Relationships between physical therapy intervention and opioid use: A scoping review

Abstract

Objective:

Synthesize available evidence that have examined the relationship between physical therapy (PT) and opioid-use.

Literature Survey:

Data sources including Google scholar, Embase, PubMed, Cochrane Library, and CINAHL were searched for English articles up to 10–24-2019 using terms (“physical therapy”[Title/Abstract] OR physiotherapy[Title/Abstract] OR rehabilitation[Title/Abstract]) AND (opiate*[Title/Abstract] OR opioid*[Title/Abstract]).

Methodology:

Included studies evaluated a PT intervention and reported an opioid-use outcome. Data were extracted to describe the PT intervention, patient sample, opioid-use measurement, and results of any time or group comparisons. Study quality was evaluated with Joanna Briggs checklists based on study design.

Synthesis:

Thirty studies were included that evaluated PT in at least one of these seven categories: interdisciplinary program (n=8), modalities (n=3), treatment (n=3), utilization (n=2), content (n=3), timing (n=13), and location (n=2). Mixed results were reported for reduced opioid-use after interdisciplinary care and after PT modalities. Utilizing PT was associated with lower odds (ranging from 0.2–0.8) to use opioid medication for persons with low back pain (LBP) and injured workers; however, guideline-adherent care did not further reduce opioid-use for persons with LBP. Early PT utilization after index visit for spine or joint pain and after orthopaedic surgery was also associated with lower odds to use opioid medications (ranging from 0.27–0.93). Emergency department PT care was not associated with fewer opioid prescriptions than standard emergency department care. PT in a rehabilitation center after total knee replacement was not associated with lower opioid use than inpatient PT.

Conclusions:

The relationship between timing of PT and opioid use was evaluated in 13 of 30 studies for a variety of patient populations. Eight of these 13 studies reported a relationship between early PT and reduced subsequent opioid use, making the largest sample of studies in this scoping review with supporting evidence. There is limited and inconclusive evidence to establish whether the content and/or location of PT interventions improves outcomes due to heterogeneity between studies.

1. Introduction

Rates of opioid abuse have skyrocketed in the U.S. over the past decade with reports estimating that approximately 8–12% of patients prescribed opioids for chronic pain will develop some form of addiction.¹ In response, the Centers for Disease Control in 2016 began endorsing non-pharmacologic alternatives to opioids as an urgent priority to combat this epidemic.² Since then, various non-pharmacologic pain management strategies have been proposed to reduce the dependence on opioids for pain control.^3–6 Physical therapy (PT) has received a lot of attention as one of these strategies with the potential to empower patients to manage their pain, reduce their opioid use, and lower the risk of addiction.

Public opinion shows that PT is a popular treatment option for pain management. A Gallup poll of approximately 6,200 people found that 68% of the American public viewed PT as a very safe intervention, and 86% reported PT as at least somewhat effective to treat neck and back pain.⁷ Physical therapy can improve a patient’s function and reduce their pain through education and exercise,⁸ which may mitigate their need for opioid medication.⁹ Despite the increasing awareness of PT as a non-pharmacologic pain management alternative, its relationship to opioid use is relatively understudied.

Physical therapy was advocated as a non-pharmacologic alternative to opioids by the Centers for Disease Control and Prevention² and the United States Surgeon General.¹⁰ These recommendations do not detail which patients, in which settings, stand to benefit most from using PT as an alternative to opioids. Physical therapy can be used postoperatively,^11–16 preoperatively,^17–21 as part of a multidisciplinary program,^22–28 and in a variety of settings²⁹ with diverse interventions.³⁰ The breadth of PT interventions and the various applications of PT highlight the need for a broad review of PT and its relationship to opioid use. The purpose of this scoping review was to synthesize available evidence that examined the relationship between various PT intervention and opioid use within different patient populations.

2. Methods

This scoping review developed a research question, search strategy, and study selection criteria as an a priori protocol (Appendix A) shared amongst co-authors; this protocol was not registered publicly. Amendments were made to this protocol after iterations of peer review. The Population, Intervention, Comparison, and Outcome (PICO) characteristics of this scoping review were defined in this protocol as persons with pain (P), skilled physical therapy treatment or modality (I), unrestricted comparison or lack thereof (C), and opioid use, consumption, or prescription (O).

2.1. Eligibility Criteria

Studies were included if (1) they studied a PT intervention, a PT modality, or referral to, or utilization of, PT services; and, (2) they reported degree of opioid use after the intervention. Physical therapy intervention was defined as care with a physical therapist, supervised exercise with a physical therapist, manual therapy from a physical therapist, or education by a physical therapist. Physical therapy modalities included taping, dry needling, electrical stimulation, ultrasound, and traction. Acupuncture was also included to avoid exclusion of Western medical acupuncture³¹ performed by physical therapists. Acupuncture studies were excluded if acupuncture was not applied by a physical therapist. Websites, conference proceedings, case study/series, perspective pieces, reviews, studies published in a language other than English, studies without full-text availability, and animal studies were excluded.

2.2. Data Sources and Searches

Google scholar, Embase, PubMed, Cochrane Library, and CINAHL Complete were searched on October 24, 2019 using the following search terms: (“physical therapy”[Title/Abstract] OR physiotherapy[Title/Abstract] OR rehabilitation[Title/Abstract]) AND (opiate*[Title/Abstract] OR opioid*[Title/Abstract]) Limits: English. This combination of PubMed search terms was modified for specific system requirements in the remaining databases (Appendix B).

