Influence of Initial Health Care Provider on Subsequent Health Care Utilization for Patients With a New Onset of Low Back Pain: A Scoping Review

Abstract

Objective

The aim of this research was to examine the scope of evidence for the influence of a nonmedical initial provider on health care utilization and outcomes in people with low back pain (LBP).

Methods

Using scoping review methodology, we conducted an electronic search of 4 databases from inception to June 2021. Studies investigating the management of patients with a new onset of LBP by a nonmedical initial health care provider were identified. Pairs of reviewers screened titles, abstracts, and eligible full-text studies. We extracted health care utilization and patient outcomes and assessed the methodological quality of the included studies using the Joanna Briggs Institute checklist. Two reviewers descriptively analyzed the data and categorized findings by outcome measure.

Results

A total of 26,462 citations were screened, and 11 studies were eligible. Studies were primarily retrospective cohort designs using claims-based data. Four studies had a low risk of bias. Five health care outcomes were identified: medication, imaging, care seeking, cost of care, and health care procedures. Patient outcomes included patient satisfaction and functional recovery. Compared with patients initiating care with medical providers, those initiating care with a nonmedical provider showed associations with reduced opioid prescribing and imaging ordering rates but increased rates of care seeking. Results for cost of care, health care procedures, and patient outcomes were inconsistent.

Conclusions

Prioritizing nonmedical providers at the first point of care may decrease the use of low-value care, such as opioid prescribing and imaging referral, but may lead to an increased number of health care visits in the care of people with LBP. High-quality randomized controlled trials are needed to confirm our findings.

Impact

This scoping review provides preliminary evidence that nonmedical practitioners, as initial providers, may help reduce opioid prescription and selective imaging in people with LBP. The trend observed in this scoping review has important implications for pathways of care and the role of nonmedical providers, such as physical therapists, within primary health care systems.

Lay Summary

This scoping review provides preliminary evidence that nonmedical practitioners, as initial providers, might help reduce opioid prescription and selective imaging in people with LBP. High-quality randomized controlled trials are needed to confirm these findings.

Introduction

Low back pain (LBP) is the number 1 cause of years lived with disability worldwide.¹ The personal, institutional, and societal-level health impacts of LBP contribute to substantial economic burden globally.² The preponderance of low-value care, defined as care that has minimal or no evidence of benefit when considered against potential harms, costs, or health care utilization,³ is thought to be central to the problem. International guidelines advocate for a demedicalization approach to nonspecific LBP, with a focus on higher-value care, including patient education, multidisciplinary rehabilitation, and exercise prescription.^4–6 However, compliance with clinical guidelines across primary care is highly variable for LBP.⁷

The high prevalence of low-value care for LBP may be attributed to a multitude of factors, including limited access to care, lack of funding,⁸ patient care-seeking behavior,⁹ practitioner beliefs and treatment preferences,¹⁰ and health care systems that support a medicalization model of care.¹¹ Patients’ entry point into the health care system and the type of practitioner patients see at the initial point of care may also contribute to low-value care decisions. As the gatekeepers to the health care system, the initial provider shapes patients’ understanding of LBP, provides management strategies and reassurance, and identifies a course for recovery.¹² The risk of patients receiving low-value care is arguably most critical at this stage because it can alter the patients’ journey through the health care system. For example, patients with LBP without indicators of serious pathology (eg, cancer, cauda equine syndrome) who are referred for early imaging or prescribed opioid medication are more likely to develop prolonged work disability,¹³ incur higher medical costs, and undergo surgery.¹⁴ When low-value care is received (ie, opioid prescription, imaging, and medical subspecialty referral) within 3 weeks of initial visit, patients with acute LBP are more than twice as likely to develop chronic pain compared with those who receive none.¹⁵

Current research indicates that one-quarter of patients with LBP will be referred to imaging and 30% prescribed opioids when initiating care with primary care physicians in general practice.¹⁶ Initiating care with health care practitioners who provide musculoskeletal care (eg, chiropractors and physical therapists) or nonphysician providers who can perform triage roles for musculoskeletal conditions in the public health system (eg, nurses) offers an alternative encounter that may favorably influence health care utilization and patient outcomes. These practitioners have been collectively referred to as “nonmedical” health care providers^17,18 in this scoping review to distinguish them from medical physicians. Recent systematic reviews comparing physician-led care to direct-access physical therapy,¹⁹ or nonmedical models of care²⁰ (triage, self-referral, and direct access), provide evidence that nonmedical initiated care may improve health care utilization and costs but not patient outcomes such as pain. These reviews focused on patients with musculoskeletal pain of any duration. By focusing on a LBP population with a new episode of care, this review seeks to highlight the impact of the initial encounter at a critical point of care.

To the authors’ knowledge, no comprehensive review has been conducted on this topic. The objective of this review is to examine the scope of evidence for the influence of nonmedical initial provider on health care utilization and outcomes in patients with LBP and identify gaps in the literature to guide future research directions.

Methods

This review follows the methodological framework set out by the Joanna Briggs Institute for scoping reviews²¹ and was conducted and reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews.²² The protocol was registered with the Open Science Framework. For deviations from the protocol see Figure 1.

Figure 1.

Deviations from the protocol.

Search Strategy

An academic librarian helped construct the search strategy following identification of relevant articles and key terminology in a preliminary database search. A combination of MeSH and search terms served to capture initial provider interactions. Terms were intentionally broad and covered provider types, health care pathways, and more direct terminology such as “initial provider” that had been highlighted during the preliminary search (see Suppl. Material 1 for full search strategy). We conducted an electronic search in MEDLINE from inception to April 2020 and adapted for CINAHL, Embase, and AMED. Examination of the reference lists for pertinent studies and citation tracking was conducted to detect additional studies. The searches were updated in June 2021 using the same search strategy.

