Advanced practice physiotherapy surgical triage and management of adults with spinal disorders referred to specialized spine medical care: a retrospective observational study

ABSTRACT Introduction In this novel advanced practice physiotherapy (APP) model of care, advanced practice physiotherapists (APPTs) assess, triage, and manage adults with spinal disorders to alleviate the growing demands in specialized spine medical care. Objectives To describe this APP model of care, to assess change in disability 3 months after rehabilitation care and to assess surgical triage and diagnostic concordance between APPTs and spine surgeons. Methods In this retrospective observational study, consecutive patients who completed the 3-month follow-up data were analyzed. Sociodemographic, clinical characteristics, and self-reported disabilities including the Oswestry Disability Index (ODI) and Neck Disability Index (NDI) at baseline and 3 months were extracted. Paired t-tests were used to assess changes in disability. Surgical triage and diagnostic concordance between APPTs and surgeons were measured with raw agreement, Cohen’s Kappa, and PABAK. Results In this model, trained APPTs triaged surgical candidates and provided rehabilitation care including education and exercises to patients with spinal disorders. The APPTs referred only 18/46 participants to spine surgeons. Surgical triage and diagnostic concordance were high with raw agreement of 94% and 89%. At the 3-month follow-up, significant improvements in disability were observed among nonsurgical candidates with back (mean difference (MD): −13.0/100 [95%CI: −19.8 to −6.3], n = 23) or neck disorders (MD: −16.0/100 [95%CI: −29.6 to −2.4], n = 5), but not among surgical candidates referred by APPTs to spine surgeons. Conclusion In this limited sample, adults with spinal disorders that were initially referred to a spine surgeon by family physicians were effectively assessed, triaged, and managed by an APPT.


Introduction
Neuromusculoskeletal disorders, including spinal disorders, are the most common cause of long-term pain and physical disability and affect hundreds of millions of people around the world (Cieza et al., 2020;Woolf, Erwin, and March, 2012).Although most spinal pain patients can be managed in primary care, several patients with complex presentations responding to first-line medical care will be referred to specialists such as a spine surgeon.Access to specialized secondary care has been especially problematic in the last few decades as there has been a consistent increase in the number of referrals to orthopedic surgeons and neurosurgeons in Canada (Moir and Barua, 2021).With the aging of the population, the number of referrals is expected to increase, leading to even longer wait times to access appropriate care (Barua, Palacios, and Emes, 2017;Cieza et al., 2020;Dall et al., 2013;Denton and Spencer, 2010).In addition, the COVID-19 pandemic has forced cancellation or postponement of consultations and surgeries including in orthopedic or neurosurgery settings (Moir and Barua, 2021).For spine surgery, evidence shows that longer wait times lead to poorer outcomes in terms of pain and disability (Braybrooke et al., 2007).Moreover, a large proportion of patients with spinal disorders referred to spine surgery are ultimately not treated surgically and, instead, are offered rehabilitation care and often referred to physiotherapy (Booth, 2019;Lafrance et al., 2021;Robarts et al., 2017).New models of care to improve access, the clinical pathway and, ultimately, outcomes for adults with spinal pain are therefore needed.
The use of advanced practice physiotherapy (APP) models of care where physiotherapists assess, triage, and manage patients with neuromusculoskeletal disorders has been proposed as a potential solution to alleviate the growing demand for specialized medical care in spine surgery (Chang et al., 2018;Desmeules et al., 2012Desmeules et al., , 2013;;Harding, Prescott, Sayer, and Pearce, 2015;Lafrance et al., 2021;Saxon, Gray, and Oprescu, 2014;Stanhope et al., 2012).Although APP models of care were initially developed in the 1970s in the US Army (Greathouse, Schreck, and Benson, 1994;James and Stuart, 1975) they are still absent or only emerging in several countries (Booth, 2019).APP models of care for patients with spinal pain have been found to significantly reduce wait times for an initial consultation and to result in high satisfaction for patients and stakeholders (Lafrance et al., 2021).APPTs and surgeon diagnostic concordance for adults with shoulder, hip, and knee MSKDs are reported as being good (Aiken and McColl, 2008;Ashmore, Smart, O'Toole, and Doody, 2014;Décary et al., 2017;Desmeules et al., 2013;Jovic, Mulford, Ogden, and Zalucki, 2019;Lowry et al., 2020;MacKay, Davis, Mahomed, and Badley, 2009;Madsen et al., 2021;Marks et al., 2016;Razmjou et al., 2013) but as yet to be evaluated for spinal pain.Regarding surgical triage, previous studies reported moderate (Lowry et al., 2020) to good (Aiken and McColl, 2008;Décary et al., 2017;Desmeules et al., 2013;Jovic, Mulford, Ogden, and Zalucki, 2019;MacKay, Davis, Mahomed, and Badley, 2009;Napier, McCormack, Hunt, and Brooks-Hill, 2013;Razmjou et al., 2013;Yin et al., 2019) concordance between APPTs and surgeons for adults with shoulder, hip, and knee MSKDs, while one study reported good concordance between APPTs and surgeon for adults with spinal pain (Robarts et al., 2008).Based on low certainty evidence, the authors report that APP models of care are as effective as usual medical care to reduce pain and disability in patients with various MSKD (Lafrance et al., 2021).
The CareAxis network, a nonprofit organization involving spine surgeons at the McGill University Health Center (Montreal, Quebec, Canada) has recently developed and implemented an APP model of care for patients referred to spine surgery.The aim of this model is to use trained advanced practice physiotherapists (APPTs) to reduce wait times to consult a spine surgeon, to triage potential surgical candidates more efficiently (i.e. increase surgical conversion rate for surgeons) and also to improve nonsurgical management for patients with spinal disorders by offering basic rehabilitation care including education and exercise.In the province of Quebec, access to outpatient rehabilitation for spine conditions is available mostly in private clinics, which is financially not accessible to all patients.In this model, patients are seen by a CareAxis affiliated APPT working in the private sector, but the assessments were free of charges for patients.This is the first APP model of care for adults referred to spine surgery in the province of Quebec, Canada and one of the few models at the international level that have been presented and appraised.The aims of this study observational study are: 1) to describe patients' characteristics and care offered through the CareAxis program in this retrospective cohort; 2) to evaluate change in health-related outcomes including change in disability assessed at 3 months after the initial evaluation by APPTs in surgical and nonsurgical patients; and 3) to assess surgical triage and diagnostic concordance between APPTs and spine surgeons.

