Background
Musculoskeletal disorders are very common in the general population worldwide [
1‐
3] causing disability for the individual and high costs for the society [
4‐
6] The Global Burden of Disease study reported in 2012 that low back pain and neck pain (NP) was one of five top ranked causes for years lived with disability [
7] and in European countries, individuals with back and NP constitute a large proportion of health care seeking in primary care [
8,
9]. Low back pain increases the risk of general muscle pain, spinal pain and multiple health complaints [
10] and to have low back pain together with neck-shoulder pain is associated with higher risk of long-term sickness absence [
11].
About 85% of all low back pain and NP is classified as nonspecific, where the specific underlying disease or pathology remains unknown [
12]. Most individuals with acute low back pain usually improve rapidly and return to work within 1 month [
13] but after 12 months, about 60% still experience relapses of pain [
14]. In a Swedish cohort of individuals seeking care for nonspecific low back pain or NP about half of the population reported pain and disability 5 years after onset [
15].
Back pain (BP) is multi-factorial in both etiology and management [
16]. Treatments should therefore be tailored based on the patient’s needs [
17] and also equally distributed in relation to the patient’s needs [
18]. A wide range of treatment options are available within primary care [
19,
20] but there is still insufficient knowledge on how to direct individuals with BP to the right treatment option at the right time [
21,
22] and how to prevent acute BP and NP from becoming chronic [
23]. Consequently, a clinical and research priority is to, at an early stage, identify subgroups of patients with nonspecific BP and NP who are at risk of developing long-standing disability, in order to optimize treatment [
17,
24].
Psychological risk factors have a key role in the transition from acute to chronic pain and the development of long-term disability [
25‐
27]. The STarT Back Screening Tool (SBT) [
28] is a validated risk stratification tool that includes questions on modifiable physical and psychosocial risk factors for long-term BP, in order to match individuals to appropriate treatments according to their prognostic profile. Patients are classified into three risk groups; low, medium or high risk for long-term pain and disability. Patients at low risk of poor outcome are directed to supported self-management, education and advice including pain relief, encouragement to stay active and are also informed about an overall good prognosis. Those at medium risk are offered evidence-based physiotherapy interventions such as manual therapy and exercise. For those at high risk, a combined physical and psychological approach is recommended [
29]. Using the SBT together with targeted treatments has shown improved efficiency regarding patients’ clinical outcomes and reduced health care costs in the United Kingdom [
30]. The SBT’s psychometric properties have been tested in several countries and it is now used in a number of different international settings [
31‐
35]. The SBT was recently cross-culturally adapted and validated in Swedish in a small low back pain population (
n = 62) [
36].
The Örebro Musculoskeletal Pain Screening Questionnaire (ÖMPSQ) aims to identify patients at risk for developing work disability due to BP and NP [
37]. ÖMPSQ is one of the most widely used screening questionnaires and several studies demonstrate the utility of the ÖMPSQ, both in research and clinical settings [
38‐
41]. The ÖMPSQ was therefore considered the most appropriate reference standard against to validate the SBT. A short form of the ÖMPSQ (ÖMPSQ-short) has been developed to further increase the clinical utility of the ÖMPSQ [
42]. The ÖMPSQ-short has earlier been compared with the SBT for patients with low back pain [
33,
36,
43] but not yet for a population with patients applying for physiotherapy treatment due to BP and/or NP. The SBT has neither been compared or validated against the ÖMPSQ-short nor for a large primary care population in Sweden. Back and neck pain is common symptom in the general population and the pain from the back and neck often occurs concurrently [
44]. Therefore clinicians need brief, practical tools for both BP and NP to identify patients at risk of poor outcome in order to decide the most appropriate interventions at an early stage [
28,
42]. Since there is a lack of short and clinically useful instruments to guide clinicians in the management of patients with non-specific BP and/or NP, more research is needed. Therefore the overall aim of this study was to study the concurrent validity of the Swedish version of the SBT and the ÖMPSQ-short in a population of patients with acute or subacute BP and/or NP in a primary care setting. In addition, to investigate the agreement regarding classification into risk groups between the SBT and the ÖMPSQ-short and also to describe clinical utility of the two instruments.
