This mixed-methods scoping review was conducted to address the following questions: What are the metrologic properties of the devices used to quantify force-time characteristics during SMT and MOB and which factors potentially facilitate or limit their use within the context of research, education and clinical practice?

The search strategy was developed with the assistance of a librarian with expertise in health sciences from the Université du Québec à Trois-Rivières. The searches in databases were conducted from inception to June 30, 2021. The following databases were searched: Index to Chiropractic literature (ICL), Cochrane Central Library, Cumulative Index to Nursing and Allied Health Literature (CINAHL) and Medline. The following terms (MESH or non-MESH) were searched in combination: [(device^* OR tool^* OR instrument^* OR manikin^* OR mannequin^* OR simulat^*) AND (mobilization^* OR mobilization^* OR manipulation^* OR “manual therapy”) AND (spinal OR musculoskeletal OR osteopath^* OR physiotherap^* OR chiropract^* OR lumba^* OR back^* OR cervical^* OR thora^* OR neck)] AND [(acceptability OR appreciation OR evaluation OR view^* OR attitude^* OR opinion^* OR acceptance OR assessment OR “clinical application^*” OR learning OR feedback OR “motor skill^*” OR “motor learning” OR education OR research OR teaching^* OR training) OR (availability OR force^* OR fidelity OR accuracy OR consistency OR precision OR repeatability OR reproducibility OR metrology OR reliability OR validity OR price OR cost OR usability)]. The search strategy was first developed for Medline (see OSF registry) and subsequently adapted to the other databases. References from relevant studies and from identified reviews were manually searched. Results from databases searches were first imported into Endnote (EndNote X9.3.3, Clarivate™) to remove duplicates. References were then imported into the Covidence software to manage the reviews (Covidence systematic review software, Veritas Health Innovation, Melbourne, Australia).

To be included, the studies had to: (1) be published in a peer-reviewed journal, (2) be written in either French or English, (3) have used, or present a device measuring SMT and/or MOB force-time characteristics on humans or with a potential use on humans, and (4) report any metrologic property of the device and/or discuss factors facilitating or limiting the use of devices within the context of research, education or clinical practice. Included studies had to present an observational, laboratory or experimental design. Technical or validation reports were also included. Reviews were initially included in case they identified additional facilitating or limiting factors. Studies were excluded if the modality was applied to non-spinal locations (i.e., upper or lower limbs or skull) or if it focused on other manual therapy techniques such as distraction, soft tissue manipulation or patient self-manipulation. Studies not conducted on humans were included if the device was intended for future use on humans (e.g., technical reports). The inclusion and exclusion criteria are detailed in Table 1.

Following the removal of duplicates, a two-phase screening process was conducted. First, title and abstracts from all studies were screened by two independent reviewers (M-A.M. and P.R.) and classified as relevant, possibly relevant or irrelevant. Next, the full texts of relevant and possibly relevant studies were independently screened by the same two reviewers to identify the final studies to be included in this scoping review. If a study was excluded only by one reviewer during phase I or II, disagreement was resolved by discussion between the reviewers. If no consensus was reached, a third reviewer (M.D. or I.P.) was involved.

The following data were extracted from the included studies: authors and years of publication, objectives, study design, measurement interface (i.e., clinician-patient or patient-table), device description and whether it was the same device used in another study, metrologic properties, and factors facilitating/limiting the use of the device. All data were extracted using a standardized form by M.A.M. and reviewed by I.P. to minimize potential errors.

A descriptive numerical analysis was first conducted to provide an overview of the results. Metrologic properties for each device were then classified by type, and data were extracted whenever available. Exact sentences regarding facilitating or limiting factors were extracted when available. One reviewer (M.A.M.) coded each sentence and defined thematic categories. Another reviewer (I.P.) initially reviewed the codes and themes. The two investigators met via teleconference to reach coding consensus and refine thematic categories. Facilitating and limiting factors were finally summarized based on themes and device interfaces (i.e., measuring at the clinician-patient interface or at the patient-table interface).

