Article Text
PDF
Abstract
Background To describe the diagnoses of people who present to the emergency department (ED) with low back pain (LBP), the proportion of people with a lumbar spine condition who arrived by ambulance, received imaging, opioids and were admitted to hospital; and to explore factors associated with these four outcomes.
Methods In this retrospective study, we analysed electronic medical records for all adults presenting with LBP at three Australian EDs from January 2016 to June 2018. Outcomes included discharge diagnoses and key aspects of care (ambulance transport, lumbar spine imaging, provision of opioids, admission). We explored factors associated with these care outcomes using multilevel mixed-effects logistic regression models and reported data as ORs.
Results There were 14 024 presentations with a ‘visit reason’ for low back pain, of which 6393 (45.6%) had a diagnosis of a lumbar spine condition. Of these, 31.4% arrived by ambulance, 23.6% received lumbar imaging, 69.6% received opioids and 17.6% were admitted to hospital. Older patients (OR 1.79, 95% CI 1.56 to 2.04) were more likely to be imaged. Opioids were less used during working hours (OR 0.81, 95% CI 0.67 to 0.98) and in patients with non-serious LBP compared with patients with serious spinal pathology (OR 1.65, 95% CI 1.07 to 2.55). Hospital admission was more likely to occur during working hours (OR 1.74, 95% CI 1.48 to 2.05) and for those who arrived by ambulance (OR 2.98, 95% CI 2.53 to 3.51).
Conclusion Many ED presentations of LBP were not due to a lumbar spine condition. Of those that were, we noted relatively high rates of lumbar imaging, opioid use and hospital admission.
- emergency department
- pain
- health services research
Statistics from Altmetric.com
Introduction
Low back pain (LBP) is the leading cause of years lived with disability worldwide.1 From 2005 to 2015, the global prevalence of activity-limiting LBP increased 17%, currently affecting 540 million people.2 LBP usually has a favourable prognosis,3 and only a very small proportion of cases are attributable to serious pathology, so guidelines recommend that the condition be managed in primary care.4 However, many patients with LBP present to the ED across the globe. For example, LBP is responsible for 2.63 million ED visits annually in the USA, with direct costs estimated at US$819 million.5
There is limited research describing the profile of patients presenting to the ED with LBP and the care that is provided to them. The largest and most comprehensive studies are from the USA,5–7 and showed high rates of opioid (61.7%) and imaging use (30.5% for simple radiographs).5 These results, however, may not generalise to countries with markedly different healthcare systems, such as Australia. In addition, most studies looking at ED presentations of LBP have not provided a comprehensive description of pharmacological management of these patients. Studies conducted in other countries, such as Australia, have either only quantified imaging use,8 or described limited characteristics of the presentation (eg, ED length of stay).9 To date, little is known about opioid use for LBP in Australian EDs.
Current literature has given little attention to the characteristics of patients with LBP transported to the ED by ambulance. There has been an increase in the use of ambulance services for people with low-acuity health conditions, which reduces the availability of the service for managing a high-acuity caseload, further contributing to ED overcrowding and increasing costs of ambulance services.10 11 Characteristics of presentations to ED leading to inpatient admission to hospital due to LBP have also been overlooked, even though rates of hospitalisation due to LBP in Australia have increased 10% from 2005 to 2015.12
The goals of this study were to describe the diagnoses, patient and care profile of people who presented to ED with LBP. Focusing on those diagnosed with a lumbar spine condition, we examined the proportion of people arriving at the ED by ambulance, receiving imaging, opioids and being admitted to hospital; and explored factors associated with these four outcomes.
Methods
Participants and settings
This is a retrospective analysis of routinely collected electronic medical record data reported in accordance with the STROBE guidelines.13 The Sydney Local Health District (Royal Prince Alfred Hospital zone) Ethics Committee granted approval for this study (protocol no. X17-0419 & LNR/17/RPAH/631).
We examined electronic medical records for all patients aged 18+ years who presented to the ED with a complaint of back pain to one of three hospitals within the Sydney Local Health District from January 2016 to June 2018: Royal Prince Alfred Hospital (RPA), Concord Repatriation General Hospital (CRGH) and Canterbury Hospital. The estimated catchment population living in Sydney Local Health District is 670 000 people.14 All three hospitals are public, located in the metropolitan Sydney, New South Wales, Australia, and have 1679 beds in total. Canterbury is a major metropolitan hospital, CRGH provides tertiary services, and RPA provides tertiary and quaternary level services.
