Referral Patterns for People With Musculoskeletal Complaints in General Practice: An Analysis From an Australian Primary Care Database

Our objective was to examine referral patterns for people with musculoskeletal complaints presenting to Australian general practitioners (GPs).


INTRODUCTION
Musculoskeletal complaints pose a major burden to individuals, health systems, and society.In Australia, in 2019 to 2020, musculoskeletal complaints were the most expensive disease group, comprising 10% of the total health expenditure. 1One of every five general practitioner (GP) consultations are reported to be for a musculoskeletal complaint, most commonly low back, neck, shoulder, and knee complaints. 2 These are also the most common reason for seeking first-line care from an allied health practitioner such as a physiotherapist. 3ost regional musculoskeletal complaints are self-limiting, and recommendations across high-quality primary care clinical practice guidelines consistently indicate that most patients can be safely and effectively managed in general practice. 4Early referral to allied health practitioners is appropriate for those at high Supported by Arthritis Queensland, Arthritis South Australia, and the Allan and Beryl Stephens Grant from Arthritis Australia.Supported by the Australian National Health and Medical Research Council (NHMRC) Program (grant APP-1113532).Dr Buchbinder's work was supported by an NHMRC Investigator fellowship.
risk of persistent symptoms. 5GPs may also refer to allied health practitioners for provision of education and advice about physical activity and/or exercise or to facilitate behavioral change.Specialist advice may be sought for people with severe or progressive symptoms that fail to improve with time or when there is clinical suspicion of serious pathology such as malignancy, infection, or inflammatory arthritis or for consideration of specialist treatments such as knee joint replacements. 57][8] None have examined variation in referrals or their timing, factors associated with different referral types, or trends over time.The aim of this study was to investigate GP referral patterns for people with low back, neck, shoulder, and knee complaints using prospectively collected longitudinal data from GP practices.We determined the number of patients with referrals and number of GPs who made referrals, and we examined their timing, associations, and trends over time.

METHODS
Study design.This is a retrospective longitudinal analysis of Australian general practices participating in the Population Level Analysis and Reporting (POLAR) database.POLAR extracts deidentified patient-related data from all electronic medical records of consenting general practices within the three Primary Health Networks (PHNs) of Eastern Melbourne, South-Eastern Melbourne, and Gippsland within Victoria, Australia.Approximately 30% of general practices within these PHNs participate in POLAR, and they provide services to more than half of Victoria's population (3.46 million people). 9The study protocol outlining the database, setting, eligibility criteria, and data handling processes has been published previously. 10udy cohort.Our cohort included all patients with at least one GP face-to-face consultation and an eligible low back (≥18 years old) and/or neck, shoulder, or knee complaint (≥45 years old) between January 1, 2014, and December 31, 2018.These criteria were chosen because the prevalence of most musculoskeletal conditions increases markedly after the age of 45 years, except for low back pain, which increases after 18 years.We excluded traumatic diagnoses and inflammatory and autoimmune rheumatic diseases.The code lists of all included complaints are available at https://clinicalcodes.rss.mhs.man.ac.uk/medcodes/article/174/.
Variables.Variables extracted from the POLAR database were patient demographics (birth year, gender, residence postcode), practice PHN, dates and diagnoses of eligible musculoskeletal complaints, and referrals to other health care practitioners likely to be for a musculoskeletal complaint.We used the first recorded diagnosis of an eligible musculoskeletal complaint as a proxy for the time when GP care was first sought for the complaint.We included referrals to allied health practitioners (physiotherapists, psychologists, osteopaths, chiropractors, exercise physiologists, or massage therapists), medical specialists (rheumatologists, neurologists, pain specialists, sports medicine doctors, or rehabilitation physicians), and orthopedic surgeons or neurosurgeons.We also included referrals to psychologists because they may be involved in care for these complaints 4 but elected to exclude referrals to dieticians and occupational therapists because we could not be certain about the reason for the referral from the POLAR database.
Data analysis.Data were extracted from the POLAR database and exported into Stata version 15.1 (STATACorp).The primary analysis included referrals during the first year of followup for each patient, and a sensitivity analysis included referrals during the entire follow-up period (until December 31, 2018).The same age restrictions applied for both analyses.
We determined the number and category of referrals (allied health, medical and surgical specialist), and number of GPs who made referrals and presented results by anatomic site or multisite.We included the first referral to each type of health practitioner.The median (interquartile range [IQR]) time (in days) from the first eligible musculoskeletal complaint until the first referral within each category according to site was also calculated.
We used multivariable logistic regression to examine the association between referral category (only allied health, medical specialist, or surgical specialist and multiple referral categories) and patient-and practice-related characteristics including patient gender, socioeconomic status (lowest quintile or other), residential remoteness (metropolitan or other), practice PHN (Eastern Melbourne, South-Eastern Melbourne, or Gippsland), site (low back, neck, shoulder, or knee), and single versus multiple sites affected.We reported odds ratios (ORs) with a 95% confidence