Participants

This review considered adult patients with a new onset or new episode of LBP as defined by the authors of the original studies. Studies on patients with traumatic, postsurgical injuries or chronic LBP as well as more serious pathologies requiring specific treatment (eg, malignancy/tumor, inflammatory arthritis such as ankylosing spondylitis, or cauda equina/spinal cord injuries) were excluded from the review.

Concept

Nonmedical health care providers were defined as any provider who is not a medical specialist (ie, not a primary care physician, general practitioner, orthopedic surgeon, sports physician, or physiatrist). Studies were included if they met the following 3 criteria: (1) assessment or treatment of patients with LBP initiated by a nonmedical health care provider delivering usual care; (2) inclusion of any comparator group (ie, usual care, preintervention or historical data, or specialty or medical care); and (3) a measure of health care utilization or patient outcome including, but not limited to, medication intake, use of imaging, care-seeking behavior, use of health care procedures, cost of care, pain, disability, and patient satisfaction measures.

Context

There was no restriction on country or health care setting.

Sources of Evidence

Published studies in peer-reviewed scientific journals were eligible for inclusion. Commentaries and editorials were excluded.

Study Selection

All studies identified in the search strategy were exported into reference manager software and deduplicated. Title and abstracts were independently screened by 2 pairs of reviewers. Reconciliation of screening was performed at regular intervals throughout the screening process. Full-text articles were retrieved and assessed for eligibility by 2 pairs of reviewers. If consensus could not be reached within the review team, the senior author (P.F.) was consulted to assist with any disagreements.

Data Extraction

Data were independently extracted by 2 reviewers using a customized data extraction form developed for this review (Tab. 1). When applicable, statistical findings relating to key outcomes were included in the data extraction process.

Table 1.

Draft Extraction Template

Study	Country	No. and Source of Participants	Setting	Objectives	Provider Type (Nonmedical)	Comparator	Definition of Episode of Care	Duration	Outcomes Relating to Health Care Utilization	Secondary Outcomes Measured	Key Findings Relating to Primary Outcome	Key Findings  Relating to  Secondary  Outcome
Carey et al23 (1995)	United States	1555 consecutively selected patients 208 randomly selected primary care providers (87 primary care physicians, 64 chiropractors, 29 orthopedic surgeons, 28 health maintenance organization providers)	Primary care	To determine whether outcomes and charges vary according to type of provider initially seen for an episode of acute low back pain	Chiropractor	Primary care physician, orthopedic surgeon	No low back pain in 2 mo preceding current episode; people who had not previously seen a health care provider for current episode	6 mo from initial visit	Medication prescription, radiography, advanced imaging, cost of care, hospitalization, frequency of visits	Function measured by Adapted Sickness Impact Profile; patient satisfaction with care	Average no. of prescriptions or over-the-counter medications was lower among patients initiating care with chiropractors than with medical practitioners (2.3 vs 3.5 medications; P < .001)	No clinically significant differences in time to functional recovery (patient reported) measured among different initial provider entry points

Analyses and Data Presentation

Analyses followed 3 stages as set a priori. First, a descriptive analysis of study characteristics including primary objectives, population sample, study design, setting, type of initial health care provider, and comparator was conducted. Second, health care utilization and patient outcomes pertaining to the initial health care provider were analyzed. Two reviewers analyzed the data from included studies and categorized findings into health care utilization outcomes (subcategorized on the basis of common outcome measures) and patient-related outcome measures. Third, key findings related to gaps in the literature and future research directions were discussed with the senior author (P.F.), and the written summary was reviewed by the research team.

Quality Assessment

The Joanna Briggs Institute critical appraisal checklist for cohort studies was used to determine potential bias in the design, conduct, and analyses of individual studies. It consists of an 11-item checklist with possible responses, including “yes,” “no,” “unclear,” or “not applicable,” for each criterion.²¹ Eligible studies were independently reviewed by 2 authors, and any discrepancies were resolved by consensus or consultation with the senior author (P.F.). Studies were not excluded based on quality appraisal findings.

Results

Search Results

The search yielded 31,883 citations, reduced to 26,462 with duplicates removed (Fig. 2). After title and abstract screening, 138 articles were eligible. A full-text review identified 11 articles pertaining to a nonmedical initial provider, and these were included in this review.

Figure 2.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) search flowchart of included studies. LBP = low back pain; MSK = musculoskeletal conditions.

Quality Assessment

Results of the quality assessment are summarized in Tab. 2. Four of the 11 studies satisfied all criteria in the Joanna Briggs Institute checklist, indicating low risk of bias.^24–27 The primary quality flaws of the remaining studies were lack of clarity of baseline comparisons between groups,^28–31 not explicitly identifying cofounding factors,^29,30 and participants not being free of a measured outcome (pain medication) at the beginning of a study.³⁰ Nine of the 11 studies^{23–28,31–33} made adjustments for confounding variables.

Table 2.

Quality Appraisal of Included Studies Using the JBI Critical Appraisal Checklist for Cohort Studies^a