Study Design
This retrospective observational longitudinal study of the CareAxis APP model of care was approved by the Maisonneuve-Rosemont Hospital Research Center and McGill University Health Center research ethic boards (#FWA00001935 and IRB00002087).

Setting and Model of Care
In this model, patients are referred to CareAxis either directly through family physicians in the community, by another medical specialist or from the Center de Répartition des Demandes de Service de Montréal, the centralized government regional dispatch medical request service.
The APPTs (n = 10) in this model of care had at least 10 years of experience as a physiotherapist and received a 4-day training with spine surgeons on diagnostic and surgical triage of various spinal disorders.The objective was to train APPTs to diagnose and triage potential surgical candidates as a spine surgeon would do, using criteria from the history, subjective, and objective patient's assessment.
In this model, the APPTs assess and triage surgical candidates, provide education and prescribe a self-management exercise program.During the assessment, the APPTs filled a standardized from to guide the diagnosis and triage decision.The forms included questions related to: symptoms (i.e.back/neck pain history, numbness, and pain location/level); red flags screening; neurological exam (i.e.reflexes, upper motor neuron tests, myotomes, and dermatomes); spinal movement; and functional exams (i.e.gait assessment, special physical tests, and palpation).These services are offered to all patients in a one-hour session.Only patients deemed to be surgical candidates by the APPTs are then referred to a spine surgeon (orthopedic surgeon or neurosurgeon, n = 3) at the McGill University Health Center in Montreal, Quebec, Canada.Since the main goal of the CareAxis program is to triage potential surgical candidates to decrease wait time for patients as well as the number of nonsurgical candidates referred to the spine surgeons, no follow-up treatment sessions with the APPTs are offered within this model of care.However, patients may be recommended to seek further physiotherapy care, which are then offered outside of this APP model of care.

Participants
Patients with spinal disorders referred to CareAxis between April and November 2019, who responded to the 3-month follow-up questionnaires and who consented to share their data were included in the current study.