2.3. Study Selection and Management

Total resultant citations from each data source were recorded, then exported to Mendeley, a reference manager (© 2020 Mendeley Ltd.), which utilized an automatic removal of duplicate entries. At each stage of review (title, abstract, full-text), citations were organized in a separate folder and the first author reviewed all citations. The “check for duplicates” tool was then used to review potential duplicates and remove citations accordingly. Remaining citations were then organized in alphabetical order by title to manually search for, and remove, duplicates. The citations were also exported to a spreadsheet at each stage of review and 10% of citations (every 10^th title/abstract/full-text) were reviewed by the second author as a verification quality check. This methodology for one reviewer with one verifier has been previously used for scoping reviews.^32,33 The specific 10% quality check was modeled off of published work by Gonzalez-Seguel and colleagues.³⁴ All citations were reviewed by the primary reviewed and the verifier based on an a priori-developed decision tree (Figure 1). Any citations excluded by the first author, but included by the second author underwent re-review by the first author as prescribed by the “include rather than exclude” methodology.^34,35 The first author also re-reviewed the other nine citations in the discrepant article’s 10-citation allocation. Any citations meeting inclusion criteria after re-review were moved to the respective review folder for the next stage.

Figure 1.

Study screening process

Articles at each stage (title, abstract, full-text) were reviewed used a decision tree; each decision tree has been combined here for simplification.

TENS: transcutaneous electrical nerve stimulation; NMES: neuromuscular electrical stimulation.

2.4. Data Extraction and Quality Assessment

The first author extracted the following data from included articles: sample demographics (sample size, age, sex distribution), intervention details, opioid measurement reporting, and study design. The first author attempted to contact corresponding authors with missing data to complete the compiled dataset. A single co-author confirmed data extraction for accuracy.

The Joanna Briggs Institute critical appraisal tools³⁶ for cross sectional studies, cohort studies, and randomized controlled trials were used to assess the quality of included articles. Two reviewers independently graded the included studies using these checklists. These reviewers discussed discrepancies to reach consensus, if agreement could not be achieved a third reviewer would settle the discrepancy.³⁵ An overall quality score was assigned in addition to reporting the individual checklist items for each included article. Each quality determinant evaluated by the checklist was marked as present, unclear, or missing. Quality determinants marked as present received a value of 2, determinants marked as unclear received a value of 1, and determinants marked as missing received a value of 0. The overall quality score was calculated as the sum of quality determinants divided by the total possible score (total evaluated determinants multiplied by two). Articles with a score greater than or equal to 0.75 were categorized as good quality, those with scores between 0.50–0.75 were categorized as fair quality, and the remaining articles with scores less than 0.50 were categorized as poor quality.

2.5. Data Synthesis and Analysis

Studies were described based on their investigation of a PT intervention, a PT modality, or utilization, timing, and location of PT. Operational definitions of PT interventions and modalities were maintained from the study selection phase.

3. Results:

Initial search through the five databases resulted in 3,176 citations. Titles, abstracts, and full-text were reviewed after duplicate removal (Figure 2). The second author performed 164 quality checks during title review (10% of total titles) and identified one article that met inclusion criteria but was excluded by the first author. The first author re-reviewed the discrepant article’s 10-citation allotment and confirmed inclusion criteria for these titles. The second author performed 45 quality checks during abstract review (10% of total abstracts) and identified one article that met inclusion criteria but was excluded by the first author. The first author re-reviewed the discrepant article’s 10-citation allotment and confirmed inclusion criteria for these abstracts. The second author performed 5 quality checks during full-text review (10% of total full-text articles) and did not identify any discrepancies for inclusion of articles.

Figure 2.

Selection Results

Thirty articles met inclusion and exclusion criteria (Table 1). Four lead authors were attempted to contact regarding missing data; three replied with requested data. Study designs included retrospective cohorts (n=20),^37–56 a retrospective survey (n=1),⁵⁷ prospective cohorts (n=7),^58–64 and randomized controlled trials (n=2).^65,66 Relationships between various PT interventions and opioid use were summarized based on population (Figure 3).

Table 1.