Study	Answer to the Following Critical Appraisal Question:
	Were the 2 Groups Similar and Recruited From   Same Population?	Were the Exposures Measured Similarly to Assign People  to Both Exposed and Unexposed Groups?	Was Exposure Measured in Valid and Reliable  Way?	Were Confounding Factors Identified?	Were Strategies to Deal With Confounding Factors Stated?	Were Groups/Participants Free of Outcome at Start of Study (or at Moment of Exposure)?	Were  Outcomes Measured in Valid and Reliable Way?	Was Follow-up Time Reported and Long Enough for Outcomes to Occur?	Was Follow-up Complete and, If Not, Were Reasons for Loss to Follow-up Described and Explored?	Were Strategies to Address Incomplete Follow-up Used?	Was  Appropriate Statistical Analysis Used?
Carey et al23 (1995)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	No	Unclear	Yes
Sundararajan et al33 (1998)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	No	Unclear	Yes
Liliedahl et al29 (2010)	Unclear	Yes	Yes	No	Unclear	Yes	Yes	Yes	N/A	N/A	Unclear
Henschke et al28 (2013)	Unclear	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Unclear	Yes
Fritz et al24 (2015)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	N/A	N/A	Yes
Fritz et al25 (2016)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	N/A	N/A	Yes
Frogner et al30 (2018)	Unclear	Yes	Yes	No	Unclear	No	Yes	Yes	N/A	N/A	Yes
Azad et al31 (2019)	Unclear	Yes	Yes	Yes	Yes	Yes	Yes	Yes	N/A	N/A	Yes
Kazis et al27 (2019)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	N/A	N/A	Yes
O’Reilly-Jacob et al32 (2019)	Yes	Yes	Unclear	Yes	Yes	Yes	Yes	N/A	N/A	N/A	Yes
Garrity et al26 (2020)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	N/A	N/A	Yes

^a JBI = Joanna Briggs Institute; N/A = not applicable.

Description of Included Studies

Key study characteristics are summarized in Tab. 3. All studies identified were observational and primarily used retrospective claims-based data. Ten studies were conducted in the United States and 1 study completed in Australia.²⁸ In all 11 studies, initial health care consultation was defined as the entry visit or first contact between patient and health care practitioner following a new episode of LBP. The most common nonmedical providers identified were physical therapists and chiropractors, with 1 study investigating nurse practitioners.³² Primary care physician was the main comparison group in 10 of the 11 studies. Frogner et al³⁰ used a comparison group with a mixed pool of providers (chiropractors, orthopedists, and acupuncturists).

Table 3.

Initial HCP Primary Study Characteristics^a

Study	Country	No. and Source of Participants	Setting	Research Aims	Nonmedical Provider Type	Comparator	Definition of Episode of Care	Duration	Outcomes Relating to  Health Care Utilization	Secondary Outcomes Measured
Carey et al23 (1995)	United States	1555 consecutively selected patients; 208 randomly selected PCPs (87 PCPs, 64 chiropractors, 29 orthopedic surgeons, 28 health maintenance organization providers)	Primary care	To determine whether outcomes and charges vary according to the type of provider initially seen for an episode of acute LBP	Chiropractor	PCP, orthopedic surgeon	No LBP in 2 mo preceding current episode; people who had not previously seen HCP for current episode	6 mo from initial visit	Medication prescription, radiography, advanced imaging, cost of care, hospitalization, frequency of visits	Function measured by Adapted Sickness Impact Profile; patient satisfaction with care
Sundararajan et al33 (1998)	United States	1580; secondary analyses of data from Carey et al23 (1995)	Primary care	To describe patterns of provider use associated with an acute episode of nonspecific LBP and their impact on cost	Chiropractor	PCP, orthopedic surgeon	No LBP in 2 mo preceding current episode; people who had not previously seen HCP for current episode	6 mo from initial visit	Seeking care from multiple providers and cost of care	None
Liliedahl et al29 (2010)	United States	85,402; Blue Cross Blue Shield of Tennessee claims database of fully insured patients (unrestricted access to primary and specialty providers); October 1, 2004–September 30, 2006	Primary care, ED	To determine whether there are differences in the cost of LBP care when a patient is able to choose a course of treatment with a medical doctor vs a doctor of chiropractic	Chiropractor	PCP (including osteopaths)	All reimbursed care delivered between first and last encounters with HCP for LBP; clean period of 60 d of no HCP use prior to current episode	Unspecified	Cost of care	None
Henschke et al28 (2013)	Australia	1328; secondary analyses of cohort study with consecutive patient selection	Primary care	To determine predictors of satisfaction with care and symptoms in patients with acute LBP 1 y after presentation to primary care	Physical therapist, chiropractor	PCP	No experience of an episode of back pain in month prior to presentation	12 mo after index visit	None	Satisfaction with care
Fritz et al24 (2015)	United States	2289; University of Utah Healthy U Medicaid managed health plan database; January 1, 2012–December 30, 2012	Primary care, ED	To examine physical therapy utilization by Medicaid enrollees with new LBP consultations and to evaluate associations with future health care utilization and LBP-related costs	Physical therapist	PCP and ED entry	No LBP-related claims in preceding 90 d	12 mo from initial visit	Radiography, advanced imaging, ED visits, epidural injections, surgery, cost of care	None
Fritz et al25 (2016)	United States	747; University of Utah database enrollees with private employer–based coverage; January 1, 2012–January 31, 2013	Primary care and secondary care	To explore associations between entry settings and future LBP-related utilization and costs	Physical therapist, chiropractor	PCP	New consultation for LBP in which no charges associated with LBP received in 90 d prior to entry visit	12 mo from initial visit	Radiography, advanced imaging, ED care, epidural injections, surgeon visit, surgery, cost of care	None
Frogner et al30 (2018)	United States	148,866; Healthcare Cost Institute claims database of patients with private insurance; 2009–2013	Primary care, secondary care, ED	To compare differences in opioid prescription, health care utilization, and costs among patients who had LBP and who saw a physical therapist at the first point of care, at any time during the episode, or not at all	Physical therapist	Delayed physical therapy or no physical therapy (including a mix of chiropractors, orthopedists, and acupuncturists)	Patients with new LBP diagnosis from “index date” that followed 6-mo clean period in which no visits for LBP	12 mo from initial visit	Opioid prescription, radiography, advanced imaging, ED care, hospitalization, costs	None
Azad et al31 (2019)	United States	478,981; Truven Health MarketScan Commercial claims and Encounters database (primarily comprising patients with private insurance); June 1, 2009–December 31, 2011	Primary care, secondary care, ED	To determine whether provider specialty influences patterns of opioid use long after the initial diagnosis	Physical therapist, chiropractor, nurse practitioner, acupuncturist, physician assistant	PCP	Defined as first visit at which enrollee had claim with ICD-9 code that met criteria of LBP or lower extremity radicular pain, without any of these codes in 12 mo prior to presentation	12 mo from initial visit	Opioid prescription	None
Kazis et al27 (2019)	United States	216,504; OptumLabs Data Warehouse. Database of commercial and Medicare Advantage enrollees; 2008–2013	Primary care, secondary care, ED	To examine the association of initial provider treatment with early and long-term opioid use in a national sample of patients with new-onset LBP	Physical therapist, chiropractor, acupuncturist	PCP	New-onset LBP defined as no diagnosis of LBP or back procedures, including spinal surgery, spinal injections, and spinal nerve stimulators, during 12-mo period prior to index event	12 mo from initial visit	Opioid prescription	None
O’Reilly-Jacob et al32 (2019)	United States	45,295; Medicare claims database from 18 Dartmouth Atlas of Healthcare hospital referral regions; 2012–2013	Primary care	To compare the rates of low-value back images ordered by primary care physicians and primary care nurse practitioners after primary care visits for first-time back pain	Nurse practitioner	PCP	Patients without recent history of back pain or back pain–related hospitalizations	30 d from initial visit	Radiography, advanced imaging	None
Garrity et al26 (2020)	United States	60,598; OptumLabs Data Warehouse. Database of commercial and Medicare Advantage enrollees; 2008–2013	Primary care	To evaluate the association of level of access to physical therapist services with LBP-related health care utilization and costs	Physical therapist (in restricted- and unrestricted- access states)	PCP	New-onset LBP defined as primary LBP claim and no prior claims with diagnosis of LBP or prior back procedures, including spinal surgery, spinal injections, and spinal nerve stimulators, during 12-mo period prior to index event	90 d from initial visit	Cost of care (physical therapist vs PCP); physician visits, radiography, advanced imaging, and back procedures analyzed between access levels only	None