Database Data
CareAxis systematically collects sociodemographic and clinical data on all participants using their web data collection portal.Clinical data include the APPT's diagnosis and management plan, the validated, and reliable STart Back Screening Tool (Beneciuk et al., 2013;Fritz, Beneciuk, and George, 2011) or a modified version for neck disorders and the Oswestry Disability Index (ODI) for back disorders or the Neck Disability Index (NDI) for neck disorders (scores 0-100; 0 = best).Both the ODI and NDI are considered valid, reliable, and responsive to change with spinal disorders populations (Chiarotto et al., 2016;Fairbank and Pynsent, 2000;MacDermid et al., 2009).APPTs diagnoses were extracted from the patient's file and associated to a diagnosis from a predefined list by a pair of evaluators (SL and MC).The list of predefined diagnosis was adapted from diagnostic classification from clinical practice guidelines (Blanpied et al., 2017;Delitto et al., 2012;Geroge et al., 2021).The diagnostic list includes: 1) neck pain without radiating pain; 2) neck pain with radiating pain (radicular); 3) neck pain with headaches; 4) thoracic pain; 5) low back pain without radiating pain; 6) low back pain with radiating pain (sciatica); 7) fibromyalgia; 8) fracture; 9) vascular pain; 10) tumor; 11) inflammatory arthritis; 12) neurological disorder; 13) autoimmune disorder; 14) peripheral neuropathy; 15): other bone pathology; and 16) other diagnoses.

Medical Chart Review of Surgical Candidates
For patients referred by APPTs to spine surgeons, the date of the consultation, spine surgeon's diagnosis and recommended or performed surgical intervention were collected through medical charts review at the McGill University Health Center.Spine surgeons' diagnoses were extracted from the patient's file and associated to a diagnosis from the predefined list by a pair of evaluators (LE and CS) without knowledge of the APPT's diagnosis.

Data Analysis
Descriptive statistics were used to present the participants' characteristics and care provided.For healthrelated outcomes, the Kolmogorov-Smirnov and Shapiro-Wilk tests were used to confirm the normality of data.Paired t-tests were used to determine differences in health-related outcomes between baseline and 3month data.For normally distributed sample, 95% confidence intervals [95%CI] were reported.Ranges were reported for very small sample size (n < 3).The proportion of participants with a clinically important improvement or worsening of their disability was also calculated.Minimal clinically important differences of 10 points were used for the ODI and NDI as previously reported in the literature (Carreon, Glassman, Campbell, and Anderson, 2010;Schwind et al., 2013;Young et al., 2009).To compare nonsurgical and surgical candidates referred to surgeons unpaired sample t-tests were used to determine the difference in the ODI and NDI scores and a Mann-Whitney test was used to determine the difference in the proportion of participants with a clinically important change.Surgical candidates who did undergo surgery within three months after initial APPT assessment were excluded from these analyses, as these patients had their surgical interventions before the 3-month follow-up, making the change from baseline analysis not valid for the purpose of comparison with the other group.Homogeneity of the variances was tested with Levene's test for t-tests.Alpha level was set at 0.05.
For participants referred to a spine surgeon, diagnostic concordance between the APPT and the surgeon was calculated with raw agreement, Cohen's Kappa, and PABAK.We interpreted the strength of agreement for Cohen's Kappa and PABAK as follows: 0.00-0.20 = weak; 0.21-0.40= slight; 0.41-0.60= moderate; 0.61-0.80= good; 0.81-0.90= very good; and 0.91-1.00= excellent (Décary et al., 2017;Lowry et al., 2020;Razmjou et al., 2013).APPT and surgeon diagnosis concordance were assessed by a pair of raters (SL and MC or FD) as described in Table S3.Surgical triage agreement was calculated based on the proportion of patients for which a surgical intervention was recommended by the spine surgeon over the total of patients referred by APPTs to surgeons.As such, Cohen's Kappa and PABAK could not be calculated for surgical triage concordance as a 2 × 2 table could not be generated.Data analyses were performed with the Kappaetc module on StataCorp (Stata Statistical Software, College Station, TX, StataCorp LLC).

Participants
Among the 105 participants who responded to the 3month follow-up questionnaires, 66 accepted to be contacted by the research team while 46 signed and returned the information and consent form.consulted for back (n = 39) or neck (n = 7) disorders.The study flow chart is presented in Figure 1 and the summary of the results is presented in Figure 2. The mean age was 66.9 ± 12.5 years old and 50% of patients were female.Most of the included participants had symptoms for at least 3 months (n = 39, 89%).Based on the Start Back Screening Tool most of the included patients were considered at moderate (n = 12, 31%) or high (n = 22, 56%) risk of poor outcomes.Participant characteristics are presented in Table 1, while results of the physical examination and APPTs' diagnoses are presented in Table S1 in supplementary material.
The 18 patients referred by APPTs were seen by spine surgeons at the McGill University Health Center.Surgery was recommended by the surgeon for 16 of these patients.Fourteen patients ultimately underwent surgery, as two patients declined surgical intervention.Surgery was performed prior to the 3-month follow-up for five patients.Surgical interventions performed included: lumbar spine short fusion (n = 6), lumbar spine laminectomy (n = 6) or lumbar spine discectomy (n = 2).No cervical surgery was performed.