SIGNIFICANCE & INNOVATIONS
• This study used routinely collected data to investigate general practitioner (GP) referrals to a wide range of health care practitioners over time.• One-third of patients in our cohort received at least one referral for a regional musculoskeletal complaint, and these were made by two-thirds of GPs in our cohort.• Surgical referrals were most common for people with knee complaints, whereas allied health referrals were most common for people with low back, neck, and shoulder complaints.• Understanding the reasons underpinning GP referrals to other health care practitioners may identify factors that explain variations in practice.
interval (CI) adjusted for age and time since diagnosis.P < 0.01 was interpreted as statistically significant to account for multiplicity, and a change in OR of ≥10% was considered clinically relevant.Gippsland was chosen as the reference PHN because it includes predominantly regional and remote areas compared to the other predominantly metropolitan PHNs.Patients with low back complaints were chosen as the reference because this was the only site that included the 18-to 45-years age group.
Temporal trend analysis was used to examine annual changes in that proportion of patients with allied health, medical specialist, and surgical referrals and changes in referral types as a proportion of the total number of referrals.Both trend analyses were conducted by site.P < 0.05 was interpreted as statistically significant, and a change in either direction of ≥1% per year was considered clinically relevant.
This study was approved by the Cabrini Health and Monash University Human Research Ethics Committees (reference numbers 02-21-01-19 and 16975, respectively).It was conducted in accordance with the Declaration of Helsinki.We did not obtain participant consent because all data were anonymized.
Association between referral category and patientand practice-related characteristics.Site was the strongest predictor of referral category (Table 2).Compared to patients with a low back complaint, those with knee complaints had an OR of 4.13 (95% CI 3.87-4.39)for receiving a surgical referral but lower odds for medical specialist or allied health practitioner referral (OR 0.37, 95% CI 0.33-0.41and 0.38 95% CI 0.35-0.40,respectively).Compared with women, men had an OR of 1.55 (95% CI 1.48-1.63)for receiving a surgical referral but ORs of 0.90 (95% CI 0.83-0.97)and 0.79 (95% CI 0.76-0.82)for a medical specialist or allied health referral, respectively.Compared to those living in an area of socioeconomic advantage, patients living in an area of socioeconomic disadvantage had an OR of 1.26 (95% CI 1.17-1.35)for an allied health referral but an OR of 0.81 (95% CI 0.75-0.89)and 0.78 (95% CI 0.69-0.89)for a surgical and medical specialist referral, respectively.
Trends over time.There was no appreciable change in the proportion of patients who received referrals over time (Supplementary Figure 2).For example, approximately 9% of patients with low back complaints received a surgical specialist referral during each study year.The only exception was a 1.2% (95% CI 0.6-1.9)annual increase in the proportion of people with knee complaints who received allied health referrals (13% in 2014 to 17% in 2017).
The relative proportion of allied health and surgical referrals varied over time, with a 2.2% (95% CI 1.9%-2.4%)annual increase in the proportion of allied health referrals and a concomitant 1.9% (95% CI 1.6%-2.1%)annual decrease in proportion of surgical referrals (Figure 1).These changes were consistent across all four sites.For example, in people with knee complaints, referrals to surgical specialists decreased by 2.3% per year, from 62% of all referrals in 2014 to 52% in 2018.There was a concomitant increase in referrals to allied health practitioners from 30% of all referrals in 2014 to 40% in 2018.The proportion of referrals to medical specialists was constant over time.Sensitivity analyses including all eligible referrals made during the study period (n = 121,087 referrals, 42% patients) did not appreciably alter the results (Supplementary Table 1).

