Musculoskeletal disorders (MSDs) are injuries or disorders affecting the body movement or the musculoskeletal system ^[1]. MSDs are the second most common cause of disability worldwide ^[2], with a prevalence comparable to the sum of all cardiovascular and chronic respiratory diseases. MSDs resulted in an economic burden of more than USD 800 billion (US) in 2015 ^[3]. Due to the significant impact of MSDs, it is essential to consider value-based care and work towards examining new approaches to manage MSDs more efficiently. Direct access to physiotherapy (DAPT) occurs when a patient can self-refer to a physiotherapist without having to see another health professional for a medical prescription ^[4]. Preliminary evidence suggests that DAPT could offer a promising option compared to other traditional models of care, particularly the physician-led model of care [5,6,7,8,9]^{[5][6][7][8][9]}. The effectiveness of DAPT has been reported by authors in several areas: reduction in direct and indirect costs for the patient and the national health system ^[6], reduction in work overload for general practitioners (GPs) ^[7], and improvement in health indicators for patients (e.g., health-related quality of life, quality-adjusted life years) ^[9]. World Physiotherapy itself, a global organization that represents the profession of physiotherapists at an international level, advocates for the growing responsibilities of the profession. This phase shift for the profession of physiotherapy is why clarifying the efficacy and safety of DAPT for patients with musculoskeletal disorders is needed [4,10,11,12,13,14]^{[4][10][11][12][13][14]}.

2. Population

A total of 32,742 patients underwent assessment and management through the DAPT care model, while 9900 patients underwent assessment and management through the physician-led care model, with a total number of 42,642 patients. Characteristics of the population are summarized in Table 31. Age, sex, and gender were not available for four studies [19,27,31,48]^{[15][16][17][18]}; seven studies did not provide additional details regarding the type and location of musculoskeletal disorder [19,27^{[15][16][17][19][20][21][22]},28,31,32,36,43], while seventeen studies did not specify pathology onset [9,18,19,26,27,28,29,31,32,36,37,38,39,40,41,45,48]^{[9][15][16][17][18][19][20][21][23][24][25][26][27][28][29][30][31]}.

Age (mean)	49 Years
Sex (%)	57% F–43% M
Type of MSDs (%)	Upper Limb 15% Spine 50% Lower Limb 25% Widespread/Mixed Pain 7% Rheumatological Disease 3%
Onset (%)	Acute (<6 weeks) 40% Sub-acute (6 weeks to 3 months) 25% Chronic (>3 months) 35%

All experimental interventions provided traditional DAPT, but two studies evaluated patients remotely by phone [34,36]^[21][32]. Characteristics and qualifications of the involved physiotherapists in DAPT are summarized in Table 42.

Type of Specialization	Nr. of Studies
Post-graduate specialization/ doctorate or musculoskeletal certification	2 studies
1 day out for training; direct access to primary care and mentoring	1 study
From 6 to 28 years of experience	5 studies
At least 3 years of experience in primary care + at least 1 orthopedic manipulative physiotherapist specialization	4 studies
Extended or advanced scope practitioner	11 studies
Degree of physiotherapy	2 studies
Specialization not specified	3 studies

Seven studies [16,19,26,29,39,40,41]^{[15][24][25][28][29][30][33]} did not include a control. Interventions for the control group were delivered by a physician in eight studies [27,28,31,34,43,45,46,48]^{[16][17][18][19][22][31][32][34]}; by a general practitioner in eight studies [9^{[9][20][21][26][27][35][36][37]},32,35,36,37,38,44,47], and by an orthopedic physician in five studies [17,18,30,33,42]^{[23][38][39][40][41]}. The different setting of care is reported in Figure 21: eight studies were performed in an outpatient setting [16,19^{[15][22][24][30][32][33][34][41]},26,34,41,42,43,46], one study in a military service setting [48]^[18], and twenty in a primary care hospital setting [8,9,17,18,27,28,29,30,31,32,33,35,36,37,38,39,40,44,45,47]^{[8][9][16][17][19][20][21][23][25][26][27][28][29][31][35][36][37][38][39][40]}.