Study characteristics

Study	Design	Population	Baseline opioid status	N	N Groups	Age (y)*	Female (%)	Intervention details	Duration
Interdisciplinary programs
Meineche-Schmidt 2012	Prospective cohort	Chronic pain	Mixed	306	141  Individual-treated 165  Group-treated	53 49	60 72	Individualized combination of pharmacy, psychology, relaxation, socioeconomic counseling, pain education, CBT, and PT.	237* d
Vines 1996	Prospective cohort	Chronic pain	Users	23	---	46(14)	73	Inpatient pain program of relaxation, exercise, stress management, pain coping, therapeutic integration, and vocational counseling.	4 w
Stein 2013	Prospective cohort	Chronic pain	Mixed	51	---	48(7)	86	Outpatient group sessions for pain education, pharmacy, posture, stretching, ergonomics, mindfulness, breathing, CBT, diet & fitness, and insurance & employment counseling.	6 w
Murphy 2016	Retrospective cohort	Chronic pain	Mixed	324	67  Females 257  Males	47(9) 52(11)	---	Inpatient program including PT, OT, aquatic & recreational therapy; daily exercise; relaxation training; psychotherapy; group education; family intervention; and medication management.	3 w
Anamkath 2018	Retrospective cohort	Chronic pain	Mixed	35	---	56(7)	29	Intensive, pain rehabilitation including acceptance & commitment therapy, CBT, PT, pain education, and pharmacy counseling.	12 w
Bruce 2017 b	Prospective cohort	Pediatric chronic pain	Mixed	960	---	15(2)	74	Outpatient pain rehabilitation with group CBT, PT & OT, relaxation, and biofeedback. Individual treatment provided prn.	3 w
Bruce 2017 a	Prospective cohort	Pediatric chronic pain	Mixed	171	---	15(1)	71	Outpatient pain rehabilitation with group CBT, PT & OT, relaxation, and biofeedback. Individual treatment provided prn.	3 w
Crisostomo 2008	Retrospective cohort	Chronic LBP	Mixed	383	125  Fusion 62  Non-fusion 196  Non-operative	47(14)	62	Inpatient interdisciplinary rehabilitation with daily PT, OT, and cognitive-behavioral group education.	3 w
Modalities
Chabal 1998	Retrospective survey	Chronic pain	Mixed	376	---	47(14)	62	Long-term (≥6mo) use of a prescribed TENS device.	6 m
Castellano 2016	RCT	THR	NP	29	15  NMES 14  Subclinical NMES	57 53	60 64	Two 20-minute sessions of NMES during weightbearing exercise with inpatient PT. Subclinical NMES was used as a control. Devices also used during home exercise programs.	6 w
Collinsworth 2019	RCT	Shoulder surgery	NP	40	21  BFA+usual care 19  Usual care	20(1)	22	Post-operative PT for shoulder surgery (usual care) with BFA (≤5 needles in each ear 3–5 days; follow-up BFA based on response).	6 w
Physical Therapy Treatment
Pullen 2017	Retrospective cohort	HIV with chronic pain	Mixed	46	---	47	39.1	Receipt of PT at multidisciplinary HIV clinic.	DC or 4 m
Pullen 2019	Retrospective cohort	HIV with pain	NP	225	46  PT 179  No  PT	54 52	30 26	Receipt of PT at multidisciplinary HIV clinic.	DC 12–20w
Thackeray 2017	Retrospective cohort	New onset LBP	Mixed	454	239  No PT referral 81  PT participant 134  PT nonparticipant	40(12)	71	PT referral used (participant) or not (non-participant) within 90 days.	≥1 visit
Fritz 2012	Retrospective cohort	New onset LBP	Mixed†	2,077	1,102  Early PT 975  Delayed PT 413  Adherent PT 1,504  Nonadherent PT	43(10)	53	PT within 14 days (early) or 15–90 days (delayed) from index visit with primary care. PT episodes with ≥1 visit were categorized as adherent if each visit had 1 active billing code and 75% of codes during the episode were active.	6.4(5.1)* visits
Childs 2015	Retrospective cohort	New onset LBP	Mixed	753,450	630,727 No PT 72,641  Early PT 50,082  Delayed PT 30,917  Adherent PT 40,642  Nonadherent PT	36(12)	46	PT within 14 days (early) or 15–90 days (delayed) from index visit with primary care. PT episodes with ≥1 visit were categorized as adherent if each visit had 1 active billing code and 75% of codes during the episode were active.	7.1 (12.2)* visits
Lentz 2018	Prospective cohort	Joint pain	NP	246	---	46(16)	65	Outpatient PT treatment provided at therapist discretion.	NP
Kwok 2019	Retrospective cohort	Injured workers	NP	9,596	PT utilization: 2,397  Very high 2,398  High 2,314  Medium 1,070  Low 1,417  None	42(12)	34 37 30 38 34	PT utilization categorized based on number of PT claims as very high (148+), high (38–147), medium (8–37), low (1–7).	210 (346)*d
Timing of Physical Therapy
Sun 2018	Retrospective cohort	Joint pain	Naïve (12m)	88,985	26,096 Early PT 62,889 No early PT	45(10) 46(11)	44 42	PT within 90 days of index date (initial physician office or emergency department visit for musculoskeletal pain).	≥1 visit
Rhon 2018	Retrospective cohort	Hip scope	Mixed	1,679	364  PT-only 266  Opioid-only 678  PT-first 371  Opioid-first	33(8) 32(8) 31(7) 30(7)	39 47 46 46	Post-operative PT evaluation in relation to prescription dates for post-operative opioid medication.	2 y
Sarpong 2019	Retrospective cohort	TKR	Naïve	687	295  PT on POD0 392  PT on POD1	69(10) 68(10)	70 73	30-minutes of PT 1x/day on POD0 and 1–2x/day on POD1 included education, knee exercises, and ADL training.	2.7(1.9)*d 3.2(1.9)*d
Witcher 2015	Retrospective cohort	MV ≥24hr in Neuro- ICU	NP	68	31  Pre-EM 37  Post-EM	61 60	63 45	Early mobilization (EM) protocol for range of motion, bed-level activities, standing, and ambulation.	10* d 13* d
Karvelas 2017	Prospective cohort	New onset LBP	Mixed	4,723	628  Early PT 4,095  No early PT	73(6)	64	PT within 28 days of index primary care appointment.	≥1 visit
Fritz 2018	Retrospective cohort	New onset LBP	Naïve (90d)	707	91  PT 616  No PT	38(11)	67	PT within 30 days of index physician visit.	≥1 visit
Frogner 2018	Retrospective cohort	New onset LBP	Mixed	148,866	12,906 PT first 17,135 PT later 118,825 No PT	NP	64 55 53	Index visit was with a physical therapist (PT first), index visit was with another provider & PT received within year (PT later), or no PT received during the year (no PT).	≥1 visit
Horn 2018	Retrospective cohort	New onset neck pain	Mixed§	308	160  Early PT 74  Delayed PT 74  Late PT	48(14)	69	PT visit within 14 days after index visit for neck pain (early PT), within 15–90 days after index visit (delayed PT), or 91–365 days after index visit (late PT).	≥1 visit
Kazis 2019	Retrospective cohort	New onset LBP	Naïve (12m)	216,504	3,499  PT 192,961 Other HCP	48(15)	50	PT as index visit.	1 visit
Smith 2019	Retrospective cohort	Lumbar fusion	Mixed	230	123  Pre 96  Post	60(12) 61(13)	57 50	Pre- and post-implementation of Enhanced recovery after surgery (ERAS): preoperative education & postoperative PT on POD1.	3 d
Location of Physical Therapy
Tse 2017	Retrospective cohort	TKR	NP	28	14  Inpatient care 14  Rehab center	68 68	0 0	2x/day PT including assisted ambulation with a 2-wheeled walker.	4* d
Kim 2019	Retrospective cohort	Back & neck pain	NP	464	390  Usual ED 74  PT in the ED	48	58	PT consultation in the ED: activity progression education, body mechanics, HEP, and further PT referral.	1 visit

^*indicates a mean (standard deviation) provided;

^†pre-study narcotic use provided, not specific to opioid medication;

^§pre-study substance abuse, not specific to opioid medication; NP not provided; N number of participants; CBT cognitive-behavioral therapy; PT physical therapy: OT occupational therapy; LBP low back pain; TENS transcutaneous electrical nerve stimulation; THR total hip replacement; NMES neuromuscular electrical stimulation; BFA battlefield acupuncture; TKR total knee replacement; CPM continuous passive motion; POD postoperative day; HIV human immunodeficiency virus; Scope=arthroscopy; ED emergency department; HEP home exercise program; PM&R physical medicine & rehabilitation; DC discharge; prn=as needed; ADL activities of daily living; HCP health care provider; MV mechanical ventilation; RCT randomized controlled trial

Figure 3.

