Low back pain presentations to rural, regional, and metropolitan emergency departments

Abstract Objective To describe the context of low back pain (LBP) presentations to emergency departments (EDs) by remoteness areas, hospital delineation level and staffing portfolios. Design A retrospective observational study using routinely captured ED and admission data over a 5‐year period (July 2014–June 2019). Settings Thirty seven EDs across a large health district in NSW, Australia, covering major cities, inner regional areas and outer regional areas. Participants Emergency department (ED) presentations with a principal or secondary diagnosis of LBP based on ICD‐10 code (M54.5). Main outcome measures ED presentation and associated admission measures, including presentation rate, referral source, time in ED, re‐presentation rate, admission details and cost to the health system. Results There were 26 509 ED presentations for LBP across the 5 years. Time spent in ED was 206 min for EDs in major cities, 146 min for inner regional EDs and 89 min for outer regional EDs. Re‐presentation rates were 6% in major cities, 8.8% in inner regional EDs and 11.8% in outer regional EDs. Admission rates were 20.4%, 15.8% and 18.8%, respectively. Conclusions This study describes LBP presentations across 37 EDs, highlighting the potential burden these presentations place on hospitals. LBP presentations appear to follow different pathways depending on the ED remoteness area, delineation level and staff portfolio.


| INTRODUCTION
Low back pain (LBP) is common and disabling and causes reduced quality of life. 1,2 Approximately 84% of individuals are likely to develop LBP at some time in their life. 3 While international guidelines recommended initial management in primary care, many patients with LBP present to hospital emergency departments (EDs). 4 LBP is the 5th most common cause for presentations to Australian EDs. 5 Numerous Australian and international studies have described ED presentations for LBP, [6][7][8][9] but most concentrate on metropolitan hospitals.
Regional and rural hospitals face different operational challenges compared with metropolitan facilities. Clinician capabilities and available resources and services in EDs are often specific to the location and hospital delineation. 10,11 Time pressures for understaffed facilities in rural areas, limited opportunities for professional development related to musculoskeletal conditions, and limited access to Fellowship of the Australasian College of Emergency Medicine (FACEM)-trained physicians may lead to higher variation in the management of LBP. Metropolitan EDs are often staffed by FACEM physicians who are specifically trained to manage presentations such as LBP in the ED. In contrast, regional and rural EDs are staffed by General Practitioner Visiting Medical Officers (GPVMO). They also work as general practitioners in the community and therefore require a broader field of knowledge. These variations may account for differences in the management of LBP in EDs across these areas. 12 Rural EDs also have more demand per capita and fewer options to support the management of patients upon discharge. 5,13 One previous study reported higher presentation rates for LBP in regional and rural areas than in metropolitan areas. 14 Australian national data show residents of remote and very remote areas are more likely to have been to an ED in the past 12 months than their metropolitan counterparts due to limited general practitioner availability. 13,15 Access to medical specialists (e.g. orthopaedic or neurosurgical teams) is also poorer in rural areas. 13,15 The differing demand, hospital facilities and external medical services for EDs in regional and rural areas mean the challenges for managing LBP may not be the same as metro EDs. Despite this, rural and regional hospitals are often held to similar service-level performance indicators. For example, all New South Wales (NSW) public hospital EDs are currently measured against the same performance indicators. The emergency treatment performance target is a statewide mandate that calls for ED processes to be completed within 4 h for most ED presentations. 16 and 89 min for outer regional EDs. Re-presentation rates were 6% in major cities, 8.8% in inner regional EDs and 11.8% in outer regional EDs. Admission rates were 20.4%, 15.8% and 18.8%, respectively. This study describes LBP presentations across 37 EDs, highlight-ing the potential burden these presentations place on hospitals. LBP presentations appear to follow different pathways depending on the ED remoteness area, delineation level and staff portfolio.

K E Y W O R D S
musculoskeletal disorders, public hospital, regional health, rural health, service improvement What is already known on this subject: • Low back pain is the 5th most common reason for presentation to Australian emergency departments • Low back pain presentations to emergency departments are well described in metropolitan areas • Higher presentation rates for low back pain have been reported in rural areas compared with metropolitan areas What does this study adds?
• There are considerable differences in presentations of patients with low back pain in different areas of remoteness, particularly related to time in ED, re-presentation rate and admission rate • Service-level outcomes including time in ED and re-presentation rate are different depending on ED staffing portfolio and inconsistent across hospital delineation level • The study shows that strategies to improve care for patients with low back in the ED likely need to consider the local context at each facility. A one-size-fits-all approach is unlikely to be effective The planning of services and support mechanisms should consider ED performance indicators and the specific challenges that facilities face. There has been no documentation of LBP presentations to EDs across different localities to guide service support for regional and rural EDs. 17 This study aimed to describe characteristics of LBP presentations and servicelevel performance indicators to EDs across different remoteness areas, hospital delineation levels and ED staffing portfolios.