3. DAPT Management Accuracy

Fourteen studies evaluated the triage proficiency of physiotherapists [16,18,19,26,29,31,33,36,40,41,42,46,47,49]^{[15][17][21][23][24][25][29][30][33][34][37][40][41][42]}. Most commonly, physiotherapists triage patients without additional medical consultation [16,18,19,26,31,36,40,41,42,46,47]^{[15][17][21][23][24][29][30][33][34][37][41]}. Three studies compared the surgical conversion rate expressed as a percentage (i.e., surgical conversion rate is considered a useful measure of appropriateness of referrals, as it is a measure of the percentage of patients that were referred to a physician and underwent surgery) [19,40,41]^[15][29][30]. Referral selection accuracy was used as a measure of management accuracy only in one study [33]^[40]. The rate of return to visit or rate of re-referral was evaluated in four studies [29,31,33,47]^{[17][25][37][40]}. The agreement between physiotherapy and medical evaluation was evaluated in three studies [18,31,49]^[17][23][42]. Management accuracy is summarized in Table 53, resulting in secondary care referral accuracy (15% average difference) in favor of DAPT. The return rate for further medical examination following physiotherapy discharge ranged between 0.9% and 9% (average 5.87%) from four studies without a control group [29,31,33,47]^{[17][25][37][40]}. Two studies estimated the agreement rate between physiotherapist and medical evaluation ranging between 74% and 87% (average 80.5%) [18,31]^[17][23]. Only one study evaluated the agreement of a chief radiologist toward physiotherapist prescription of radiographs for patients directly managed by physiotherapists [27]^[16].

4. Cost-Effectiveness

Sixteen studies evaluated cost-effectiveness between models using different outcomes [7,9,27,28,31,33,34,35,36,37,42,43,44,45,47,48]^{[7][9][16][17][18][19][21][22][26][31][32][35][36][37][40][41]}:

-

Use of health system resources, calculating the intervention costs: medication use and number of imaging referrals [9,27,31,33,34,35,37,42,45,46,47,48]^{[9][16][17][18][26][31][32][34][35][37][40][41]};

-

Cost sustained by the patient [27,34,46]^[16][32][34];

-

Patient savings [36,48]^[18][21];

-

Incremental Cost Effectiveness Ratio (ICER) [9,43,45]^[9][22][31];

-

Benefits estimated by the Quality-Adjusted Life Years (QALYs) [9,34,43,45]^{[9][22][31][32]};

-

Time needed to deliver the triage process [28]^[19];

-

Waiting time rate per visit [32,34]^[20][32];

-

Timeframe before discharge [28]^[19];

-

Number of visits needed for discharge [44]^[36].

Health system resources use was found to be the most widely used method among the included studies to estimate cost-effectiveness [9,27,31,33,34,35,37,42,45,46,47,48]^{[9][16][17][18][26][31][32][34][35][37][40][41]}. Intervention types commonly prescribed in the included studies are summarized in Table 64.

Table 64.

Type of treatment prescribed.

	DAPT Mean (Min–Max)	Medical Model Mean (Min–Max)	Mean Difference
Imaging	21% (0–63%)	49% (27–86%)	28%

]^[9][35]. A summary of compared wage costs and treatment options [42,45]^[31][41] are reported in Table 75.

Table 75.

Comparison between wage costs and treatment options and QALYs.

	DAPT Mean (Min–Max)	Medical Model Mean (Min–Max)	Mean Difference
Cost for episode of care	EUR 301.5 (255.55–628.24)
Medication	22.3% (8–50%)	63.5% (60–73.1%)	41.2%
Referral	9.3% (2.9–19.3%)	30% (14–40%)	20.7%

DAPT resulted in EUR 39.370 and EUR 62.867 of savings [9,27,35]^[9][16][35] in a timeframe ranging from 6 months [27]^[16] to 1 year [9,35

EUR 743.44 (498.38–988.51)

EUR 441.9

One study found no significant differences between the total savings for the healthcare system between models [34]^[32], while another one found no statistically significant difference across the average costs of professionals [46]^[34].