Figure 1 presents the PRISMA diagram of the current scoping review. The literature search was initially conducted from inception to June 19, 2020, and yielded 11,415 references, of which 8,997 were screened after removal of duplicates. Screening of the titles and abstracts resulted in the exclusion of 8,828 studies and an additional 132 studies were excluded after full-text screening. The reference list of the included studies was searched and yielded the identification of 6 additional original studies. All disagreements were resolved by consensus between the two reviewers. The literature search was updated on June 30, 2021, and led to the identification of three additional articles fulfilling the inclusion criteria. Forty-six studies were deemed relevant and were included in this review (7–12, 15–54). The literature search also revealed seven literature reviews from which no additional original studies were identified (2, 6, 55–59). No additional facilitating or limiting factors were identified from the initially included reviews and, therefore, all reviews were excluded.

The characteristics of the included studies and both quantitative and qualitative data are reported in Supplementary Table 1 and visually presented in Figure 2. From the 46 included studies, 20 (43.5%) used a device to measure SMT or MOB force-time characteristics at the clinician-patient interface (7, 11, 12, 24, 26, 28–31, 37, 39, 40, 44, 46–52), 23 (50%) at the patient-table interface (9, 10, 15–23, 25, 27, 32–36, 38, 42, 45, 53, 54), and 3 (6.5%) at both interfaces (8, 41, 43). A total of 10 different devices measured SMT/MOB force-time characteristics at clinician-patient interfaces and 6 at patient-table interfaces. In light of the quantitative data, 37 (80.4%) studies reported at least one metrologic property (7–10, 12, 15–29, 31–34, 36–38, 40, 42–45, 47, 49, 50, 53, 54), with 16 (34.8%) reporting previously published results or referring to other studies (8, 12, 15, 22, 23, 27, 33, 34, 36, 38, 42, 43, 45, 50, 53, 54). On the other hand, facilitating and limiting factors were identified in 18 (39.1%) studies measuring forces at the clinician-patient interface (7, 11, 12, 24, 26, 28–31, 37, 39, 40, 44, 46–48, 51, 52), 13 (28.3%) studies measuring forces at the patient-table interface (10, 17, 19, 20, 22, 23, 25, 32–36, 38), and in 2 (4.3%) studies including measurement at both interfaces (8, 41).

Supplementary Table 2 presents the metrologic properties reported for each device as well as the modality evaluated and the targeted spine region. With exception of device xvi, at least one metrologic property was reported for each device. Eight properties were reported for device xiii, a standard mobilization couch adapted with load cells, four properties were reported for devices ix and xi, three for devices v, viii, and xiv, two for devices ii, vi, x, and xv, and one property was reported for devices i, iii, iv, vii, and xii.

All authors stated that their device was deemed appropriate for their purpose. The most common evaluated property was the calculation of the measurement error (raw or percentage) of a force measured simultaneously by the device and a gold standard, which was usually a load cell or force plate (devices ii, iv, v, x, and xiv). Measurement error was also assessed by comparing the force/pressure measured by the device while a known weight was directly placed on the device (devices viii, ix, xi, xiv, and xv). In both cases, authors referred either to the device accuracy (devices ii, iv, viii, ix, xi, xiv and xv), validity (devices v), fidelity (device xiv) or calibration (devices x). Accuracy was also reported for other devices, but how this property was determined was not described (devices i, v, vi, vii, xii, and xiii). The second most common property reported was the error between several measurements of the same event and referred to the device reliability (devices x, xi, xv, ix, and xiii) or repeatability (device viii). The device sensitivity was reported for four devices (device ii, vi, ix, and xiii). This property was defined as the minimal detectable signal over the noise inherent to the device for device xiii and as the smallest measurable output and output changes for device ix. Static sensitivity, the slope of the least square calibration regression line, was also reported for device ix. How the sensitivity of devices ii and vi was measured was not reported. Linearity, described as the correlation coefficient computed between different applied force/pressure and the measured value by the device, was reported for three devices (devices iii, xi and xiii). Two devices reported the signal drift, which was defined as the variation in the signal over time when a known load laid steadily on the sensor (device ix and xiii). Coupling or crosstalk effect was also reported, and was described as the percentage of the load measured in the caudad-cephalad and medial-lateral axes when a known load is applied vertically on the load cells (device xi and xiii). In line with the coupling effect, device v, a sensing polyester film, was reported to have some shear components when the force was applied perpendicular to the surface. Variability, that is the variation in the load measured by the device in response to changes in the point of load application, was reported for the mobilization couch adapted with load cells (device xiii). Finally, the calibration procedure was reported for three devices (devices v, viii, and xiii). Reporting frequency of each metrologic property can be visualized on Figure 3.