^a ED = emergency department; HCP = health care provider; ICD-9 = International Classification of Diseases, Ninth Revision; LBP = low back pain; PCP = primary care physician.

Study outcomes were categorized into 6 areas: medication (5 studies), imaging (5 studies), care seeking (3 studies), cost of care (6 studies), health care procedures and hospitalizations (3 studies), and patient outcomes (2 studies). Table 4 presents the key findings regarding the influence of initial provider on primary and secondary health care outcomes.

Table 4.

Key Findings Regarding Influence of Initial Provider on Primary and Secondary Health Care Outcomes^a

Category	Study	Health Care Outcomes per Episode of Care
Care seeking	Carey et al23 (1995)	Mean no. of visits associated with initiation of care significantly higher with DCs than with medical providers (mean visits in rural/urban areas = 10.1/15 for DCs vs 4.6/4.4 for PCPs; P = .001)
	Sundararajan et al33 (1998)	Patients who saw HMO providers least likely to see multiple providers during an episode of care for LBP (9% [95% CI = 5% to 14%]), whereas patients who initiated care with orthopedic surgeons most likely to do so (30% [95% CI = 23% to 37%]); of those who initiated care with DCs and PCPs, 19% (95% CI = 16% to 23%) and 14% (95% CI = 11% to 17%), respectively, sought care from other health care providers
	Fritz et al25 (2016)	Duration of episode of care with initial visit in DC setting longer than that with initial visit in primary care setting (standardized ß = 0.51 [95% CI = 0.27 to 0.76]; P < .001)
Medication use/prescription	Carey et al23 (1995)	Average no. of prescriptions or over-the-counter medications lower among patients initiating care with DCs than among those initiating care with medical practitioners (2.3 vs 3.5 medications, respectively; P < .001)
	Fritz et al24 (2015)	Odds of receiving opioid medication early significantly lower for patients entering via physical therapy (OR = 0.5 [95% CI = 0.28 to 0.89]; P = .02) than for those entering via primary care (combined PCP, ED, and physiatry)
	Frogner et al30 (2018)	Patients who had physical therapy first had an 89.4% (SE = 0.053; P < .001) reduced probability of receiving an opioid prescription than those who had no physical therapy or physical therapy later
	Azad et al31 (2019)	Compared with patients who initiated care with PCP, those who initiated care with nonmedical provider less likely to receive an early opioid prescription (HR = 0.5 [95% CI = 0.49 to 0.5]; P < .0001) and less likely to receive a third opioid prescription (HR = 0.45 [95% CI = 0.43 to 0.46]; P < .0001)
	Kazis et al27 (2019)	Compared with patients who initiated care with PCP, patients who initiated care with DC (OR = 0.1 [95% CI = 0.09 to 0.1]; P < .01), acupuncturist (OR = 0.09 [95% CI = 0.07 to 0.12]; P < .01), or physical therapist (OR = 0.15 [95% CI = 0.13 to 0.17]; P < .01) had significantly lower odds of early opioid use; this result also seen with long-term opioid use (for DC: OR = 0.22 [95% CI = 0.15 to 0.48] [P < .01]; for acupuncturist: OR = 0.07 [95% CI = 0.01 to 0.48] [P < .01]; for physical therapist: OR = 0.27 [95% CI = 0.15 to 0.48] [P < .01])
Imaging	Carey et al23 (1995)	No. of radiographs higher per episode of care for patients initiating care with DCs and orthopedic surgeons (67%–72% of patients) than for those initiating care with PCPs (26%–32% of patients) (P = .001); use of advanced imaging lower for patients seeing DCs (7%–8%) and going to HMO (6%) than for patients seeing an orthopedist (17%) (P = .004)
	Fritz et al24 (2015)	Relative to primary care, physical therapy as entry setting associated with lower odds of radiography (OR = 0.32 [95% CI = 0.15 to 0.65]; P < .001); no statistical difference seen in advanced imaging rates
	Fritz et al25 (2016)	Relative to primary care, physical therapy as entry setting associated with decreased risk of radiography (OR = 0.39 [95% CI = 0.18–0.84]), but no statistical difference in advanced imaging rates; chiropractic as entry setting associated with decreased risk of advanced imaging rates (OR = 0.21 [95% CI = 0.08 to 0.50]), but no statistical difference in radiography
	Frogner et al30 (2018)	Patients who had physical therapy as first point of care had 29.7% (SE = 0.045; P < .001) reduced probability of having advanced imaging and 16.6% (SE = 0.056; P < .001) reduced probability of having radiography than those who had no physical therapy or physical therapy later
	O’Reilly-Jacob et al32 (2019)	No significant difference in rates of low-value back images between primary care medical doctors (24.5% [IQR = 11%–38%]) and primary care nurse practitioners (26.5% [IQR = 7%–40%]) after initial consultation
Cost of care	Carey et al23 (1995)	Care initiated by urban chiropractors (adjusted mean = $783 [95% CI = $698 to $868]) and orthopedists (adjusted mean = $746 [95% CI = $633 to $858]) had highest costs per episode of LBP compared with care initiated via urban primary care providers (adjusted mean = $508 [95% CI = $418 to $598])
	Liliedahl et al29 (2010)	Mean cost per episode significantly lower for care initiated with chiropractor ($532.54 [SE = $9.56]) than initiated with medical doctor ($661.10 [SE = $29.16])
	Fritz et al24 (2015)	Physical therapy as entry point of care associated with significantly lower health care costs over 12 mo ($335 [95% CI = $241 to $429]) than primary care ($533 [95% CI = $470 to $598])
	Fritz et al25 (2016)	Care initiated via physical therapy (standardized ß = −0.21 [95% CI = −0.63 to 0.2]; P = .34) and chiropractic (standardized ß = −0.28 [95% CI = −0.058 to 0.021]; P = .07) not associated with statistically significant differences in cost compared with care initiated via primary care
	Frogner et al30 (2018)	Care initiated via physical therapy associated with higher provider costs but lower pharmacy, outpatient, and out-of-pocket costs than care not initiated via physical therapy; total costs did not differ between patients who did and patients who did not initiate care via physical therapy
	Garrity et al26 (2020)	The 90-d cost ratio higher for care initiated via physical therapy in both provisional access states (1.28 [95% CI = 1.20 to 1.36]) and unrestricted states (1.14 [95% CI = 1.05 to 1.23]) than for care initiated via primary care