ICER, or the incremental cost-effectiveness ratio, is a synthetic measure that represents the economic value of an intervention, compared with an alternative intervention. Three studies have shown the ICER to be smaller for DAPT, meaning that physiotherapist-led management is less expensive and more effective than a physician-led model of care [9,43,45]^[9][22][31]. Two studies did not find statistically significant differences for direct costs sustained by the patient between care models [34^[32][34],46], while another study estimated a favorable cost for the patient who carries out DAPT equal to EUR 29.5 per visit compared to the medical model which had an expense of EUR 63.8, for a total cost saving to the patient equal to EUR 34.3 in favor of DAPT [27]^[16]. Two studies estimated between USD 36.42 and USD 129 saved per patient per episode of care in favor of DAPT [36^[18][21],48], with an average of USD 82.71 saved favoring DAPT. Mallett et al. also calculated an amount of £84,387.80 and £124,472.06 as the projection of the total savings over a year for a physiotherapist-led service, towards a general practitioner (GP)-led pathway, initiated by telephone contact from the patient, followed by a subsequent telephone triage appointment with a physiotherapist [36]^[21]. Notably, the DAPT pathway ensures an increase of 0.07 and 0.047 (average 0.05) of QALYs [9,34,43,45]^{[9][22][31][32]}. Although not statistically significant, Bornhöft et al. agree that DAPT has the potential to be a better cost-effective option (9). Regarding the rate of presence to visit (i.e., the effective presence to visit of a patient after a phone scheduled appointment) results [34,36]^[21][32] are summarized in Table 86.

Table 86.

Rate of presence to visit.

	DAPT Mean (Min–Max)	Medical Model Mean (Min–Max)	Mean Difference
Presence to visit	93.5% (90–97.1%)	87.5% (86–89%)	6%

Interestingly, the lack of missed appointments due to patient no-show would allow the health system to save between £84,387.80 and £124,472.06 in one year [36]^[21]. As for the outcome of total time needed to triage and to discharge the patient, results are summarized in Table 97.

Table 97. Time to triage/discharge the patient.

	DAPT Mean	Medical Model Mean	Mean Difference
Time to triage in minutes	108 min	148 min	40 min
Percentage of patients discharged within 4 h in primary care	93%	75%	18%
Number of sessions/days to discharge (Ankle MSDs)	5.6 sessions/ no difference	6.7 sessions/ no difference	1.1 sessions
Number of sessions/days to discharge (Knee MSDs)	6.3 sessions/ 49.7 days	9.1 sessions/ 60.2 days	2.8 sessions/ 10.5 days

5. Work Related Outcomes

The impact of DAPT on patient’s ability to work was evaluated in 10 studies [9,17,32,34,37,39,43,45,46,48]^{[9][18][20][22][26][28][31][32][34][38]} which focused on:

-

Self-administered questionnaire on patients’ work productivity [46]^[34] or questionnaire intended to measure self-efficacy in the workplace [43]^[22];

-

Percentage of patients prescribed sick days [37,45]^[26][31];

-

Number of sick days [9,17,45,48]^{[9][18][31][38]};

-

Labor participation measured on a 3-point scale (1 = did not return to work; 2 = returned to work with adaptations; or 3 = returned to work without adaptations) [39]^[28];

-

Time off work directly reported by the patients [32,34]^[20][32] or the amount of sick hour leaves ^[9].