Although previous literature reviews synthesized studies using devices to measure manual therapy force-time characteristics such as total peak force or rate of force application, the main interest of these reviews were the results relevant to manual therapy biomechanics or learning (2, 6, 55–59). The current review is therefore the first to report the metrologic properties and the factors facilitating and limiting the use of the devices that have been previously used in the literature related to manual therapy. A focus was placed on devices developed or adapted to measure SMT or MOB force-time characteristics delivered on human spine. Devices were categorized into those measuring the force/pressure/load directly at the site of application (clinician-patient interface) and those measuring indirectly through the reaction forces of the patient's body on the sensor (patient-table interface). Interestingly, almost the double number of devices measuring at the clinician-patient interface (n = 10) than at the patient-table interface (n = 6) were identified. This may reflect not only researchers' creativity in adapting commercially available devices to measure SMT and MOB, but also the variety and availability of such devices. In contrast to devices at the clinician-patient interface that were first the subject of publication in the early 1990s (49), the first publications including a device at the patient-table interface were published in 1995 (16, 17). However, 57% (26/46) publications included a device at the patient-table interface (23/46 or 50% for the clinician-patient interfaces) highlighting the higher number of publications using the same device at the patient-table interface. Indeed, one of the instrumented tables with force plate (device xiv) was reported in 11 studies (15, 19, 23, 25, 27, 41–43, 45, 53, 54). Future investigations should focus on the identification of the most appropriate technology depending on the assessment context. This could be achieved through expert consensus and could lead to the development of a theoretical framework for future development on this topic.

Metrologic properties include different concepts that are often misused or used interchangeably within the literature (61). In the current review, as many as four different terms were used for the description of the same measurement. Specifically, accuracy, validity, fidelity, and calibration were all defined as the measurement error between either the evaluated device and a gold standard device, or with a known weight lying on the device. An attempt was made to regroup metrologic properties evaluated in a similar manner so that the data could be synthesized comprehensively. However, how these properties were measured was not always defined (devices i, vi, xii, and xiii for their accuracy; devices ii and vi for their sensitivity) and few authors only refer to a study evaluating the device in a context not related to SMT or spinal MOB (devices v and vii for their accuracy). For this reason, it was not possible to contrast metrologic properties between devices. It is therefore crucial to define a common and standardized terminology to be consistently used in manual therapy studies. It is also fundamental that future studies properly describe how each metrologic property is measured.

Squara et al. (62, 63) recently published a framework based on the International Bureau of Weights and Measures (64), but adapted to perioperative and intensive care medicine. This framework defines the terms related to quantities and units, properties of measurements, devices for measurement, properties of measuring devices, and measurement standards. Considering that Squara et al. (62, 63) framework is adapted to the medical field, it seems reasonable to use it as a reference point for a common and standardized terminology in the manual therapy field.