(continued)

Review Findings

Medication

Five studies reporting data on medication showed an associated reduction in medication prescription when LBP patients initiated care with a nonmedical provider compared with a medical provider^23,24,27,31 or a mixed group of providers.³⁰ Studies demonstrated that initiating patient care with a nonmedical provider was associated with a reduction in both short-term^24,27,31 and long-term^27,31 opioid prescriptions compared with initiating care with a medical provider. For example, Azad et al³¹ observed a 50% lower risk (hazard ratio = 0.5 [95% CI = 0.49–0.5]; P < .0001) of receiving an early opioid prescription and a 55% lower risk (hazard ratio = 0.45 [95% CI = 0.43–0.46]; P < .0001) of receiving a long-term opioid prescription when LBP patients initiated care with a nonmedical provider versus a primary care physician. Carey et al²³ observed that the average number of prescriptions or over-the-counter medications (ie, nonsteroidal antiinflammatories) per episode of LBP was lower for patients initiating care via a chiropractor than for those initiating care via a medical provider (2.3 vs 3.5 medications; P < .001). Table 4 presents the key findings regarding the influence of the initial provider on primary or secondary health care outcomes.

Care Seeking

Three studies reported measures of care seeking, including number of visits to a health care provider per episode of care,²³ number of multiple providers used,³³ and duration of episode of care.²⁵ Results from these studies showed that care initiated by a chiropractor was associated, on average, with 5 more visits (mean number of chiropractor visits = 10; mean number of primary care physician visits = 5) (P = .001)²³ and increased care duration (standardized ß = .51 [95% CI = 0.27–0.76]; P < .001)²⁵ per LBP episode than care initiated by a primary care physician. Using secondary analysis of previous study data, Sundararajan et al³³ observed that 19% (95% CI = 16%–23%) of patients initiating care with a chiropractor sought care from multiple providers compared with 14% (95% CI = 11%–17%) for private primary care physicians and 9% (95% CI = 5%–14%) for network primary care physicians. Initiating care with a physical therapist showed no differences in care duration compared with initiating care with a primary care physician.²⁵

Cost of Care

The components of the cost of care per episode of LBP were consistent across the 6 studies, including this outcome, and included a sum of direct costs relating to imaging, medication, consultations, related health care procedures, and inpatient charges. Costs were calculated using state-wide averages²³ or recorded costs from claims databases.^{24–26,29,30} When chiropractor- and physical therapist–initiated care for LBP patients were compared with primary care physician data, the results were conflicting. Two studies indicated an increased cost^23,26; for example, Garrity et al²⁶ showed 14% higher (95% CI = 5%– 22%) LBP-related costs for patients initiating care with a physical therapist than for those initiating care with a primary care physician. Two studies found reduced costs^24,29; for example, Fritz et al²⁴ showed that physical therapists incurred reduced costs ($335 [95% CI = $241–$429]) compared with primary care physicians ($533 [95% CI = $470–$598]). Two studies found no difference in total costs of care between nonmedical providers and primary care physicians.^25,30

Imaging

Five studies reported on the use of radiography and advanced imaging (computerized tomography and magnetic resonance imaging). Imaging rates differed by the type of nonmedical provider. There was an associated reduction in advanced imaging rates for patients initiating care with a chiropractor compared with a medical provider. Fritz et al²⁵ observed that patients seeing a chiropractor were 79% less likely (odds ratio = 0.21 [95% CI = 0.08 –0.50]) to receive advanced imaging than patients seeing a primary care physician. Carey et al²³ observed a reduction (8% vs 17%; P = .004) in advanced imaging when comparing chiropractor-initiated care with orthopedist-initiated care. The same studies had conflicting results for plain-film radiography rates.