No major difference was found for work productivity between the care models [46]^[34]. However, one study (without control group) mentioned better work performance favoring DAPT [43]^[22]. Samsson et al. found no significant differences in days lost from work due to MSDs between the DAPT group and the physician-referred medical model [17]^[38]. One longitudinal observational study without a control group and with a 10-year follow-up, evaluated the ability of the DAPT (without control group) on the restriction to work participation, showing that out of 423 patients visited through DAPT, 168 patients (39.7%) did not return to work, 123 patients (29.1%) returned to work with adjustments, and 132 (31.2%) returned to work without adjustments [39]^[28]. Finally, both models showed similar results regarding lost working days (21.27) and lost working hours [9,32,34]^[9][20][32]. The comparison of sick leave and number of days lost from work is summarized in Table 108 [9,37,45]^[9][26][31].

Table 108.

Percentage of sick leave prescriptions and number of sick leave days.

	DAPT Mean (Min–Max)	Medical Model Mean (Min–Max)	Mean Difference
Percentage of sick leave prescriptions	9% (3–15.1%)	12.16% (7.3–23.5%)	5%
Number of sick leave days prescribed	13.5 days (0–27 days)	50.5 days (26–75 days)	37 days

6. Patient Satisfaction

Ten studies evaluated the patient, using:

-

A 10- and 7-point Likert scale [7,42]^[7][41];

-

Satisfaction questionnaires [16,18,27,31,33,36,47]^{[16][17][21][23][33][37][40]};

-

Qualitative surveys [34]^[32].

Different satisfaction questionnaires were used:

-

A questionnaire related to the satisfaction in care received [47]^[37];

-

A modified and adapted questionnaire for assessing the quality of direct remote-access care (telephone) [36]^[21];

-

The Quality from the Patient’s Perspective Questionnaire (QPP) [33]^[40];

-

A patient satisfaction questionnaire and a physician satisfaction questionnaire related to how the physiotherapist performed the triage [18]^[23];

-

The Perceived Improvement Evaluation (PIVAS) questionnaire [50]^[43] and the Deyo and Diehl (DD) [51]^[44] questionnaire [16]^[33];

-

A questionnaire on patients’ experience of care [31]^[17];

-

A questionnaire on patient satisfaction/dissatisfaction with being referred to another professional or additional diagnostic investigations (e.g., X-ray) [27]^[16].

Comparison of the results was not suitable because of the heterogeneity of the outcome measures. Of the ten studies reporting patient satisfaction, only one study [46]^[34] did not report significant differences between DAPT and the medical model of care. Another study found that DAPT resulted in higher quality of the perceived treatment for the following [42]^[41]:

-

patient dissatisfaction with staff communication;

-

patient dissatisfaction with the quality of treatment received;

-

patient dissatisfaction with the facilities.

In addition, patients evaluated by a physiotherapist were more satisfied with the care received than those assessed or managed within the physician-led pathway [16,19,31,33,36,47]^{[15][17][21][33][37][40]}. Another study also evaluated the physician’s satisfaction regarding a physiotherapist-led service: reporting high level of satisfaction, with an average score of 1.9 on a scale ranging from 1 to 3 [18]^[23]. One study investigated patient satisfaction/dissatisfaction for additional imaging referral [27]^[16]. A total of 91% of patients reported being very satisfied with the referral and 84% reported being very happy with the feedback received from the physiotherapists about the need for further diagnostic investigation.

7. DAPT Safety

Five studies [27,32,34,38,46]^{[16][20][27][32][34]} investigated DAPT safety by the number of adverse events (i.e., unexpected events that occur following an intervention without evidence of causality). As an example, an increase in pain after a physiotherapist intervention occurred. Only one study reported the occurred severity of the adverse events that occurred by categorizing the event as none, mild, moderate, or severe [46]^[34]. DAPT safety is reported in Table 119.

Table 119.

DAPT safety.