Discussion
This is the first time that the SBT has been compared with the ÖMPSQ-short for a large group of patients with acute or subacute BP and/or NP in primary care in Sweden. This study demonstrated moderately strong correlations between the SBT and ÖMPSQ-short total scores for individuals with back and/or neck pain with short duration. The observed classification agreement was 70.2%. In subgroup analyses, we found the correlation was lower among females than for males and also lower among older age groups, especially females aged over 50 years. Clinicians made less miscalculations with the SBT compared to the ÖMPSQ-short.
The moderately strong correlation between SBT and ÖMPSQ-short total scores (0.62) in this study is similar to the results in a previous Swedish study (0.61) [
36] and a Brazilian study (0.73) [
43] that also compared SBT and ÖMPSQ-short total scores. Both these studies were in a low back pain population. Higher correlations have been reported when comparing the SBT with the original ÖMPSQ (24 item) for patients with low back pain in an English population (0.80) [
49] and in a French population (0.74) [
31] but also lower correlations have been found in a Finnish population (0.45) [
33]. Cross-cultural differences may be one factor that can explain the differences between different study results. In contrast to the above mentioned studies, the study population in this study also includes patients that can have NP, but this does not seem to affect the correlations substantially.
In the subgroup analyses, we found the correlation was lower in females than for males and was further reduced among increasing age groups. We have also used other age groups in the analyses but regardless of which type of age group we used, we found the same results. We unexpectedly found the correlation for females ≥50 years as poor while the correlation for males ≥50 years were still moderately strong. We can’t rule out that this difference between males and females ≥50 years might be random but there may also be biological differences. There is a gender gap concerning low back pain, where females are having more prevalent comorbidity of neck and shoulder pain and psychosocial distress than males [
44]. In this study there were more females than males and a greater share of females that had neck pain (68.5 vs 50%). On the other hand, when studying different age groups, a greater share of younger males had neck pain (51.3%) compared to the oldest age group (44.4%). This difference was not seen in females. In other studies, comparing the SBT and the ÖMPSQ, the percentage of females varies between 55.6 and 80.8% but no results on the percentage of participants with neck pain, nor subgroup analyses for correlation based on gender and age have been reported in these studies [
31,
33,
36,
43,
49].
Participant classification to low or high risk for long-term pain and disability by the SBT and the ÖMPSQ-short questionnaires had moderate agreement (κ = 0.42). Similar results were also found in the Brazilian study (κ = 0.49) [
43] and in a study where they compared the SBT with the original ÖMPSQ (24 item) in an English population (κ = 0.57) [
49]. The English study also showed significant differences between the SBT and the ÖMPSQ (24-item) regarding the threshold for high risk. The proportion of patients allocated to the high risk group in our study was higher for the SBT (53.7%) than for the ÖMPSQ-short (36.5%) which also was found by Fuhro et al. [
43]. In contrast, Hill et al. [
49] found that the SBT allocated fewer (35%) patients to the high risk group than the original ÖMPSQ (24-item) (38%). The 24-item ÖMPSQ has three risk groups, as the SBT. The higher proportion of patients allocated to the high risk group by the SBT in our study might have been influenced of the high risk classification we used when we merged the medium and high risk group for the SBT, but this method was used both in our and in the Brazilian study [
43] but not in the UK-study [
49]. Thus, when clinicians use the SBT instead of the ÖMPSQ-short, they will likely find more patients identified as medium or high risk by the SBT compared to the ÖMPSQ-short.
When choosing a classification instrument, clinicians and organizations need to be aware of, that patients at medium and especially patients at high risk need a more extensive treatment compared to those at low risk who can be reassured and offered less intensive treatment [
29]. Patients at high risk are especially important to identify at an early stage as this group of patients will benefit most from psychological informed physiotherapy approaches. But, it is also important to be aware of the potential of misclassifying high risk and that there may be patients not being high risk and not being appropriate for the enhanced intervention approach.
However, patients with medium and high risk are those who will benefit most from physiotherapy [
30] and to identify them at an early stage, will maximize treatment benefit, reduce harm and increase health-care efficiency by offering the right treatment to the right patient at the right time [
17].