To identify LBP presentations, we identified ED patients with a ‘visit reason’ related to a complaint of back pain using relevant keywords (eg, ‘back pain’, ‘lower back pain, ‘back discomfort’, ‘flank pain’). The discharge diagnoses were coded to the Systematized Nomenclature of Medicine Clinical Terms—Australian Version (Emergency Department Reference Set) (SNOMED CT-AU (EDRS)) codes (online supplementary file 1).15 We collapsed the SNOMED CT-AU (EDRS) codes into those beyond the lumbar spine (see online supplementary file 2 for a description of all diagnostic categories) and three lumbar spine condition categories: LBP with non-specific cause (non-specific LBP), LBP with neurological signs and symptoms (radicular LBP), and LBP due to serious spinal pathology (serious LBP). Disagreements were resolved by consultations with a clinician (CN) with experience in using SNOMED CT-AU (EDRS).
Supplemental material
Data sources
All data were extracted from the Sydney Local Health District Targeted Activity and Reporting System (STARS). STARS is a business intelligence programme (trade name Qlik Sense) that provides reporting and analytical functions across multiple datasets. All patients seen in Sydney Local Health District EDs have their episode of care recorded in the electronic medical records stored in a range of datasets stored separately in the Health Information Exchange database, and we used STARS to link them using the patient’s unique identifier.16
Data collection
We extracted the following data from STARS: demographic information (eg, age, gender and postcode), presentation year (2016 to 2018), mode of arrival (eg, ambulance, public transport, walked in or private car), length of stay in the ED (hours), presentation day and hour (presentation during working hours were those from Monday to Friday between 8:00 and 17:00), hospital length of stay for those admitted to hospital (days), representation to the same ED within 48 hours, imaging and medications for pain. We also extracted the triage category assigned by the triage nurse based on the Australasian Triage Scale (ATS).17 The ATS is a 5-point scale (1–5) where 1 represents life-threatening condition requiring immediate response and resuscitation, 2 are emergency cases, 3 are urgent cases, 4 are semi-urgent cases, and 5 represents less urgent or clinical administrative problems requiring assessment and management to start within 2 hours. Age was dichotomised, with age ≥65 years set as the cut-off value for being older. Socioeconomic status was derived from the postcode of the patient from the Australian Bureau of Statistic’s Socio-Economic Indexes for Areas 2016 and reported as deciles, with the lowest decile designating areas with the highest socioeconomic disadvantage.18 Medications for pain administered in the ED were classified according to the Anatomical, Therapeutic and Chemical Classification System19 (online supplementary file 3).
Outcomes
Outcomes of this study were the discharge diagnoses, and, in those with a diagnosis of a lumbar spine condition, we computed the proportion of presentations that (1) arrived at the ED by ambulance, (2) underwent any lumbar spine imaging (spinal radiographs, CT and MRI) while in the ED, (3) used any opioid analgesic in the ED and (4) resulted in inpatient hospital admission. We reported imaging as simple radiographs and advanced imaging (MRI and CT). We reported opioids as the proportion of any opioid intake in the ED. Opioids included opioid anaesthetics (eg, fentanyl), opioids (eg, oxycodone) and opioids in combination with non-opioid analgesics (eg, codeine–paracetamol).20 Presentations resulting in admission to hospital were those where a patient had been allocated to an inpatient hospital ward.
Statistical methods
Continuous variables were presented as mean (SD) or median (IQR) where appropriate. Categorical variables were presented as frequency (%). Multilevel mixed-effects logistic regression models were used to account for the clustering effect of the hospital participants presented to.20 For the multivariate analysis, we grouped patients with non-specific LBP and radicular LBP together under the label ‘non-serious LBP’, as guideline recommendations for imaging and opioid prescription are similar for both conditions.21 Patient characteristics (gender, age and socioeconomic status) and characteristics of the presentation (triage category,17 mode of arrival, duration of ED presentation and whether or not the current visit was a representation) were forced-entered as predictors in the multivariate models to explore its association with the three outcomes. We used the least urgent triage category (ATS 5) as the reference category when entering triage category in the multivariate model. The selection of predictors was theory driven, based on the available literature and on the expertise of the study team. For example, previous studies have shown that factors such as male gender, age <65 years and lower socioeconomic status were associated with increased utilisation of opioids in primary care22 and that older age increased odds of hospital admission following an ED visit.23 We hypothesised that people arriving by ambulance, assigned to more urgent triage categories and with serious spinal pathology would have higher rates of imaging, use of opioids in the ED and admission to hospital. We reported data as ORs and 95% CIs. Analyses were carried out in STATA V.15.0.