DISCUSSION
Our findings, based on real-time general practice data, confirm that one-third of people with low back, neck, shoulder, and knee complaints receive referrals to other health care practitioners and that, in general, these occur early in their clinical course.One-third of GPs in our cohort made no referrals over the five-year period.Patients with low back, neck, and shoulder complaints most commonly received early referrals to allied health practitioners, whereas one-quarter of patients with knee complaints were referred to an orthopedic surgeon at a median of two weeks after diagnosis.We also observed referral disparities based upon both patient gender and socioeconomic status.   1) 111 (1) a Multisite: 13,049 (10%), 1,940 (1%), and 187 (.14%) patients had 2, 3, and 3 body region complaints, respectively.b Each patient may have one or more referrals to one or more health care practitioners.One notable finding was that orthopedic referrals were disproportionally higher for people with knee pain.Although we were unable to ascertain the reasons underpinning these referrals, the lifetime risk of knee replacement in Australia is <20%, suggesting that at least some of these referrals may have been premature. 12his is consistent with an interview survey of 201 older adults with knee pain in England that observed surgical referral was commonly initiated before attempting nonoperative options. 13pproximately one in five patients within our cohort received an allied health referral, most commonly within two to three weeks of the initial diagnosis.Because patients can attend physiotherapists without a GP referral in Australia, this is likely an underestimate of the proportion of patients who seek physiotherapy for regional musculoskeletal conditions.For GPs who made referrals, this may reflect a preference for early involvement of physiotherapists in managing regional musculoskeletal complaints.Reasons could include consultation time constraints, a lack of relevant formal training or confidence in managing these conditions, a perception that exercise prescription is outside their scope of practice, and beliefs that physiotherapists may be able to offer a more thorough assessment, management, and review than they are able to provide. 14elatively few patients with musculoskeletal complaints were referred to rheumatologists, although their scope of practice includes specialist physician assessment and management of people with all forms of arthritis and musculoskeletal disease.These findings are in keeping with previous studies examining rheumatology referrals for musculoskeletal conditions. 6,7Accessibility issues including long wait times may be one factor accounting for this low number.Other reasons could include GP misconceptions that rheumatologists manage only inflammatory arthritis and autoimmune conditions 15 and patient preferences for surgical consultation. 16ur findings of gender disparities in surgical referrals is consistent with known gender and socioeconomic disparities in care identified globally. 17Women may be more averse to surgical risk, more concerned about burdening their family, and less likely to discuss surgery with a GP than men. 18However, these reasons do not appear to account for the significantly fewer referrals to allied health and nonsurgical medical specialists among men.Although our finding of significantly fewer surgical and medical specialty referrals among people in socioeconomically disadvantaged areas is also consistent with known socioeconomic disparities in care, our observation that they had significantly more allied health referrals was surprising.
A unique finding of our study was the relatively high proportion of referrals to psychologists-ranging from 8% of all referrals for those with knee complaints to 14% for low back complaints.Although this may be for appropriate assessment and management of psychosocial factors, 4 these findings may also be explained by the relatively high prevalence of comorbid anxiety and/or depression (25%) among our cohort. 10he strengths of our study are its use of routinely collected longitudinal data, its large sample size, and the investigation of referrals to a wide range of health care practitioners.Examination of the timing of referrals and trends over time is also unique to this study.Limitations include that we may have misclassified potentially eligible patients, resulting in an undercount of patients with regional musculoskeletal complaints.It is also possible we underestimated referrals given some patients may have sought GP care for the same complaint before and after our study period.We were also unable to capture reasons for referral or clinical information outside of general practice.Although our study cohort is broadly representative of the wider Australian population, 11 people living in remote and regional areas and those living in low socioeconomic areas are not well captured.It is therefore possible that referral patterns may differ in other areas with different accessibility and funding arrangements.Furthermore, referral patterns may have changed since 2018, especially in view of the COVID-19 pandemic, and our estimates of referral timing may not be entirely accurate because GPs may record a diagnosis for a complaint at the first presentation or at a later time when the diagnosis is "confirmed."Finally, we were not able to study the influence of GP characteristics such as years since graduation, gender, group or solo practice, and so forth, on referrals because these data are not captured within POLAR.
In conclusion, one-third of patients receive, and two-thirds of GPs make, referrals for musculoskeletal complaints.Understanding the reasons for referral and differences between GPs who refer more and less frequently may identify factors that explain variations in practice.

Figure 1 .
Figure 1.Proportion (%) of the total number of all referrals by site and referral category over time.

Table 1 .
Number (%) of patients who received a referral to allied health, surgical, and/or medical specialist overall and by anatomic site within one year of diagnosis and total number (%) of referrals overall and by type and anatomic site

Table 2 .
Regression analyses presenting association between referral categories and patient-and practice-related variables* * All models are adjusted for age and time since index diagnosis.Bold values represent statistical significance at P < 0.01.CI, confidence interval; PHN, Primary Health Network.