Study populations and topics

Included articles were described based on population (levels of the pyramid). These populations were categorized as patients with spine pain, chronic pain, joint pain, or as special populations. Arrows indicate what aspect of physical therapy was evaluated for each population (each level of the pyramid).

3.1. Quality Assessment of Included Studies

Fifteen (50%), ten (33%), and five (17%) of the included studies received a good, fair, or poor overall quality score, respectively, based on the previously described assessment of the Joanna Briggs Institute checklist evaluations. Individual quality determinants from the checklists were described in Table 2.

Table 2.

Quality Assessment

	True Randomization	Concealed allocation	Blinded participants	Blinded interventionist	Blinded assessor	Same treatment (other than test intervention)	Same outcome assessment	ITT	Appropriate design	Clear inclusion criteria	Groups similar at baseline	Exposure assigned similarly to groups	Detailed subject & setting descriptions	Valid & reliable exposure measurement	Objective measurement of the condition	Valid & reliable outcome assessment	Groups free of outcome at baseline	Sufficient follow-up	Complete follow-up (or explained)	Follow-up strategies	Identified confounders	Strategies for confounders	Appropriate statistics	Overall quality
Interdisciplinary Programs
Meineche-Schmidt 2012	-	-	-	-	-	-	-	-	-	-	U	Y	-	Y	-	N	N	Y	N	N	Y	N	Y	Fair
Vines 1996	-	-	-	-	-	-	-	-	-	-	NA	NA	-	Y	-	N	U	U	N	N	Y	N	N	Poor
Stein 2013	-	-	-	-	-	-	-	-	-	-	NA	NA	-	Y	-	Y	U	Y	N	N	N	N	Y	Fair
Murphy 2016	-	-	-	-	-	-	-	-	-	-	NA	NA	-	Y	-	U	N	Y	U	U	Y	Y	Y	Fair
Anamkath 2018	-	-	-	-	-	-	-	-	-	-	NA	NA	-	Y	-	Y	N	Y	N	N	Y	Y	Y	Fair
Bruce 2017 b	-	-	-	-	-	-	-	-	-	-	NA	NA	-	Y	-	N	N	Y	Y	N	N	N	N	Poor
Bruce 2017 a	-	-	-	-	-	-	-	-	-	-	NA	NA	-	Y	-	N	N	Y	Y	N	N	N	N	Poor
Crisostomo 2008	-	-	-	-	-	-	-	-	-	-	NA	NA	-	Y	-	U	N	Y	N	N	Y	Y	N	Fair
Modalities
Chabal 1998	-	-	-	-	-	-	-	-	-	N	-	-	U	Y	N	N	-	-	-	-	Y	N	Y	Poor
Castellano 2016	Y	U	Y	Y	Y	U	Y	Y	Y	-	Y	-	-	-	-	Y	-	-	N	-	-	-	Y	Good
Collinsworth 2019	Y	Y	N	N	U	Y	Y	Y	Y	-	Y	-	-	-	-	Y	-	-	N	-	-	-	Y	Fair
Physical Therapy Treatment
Pullen 2017	-	-	-	-	-	-	-	-	-	-	NA	NA	-	Y	-	N	N	N	Y	NA	N	N	N	Poor
Pullen 2019	-	-	-	-	-	-	-	-	-	-	N	U	-	Y	-	Y	U	N	Y	NA	Y	N	N	Fair
Thackeray 2017	-	-	-	-	-	-	-	-	-	-	N	Y	-	Y	-	Y	U	Y	Y	NA	Y	Y	Y	Good
Fritz 2012	-	-	-	-	-	-	-	-	-	-	Y	Y	-	Y	-	Y	U	Y	Y	NA	Y	Y	Y	Good
Childs 2015	-	-	-	-	-	-	-	-	-	-	N	Y	-	Y	-	Y	U	Y	Y	NA	Y	Y	Y	Good
Lentz 2018	-	-	-	-	-	-	-	-	-	-	NA	NA	-	Y	-	N	U	Y	N	Y	Y	Y	Y	Fair
Kwok 2019	-	-	-	-	-	-	-	-	-	-	N	Y	-	Y	-	Y	U	Y	Y	NA	Y	Y	Y	Good
Timing of Physical Therapy
Sun 2018	-	-	-	-	-	-	-	-	-	-	N	Y	-	Y	-	Y	Y	Y	Y	NA	Y	Y	Y	Good
Rhon 2018	-	-	-	-	-	-	-	-	-	-	N	Y	-	Y	-	Y	U	Y	Y	NA	Y	Y	Y	Good
Sarpong 2019	-	-	-	-	-	-	-	-	-	-	Y	Y	-	Y	-	Y	Y	Y	Y	NA	Y	Y	Y	Good
Witcher 2015	-	-	-	-	-	-	-	-	-	-	N	Y	-	Y	-	Y	U	N	Y	NA	Y	N	Y	Fair
Karvelas 2017	-	-	-	-	-	-	-	-	-	-	Y	Y	-	Y	-	Y	U	Y	Y	NA	Y	Y	Y	Good
Fritz 2018	-	-	-	-	-	-	-	-	-	-	Y	Y	-	Y	-	Y	Y	Y	Y	NA	Y	Y	Y	Good
Frogner 2018	-	-	-	-	-	-	-	-	-	-	N	Y	-	N	-	Y	U	Y	U	U	Y	Y	Y	Fair
Horn 2018	-	-	-	-	-	-	-	-	-	-	Y	Y	-	Y	-	Y	U	Y	Y	NA	Y	Y	Y	Good
Kazis 2019	-	-	-	-	-	-	-	-	-	-	Y	Y	-	Y	-	Y	Y	Y	Y	NA	Y	Y	Y	Good
Smith 2019	-	-	-	-	-	-	-	-	-	-	Y	Y	-	Y	-	Y	U	Y	Y	NA	Y	NA	Y	Good
Location of Physical Therapy
Tse 2017	-	-	-	-	-	-	-	-	-	-	Y	Y	-	Y	-	Y	U	Y	Y	NA	Y	Y	Y	Good
Kim 2019	-	-	-	-	-	-	-	-	-	-	N	N	-	Y	-	Y	U	Y	Y	NA	Y	Y	Y	Good