Results
Sample characteristics
From January 2016 to June 2018, there were 409 405 presentations to the three EDs. From these, we identified 14 024 (3.4%) presentations with a presenting complaint of ‘back pain’—6393 (45.6%) with a discharge diagnosis specific to the lumbar spine and 7631 (54.4%) with a diagnosis of a disorder beyond the lumbar spine. One hundred and six patients left without receiving any treatment and another eight left against medical advice.
Figure 1 shows the discharge diagnoses of those presenting with back pain. Of the non-lumbar spine diagnoses, renal disorders were the most prevalent, with 2644 presentations (18.8%). Among these, renal colic was the most common diagnosis (2072 presentations, 14.8%). A diagnoses of flank pain was given to 896 (6.4%). We identified 757 (5.4%) presentations that were given a discharge diagnosis of a generic musculoskeletal condition, for example, ‘musculoskeletal pain’ (n=301) that did not allow us to ascertain if the diagnosis was related to LBP or another musculoskeletal condition (eg, neck pain). An additional 652 (4.6%) presentations received a discharge diagnosis of other musculoskeletal conditions (eg, neck pain, hip pain). A cancer diagnosis was given to 35 presentations (0.2%). The most prevalent cancer diagnosis was secondary malignant neoplastic disease (eight presentations, 0.05%). Vascular disorders, including abdominal aortic aneurysm, accounted for 24 (0.2%) presentations. Online supplementary file 2 presents data for all 7631 presentations diagnosed with a pathology beyond the lumbar spine.
Diagnostic categories based on the discharge diagnosis given to patients (n=14 024). Grey bars represent disorders beyond the lumbar spine (n=7631). Black bars represent lumbar spine conditions (n=6393).
The 6393 ED presentations that were given a discharge diagnosis of a lumbar spine condition represented 45.4% of all presentations that had an initial ‘visit reason’ of ‘back pain’. Among these 6393 presentations with a lumbar spine diagnosis, 5461 (85.4%) were non-specific LBP, 642 (10.1%) radicular LBP and 290 (4.5%) serious spinal pathology. The characteristics and interventions used in this population are described in table 1. The mean age of the lumbar spine condition subsample was 52.4 (SD 20.2) years, with older adults representing 30.8% of these presentations. Women represented 50.1% of all presentations. Most presentations arrived by private car (55.5%) followed by ambulance (31.4%). Most presentations were categorised as ATS 3 (36.2%) and ATS 4 (58.7%) and happened outside working hours (58.7%). During the 30-month data collection period, there were 63 representations within 48 hours to the same ED (0.9%) (table 1).
Demographic and ED presentation characteristics
Electronic data on pain medications were not available for Canterbury Hospital and became available for RPA from 1 November 2017. Hence, we present pain medication data for 2539 of the 6393 patients with a lumbar spine condition diagnosis. Opioids were the most common pain medication given in the ED, with 1767 patients receiving at least one opioid (69.6%). Opioids alone were the most commonly used types of opioids, used in 1559 (61.4%) presentations. Opioids in combination with non-opioids were used in 434 (17.1%) of presentations, and opioid anaesthetics were used in 46 (1.8%) presentations. Among all opioids, oxycodone was the most commonly used (1436 presentations, 56.6%), followed by paracetamol–codeine (331 presentations, 13.0%) and morphine (257 presentations, 10.1%). A full list of all opioids used in our sample of patients is described in online supplementary file 4. Simple analgesics (paracetamol or acetylsalicylic acid) were given for 1421 patients (55.9%), and non-steroidal anti-inflammatory drugs (NSAIDs) were given for 956 presentations (37.6%). Skeletal muscle relaxants (baclofen or orphenadrine) were given to only five patients (0.2%), whereas benzodiazepines were given to 237 patients (9.3%) (table 2). Non-opioid analgesics (ie, analgesics, NSAIDS, skeletal muscle relaxants or benzodiazepines) used in isolation accounted for 766 (30.1%) presentations, whereas the majority of presentations (1767, 69.5%) used at least one opioid in conjunction with at least one non-opioid analgesic.