Accuracy is defined by Squara et al. (62, 63) as the closeness of agreement between a single value measured by the device and the real value, and is usually expressed as an error of measurement. A distinction between accuracy and trueness should also be done. Trueness is determined by the systematic measurement error or difference between the averaged measured value and the reference value obtained when several measurements are compared to the true value (e.g., several measurements of a known weight laid on the device). In the current study, the error of measurement was also associated with the terms validity, fidelity, and calibration, but no study used the term trueness, which, based on the measurement descriptions, would have been the most appropriate term. Interestingly, Waddington et al. (47) reported that the force measured by their modified dynamometer never exceeded 1 N of the real force, and referred to this assessment as their device calibration. However, this assessment would better reflect the device trueness. By definition, calibration is the comparison between the measure obtained by the device to a known reference standard. In case of improper calibration, a procedure of adjustment can be then performed.

Precision is another metrologic property, which can be expressed as variability (%) or a random measurement error. This property refers to the variability of replicate measurements of a same procedure without reference to the reference value, and includes both repeatability (several measurements of the same procedure done in a short period of time in the same exact condition) and reproducibility (several measurements of the same procedure done in variable conditions). In the studies included in the current review, authors seem to have used the term reliability instead of repeatability or reproducibility. It is worth noting that Harms et al. (20) assessed the variation in the load measured by their mobilization couch adapted with load cells (device xiii) in response to changes in the point of load application and referred to it as the device variability, but, based on Squara et al. (62), the term reproducibility would have been more appropriate.

Sensitivity is defined by Squara et al. (62) as the quotient of the change in an indication and the corresponding change in the quantity intended to be measured. In other words, sensitivity corresponds to the change in the load measured by the device in comparison with the real load variation. In the studies included in this review, sensitivity was defined as the smallest measurement output change (device xiv) or as the minimal signal detectable over the noise inherent in the system (device xiii). However, in both cases the measurements more appropriately reflect the device resolution, which corresponds to the smallest change in the measure that can be detected by the device. It is important to note that sensitivity can be measured throughout the measurement interval relevant for the context. How sensitivity varies throughout this interval refers to the linearity, which is, in fact, a mathematical property and not a metrologic one. Linearity was reported for three devices (iii, xi, and xiii) identified in this review, which all showed almost perfect linearity (r ≈ 0.99).

The measure of the device drift, reported for devices ix and xiii, corresponds with the definition provided by Squara et al. (62), i.e., the change over time of the value measured by the device when the value remains the same in reality. A device with no drift has an excellent or perfect stability. Coupling or crosstalk effect is not reported as a metrological property in Squara et al. (62, 63). However, this property is deemed important in the context of manual therapy, since part of a vertically applied load can be falsely measured on the cephalad-caudal and medial-lateral axes, as reported for devices ix and xiii. Finally, it is noteworthy to mention that validity is not considered a metrologic property, but a research construct. To determine the validity of a device, the relevant metrologic properties for the specific purpose have to be considered as a whole (62, 63). Specific instructions on how to measure each metrological property, with examples, are described in Squara et al. (62, 63).

Potential factors facilitating and limiting the use of the devices included in this scoping review were collected qualitatively. With the exception of “comfort,” all other factors were subjective comments made from authors' experiences rather than an objective assessment. Technology acceptance models have been developed to predict technology usage by assessing factors that influence its acceptance (65, 66). While these models are often focused on the technology itself, two models have been used in health care to assess acceptance of health care information systems and technology: the Technology Acceptance Model (TAM) and the Unified Theory of Acceptance and Use of Technology (UTAUT) (67). While both models include the assessment of the t perceived usefulness and ease of use of the technology, UTAUT also includes social influence and facilitating conditions.

Perceived usefulness corresponds to the user's perception that using the technology will increase the task performance, and aligns with the metrologic/intrinsic properties and feedback themes identified in this study. Specifically, the metrological/intrinsic properties theme included comments related to the importance of the technical characteristics of the devices, and how these enable or restrict the proper measurement of manual therapy forces. The feedback theme emphasized the potential application of manual therapy real-time feedback in different areas, such as in education context.