Wait Times in the Care Axis Model of Care
Median wait time between APPT's referral to the surgeon and the initial consultation with the spine surgeon was 37.5 days (interquartile range: 29.5-62.75).Among the 14 patients who underwent surgery, the median wait time between the APPT's referral to the surgeon and the surgery was 117.5 days (interquartile range: 85.25-195.5).

Health-Related and Disability Outcomes for Patient in the Care Axis Model of Care
For patients with back disorders, a statistically significant improvement on the mean ODI score from baseline to the 3-month follow-up was observed among nonsurgical patients (MD: −13.0 out of 100 points [95% confidence intervals (CI): −19.8 to −6.3], n = 23), but not for potential surgical candidates (mean difference (MD): 2.5 out of 100 points [95%CI: −7.8 to 12.9], n = 11).Clinically important reductions in disability were observed for 13 (56.5%) of the nonsurgical patients and 3 (27.3%) of the potential surgical candidates (Table 2).
For patients with neck disorders, a statistically significant improvement from baseline to the 3-month followup on the mean NDI was observed among nonsurgical patients (MD: −16.0 out of 100 points, [95%CI: −29.6 to −2.4], n = 5), but not for potential surgical candidates (MD: 5.0 out of 100 points, range: 2.0 to 8.0, n = 2).Clinically important reductions in disability were observed for 3 (60%) of the nonsurgical patients, but in none of the potential surgical candidates (Table 2).
Nonsurgical candidates had significantly better scores on the STarT Back at 3-month when compared to baseline (p = .002),while no significant change was observed among surgical candidates (p = .688).Details and results for the modified STarT back for neck disorders are presented in Table S2 in supplementary materials.

Concordance between Advanced Practice Physiotherapists and Spine Surgeons
Regarding APPT-surgeon diagnostic concordance, a raw agreement of 89% was observed with a Cohen Kappa of 0.63 [95%CI: 0.13-1.0]and a PABAK of 0.85 [95%CI: 0.64-1.0],as presented in Table 3. Regarding surgical indication, a raw agreement of 94% was observed (Table 3).Individual diagnoses are presented in Table S3 in supplementary material.

Main Results
Adults with spinal disorders referred by a family physician to a spine surgeon were effectively assessed, triaged, and managed by APPTs in this novel model of care implemented in one health-care center in the province of Quebec, Canada.This is one of the first studies on APP model of care in spine surgery and the first one to report APPT and spine surgeon diagnostic concordance.

Interpretation of the Results, Comparison with the Literature and Future of Advanced Practice Physiotherapy
APPT and spine surgeon's triage and diagnostic concordance was good to excellent based on the raw agreement  and PABAK, although the Kappa confidence interval was wide.Other studies have reported good to very good diagnostic concordance between APPTs and orthopedic surgeons for shoulder, hip, and knee disorders (Aiken and McColl, 2008;Ashmore, Smart, O'Toole, and Doody, 2014;Décary et al., 2017;Desmeules et al., 2013;Lowry et al., 2020;Razmjou et al., 2013).For triage concordance, previous studies have also reported good triage concordance between APPTs and surgeons for spinal disorders (Robarts et al., 2017) as well as for shoulder, hip, and knee disorders (Desmeules et al., 2013;MacKay, Davis, Mahomed, and Badley, 2009) while some studies reported a slight to moderate concordance (Lowry et al., 2020;Oldmeadow et al., 2007).These results suggest that APPTs are adequately trained to diagnose and triage adults with neuromusculoskeletal disorders including spinal pain that are referred to specialized medical care.
In this APP model of care, median wait time to see a spine surgeon was 37.5 days, which is largely below previously reported median wait times to see an orthopedic surgeon (77 days) or a neurosurgeon (84 days) in the province of Quebec (Moir and Barua, 2021).Similarly, median wait time from referral to surgery was lower in this APP model of care (117.5 days) compared to provincial median wait time (233.1-346.5 days) (Moir and Barua, 2021).Mean wait time from referral to initial APPT consultation was estimated by CareAxis at 5.8 business days although it was not systematically collected for all patients.These numbers suggest that larger implementation of this model may increase access to specialized spine medical care, as previously reported in the literature (Lafrance et al., 2021).
Most of the included patients were older adults with chronic spinal pain with a moderate to high risk of poor outcomes and who were referred by a primary care physician for specialized care.Somewhat surprisingly, the group of patients triaged by an APPT and deemed as nonsurgical candidates significantly improved at the 3month follow-up with most patients showing clinically important improvements.These improvements were not observed among potential surgical candidates referred to a spine surgeon.Therefore, even among a population of chronic patients with a poor prognostic and in which a surgical intervention was considered by the referring physician, we can see improvement in their conditions, that is likely due to education including reassurance that no surgical intervention was needed, the prescription by APPTs of an exercise program, and the addition of other physical modalities for some patients.The education and reassurance that their conditions do not require a surgical intervention could have been an important factor to decrease pain and disability by decreasing catastrophization and/or kinesiophobia.Although this needs to be confirmed, physical modality treatments may also have been a factor as the APPTs recommended passive modalities for a small proportion of patients that was low and nearly equivalent among the two groups (i.e.15% for nonsurgical and 20% for surgical candidates).Although most nonsurgical candidates reported a clinically important improvement, there is still 44% of nonsurgical candidates that did not report such an improvement.It is unclear if follow-up sessions with the APPTs and/or other treatments could have been beneficial to these patients; this should also be evaluated in future studies.These results are in contrast with previous studies, which reported comparable disability reduction in patients cared nonsurgically by APPTs or medical specialists such as surgeons (Daker-White et al., 1999;Lafrance et al., 2021;Samsson and Larsson, 2015).
Similarly, significant reductions in the STarT Back scores were also observed, suggesting that nonsurgical candidates cared by an APPT had a reduction in their risk of poor outcomes at 3-months compared to their initial score.This should, however, be interpreted with caution as the STarT Back Screening Tool was not developed as an outcome measure but as a tool to assess the initial risk of poor outcomes (Fritz, Beneciuk, and George, 2011).