Three studies showed a reduction in radiographs when patients initiated care with physical therapists compared with primary care physicians^24,25 or a mixed group of initial providers.³⁰ For example, Fritz et al²⁴ observed reduced odds of radiographs (odds ratio = 0.32 [95% CI = 0.15–0.65]; P < .001) for patients initiating care with a physical therapist than with a primary care physician. Results from these studies were inconsistent for the use of advanced imaging. One study compared rates of imaging between primary care physicians and nurse practitioners and observed similar rates of low-value imaging after an initial visit for LBP.³²

Health Care Procedures

Health care procedures described in original studies included epidural injections, emergency department visits, surgery, spine surgeon consultations, and hospitalizations. Two studies comparing patients initiating care with a nonmedical provider versus a primary care physician showed no difference in the odds of receiving an epidural injection or visiting an emergency department.^24,25 Frogner et al³⁰ found that patients who initiated care with a physical therapist showed a 38.3% (SE = 0.052; P < .001) reduction in their probability of emergency department visits but a 19.3% (SE = 0.03; P < .001) increase in their probability of hospitalization associated with physical therapist care. Two studies investigated surgical rates and observed that no patients who had LBP and initiated care with a physical therapist underwent surgery in the 12-month follow-up period.^24,25 By contrast, patients initiating care with a physical therapist were 4.75 (95% CI = 1.42–16.0; P = .012) times as likely to undergo surgery as those initiating care with a primary care physician. Fritz et al²⁵ also observed reduced odds (odds ratio = 0.18 [95% CI = 0.03–0.53]; P = .005) of surgical consultations for patients who initiated care via a chiropractor compared with those who initiated care via a primary care physician.

Table 4.

Continued

Category	Study	Health Care Outcomes per Episode of Care
Health care procedures and hospitalizations	Frogner et al30 (2018)	Patients who had physical therapy first had 38.3% (SE = 5.2%; P < .001) reduced probability of ED visits but 19.3% (SE = 3.0%; P < .001) increased probability of hospitalization
	Fritz et al24 (2015)	Physical therapy as initial management not associated with statistically significant difference in odds of receiving an injection (OR = 0.36 [95% CI = 0.04 to 2.99]; P = .35) or having ED visit (OR = 0.46 [95% CI = 0.18 to 1.18]; P = .11) compared with primary care; no patients who initiated care via physical therapy had surgery
	Fritz et al25 (2016)	Care initiated via physical therapy (OR = 0.36 [95% CI = 0.05 to 2.82]) and chiropractic (OR = 0.42 [95% CI = 0.10 to 1.80]) had no significant effect on ED care or odds of injection compared with care initiated via primary care; care initiated via chiropractor had reduced odds (OR = 0.13 [95% CI = 0.03 to 0.53]) of spinal surgeon consultation; no patients who initiated care via physical therapy had surgery
Patient outcomes	Carey et al23 (1995)	No clinically significant differences in time to functional recovery (patient reported) measured among different initial provider entry points; patient satisfaction levels for initiation of care with chiropractor vs PCP or orthopedic or HMO provider were: information given (47.1% vs 30.2%; P < .001), treatment of back problem (52.1% vs 31.5%; P < .001), outcomes (42.1% vs 26.5%; P < .001), detailed history of back pain (88.4% vs 68.4%; P < .001), careful examination of back (96.1% vs 79.9%; P < .001), and cause of problem clearly explained (93.6% vs 74.6%; P < .001)
	Henschke et al28 (2013)	No statistically significant differences in satisfaction with care (OR = 0.81 [95% CI = 0.58 to 1.12]; P = .2) or satisfaction with symptoms (OR = 0.98 [95% CI = 0.73 to 1.33]; P = .916) between care initiated via PCP and that initiated via physical therapist

^a DC = chiropractor; ED = emergency department; HCP = health care practitioner; HMO = health maintenance organization; HR = hazard ratio; IQR = interquartile range; LBP = low back pain; OR = odds ratio; PCP = primary care physician.

Patient Outcomes

Two studies assessed patient outcomes including satisfaction with care^23,28 and functional recovery.²³ When patients initiated care with a physical therapist compared with a primary care physician,²⁸ no difference in satisfaction with care was observed. In contrast, another study²³ reported that patients who initiated care via chiropractors were more satisfied with all aspects of care than patients who initiated care via medical providers (satisfaction with care provided by chiropractors was 42.1%, and that with care provided by physicians was 26.5%; P < .001). In a study of functional recovery in LBP,²³ no statistical difference in time to functional recovery among the 6 different entry providers (urban and rural chiropractors, urban and rural primary care physicians, orthopedists, and network primary care physicians) was observed for patients with LBP.

Discussion

This scoping review identified 11 observational studies that investigated the influence of initial management of LBP patients by nonmedical health care providers, primarily compared with physicians, on the outcomes of medication prescription, care seeking, cost of care, health care procedures, imaging, and patient outcomes.

