Secondary Logo

Journal Logo

Original Research Article

Assessment of Functionality of Multifunction Prosthetic Hands

Kyberd, Peter J. PhD

Author Information
Journal of Prosthetics and Orthotics: July 2017 - Volume 29 - Issue 3 - p 103-111
doi: 10.1097/JPO.0000000000000139
  • Free


Starting with the commercial release of the Touch Bionics hand in 2005, new designs of prosthetic hands have changed the expectations of function for upper-limb prostheses. These “multifunction” hands have the defining feature of separate powered movement of individual digits. They can form different grasp shapes and have the possibility of a grip that adapts to the shape of the target. The potential number of grips formed is greater than the standard single degree of freedom (sDoF) hand. This form was first seen in the “Russian hand” in the 1950s.1 The consensus from manufacturers and users alike is that this change to more available motions will result in increased function. This assertion needs to be supported by evidence. This article is part of an effort to answer this question by using a test of function to assess the function of these hands.


Despite apparent limitations, users of earlier hands are able to perform a wide range of tasks. The hands are robust; they are set in a precision grip (the most common grip in everyday use2). The control is simple, with one or two input channels (generally electromyogram [EMG]) controlling the flexion and extension. Although many users pragmatically accept these devices, the wish for a hand with greater function and a more lifelike appearance has remained.3,4

Throughout the history of prosthetics, anthropomorphic hand designs have been sought,5 and many designs have been proposed.6 Few research designs have been tried by the limb-absent,7,8 fewer have been worn by more than a small number of users, and fewer become products.9 The i-limb hand, derived from the research program in Edinburgh directed by David Gow,10 was the first multifunction hand of the modern era to be released as a commercial product. It led a dramatic change in commercial hand design. The i-limb hand had individually driven fingers, based on the ProDigits design.10 Since the release of the i-limb hand, both Steeper and Ottobock have produced new designs that are capable of multiple motions.

Similarly to the Touch Bionics hand, the fingers in the Steeper BeBionic hand also curl at a second joint. It also has a passive rotation of the thumb. The Ottobock Michelangelo hand represents a design intermediate between the earlier sDoF designs and the other multifunction hands. It does not have curling fingers; all fingers are driven together, but it can change the relative motion of fingers and thumb to give two distinct prehension patterns (tips and lateral opposition). As the fingers adapt to the grasped object, all hands are capable of a number of different grips (see Table 1). It is this feature that makes the hands distinct from earlier hands. One consideration to be made is if users must perform compensatory motions to overcome the limitations of their prostheses.11–13

Table 1:
Summary of grips the different hands are capable of achieving

With fewer joints in the arm, the wearer has to use bodily motions to place the hand in the correct position to grasp an object. For example, without prosupination of the forearm, the person may use shoulder abduction.13 This increased range of motion may cause overuse injuries later. Overuse can only be confirmed through long-term observations. One factor to influence the compensations used is how much the hand adapts to its environment. For a fixed-geometry hand, it is possible that the number of positions in which the hand can be placed to pick up an object securely are fewer (compared with an adaptive hand), thus restricting the approach to the object.14 For example, a parallel-sided gripper may only pick up very regular objects and only do so reliably when an axis of the “fingers” is parallel with the orientation of the axis of the object. Alternatively, if the hand has some adaptability in the grip, this limitation may be overcome. This adaptability is a commercial possibility for the first time.


An important consideration for someone prescribing the new hands is to determine if the user will be more functional with the device. All the hands considered are generally more anthropomorphic and, anecdotally, thus attractive to the user population. However, a clinician needs to balance interest from users with an effective clinical outcome. The way to achieve this is through objective measures of function. The Upper Limb Prosthetics Outcomes Measures group (ULPOM) outlined a method to create consensus and understanding between clinicians by using the World Health Organization's International Classification of Function (WHO-ICF).15,16

They identified measures that have sufficient psychometric properties to objectively measure the three different domains: function, activity, and participation. This study uses one tool identified to operate in the function domain. However, it is important to bear in mind that appearance is an essential function for some users, so a hand that moves more anthropomorphically will be important to some users. This factor is not considered objectively in the current study.


This study is part of a program to investigate the function of prosthetic arm systems.17,18 It aims to assist in the decision-making process of clinicians when prescribing an advanced hand by providing objective information on its functional capabilities. An earlier study concerned sDOF myoelectric hands and their controllers. This forms a baseline with which to compare the recent hand designs in this article.