| Sample
The health district is a large NSW public health service, encompassing 131 785 km 2 , with a total patient catchment area of 912 352 people in 2016. 18,19 The Australian Statistical Geography Standard Remoteness Structure defines five categories of remoteness: major cities, inner regional areas, outer regional areas, remote areas and very remote areas. 20,21 Approximately 51% of the health district population is classified as residing in a major city area, 35% in an inner regional area, and 14% in outer regional and remote areas. 22

| Data capture
The Emergency Department Data Collection, which forms part of the Health Information Exchange, contains records of all ED presentations across NSW Health jurisdictions. The data originate from the Patient Administration System, which tracks the patient from their arrival, through their ED visit (+/-hospital admission), and includes basic discharge information. Upon discharge from ED, the treating clinician enters a principal diagnosis code (+/− a secondary diagnosis code), which documents the reason for the ED presentation. If the patient is admitted, coders review the clinical notes and determine the diagnosis codes to assign. Available diagnostic codes are based on a subset of the International Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD-10) codes. 23 We identified and extracted data from the health information exchange for all presentations to the EDs within the health district for LBP within the study period based on a principal or secondary diagnosis code of LBP (M54.5). An independent data analyst confirmed the code captured relevant LBP presentations and omitted specific causes (e.g. fracture, spinal cord injury, cauda equina syndrome and radicular low back pain). To ensure no relevant presentations were missed, we completed a keyword search of all ED presentations using the keywords of 'back' AND 'pain' but NOT 'back of'. We queried admitted patient data to capture information from any admissions (inpatient wards and emergency short-stay units) directly following ED presentation for the patients identified with LBP through ED records.
We extracted data for the variables described in Table 1. We coded arrival time as AM or PM, and arrival in or out of business hours (business hours 09:00-17:00). We calculated time in ED using presentation and discharge dates/times. We calculated the cost of presentations using the National Weight Activity Unit (NWAU) for each presentation or admission, multiplied by the national efficient price (NEP) for the relevant NWAU version. We based NWAU figures on all relevant variables for the year of ED presentation and associated admission (e.g. for ED presentation, Indigenous status, compensable status, funding eligibility indicator, establishment remoteness area and Urgency Disposition Group [UDG] code). 24 To maintain a homogeneous sample, presentations were limited to 12 years of age and older. We based this decision on consensus from the investigator group considering the rising prevalence of non-specific low back pain from age 12. [25][26][27] An audit of presentations for patients less than 12 years revealed that most were trauma-related or incorrectly coded.

| Analysis
We summarised descriptive data using mean (standard deviation [SD]) or median (inter-quartile range [IQR]) for continuous variables, depending on distribution, and n (%) for categorical variables.
We used three hospital classification methods to describe the facilities.
1. Hospital postcode to categorise according to remoteness. There were 5 EDs in a major city, 13 in an inner regional area and 19 in an outer regional area. There are no facilities in remote or very remote areas within the health district. 2. Delineation levels to categorise according to hospital service capabilities. All EDs in NSW are classified according to a delineation level, which describes minimum workforce and service requirements for safe clinical service delivery. 28

| Baseline characteristics
The mean age of patients was 49.2 years (SD 20.0), 51.7% were female, and 10.3% identified as Aboriginal or Torres Strait Islander (7.1% of residents in the health district identify as Aboriginal or Torres Strait Islander 30 ). The majority of patients resided in major city areas (44.1%) or inner regional areas (38.5%) and were triaged as semiurgent (60.5%) and urgent (25.6%). They were primarily self-referred (93.4%), with a small portion referred by a general practitioner (2.8%). 8.0% were re-presentations to the ED within 5 days, and 18.4% were admitted to the hospital at the end of their ED visit (Table 2). Table 2 shows ED and related admission data by facility remoteness classification. 20, 21 We did not include remote or very remote classifications as there are no facilities in the health district that are classified as this. Across the study period, there were 12 399 (46.8%) ED presentations in a major city area, 9824 (37.1%) in an inner regional area and 4286 (16.2%) in an outer regional area. Table 3 shows ED and related admission data by NSW Health delineation level. 28