3.2. Interdisciplinary Programs that Include PT

Eight studies investigated the relationship between interdisciplinary pain programs and opioid use. All studies reported inter- or multi-disciplinary interventions including at least PT and psychological consultation (most often, cognitive behavioral therapy). Specific disciplines included in these pain management programs were previously detailed (Table 1). All eight studies, of fair or poor quality, reported at least part of their sample had a history of opioid use (Table 1). Six of these studies evaluated adult patients with chronic pain using either prospective^58–60 or retrospective methods,^37,38,56 including Crisostomo and colleagues (2008) who specifically evaluated adult patients with chronic low back pain.⁵⁶ Four studies with adult patients evaluated opioid use from admission to discharge of an interdisciplinary pain program, two of which reported a significant reduction in the percent of participants using opioids ranging from 24 to 65% reduction for various subgroups within the studies.^37,56 Whereas, Meineche-Schmidt et al. and Anamkath et al. found no significant reductions in the percent of participants using opioids at discharge from an interdisciplinary pain program (Table 3).^38,58 Both Meineche-Schmidt et al. and Anamkath et al. reported significant reductions in the opioid doses used from admission to discharge (102→83mg/day and 69→62mg/day, respectively); however, the mean daily morphine milligram equivalent (MME) doses at discharge were above clinically-established safe dose levels.⁶⁷ The relationship between PT, as part of an interdisciplinary program, and immediate post-treatment opioid use is unclear based on these adults studies.

Table 3.

Study results

Study	Intervention & population	Opioid measure*	Groups	Baseline Value	Follow-up value	Follow-up
Interdisciplinary programs
Meineche- Schmidt 2012	Chronic pain program	Self-reported use (%)	---	26%	22% 25%	Discharge 12m
Vines 1996	Chronic pain program	Self-reported use (%)	---	100%	30%	4w
Stein 2013	Chronic pain program	MME ┼	---	1828	1382	12m
Murphy 2016	Chronic pain program	Record-indicated use (%)	Females v Males Females v Males	24 v 38%	0 v 0% 18 v 17%	Discharge 3m
Anamkath 2018	Chronic pain program	Record-indicated use (%)	---	74%	71%	Discharge
Crisostomo 2008	Chronic LBP program	Record-indicated use (%)	Fusion Non-fusion Non-operative	65% 70% 48%	18% 5% 10%	Discharge
Bruce 2017 b	Pediatric chronic pain program	Self-reported use (%)	---	16%	2%	Discharge
Bruce 2017 a	Pediatric chronic pain program	Self-reported daily use (%)	---	13%	0% 0%	Discharge 3m
Modalities
Chabal 1998	TENS for chronic pain	# medications reported by sample	---	206	101	6m
Castellano 2016	NMES after THR	Days on medication	NMES v control	---	3.13 v 3.36	6w
Collinsworth 2019	BFA after shoulder surgery	Group differences in self- reported daily # of pills	BFA+usual v Usual care	---	0.29 3.74	24hr 6w
Physical Therapy Treatment
Lentz 2018	PT for joint pain	Self-reported use (%)	---	---	18%	12m
Pullen 2017	PT for chronic pain in persons with HIV	Record-indicated use (%)	---	24%	24%	Discharge
Pullen 2019	PT for pain in persons with HIV	% increase in use	PT No PT	---	0% 9%	Discharge 12–20w
Kwok 2019	PT for injured workers	Odds of ≥1Rx ≥90d post-injury‡	PT utilization: Very high High Medium Low		2.71(2.28,3.23) 0.80(0.67,0.94) 0.2 (0.21,0.32) 0.20(0.14,0.27)	210(346)d
Fritz 2012	PT for LBP, adherent or nonadherent to CPG	Record-indicated use (%)	Early v delayed Adherent v non	---	49 v 55% 49 v 53%	18m
Childs 2015	PT for LBP, adherent or nonadherent to CPG	Record-indicated use (%)	Early v delayed Adherent v non	34 v 34%§ 35 v 36%§	59 v 70% 65 v 66%	2y
Thackeray 2017	PT referral use and participation for LBP	Odds of opioid Rx‡			0.65(0.43,1.00) 0.47(0.24,0.92)	1y
Timing of Physical Therapy
Sun 2018	PT visit within 90 days for joint pain	Odds of ≥1 opioid Rx‡ after 90 days	Shoulder pain Neck pain Knee pain LBP	---	0.85(0.77,0.95) 0.92(0.85,0.99) 0.84(0.77,0.91) 0.93(0.88,0.98)	9m
Rhon 2018	PT v opioid use after hip arthroscopy	Receipt of 3+ opioid Rx (%)	PT-first Opioid-first		50% 65%	2y
Sarpong 2019	Inpatient PT on POD0 or 1 after TKR	MME / LOS ratio	PT on POD0 PT on POD1	---	60.5 (35.9) 68 (53.1)	Discharge
Witcher 2015	Early mobilization (EM) in neurologic ICU	Daily fentanyl equivalent (μg)	Pre- EM Post- EM	---	50 173	Discharge
Karvelas 2017	PT visit within 28 days for back pain	Odds of opioid Rx		---	1.13(0.90,1.43)	1yr
Fritz 2018	PT visit within 30 days for back pain	Odds of ≥120 d supply or >90 days & ≥10 fills	---	---	0.44(0.22,0.89)	1yr
Frogner 2018	PT visit as index or follow-up in 1st year of new onset LBP	Record-indicated use (%)	PT first PT later No PT	---	20% 31% 25%	1yr
Horn 2018	Early, delayed, or late PT after index visit for neck pain	Record-indicated use (%)	Early PT Delayed PT Late PT	---	55% 59% 82%	1yr
Kazis 2019	PT visit as index visit for LBP	Odds of Rx ≤60 days from index & ≥120 d supply -or- ≥90 d supply & ≥10 Rxs	---	---	0.27(0.15,0.48)	1yr
Smith 2019	ERAS for lumbar fusion	Record-indicated long-acting use (%)	Pre-ERAS Post-ERAS	3% 2%	14% 5%	1st follow-up (≤30d)
Location of Physical Therapy
Tse 2017	Ambulation-focused PT after TKR	Daily MME	Inpatient PT Rehab center PT	---	38 53	2d
Kim 2019	PT visit in the ED for neck or back pain	Opioid Rx (%)	Usual v PT in ED	---	41 v 50%	Discharge