To the best of our knowledge, this is the first time that clinical utility has been focused when comparing the two instruments. It was a difference regarding completion rates from patients with SBT having some incomplete questionnaires (n = 11) while the ÖMPSQ-short had none. One contributing factor might be the order in which the questionnaires were completed. Patients completed the ÖMPSQ-short first, as the ÖMPSQ-short was used as an inclusion criteria to the clinical trial (ClinicalTrials.gov ID: NCT02609750). The physiotherapists might have checked all questions and calculated the ÖMPSQ-short scores more carefully. Physiotherapists in Sweden are also more used to score the ÖMPSQ-short than the SBT. The rates of missing calculations of total and subscale scores in the SBT might be due to the instructions to the physiotherapists. They were not explicitly told to do these calculations. The fact that it was the treating physiotherapist that administered and scored the SBT and that they had different experiences of using the two questionnaires may have influenced the rates of miscalculations of scores. But, at the same time, this might have had minimal impact of the results as there were so many primary care centers (35) in different regions included. The higher rate of miscalculations of total scores in ÖMPSQ-short (54/315) compared to SBT (13/315) indicates that the SBT is easier to score for clinicians. A potential benefit of using the SBT instead of the ÖMPSQ-short might be that the SBT seems to be more clinical feasible to use in routine clinical practice. More miscalculations in the ÖMPSQ-short might be due to the more complex scoring with 0–10 points for each item and also the reversed scoring in three items (3, 4 and 8). However, even though there were more miscalculations of total scores in the ÖMPSQ-short than in the SBT, there were no difference in misclassification to either a higher or lower risk group between the ÖMPSQ-short (n = 7) and the SBT (n = 5) questionnaires. Consequently, miscalculations of the SBT and the ÖMPSQ-short total scores do not seem to substantially influence the risk classification. Clearly introducing electronic questionnaires with automatic summations have the potential to eliminate errors even further.
Strengths and limitations
The main strength of the present study is the size of the study population (n = 315) at a great number (35) of different real world primary care settings. The same individuals completed both the SBT and ÖMPSQ-short at the same time. We have thoroughly checked and validated all questionnaires and also studied psychometric properties and clinical utility in a primary care setting. The population studied was relatively homogenous including only patients with short duration of pain, not participants with chronic pain. This means that we can identify patients at risk of poor prognosis at an early stage where it is still possible to do brief interventions and influence outcome by treating modifiable prognostic factors and stratify care.
To our knowledge, this is the first time SBT is validated in a population with both NP and BP. A limitation in this study is that we have no diagnoses registered for the study population and therefore we were not able to distinguish patients diagnosed with only NP. On the other hand, having NP with or without comorbid BP is common [
44,
50] and thus makes the results of this study applicable to a common clinical situation. Another limitation was that we merged the medium and high risk group for the SBT in the analysis to allow a comparison with the ÖMPSQ-short. But regardless of merging the medium and high risk group or merging the low and medium risk group for the SBT, we found significant differences in observed agreement between the two instruments. The observed classification agreement was 70.2% but there was at the same time a disagreement rate of 29.8%. The agreement analysis strategy used in this study by merging the medium and high risk group for the SBT may limit the ability to evaluate for false positive or negative SBT medium risk classification. However, the findings in this study with a moderately strong correlation between the SBT and the ÖMPSQ-short scores support that the SBT can be used as a clinical tool for patients with acute and subacute BP and/or NP in primary care and that the SBT will provide clinicians with more additional guidance in the level of care compared to the ÖMPSQ-short. The SBT is designed for stratified care, which involves targeting treatment to subgroups of patients based on their key characteristics [
17] and we think this is an advantage in primary care because clinical intuition does not always match consistently to patient prognosis [
51]. When using stratified care, clinicians can minimize the risk of overtreatment for low risk patients and give more appropriate treatment for medium and high risk patients [
30]. The SBT may help the physiotherapists, how to prioritize between different treatments pathways. Future studies are needed to study how the SBT can predict chronic pain and work disability for the target group of patients with BP and/or NP in primary care and especially in the group of elderly females.
Acknowledgements
The authors would like to thank senior statistician Mikael Åström for statistical support, the research team at Epidemiology and Register Centre South, Region Skåne, Lund, Sweden, the research team at the Institute of Primary Care and Health Sciences, Keele University, Stoke-on-Trent, UK, the staff at Blekinge Centre of Competence, Karlskrona, Sweden for help with support and also physiotherapists and patients involved in the WorkUp research project in Region Skåne, Region Kronoberg and Landstinget Blekinge, Sweden, for help with data collection.