damme S, Vlaeyen JWS, Karoly P. Fear-avoidance model of chronic pain. The next generation. The Clin J Pain. 2012;28(6):475-83 Available from http://www.ncbi.nlm.nih.gov/pubmed/22673479.
11. 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(10):1138-45.
12. Chou R, Shekelle PG. Will this patient develop persistent disabling low back pain? JAMA. 2010;303(13):1295-302 Available from: http://www.embase.com/search/ results?subaction=viewrecord&from=export&id=L358597409%0Ahttp://jama. ama-assn.org/cgi/reprint/303/13/1295%0Ahttp://dx.doi.org/10.1001/ jama.2010.344%0Ahttp://sfx.library.uu.nl/utrecht?sid=EMBASE&issn=00987484 &id=doi:10.1001%2Fj.
13. Idler EL, Benyamini Y. Self-rated health and mortality: a review of twenty-seven community studies. J Health Soc Behav. 1997:21-37.
14. Ferreira PH, Beckenkamp P, Maher CG, Hopper JL, Ferreira ML. Nature or nurture in low back pain? Results of a systematic review of studies based on twin samples. Eur J Pain. 2013;17(7):957-71.
15. Abajobir AA, Abate KH, Abbafati C, et al. Global, regional, and national disability- adjusted life-years (DALYs) for 333 diseases and injuries and healthy Life expectancy (HALE) for 195 countries and territories, 1990-2016: a systematic analysis for the Global Burden of Disease Study 2016. The Lancet. 2017;390(10100):1260-344 Available from: http://arxiv.org/abs/NIHMS150003.