% indicates the outcome measure was the percent of participants with the study-specific outcome measure. Data are presented as mean (standard deviation) unless otherwise noted. Percentages were reduced to integer values.

^*opioid use reported as % of the sample was prioritized for this table, studies may have reported additional measures of opioid use as described in the text.

^┼mean MME provided for the 12 months prior to program initiation (baseline) and 12 months after program completion.

^‡odds ratio and 95% confidence interval provided. ¤ odds ratio and 99% confidence interval provided.

^§opioid use in the 12mo prior to index visit. LBP low back pain.

^*opioid dose equivalency conversion was study-specific, not standard to MME; MME morphine milligram equivalent (mg); EHR electronic health record; TENS transcutaneous electrical nerve stimulation; NMES neuromuscular electrical stimulation; THR total hip replacement; BFA battlefield acupuncture; CPM continuous passive motion; TKR total knee replacement; LOS length of stay; PT physical therapy; HIV human immunodeficiency virus; Rx prescription; CPG clinical practice guideline; POD postoperative day; ICU intensive care unit; ED emergency department; ERAS early recovery after surgery

Four studies with adult patients provided follow-up beyond discharge, ranging from four weeks to one year.^37,58–60 The percent of participants who reported opioid use was lower from admission to 4-week follow-up in the study by Vines and colleagues (100% to 30%),⁵⁹ but this reduced use was not observed by Meineche-Scmidt and colleagues (26% to 25% at 12mo).⁵⁸ In contrast, Murphy and colleagues. identified an increase in the percent of participants using opioids from discharge (0%) to 3-month follow-up (17%).³⁷ A significant, but small reduction in daily opioid dose was identified from program discharge (83±125 MME) to 12-month follow-up (79±87 MME) for the 306 participants who underwent an individualized interdisciplinary pain program including PT.⁵⁸ Total MME in the year of treatment was also lower (1382 MME) than total MME in the year prior to treatment (1828 MME) for the 23 participants who completed inpatient multimodal pain rehabilitation.⁶⁰ Conflicting evidence, various timelines, and differences in reporting measures limit conclusions across these studies regarding the relationship between PT, as part of an interdisciplinary program, and subsequent opioid use.

The two prospective studies that reported on a 3-week outpatient pediatric interdisciplinary pain program were conducted at the same institution.^61,62 Participant-reported opioid use reduced to less than 5% of participants at program discharge in both studies. Follow-up was only provided in one of these pediatric studies, with 0% of participants reporting opioid use three months after the interdisciplinary program.⁶² Overall, this pediatric interdisciplinary pain program, which included PT, was associated with reduced opioid use.

3.3. PT Modalities

Three studies evaluated the impact of PT modalities on opioid use.^57,65,66 Two of these studies evaluated electrical stimulation, either in the form of transcutaneous (TENS)⁵⁷ or neuromuscular (NMES).⁶⁶ A retrospective survey of 376 participants identified a significant reduction in the sample’s total opioid prescriptions from 206 to 101 over six months after TENS use.⁵⁷ Castellano and colleagues studied NMES after total hip replacement via a randomized controlled trial and found no difference from subclinical NMES for days on opioid medication in the six weeks post-surgery⁶⁶; pre-operative use of opioid medication was not detailed. No reports were identified that investigated the relationship between dry needling and opioid use; however, one study considered the effect of battlefield acupuncture on opioid use after shoulder surgery. Battlefield acupuncture is a auricular-acupuncture for rapid pain-relief commonly used in the military.^68,69 No significant group differences were identified at any time in the six weeks post-surgery between participants randomized to standard post-operative PT and those randomized to additional battlefield acupuncture.⁶⁵ The impact of pre-operative opioid use was not reported for this study. This scoping review identified that minimal work has been done to evaluate the relationships between PT modalities and subsequent opioid use.

3.4. Effect, Content, and Use of PT

The relationship between opioid use and the effect of participation in PT was evaluated in three studies of participants with musculoskeletal joint pain^48,53,63; two of which were specific to persons with human immunodeficiency virus (HIV).^48,53 None of the these studies considered how pre-treatment opioid exposure may have impacted results, and each study was graded as poor or fair quality. Almost 250 participants with various musculoskeletal pain conditions were included in a prospective cohort that identified a significant increase in odds of self-reported opioid use one year post-treatment for participants who experienced an increase in pain over four weeks of PT.⁶³ Pullen and colleagues retrospectively evaluated PT for persons living with HIV who reported chronic pain; their study identified no reduction in the number of participants using opioids from pre- to post-PT treatment.⁵³ This same research group also examined persons living with HIV who engaged in PT for any musculoskeletal pain or deconditioning. This second retrospective study compared PT users to non-users and identified that opioid use did not increase for the 30 PT participants, but a 9% increase was observed for the 26 participants who did not participate in PT over the 20 week period.⁴⁸ While Lentz and colleagues identified that unsuccessful PT (increased pain after four weeks of PT) was associated with 1.7 greater odds of opioid use, Pullen and colleagues did not identify any significant relationships between PT participation and subsequent opioid use.