Number (%) of different types of pain medication used in the ED (n=2539)
Persons arriving by ambulance
A higher proportion of older people with lumbar spine diagnoses arrived by ambulance (46.4%) compared with younger people (23.6%). The proportion of presentations arriving by ambulance was similar between non-specific LBP (32.4%) and serious spinal pathology (33.1%). Multivariate regression (n=5986) identified variables independently associated with having arrived at the ED by ambulance. People over 65 years old were 2.8 (95% CI 2.52 to 3.20) times more likely to arrive by ambulance than younger adults. For each increase in the socioeconomic status, the odds of presenting to the ED by ambulance increased by 1.04 (95% CI 1.01 to 1.06). Presentations occurring during working hours were 0.78 (95% CI 0.69 to 0.87) times less likely to arrive by ambulance. There were no presentations categorised as life-threatening (ATS 1). Patients classified as ATS 2, 3 and 4 were 6.33 (95% CI 3.47 to 11.55), 3.46 (95% CI 2.19 to 5.47) and 2.48 (95% CI 1.58 to 3.90) times more likely to arrive by ambulance than patients classified as ATS 5. Patients with serious spinal pathologies were not more likely to arrive by ambulance than those with non-specific LBP or radicular LBP (OR 1.02, 95% CI 0.77 to 1.34) (online supplementary file 5).
Lumbar spine imaging
In total, 1508 (23.6%) presentations with lumbar spine diagnoses received at least one type of lumbar spine imaging (table 3). A total of 1223 (19.1%) presentations received a simple spinal radiograph, 338 presentations received a CT scan (5.3%) and 86 presentations received an MRI (1.3%). Multivariate logistic regression (n=5986) identified several factors independently associated with receiving at least one type of lumbar spine imaging in the ED (online supplementary file 6). For example, men were 0.85 times less likely to receive any lumbar spine imaging than women (0.85, 95% CI 0.75 to 0.96). Older adults (65+ years) were 1.79 (95% CI 1.56 to 2.04) times more likely to receive any lumbar spine imaging as compared with younger persons. Individuals arriving at the ED by ambulance were 1.36 (95% CI 1.19 to 1.56) times more likely to receive any lumbar spine imaging than those arriving by other means (eg, private car or public transport). Compared with less urgent presentations (ATS category 5), presentations assigned to ATS 2, 3 and 4 were 4.72 (95% CI 2.26 to 9.87), 3.58 (95% CI 1.95 to 6.57) and 3.06 (95% CI 1.68 to 5.59) times more likely to receive any lumbar spine imaging, respectively.
Number (%) of patients receiving lumbar spine imaging
Opioid analgesics
In total, 1767 (69.6%) presentations received at least one opioid. Rates of opioid use were similar between presentations with non-specific LBP (68.3%) and serious spinal pathology (63.7%), and slightly higher for patients with radicular LBP (74.9%). Opioids were used in combination with benzodiazepines in 215 (8.5%) presentations. Multiple regression (n=2395) identified several factors independently associated with receiving any opioid in the ED (online supplementary file 7). Presentations arriving by ambulance were 1.33 (95% CI 1.06 to 1.66) more likely to receive an opioid. Presentations occurring during working hours were 0.81 (95% CI 0.67 to 0.98) times less likely to receive an opioid. Considering ATS category 5 as reference, presentations assigned to ATS categories 2, 3 and 4 were 5.71 (1.97 to 16.50), 6.72 (3.20 to 14.10) and 4.62 (2.24 to 9.54) more likely to receive an opioid. Presentations deemed to have non-serious LBP were 1.65 (1.07 to 2.55) times more likely to receive an opioid. For each hour spent in the ED, the odds of receiving an opioid increased by 1.39 (1.31 to 1.48).