Authors	Evaluation of DAPT Safety	Results
Peterson et al., 2021 [27][16]	Number of adverse events	none
Bishop et al., 2017 [32][20]	Number of adverse events	none
Salisbury et al., 2013 [34][32]	Number of adverse events	none
Bornhöft et al., 2019 [38][27]	Number of adverse events	none
Ojha et al., 2020 [46][34]	Number and type of adverse events	4 mild adverse events: 2/77 DAPT group 2/73 medical group Two patients in the medical group had an accidental fall at home, and unclear diagnosis of ankle pain at one-year and two patients in the DAPT group had side effects from an emergency room medication, and unclear diagnosis of low back pain.

8. Health Outcomes

Health outcomes were investigated in 15 studies [16,17,28,32,34,35,36,38,39,42,43,45,46,47,48]^{[18][19][20][21][22][27][28][31][32][33][34][35][37][38][41]}.

-

Six studies [28,32,33,34,36,48]^{[18][19][20][21][32][40]} estimated waiting time for assessment/management of DAPT toward a physician-led model of care, of which five were estimated in days [32,33,34,36,48]^{[18][20][21][32][40]} and one was estimated in minutes [28]^[19]. Results of DAPT waiting times are summarized in Table 120;

Table 120.

DAPT waiting times.

	DAPT Mean (Min–Max)	Physician-Led Care Model Mean (Min–Max)	Mean Difference
Waiting times in days	12.31 (3.55–26 days)	35.59 (28–57 days)	23.28 days
Primary care waiting time (min)	n.a.	n.a.	31 min

-

Three studies evaluated health outcomes using district-specific questionnaires [39,42^[22][28][41],43], but there was a large heterogeneity of questionnaires used between studies;

-

Quality of life was assessed in 13 studies. The most-often used tool was the EuroQoL (EQ5D) [16^{[21][22][27][31][32][33][37][38][41]},17,34,36,38,42,43,45,47], followed by the Short Form Health Survey 36 (SF-36) [16^[33][41],42], SF-12 [43]^[22], and SF-10 [46]^[34];

-

Perception of disability was evaluated in four studies through the Pain Disability Index (PDI) [17]^[38], the Disease Repercussions Profile [42]^[41], or the Measure Yourself Medical Outcomes Profile and global improvement [34]^[32]. Koojiman et al. ^[35] analyzed and compared the percentage of patients who achieved the expected outcomes between patients who underwent DAPT and those who went to the physiotherapist following medical referral;

-

One study [38]^[27] assessed functional disability in ADLs with the Disability Rating Index (DRI) and the patient’s attitudes towards their musculoskeletal disorder through the Attitude Responsibility for Musculoskeletal disorders scale (ARM);

-

Oostendorp et al. [39]^[28] evaluated the patient’s coping through the Pain Coping Inventory (PCI) and general health with the Global Perceived Effect (GPE);

-

Two studies [38,42]^[27][41] assessed psychological health (35,39) through the Hospital Anxiety and Depression Scale;

-

Five studies measured pain, three using the VAS (Visual Analogic Scale) [16^[28][33][41],39,42], one the Numeric Pain Rating Scale (NPRS) [38]^[27], and another the 10-point Likert scale [46]^[34];

-

Risk of chronicity of musculoskeletal pathology was carried out by two studies [38,43]^[22][27] through the Örebro Musculoskeletal Pain Screening Questionnaire (ÖMPSQ);

-

Pain-related catastrophization of the patient was assessed by two studies [43,46]^[22][34] using the Pain Catastrophizing Scale (PCS);

-

Two studies [39,43]^[22][28] evaluated avoidance behavior using the Fear Avoidance Belief Questionnaire (FABQ);

-

Two studies [42,46]^[34][41] used the Pain Self Efficacy Questionnaire (PSEQ) to assess patient self-efficacy;

-

One study [46]^[34] used the Patient-Specific Functional-Scale (PSFS) for physical function and the Patient Acceptability Symptom State (PASS) to measure acceptability of symptoms.