Perceived ease of use corresponds to the user's perception that using the technology will be easy and effort-free. This aligns with the user-friendliness and versatility, as well as with technique application themes identified in this study. While the user-friendliness and versatility theme highlighted the characteristics that contributed to the device's ease of use (e.g., size, not interfering with movement, easy to transport, set up and store), technique application theme referred to the level of comfort of both clinicians and patients experienced while using the device to measure manual therapy forces and to the perceived realism of technique application by the clinician.

The social influence variable described in the UTAUT's model corresponds to the users' perception of others' believing that it is important to use this technology. This was not captured in any of the themes in our study, suggesting that social influence might not play a significant role when researchers are choosing the technology to use to measure manual therapy forces.

Facilitating conditions (from the UTAUT's model) corresponds to the organizational and infrastructure support to use the technology, and aligns with the cost and durability theme from this study. Costs and cost-benefit assessment (either objective or subjective) are most often considered when selecting the device, which was emphasized by the authors of studies included in this review.

The alignment between variables from technology acceptance models and the themes identified in this study indicates that while such models can indicate specific aspects that are important for a device's acceptance and usage to measure SMT and MOB forces, a specific model for this specific field might assist manufacturers to tailor their device's design to measure forces during manual therapy. This would also allow for a formal evaluation of the acceptance and use of force sensing devices in the context of manual therapy, which would inform the selection of the device to be used in future studies, and the design of future force sensing devices.

The results of this scoping review should inform researchers when designing future studies requiring the measurement of SMT or MOB force-time characteristics. Whenever possible, the use of an available device should be prioritized over developing a new one. This would allow resource-use efficiency, as well as measurement standardization across studies which, in turn, would facilitate combining data from different studies, and knowledge sharing.

Augmented feedback is considered a key feature of motor learning of any task, and manual therapy makes no exception (14). The quantity and type of feedback depends on several factors, including the motor task itself, the learner expertise, and the learning objective (68). When learning SMT and MOB, students often practice on their peers, following the observation of an expert (a clinician or a professor) performing the task. The expert then provides a subjective feedback to the student regarding their SMT or MOB execution (14). The inclusion of devices measuring SMT/MOB force-time characteristics into manual therapy professions curriculum offers the opportunity to provide objective feedback to the students. Early studies support the use of such devices to help students becoming more accurate and consistent during SMT or MOB delivery (12, 37, 46). In addition, students have been considering that the use of devices providing objective feedback helps them learn and retain manual skills (12, 69). Postgraduate programs, such as physiotherapy, chiropractic, naprapathy and osteopathy, looking to implement devices providing real-time quantitative feedback into their curriculum should consider the data presented in the current study.

Mechanisms underlying the clinical effects of SMT and MOB have not been entirely clarified, but researchers agree that they include both spinal and supraspinal events, as well as contextual factors (3, 70, 71). Given clinical practice recommendations to use conservative interventions, including SMT and MOB, to manage acute and chronic musculoskeletal pain, implementing tools measuring SMT/MOB force-time characteristics into clinical practice could help clinicians to adapt their treatment to each patient's individual biomechanics and needs. Safety could also be potentially enhanced by insuring the delivery of SMTs causing lower constraints on both the clinician's and patient's joints (44, 47). The use of devices into clinical practice remains to be further investigated.

The main strength of this scoping review is the use of a mixed-methods involving both quantitative and qualitative data. It is worth noting that the facilitating and limiting factors identified in this scoping review were mostly based on the subjective opinion of the authors of the included studies. This, the relative importance of these factors in the use of the devices in research, educational or clinical setting could not be determined. Other factors might also not have been identified. Finally, due to the nature of this study and the lack of study comparing different devices, it was not possible to identify a specific device that should be used over the others.