Admission to hospital
The overall rate of admission to inpatient hospital wards was 17.6%. Admission rates were similar between non-specific LBP (16.4%) and radicular LBP presentations (16.5%), but higher in those diagnosed with serious spinal pathology (43.1%). The overall median (IQR) length of stay was 6 (3–12) days. Multiple regression (n=5986) identified several variables independently associated with admission to hospital (online supplementary file 8). The total number of admissions was 1129 (17.6%) (table 1). Adults 65+ years old were 3.0 times more likely to be admitted (95% CI 2.59 to 3.59). Those who arrived at the ED by ambulance were 2.98 (95% CI 2.53 to 3.51) times more likely to be admitted. Presentations occurring during working hours were 1.74 (95% CI 1.48 to 2.05) times more likely to be admitted. Presentations due to non-serious LBP were 0.23 times less likely to be admitted (95% CI 0.17 to 0.32) compared with presentations due to serious spinal pathology. Presentations assigned to ATS categories 2 and 3 were 3.73 (95% CI 1.48 to 9.38) and 2.99 (95% CI 1.37 to 6.48) more likely to be admitted compared with ATS 5 presentations. Nevertheless, there was no difference in the odds of being admitted between patients assigned to ATS 4 versus ATS 5 (OR 1.83, 95% CI 0.84 to 3.95).
Discussion
In people presenting to the ED with back pain, about half had a condition beyond the lumbar spine. In those with a lumbar spine condition, 69.6% received an opioid, 23.6% were imaged and 17.6% were admitted. Those allocated a more urgent triage category were more likely to receive lumbar spine imaging, opioids and being admitted to hospital. Being over 65 years old and having arrived by ambulance increased the odds of imaging and inpatient hospital admission. Having a serious spinal pathology did not increase the odds of arriving at the ED by ambulance compared with having non-specific LBP or radicular LBP.
One strength of our study was the inclusion of all patients that presented to the study EDs over a 30-month period. That approach avoided selection bias, which is commonly found in retrospective studies. We identified the discharge diagnoses of all patients presenting to the ED with a complaint of ‘back pain’. In the ED setting, lumbar spine conditions account for about half the back pain presentations. Our findings are in accordance with an Australian study9 in which 53% of ED presentations with a visit reason of LBP received a diagnoses of ‘non-muscular LBP’, such as renal colic, but we provide greater details on the diagnoses.
A second strength of our study was the allocation of patients into three different categories: non-specific LBP, radicular LBP and serious spinal pathology. In previous studies, patients with non-specific LBP and radicular LBP were presented as a homogeneous group.5 24 Our approach revealed nuances in management across the three LBP subgroups, which so far had not been captured. For example, patients with radicular LBP had a lower rate of any lumbar spine imaging (16.8%) than those with non-specific LBP (23.5%). The observed rates of imaging for non-specific LBP in our study are not in line with the recommendations from the American College of Emergency Physicians Choosing Wisely campaign.25
Adults over 65 years were almost twice as likely as younger individuals to receive lumbar spine imaging and three times more likely to be admitted than their younger counterparts. Older adults represent a more complex population with more comorbidities that usually require more medical attention.26 We believe this factor might have been the driver of the higher rates of imaging in older patients observed in our study. Additionally the prevalence of serious spinal pathology in older adults was slightly higher (5.2%) than that observed in younger individuals (4.2%).
Prior to our study, data on opioid use for LBP in Australian ED settings were unavailable; and our study revealed a higher proportion of opioid utilisation (69.6%) than in previous studies conducted in the USA5 (61.7%) and Canada (34.5%).24 Despite the uniform message across guidelines for LBP4 21 27 to reduce prescription of opioids, to initiate treatment with non-opioid analgesics, and to resort to opioids only when other medicines have not been tolerated or have failed, opioids were the most common analgesic strategy adopted in our study. These findings are concerning for a number of reasons. The efficacy of opioid use for LBP in the ED setting has not been established,28 29 and about half of those with LBP who start the medicine stop taking it because of lack of effect or tolerability.28 Furthermore, recent trials suggest that the effects between opioid and non-opioid medications are smiliar.30 31 In addition, opioids were used in conjunction with benzodiazepines in 8.46% of presentations with spinal conditions in our study. Benzodiazepines have shown to be ineffective in the ED for patients with non-specific LBP,32 and their use concurrent with prescription opioids has been associated with increased risk for opioid-related overdose.33 Our data are restricted to opioid use in the ED, and very little is known about the relationship between opioid use during an ED visit and subsequent prescription at discharge, which is known to increase risk of long-term opioid use.34 One study carried out in the USA showed there were significant differences in the opioids administered in the ED versus those prescribed at discharge. Parenteral administration of hydromorphone and morphine was common in ED whereas prescription of oxycodone and hydrocodone was common at discharge.35 Oxycodone is the most commonly prescribed opioid analgesic in any clinical setting in Australia.36 This is problematic as oxycodone has a high misuse potential and is now responsible for far more deaths and poisoning hospitalisations than illegal opioids such as heroin.36 The findings from our study suggest that practitioners widely regard opioid analgesics, particularly oxycodone, as a powerful analgesic medicine for the management of severe acute low back pain. However, qualitative studies are required to validate this assumption and to understand the drivers of opioid administration and prescription in ED. Given the small expected benefit and the appreciable harms associated with opioid use for LBP, strategies16 to review its use in the ED are needed. Some non-randomised studies have suggested that education interventions, such as active dissemination of guidelines37 and interdisciplinary discussions among ED staff supplemented by a medical record decision support tool38 may decrease prescription opioids. A small study also showed that a state-wide prescription monitoring programme can alter ED physicians’ opioid prescribing behaviours.39
Admissions rates in our study (17.6%) were substantially higher than rates typically reported in the studies conducted in the USA and Canada. Those studies used a similar definition of back pain to ours, and their admission rates ranged from 2.2%5 to 2.5%,24 respectively. In contrast, our rates were very similar to that reported by Lovegrove et al 9 for Australia (17.1%) but lower than that reported by McCaughey et al 8 (34.1%) also for Australia. A recent study showed that older age (≥80 years), sciatica and vertebral fracture were significantly associated with longer inpatient length of stay,12 in accordance with our findings.