Strength and Limitations
This study provides a description of a new APP model of care in spine surgery, outcomes of patients cared for in this model and it is the first study to report data on APPTs and spine surgeons' concordance and patient reported outcomes for this population.However, some limitations need to be highlighted starting with the study design.Because data collection was retrospective, some variables such as additional private physiotherapy care or other medical co-interventions received outside of this model of care were not available.Therefore, we cannot conclude on the efficacy of a specific intervention, but on the clinical evolution of nonsurgical patients cared in this model.There is also a potential selection and information bias since the database only included data of patients who voluntarily responded to questionnaires.The sample size is also limited.Since the evaluation of models of care remains context dependent, results from our study can only be generalized to similar healthcare systems, such as a universal healthcare system in high-income countries.

Conclusion
This study provides novel evidence describing the use of an APP model of care involving APPTs and spinal surgeons to assess, triage, and manage adults with spinal disorders.APPTs' surgical triage and diagnostic were concordant with the ones of the spine surgeons while waiting time from referral to consultation and to care were largely lower than in what is reported in a usual only medical care pathway.Furthermore, patients deemed as nonsurgical candidates by an APPT reported significant reductions in disability during follow-up and after receiving rehabilitation care including education and exercise as the main program component.Although we cannot conclude on the formal efficacy of intervention given by APPTs in this model of care, evidence presented here suggests that a larger implementation of this APP model of care may improve access to specialized spine medical care.This study is based on a small sample size, especially for diagnostic and triage concordance data.Further research with larger sample sizes and using an experimental design to isolate the exact efficacious component of rehabilitation care are needed to formally conclude on the efficacy of such a model of care.

Figure 2 .
Figure 2. Summary of the results for diagnostic and triage concordance and changes in disability scores.APPT: Advanced practice physiotherapist; CI: Confidence intervals; ODI: Oswestry Disability Index; NDI: Neck Disability Index; PABAK: Prevalence-adjusted and bias-adjusted Kappa.

Clinical presentation: history and subjective exam
Symptom's duration: missing data for 2 participants; ODI: Oswestry Disability Index; NDI: Neck Disability Index; SD: Standard deviation.

Table 2 .
Disability over time and proportions of participants with a clinically important change.

Table 3 .
Advanced practice physiotherapists -spine surgeon diagnostic and surgical indication concordance.Confidence interval; PABAK: Prevalence-adjusted and bias-adjusted Kappa.a. Kappa and PABAK could not be calculated for surgical indication concordance as a 2 × 2 table could not be generated since the spine surgeon only assessed patients referred and considered as surgical candidates by the APPT.b.One patient was excluded from the surgical triage concordance as this patient was an active patient from one of the spine surgeons.