Physical therapy content for participants with low back pain was evaluated as adherent or non-adherent to clinical practice guidelines for two retrospective cohorts of good quality.^45,70 Participants who received adherent care were those who had at least one “active” billing code (therapeutic exercise or activity, self-care management, neuromuscular re-education or group therapy)⁷¹ each visit and 75% of their total codes were classified as active across their episode of care.^45,70 The percentage of participants with opioid use at follow-up was similar across groups in both studies. Fritz and colleagues identified 49% and 53% of participants with adherent and non-adherent PT, respectively, used opioids at 18 months,⁷⁰ whereas Childs and colleagues identified 65% and 66% of their respective samples used opioids at 24 months.⁴⁵ Both studies reported approximately one third of their adherent and non-adherent samples had history of opioid use before treatment. Participation in adherent PT for low back pain was not associated with a significant reduction in odds of receiving an opioid prescription in the two years following initial consultation for pain.⁴⁵ This finding was echoed in a retrospective analysis of 9,596 persons with workplace injury that identified no significant reduction in the odds of receiving a long-term opioid prescription after participating in active PT in the 90 days post-injury.⁵⁵ In summary, guideline-adherent PT care for low back pain did not have a relationship with opioid use or prescription in the 6–24 months post-treatment.

Two retrospective studies analyzed the impact of PT utilization on opioid use; these studies evaluated PT referrals and use of those referrals rather than investigating the effect or duration of PT. Opioid prescriptions at initial physician consult for low back pain were slightly more prevalent for participants who did not receive a PT referral (28%) compared to participants who received a referral (25%).⁴⁶ Receipt of a referral and use of that referral were also associated with 0.47 lower odds of receiving an opioid prescription in the year following physician consult.⁴⁶ In addition to evaluating the relationship between PT content and opioid use, Kwok and colleagues also described the relationship between PT utilization and opioid use in 9,596 persons with an acute opioid prescription who sought PT after a workplace injury.⁵⁵ Participants who utilized more than 148 billed codes of PT during their injury claim had significantly higher odds (2.71) to receive an opioid prescription beyond 90 days from initial injury. In contrast, the odds to receive this long-term prescription were lower (0.20–0.80) for participants who had lower PT utilization (ranging from 1–147 billed codes).⁵⁵

3.5. Timing of PT

Thirteen studies included in this scoping review evaluated the relationship between opioid use and timing of PT from index provider visit for workplace injury,⁵⁵ joint pain,⁴⁹ neck or back pain^{39–41,43,45,64,70}; from first mobilization in an intensive care unit⁵⁴; or from spine,⁴⁷ hip,⁵¹ or knee⁵² surgery. The previously stated relationship between very high utilization (≥148 billed codes) of PT for a workplace injury and increased risk of opioid use was upheld when only considering PT use in the 90 days post-injury; however, the relationship between early PT and reduced opioid use was not upheld in the low-high utilization (1–147 billed codes) groups.⁵⁵

Receipt of early PT was associated with significantly reduced odds, ranging from 0.15 to 0.84 lower odds of opioid prescription in five of six studies that evaluated odds of opioid use after PT for low back pain^{39,43,45,49,64,70}; two of these studies were specific to opioid naïve patients.^39,43 These six studies were all good quality based on the critical appraisal (Table 1). Definition of early PT varied from PT as the index visit (i.e., a physical therapist was the initial healthcare provider) to PT evaluation 14–90 days from index visit with a primary care provider. Odds of opioid use were evaluated in follow-up time frames ranging from 9–24 months. Sun and colleagues also identified a reduction in the odds of opioid prescription after PT within 90 days from index for persons with musculoskeletal pain including shoulder, neck, and knee pain.⁴⁹ Furthermore, Horn and colleagues found lower percentages of opioid use among participants who received PT either within 14 days (55%) or within 90 days (59%), compared to participants who received PT after 90 days from index visit (82%) for neck pain.⁴¹ Karvelas and colleagues found opposing results to these other studies. Receipt of PT within 28 days for back pain was not associated with a significant change in odds of opioid prescription within the following year; however, participants who received early PT had lower total days’ supply of opioids (38) compared to participants who did not receive early PT (52;).⁶⁴ In summary, the majority of research that evaluated relationships between the first occurrence of PT in the episode of care and subsequent opioid use has focused on spine pain and has demonstrated reduced opioid use was associated with early PT care.

Early mobilization with PT in the hospital-setting was evaluated for participants in the neurological intensive care unit,⁵⁴ opioid naïve participants who received total knee replacement,⁵² and participants after lumbar fusion.⁴⁷ Sarpong et al. and Witcher et al. both evaluated daily opioid dose for medications during the participants’ hospital stays. Sarpong et al. found lower opioid dose on postoperative day 1 (60 vs 83 MME) for participants who received PT on postoperative day 0 compared to postoperative day 1 after total knee replacement.⁵² Witcher et al. found higher opioid doses (173 vs 50μg fentanyl equivalent) for participants who received early mobilization in the neurological intensive care unit.⁵⁴ Smith and colleagues did not report opioid dose, but did report the percentage of patients using both short- and long-acting opioids after lumbar fusion. The 96 participants who received PT on postoperative day 1 had a lower percentage of participants who were using long-acting opioids at first physician follow-up compared to the 123 participants who were treated with standard care before the enhanced recovery protocol (5 vs 15%).⁴⁷ The early PT as part of the enhanced recovery did not impact use of short-acting opioids; approximately 80% of both groups were using these medications at first physician follow-up.⁴⁷ These findings from good quality studies highlight the mixed results regarding early PT in inpatient settings for a variety of conditions.