The absence of any randomized controlled trial identified in this scoping review prevents strong conclusions on the efficacy of nonmedical initiated care. However, consistent associations for reduced rates of short-term^24,27,31 and long-term^27,30,31 opioid prescriptions as well as selective imaging (radiography for physical therapists^24,25,30 and advanced imaging for chiropractors^23,25) identified in this review suggest that there may be benefits associated with initiating care with a nonmedical provider compared with a medical provider for patients with a new episode of LBP. Potential disadvantages could include the increased rate of care seeking^23,25,33; however, the strength of this finding is limited by the inclusion of 2 studies using the same data set.^23,33 There were inconsistent results for cost of care, health care procedures, or patient outcomes.

The results in this review should be interpreted within the setting that the observational studies were primarily conducted, which was the United States (10/11 studies). For example, the reduction in opioid prescriptions and imaging referrals may reflect the lack of prescribing rights ascribed to nonmedical practitioners as well as barriers to referrals for imaging compared with medical providers, with various states, provider settings, and insurance companies dictating the ability to request images directly.²⁵ Additionally, payment for nonpharmacological treatments for LBP among included studies is likely to be impacted by variable public and private insurance coverage unique to this setting.³⁴ Examining alternate pathways that patients access health care, such as referral to nonmedical care^35,36 and nonmedical models of care,^19,20 provides some context to the findings from this review.

A previous systematic review investigating the timing of nonmedical care for patients with LBP compared early physical therapy (within 30 days of the index visit to a medical provider) with both delayed physical therapy (>30 days of the index visit) and usual care.³⁷ Early physical therapy resulted in reduced imaging, opioid prescription, and downstream costs compared with delayed physical therapy. However, compared with usual care, early physical therapy did not demonstrate consistent associations with health care utilization outcomes, including imaging and opioid prescription. Two randomized controlled trials included in the review showed higher costs for patients in the early physical therapy arm.^38,39 The contrasting findings with this scoping review may be explained by both timing and access to care. It is possible that nonpharmacological pain management and education strategies delivered by nonmedical practitioners are effective in reducing opioid prescribing and imaging referrals when offered at the entry point of care compared with the same strategies offered after consultation with a medical practitioner. However, the characteristics of patients who seek out nonmedical care for LBP, such as higher education and income levels,^40,41 may differ from those of patients who are referred by medical practitioners and may favorably influence the likelihood that these patients will be prescribed opioids or referred for imaging. Supporting this argument, results from a recent systematic review investigating nonmedical models of care (triage, self-referral, and direct access) in patients with musculoskeletal pain suggest that patients who “self-referred” to physical therapy care were slightly younger, were more educated, and had a better socioeconomic status than those who initiated care via the usual medical team–led care model.²⁰

Gaps in the Research

Potential influences on access to nonmedical care such as government policy,⁴² funding,³⁴ health care systems,¹¹ remoteness, and public awareness⁴³ are important elements to consider when interpreting the pragmatic impact of the initial provider on outcomes for LBP patients. Within this review, only 1 study investigated the impact of access on medical compared with nonmedical care.²⁶ Results from this study indicated that patients living in states with provisional access to physical therapy consulted physicians at a 41% higher rate (95% CI = 1.00–1.99; P = .05) and had increased odds of plain imaging in the first 30 days after consultation (odds ratio = 1.58 [95% CI = 1.03–2.42]; P = .03) compared with patients with unrestricted access to physical therapy. With only 20 states in the United States classified as having unrestricted direct access to physical therapy care,⁴⁴ this issue may be an important consideration when interpreting results for care initiated via nonmedical personnel in different settings.

This review did not identify any studies investigating the impact of nonmedical initiated care for LBP patients outside of primary care. Nonmedical practitioners are afforded advanced scope of practice roles in countries such as Australia and the United Kingdom.^45–47 However, systematic evidence on the impact of nonmedical providers in these settings has largely grouped all musculoskeletal complaints together,^48,49 and evidence in the LBP population specifically did not control for the initial encounter.⁴⁵

The measure of influence of a nonmedical provider at the initial point of care in patients with LBP may not be limited to changes in health care utilization or patient outcomes identified in this review. Systematic reviews of direct-access physical therapy¹⁹ and nonmedical triage models in musculoskeletal patients²⁰ have reported other important outcomes, such as reduced physician visits,¹⁹ improved quality of life,^19,20 and reduced work-related absence²⁰ associated with nonmedical care, that either were not measured in the included observational studies in this review or were limited to a few studies.^25,33 Additionally, only 1 study in this review attempted to measure low-value care³² in the form of unnecessary imaging on the basis of the recommendations provided by Choose Wisely.⁸ Short-term opioid prescribing, imaging requests, and surgical referrals may be appropriate recommendations for specific LBP patients, and distinguishing these from low-value care choices is important to provide evidence of a practitioner’s ability to guide patients’ early care decisions through the health care system.

Several of the studies included in this review reported on the influence of nonmedical initiated care on rare health care outcomes such as injections, hospitalizations, and spine surgery.^24,25,30 These outcomes arguably incur some of the greatest costs in an episode of LBP care and often represent low-value care choices given clear lack of evidence for effectiveness in the majority of LBP patients.^50–52 Despite their importance, the low incidence rates of these outcomes in LBP populations⁵³ create some uncertainty around the results within included studies. For example, elective lumbar fusion rates in the United States have been estimated at 80/100,000 adults.⁵⁴ The sample size required to determine a moderate effect size between nonmedical initiated care and physician-led care in a rare outcome such as this would be extremely large and unlikely to be accurately represented by the relatively small sample sizes of some of the included studies.^24,25

Implications for Clinical Practice and Future Research

The relationship between initiation of care by nonmedical providers and decreased short- and long-term opioid prescription as well as reduced imaging in patients with LBP warrants further investigation. Opioid prescription has not been shown to be superior to nonopioid medications in LBP, has no evidence to support early return to work, is not associated with improved patient outcomes, and is associated with significant harms, including overdose and death.^55–57 Similarly, radiographs or advanced imaging (eg, magnetic resonance imaging) have been associated with potential harm, including increased health care costs,⁵⁸ radiation exposure (radiographs and computerized tomography),⁵⁹ and negative psychological consequences such as increased patient anxiety or movement-related fear.⁶⁰ Both types of imaging may be considered low-value care in the absence of signs and symptoms suggesting serious pathology.⁶¹ Efforts to reduce unnecessary early prescription of opioid medication and imaging referral may have important long-term health implications for patients and the health care system.