No significant differences between DAPT and the medical group were seen for all health outcomes except for quality-of-life assessment that showed contradictory evidence. Concerning quality of life, three studies reported the superiority of DAPT [33,34,38]^[27][32][40] and three studies noted no significant statistical difference [36,42,45]^[21][31][41].

9. Risk of Bias Assessment

Overall, three studies were judged as demonstrating “low risk of bias” [32^[20][31][34],45,46], six studies were rated as “some concern” [17^{[27][32][38][40][41]},33,34,38,42], and one as “high risk” [30]^[39]. The Quality of RCT methodological evidence is summarized in Figure 32.

Figure 32.

Rob-2 Score evaluation.

Even if the NOS scale does not allow for the estimation of a final score, each star was considered as a point to generate a score. Table 131 and Table 142 summarizes the NOS evaluation. For observational studies with a control group [18^{[18][19][21][22][23][26][35][36][37]},28,35,36,37,43,44,47,48], the mean score was 6/9 points, with a minimum score of 4 and a maximum score of 9 points, while for observational studies without a control group [19,26,27,29,31,40,41,48]^{[15][16][17][18][24][25][29][30]} a mean score of 4.75/9 was reported.

Table 113.

Newcastle Ottawa Scale evaluation for case-control studies.

Reference	Selection of Case and Controls				Comparability of Cases and Controls	Exposure			Total
	ITEM#1 Is the case definition adequate	ITEM#2 Representativeness of the cases	ITEM#3 Selection of Controls	ITEM#4 Definition of Controls Not Present at Start	ITEM#5 Comparability of cases and controls on the basis of the design or analysis	ITEM#6 Ascertainment of exposure	ITEM#7 Same method of ascertainment for cases and controls	ITEM#8 Non-Response Rate
Bird et al., 2016 [28][19]	✸		✸	✸	✸	✸	✸	✸	8/9
Bornhöft et al., 2015 [37][26]		✸		✸	✸		✸		5/9
Kooijman et al., 2013 [35]	✸	✸	✸	✸		✸	✸		6/9
Lankhorst et al., 2017 [44][36]	✸	✸	✸	✸		✸	✸	✸	7/9
Ludvigsson et al., 2012 [47][37]	✸	✸	✸	✸		✸	✸	✸	7/9
Mallet et al., 2014 [36][21]	✸	✸	✸	✸		✸	✸	✸	7/9
Oldmeadow et al., 2007 [18][23]	✸	✸	✸	✸	✸		✸	✸	7/9
Phillips et al., 2012 [43][22]	✸		✸	✸	✸				4/9
Szymanek et al., 2022 [48][18]	✸	✸	✸	✸	✸	✸	✸	✸	8/9

Table 142.

Newcastle Ottawa Scale evaluation for cohort studies.

Reference	Selection of Cohorts				Comparability of Cohorts	Outcome			Total
	ITEM#1 Representativeness of the Exposed Cohort	ITEM#2 Selection of the Non-Exposed	ITEM#3 Ascertainment of Exposure	ITEM#4 Demonstration that outcome of interest was not present at start of study	ITEM#5 Comparability of cohorts on the basis of the design or analysis	ITEM#6 Assessment of Outcome	ITEM#7 Was follow up long enough for outcomes to occur	ITEM#8 Adequacy of Follow Up of Cohorts
Caffrey et al., 2019 [26][24]	✸		✸	✸			✸	✸	5/9
Chang et al., 2018 [29][25]	✸		✸	✸				✸	4/9
Downie et al., 2019 [31][17]	✸		✸	✸			✸	✸	5/9
Kerridge-Weeks et al., 2016 [41][30]	✸		✸	✸		✸	✸	✸	6/9
Lyons et al., 2022 [19][15]	✸		✸	✸			✸	✸	4/9
O’Farrell et al., 2014 [40][29]	✸		✸	✸		✸	✸		5/9
Peterson et al., 2021 [27][16]	✸		✸	✸			✸	✸	5/9
Szymanek et al., 2022 [48][18]	✸		✸	✸			✸		4/9