3.6. Location of PT

Physical therapy services can be accomplished in a variety of settings. Two retrospective studies in this review evaluated the impact of location on opioid use. Neither of these studies reported the inclusion of previous exposure to opioids in their analyses. Tse and colleagues considered opioid dose after total knee replacement for participants who received inpatient PT at the acute surgical ward in comparison to participants who were transferred to inpatient PT in a subacute rehabilitation center.⁵⁰ The PT content was consistent across locations. No significant group differences were identified in this case-controlled study for opioid dose at postoperative day 1 (49 vs 56 MME) or day 2 (38 vs 53 MME).⁵⁰ Kim and colleagues compared opioid prescriptions in the emergency department for 74 participants with back and neck pain who received a PT consultation to 390 participants who did not receive this consultation.⁴² There was no significant difference in receiving an opioid prescription at discharge between participants who consulted with a physical therapist in the emergency department (50%) compared to participants who received standard emergency care (41%).⁴² Although these studies evaluated different populations and interventions, both identified no relationship between PT location and post-treatment opioid use.

4. Discussion

The purpose of this scoping review was to synthesize available evidence and examine relationships between various PT interventions and opioid use within different patient populations. This review identified 30 studies that evaluated relationships between PT and opioid use. Seven characteristics of PT were evaluated across these 30 studies: interdisciplinary care, modalities, treatment, content, location, utilization, and timing. Eight studies of interdisciplinary care including PT reported mixed results regarding reduced opioid use, with limited follow-up. Three different modalities were evaluated across three studies with no strong evidence to suggest reduced opioid use after PT modalities. General PT treatment for patients with HIV and musculoskeletal pain was not associated with reduced opioid use in the two studies that considered this specific population. The relationship between opioid use and active versus passive PT content was evaluated in three studies, but no significant relationships were reported. Two studies evaluated the relationship between PT location and opioid use for separate populations and reported no significant reductions in use. Importantly, 14 studies evaluated the impact of timing and utilization of PT on long-term opioid use. The majority of these studies investigated this relationship in samples of patients with spine pain. This group of studies presented the most consistent evidence that utilization of PT (especially early in the episode of care) was related to reduced odds that patients would use opioids. Overall, these results provide a summary of the limited, and heterogenous, work that has been done to evaluate PT’s potential role in multidisciplinary non-pharmacologic pain management.

Interdisciplinary pain programs that include PT are a treatment of interest for persons with chronic pain, especially low back pain. Conflicting evidence exists regarding the effectiveness of these interdisciplinary programs to improve function (physical and mental) more than standard physical therapy.^26,72–75 These interdisciplinary programs provide individualized treatment, which may partially explain the mixed results in terms of functional and medication outcomes. Some interdisciplinary programs can be intensive programs, which are increasing in popularity for pediatric treatment of pain.⁷⁶ These intensive interdisciplinary rehabilitation programs included in this review accomplished cessation of opioid use by program discharge^61,62; however, due to the setting and nature of an intensive program opioid weaning can easily be reinforced as a requirement and influenced by the reduced autonomy of their pediatric participants.

This scoping review encompassed a wide variety of PT interventions with unspecific prescriptions, for differing patient populations, with numerous study designs. Cumulatively this made interpretation of relationships between PT and opioid use difficult. For example, Pullen and colleagues (2017; 2019) did not provide detailed description of PT content, frequency, or duration when speaking to the relationship between a PT episode of care and subsequent opioid use. Furthermore, many included studies only evaluated whether participants attended PT for at least one session,^{39–43,46,49} without detailing the duration or frequency of follow-up PT care. The many study designs complicated comparisons of study quality and results. The retrospective cohorts could easily identify if a patient received PT and compare to groups from the same sample population who did not receive PT. These cohorts often benefited from a large sample size to evaluate the relationship between PT treatment and subsequent opioid use; however, due to their design they cannot speak to any causal relationships between PT and opioids. The generalizability of the collective findings in this scoping review is limited by the broad inclusion of PT interventions, patient populations, and study designs.

History of substance or opioid abuse is a known predictor of future opioid use and abuse.^77,78 Only four studies in this scoping review reported on participants who were opioid naïve. Four studies controlled for previous opioid or substance use^41,46,49,51; however, the remaining studies did not provide details regarding past opioid use, or evaluated a mixed sample without controlling for this. While this review evaluated relationships between PT intervention and subsequent opioid use, one study reported on the reverse relationship. Previous history of opioid prescription was related to use of PT referral; more participants who used their referral had not acutely used opioid medication for their back pain.⁴⁶ Receipt of an opioid prescription may inhibit some participants from pursuing more active intervention such as PT.

4.1. Limitations

This scoping review provided a broad-reaching summary of relationships explored between PT and opioid use. This review only provided a synthesis and did not include meta-analysis due to wide variety of study designs, PT characteristics, and opioid-use metrics. The protocol for this scoping review was distributed to co-authors prior to study selection but was not published publicly; however, it has been provided as an appendix. The quality of included studies was assessed using the appropriate checklist from the Joanna Briggs Institute critical appraisal tools; however, 50% of studies had poor to fair quality overall. Many studies did not account for opioid status at baseline or pre-treatment, employ strategies to deal with loss to follow-up, or use validated and reliable outcome measures for opioid use. The heterogeneity of study designs made comparisons across studies difficult, and notably the majority of studies (20 of 30) were retrospective and cannot infer causality. Specifically, the studies of interdisciplinary programs including PT do not represent any direct relationship between PT treatment and opioid use, but were included due to the increasing popularity of interdisciplinary pain treatments. Lastly, while PT modalities were included in this review as a form of PT treatment, they were not individually included as search terms. This review may exclude studies of PT modalities and subsequent opioid use that do not include PT terminology in the title or abstract.

5. Conclusions

Early PT utilization (especially for LBP) appears to be associated with a reduced odds of follow-up opioid use. However, there is limited and inconclusive evidence to establish whether the content and/or location of PT interventions improves outcomes due to heterogeneity between studies. Future work should consider details of interdisciplinary programs that include PT, additional early mobilization protocols for other post-operative populations, and patient subgroups who may benefit most from PT to reduce risk of opioid abuse.