| DISCUSSION
Our study highlights key differences in ED service indicators with respect to presentations for LBP across remoteness areas and staffing portfolios in a large NSW Local Health District. The time patients spent in ED was shorter in outer regional EDs and EDs staffed by GPVMOs. Re-presentation rates were higher in outer regional EDs and EDs staffed by GPVMOs than in major city EDs. We observed notable differences between delineation levels. For example, patients spent longer in EDs at Level 5 and 6 facilities than lower delineation levels (230 min versus 165 min, respectively), and representation rates were lower. Hospital admission rates were highest in the Level 6 facility and lowest in Level 1 facilities. A strength of our study is we examined a large number of LBP presentations to EDs across a large number of facilities with varying remoteness areas, delineation levels and staffing portfolios. We have high confidence that we sampled all LBP presentations over the 5-year study period as we used a number of data extraction mechanisms to filter these. The data quality was high, being based on clinical notes taken by ED staff at the point of care delivery, and then reviewed and entered by clinical data coders after patient discharge. We also extracted and report real costs for LBP ED presentations and admissions. A limitation is we may have missed some presentations where LBP was present but not the principal or secondary diagnosis. Our study also does not provide information about EDs in remote or very remote areas or any capital cities of Australia, as there are none within the sampled health district.
Data from our study align with previous ED presentation rates for LBP in Australia. We found a presentation rate within our sample of 1.3%, which is similar to the previously described rates, ranging from 1.9% to 3.4%. 8,9,31 Another study of LBP presentations across EDs in NSW reported the highest presentation rates in rural areas based on age-standardised rates within local government areas. 14 However, our study used remoteness area and delineation levels based on the Australian Statistical Geography Standard Remoteness Structure 20,21 and found that LBP represented a smaller proportion of all ED presentations in regional and rural areas than in major city areas. The cost of an ED visit for LBP within our sample (median AUD$444 and mean AUD$547) was comparable to that found by Coombs, Machado 32 for non-ambulance presentations (mean AUD$584). However, the cost of an admission for LBP from ED was significantly lower in our sample (median AUD$3214 and mean AUD$5265) than in the previous study sample (mean AUD$13 137). 32 This difference may be due to slightly different inclusion criteria between the studies; Coombs, Machado 33 included non-specific low back pain and lumbosacral radicular syndromes, while we only included non-specific low back pain. No other studies have compared data on time in ED, re-presentation rate, admissions, or cost across remoteness areas, staffing portfolio or delineation level. The variation in ED presentations and performance measures observed in our study across ED types highlights the importance of considering local context when developing service improvement strategies. A performance measure for NSW EDs is achieving the emergency treatment performance target of 81% of all patients discharged from ED, admitted to hospital or transferred to another hospital within 4 h. 16 Other indicators include, but are not limited to, a timely triage (<5 min) process and limited rates of re-presentation. 16 All EDs (irrelevant of remoteness location, delineation level or staffing portfolio) are currently measured against the same performance indicators. These facilities are, however, different in the way they function and the patients they see. Our study shows that hospitals with different staff and across different locations do not consistently meet the above targets. This means the general targets across sites might not be a good indicator of service performance, and any solution to improve such targets needs to be tailored for the service challenges.
When considering the health care of regional and rural areas, it is critical to consider Aboriginal and Torres Strait Islander communities. 34 Higher ED attendance rates for LBP by Aboriginal and Torres Strait Islander patients in more regional/rural EDs may reflect higher populations in those areas. 35 However, it may also be due to the fact that Aboriginal and Torres Strait Islanders are 1.2 times more likely to report having a back problem than other Australians. 36 It may also be a combination of these two factors. While all EDs should ensure they are culturally safe, the numbers of presentations highlight the need to ensure that services in regional and rural areas have embedded culturally safe practices and environments to enable equitable health care for these communities.
Hospitals, including the ED, are complex systems made up of multiple interconnected parts (e.g. staffing levels, bed availability, presentation numbers). 37 There are many reasons, internal and external to the ED, which may affect these different parts. While we have only described service indicators by three selected ED characteristics (remoteness level, delineation level and staffing portfolio), there are others that may need to be considered. These may include, but are not limited to, the support available within the community (e.g. GP appointment availability), the size of the hospital or ED, the presentation referral source (e.g. self versus GP), the mode of arrival into ED (e.g. ambulance) or the clinical presentation of the patient (e.g. duration of LBP, co-morbidities). Although the complexity of hospital systems is well described, previous studies have only concentrated on describing LBP presentations to EDs within metropolitan areas 8,9 or only considered the geographical location of the ED. 14 Due to the complexity of the ED environment, it is vital to consider other factors that may impact LBP presentations and the local context for each ED.
Currently, there is little research or health care policy focus on the prevalence and burden of LBP and other musculoskeletal conditions in regional and rural areas, even though our study shows that the service-level indicators are clearly different across facilities and areas. Therefore, context-specific research in these areas is required. Further research should describe LBP presentations across other rural and remote facilities and include information regarding treatments delivered, and patient outcome data, in EDs across locations. This information about local services will enable relevant and targeted strategies to be implemented to improve patient care. Service indicators and patient journey factors (e.g. time spent in ED) can directly impact patients and the quality of care received. 38 The differences in health care provision in metropolitan, regional and rural areas are well documented; however, little is known about the impact of staffing portfolios on service-level indicators. Our study identified differences in 'time in ED', 're-presentation in 5 days' and 'cost per admission' between staffing portfolios, but we do not know why these differences exist. The reasons for these differences across locations and staffing portfolios should be further investigated. This could involve qualitative exploration of clinician and patient experiences to better understand and manage these differences.

| CONCLUSIONS
This study describes LBP presentations across 37 EDs and highlights the potential burden these presentations may place on hospitals. A higher proportion of ED patients in regional areas re-present for LBP, and these patients also have shorter ED stays. In contrast, hospital admission rates appear variable across localities. LBP presentations appear to follow different pathways depending on the ED remoteness area, delineation level and staffing portfolio. Given the differences, any implementation or support mechanisms should consider local context.