Presentations arriving by ambulance were almost three times more likely to be admitted than those arriving by other means. To the best of our knowledge, our study is the first showing that use of ambulance services doubled and tripled the odds of receiving lumbar spine imaging and being admitted to hospital, respectively. The use of ambulance services by patients with low-acuity conditions that do not require the resources provided by ambulances is common, with data suggesting a variation between 11.3% and up to 51.7%.11 40 The high proportion of patients arriving by ambulance in our study may indicate that overutilisation of ambulance services by patients with LBP is common. Initiatives such as ambulance secondary telephone triage11 might be a suitable alternative to reduce overuse and to promote a more rational use of ambulance services resources.
A limitation of our study was the lack of patient-reported outcome measures (eg, pain, disability), as well as a data on adverse events experienced by patients during the ED presentation, a common limitation across many studies in this field.5–7 9 23 24 41 Nevertheless, we reported ATS categories assigned by the triage nurse, which offers some guidance on likely pain intensity.42 Our findings showed that ATS 2 presentations, those in very severe pain, were more likely to arrive by ambulance, receive any lumbar spine imaging technique, receive an opioid or be admitted compared with ATS 5 presentations—that is, presentations with minimal levels of pain. A second limitation of our study was the lack of data on medications prescribed to patients at discharge. Because we used routinely collected data, some presentations may have been miscoded.
In summary, in patients diagnosed with a lumbar spine condition, relatively high imaging rates were noted, as well as high rates of opioid use and hospital admission. More than half of patients presenting with a complaint of back pain were diagnosed with a disorder beyond the lumbar spine. Our findings point to the ongoing need to better adopt best practice in the management of LBP in EDs.
Acknowledgments
GEF is supported by Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) Brazil with a PhD scholarship. GCM is supported by a National Health and Medical Research Council (NHMRC) Early Career Fellowship. CAS is supported by an University of Sydney Early Career Development Fellowship. CL is funded by an NHMRC Career Development Fellowship. CGM is funded by an NHMRC Principal Research Fellowship.
We thank Hannah Storey and Noel Baidya from the Performance Monitoring, System Improvement and Innovation Unit, Sydney Local Health District for the support and development of the STARS Back Pain Dashboard.
References
Footnotes
Contributors The corresponding author (GF) attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted.
Funding This study has received a 2017 Kickstart Grant ($25,000) from Sydney Medical School, The University of Sydney
Competing interests C-WCL and CGM were investigators on the PRECISE study, an investigator-initiated trial evaluating pregabalin for sciatica, funded by the National Health and Medical Research Council of Australia with in-kind research support from Pfizer (ACTRN12613000530729). CN received fees for the provision of education seminars for the SHaPED trial, a stepped-wedge cluster randomised trial evaluating the implementation of a model of care for patients with low back pain presenting to emergency departments (ACTRN 12617001160325). All other authors declare no support from any organisation for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous 3 years; no other relationships or activities that could appear to have influenced the submitted work.
Patient consent for publication Not required.
Ethics approval Research ethics and governance office Royal Prince Alfred Hospital (reference nos. X17-0419 and LNR/17/RPAH/631).
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available on reasonable request.