Future research examining the effect of the entry point for LBP care must balance the need to determine the true effect of the initial provider under ideal experimental circumstances (efficacy) alongside the need to examine the impact of the initial provider in real-world scenarios (effectiveness). For example, using a randomized controlled trial design and randomly assigning LBP patients to nonmedical or medical providers in primary care settings would help address issues of selection bias. However, by removing the patient’s choice of initial provider, those factors known to influence a patient’s entry point of care, such as cost, access, health care funding, insurance type, a patient’s beliefs, and social recommendations, are no longer accounted for.^62–64 A potential solution would be to use a cluster design approach as demonstrated by a pilot trial in the United Kingdom, where randomization occurred at the clinic rather than patient level.⁶⁵ Practices were randomized to offer 2 models of care: the option of direct-access physical therapy alongside physician-led care in the intervention arm or usual physician-led care. The advantage of this design is that it still allows patients to choose their care while minimizing potential issues with selection bias as well as contamination between arms of the trial.

Considering the potential benefits to reducing low-value care (opioid prescription and imaging) alongside the uncertainty of improved clinical outcomes, a cost-effectiveness analysis would be an important addition to future research to help inform policy makers and patients of the efficiency of nonmedical compared with traditional physician-led care in the treatment of LBP. This would need to include direct costs attributed to patients and health care systems as well as indirect costs associated with loss of productivity (ie, work absenteeism).⁶⁶

Strengths and Limitations

To the best of our knowledge, this is the first review to analyze the influence of the initial health care provider on subsequent outcomes in patients with a new episode of LBP. This review identified the relationship between the nonmedical initial provider, opioid prescription, and imaging use as areas for future research and a potential pathway to reduce low-value care.

The results of this study must also be considered alongside its limitations. By using the term “nonmedical provider,” we sought to obtain a scoping perspective of care approaches for LBP but acknowledge that within this mixed group, practitioners such as nurses and physical therapists have different roles and levels of training in LBP care, which may influence outcomes. Nurse practitioners commonly substitute physician roles in triage, and the majority have limited training in treating musculoskeletal conditions.⁶⁷ It is possible that their lack of confidence delivering a nonmedical management approach for LBP would lead to a similar number of investigations, prescriptions, and procedures as physicians. This has been demonstrated within a broader patient population in primary care in a recent Cochrane review⁶⁸ but not specifically for patients with LBP. Although the trends reported in this scoping review would not change, future studies and systematic reviews should define providers by the type of training they receive and the type of care offered.

Most included studies used a retrospective design, with this method of examining the influence of the initial provider being limited by the information contained in the respective databases. First, all databases contained information on patients in the US health care system, limiting generalizability given the different funding arrangements and structure of health care systems that are likely to influence access to care. Second, there is a risk of bias due to unmeasured confounding variables not being accounted for within the databases. Specifically, pain severity and disability, which are known characteristics associated with health care use⁶⁹ in LBP patients, were measured in only 2 of the included studies.^23,33 It is possible that patients with greater severity of pain and disability presented to medical providers and that the lack of adjustment for these variables is a limitation of the original studies. The Joanna Briggs Institute critical appraisal tool for cohort studies was chosen given its appropriateness for assessing included study designs.⁷⁰ However, a limitation of the tool is the lack of definitive guidance provided in responding with “yes/no/unclear” to some criterion, such as item 4 (“Were confounding factors identified?”), which does not require a decision on the inclusion of “sufficient” confounders. It is likely that future trials will provide more clarity around potential confounding domains as the evidence base develops. This could be addressed in future reviews by including pre-specified confounders in the review protocol alongside the use of more detailed critical appraisal tools.^70,71

Although this is a scoping review and no pooling of data was performed, the consistent finding within included studies suggests initiating care with a nonmedical provider may be protective of opioid prescription in patients with LBP. The impact of providers on other measures such as use of imaging and care seeking is less clear. Future research should incorporate more comprehensive patient variables, cost-effectiveness analysis, and a carefully considered trial design to examine the pragmatic impact of nonmedical initiated care on opioid use and other health care outcomes in patients with acute LBP, with important implications for health care policy to reduce low-value care.

Author Contributions

Concept/idea/research design: J. Zouch, A. Bussières, C.E. Ashton-James, M. Ferreira, P. Ferreira

Writing: J. Zouch, C.E. Ashton-James, P. Ferreira, A. Bussières

Data collection: J. Zouch, J. Comachio, A.H. Salles dos Reis, Y. Chen

Data analysis: J. Zouch, Josielli Comachio, A. Bussières, P. Ferreira

Project management: J. Zouch, P. Ferreira

Fund procurement: P. Ferreira

Providing facilities/equipment: P. Ferreira

Providing institutional liaisons: P. Ferreira

Consultation (including review of manuscript before submitting): J.Zouch, J. Comachio, A. Bussières, C.E. Ashton-James, M. Ferreira, P. Ferreira

Acknowledgments

We thank Ms Kanchana Ekanayake from The University of Sydney Library for assistance with designing the search strategy.

Funding

This research received no funding.

Disclosures and Presentations

The authors completed the ICMJE Form for Disclosure of Potential Conflicts of Interest and reported no conflicts of interest.

A poster presentation of the abstract was presented at the 2021 Back and Neck Pain Forum, November 11–13, 2021; Virtual.