This study uses the Southampton Hand Assessment Procedure (SHAP) to measure the function of the hand using a single subject and repeated measures. Full details are available at


SHAP was designed to assess the function of the hand using a form-board and self-timed tasks with abstract objects and simulated activities of daily living (ADLs).

Completion supplies the tester with numbers that allow direct comparison between subjects over time and between levels of recovery.19 It was shown to have good interrater and intersubject repeatability on normal subjects and some conditions of the hand and arm.19–22 SHAP has been used with users with prostheses21 and with children.23 SHAP can cover both the Functional and Activity domains (ICF). The SHAP score is not that of the device, but of the user and the device combined; in this study, using a single user makes the resulting scores of the devices alone. Interpretation of the results from this current trial must be read as the relative functional performance of the hands user.17

The tasks are divided into two sections: abstract objects, to encourage the use of six standard grips and 14 simulated ADLs.24 The task times are compared with a normative group, and the Overall score is weighted to reflect the relative importance of the each grip type, according to the literature.2 The result is a score were 100 is “normal” (compared with a group of unimpaired subjects who score above 95).

SHAP was designed to be used with all hands, natural, impaired, or prosthetic. It emphasizes the function of the hand. It assigns a score based on the type of grip most commonly used in each task; however, for all tasks, any grip can be used. Its philosophy is essentially practical. Whatever the subject uses to perform the task is correct. The tasks are related back to the grips usually used for the task. For this experiment, the grip adopted was recorded. The hands under test did not have particular grasps explicitly created within their repertoire; all the multifunction hands could adapt to the object being grasped. The Touch Bionic and Steeper hands were adaptable, as each finger was separately driven and stopped when it encountered and obstruction. So when closing all fingers on a sphere or a cylinder they approach equally, but the fingers stop in a line when they encounter the cylinder, but each stops separately around the sphere, much as natural hands would. The fingers of the Michelangelo are linked through elastic elements, making the fingers compliant and allowing it too to achieve both grasps. This is not how an older single-degree hand would close. This enables all hands to functionally create a spherical or cylindrical grip, and this was recorded in the results. Even the single-axis hand could be described as grasping in tips opposition or spherical, depending where the object was placed within the hand. Thus all multifunction hands are described as achieving a wider range of grips than the manufacturers' basic descriptions.

The earlier tests used SHAP on conventional sDoF prosthetic hands. It showed that hand size and flexion speed have little impact on the measures of function, whereas the control format has a significant impact on the overall assessment.17 A second study showed that a passive (or compliant) wrist flexion unit has a lesser, but significant, impact18 on function. The likely reason for this finding is that the hand can be placed in a more effective position to hold the target object. This current article outlines the next stage: to test the three multifunction hands that were commercially available at the time. Although the manufacturers have continued to improve the designs, the improvements are more concerned with greater robustness than functional range, the basic comparison―conventional single-axis hand against multifunction hand―remains relevant.



The i-limb pulse was a second-generation hand from Touch Bionics. It used the ProDigits format for the fingers, curling at the MCP joint. The thumb has a fused MCP and passively rotates at its base, which allows it to oppose the tip or side of the index finger.

Control is of all the fingers at once, opening and closing as a single unit. Obstruction of a digit will stall the motor and allow the others to continue closing forming a grip around an object. Different grip formats are selected using the EMG command channel; with co-contractions or pulsed extension. Recent additions include selection of grips using personal mobile devices (www/ By the release of the i-limb pulse, it was possible to select two- and three-jaw chuck and Lateral grips; however, the rotation of the thumb was passive and needed to be set by the operator. The precise position of opposition had to be judged by the user. There is no internal feedback in the i-limb hands. Digit coordination is based entirely on open loop timing of the digit flexion. So digit coordination relies on all the fingers starting at full extension and closing towards tips opposition. Repeated flexing and extending of the fingers in grips, without returning the digits to the fully extended position, means that the digits will get out of the single alignment. This is especially true of the thumb, which will tend to not oppose the tip of the fingers once the hand has been partially opened and closed. Since this study was conducted, a powered form of thumb rotation has been launched, but internal feedback is still unavailable.


In the Steeper BeBionic hand, each finger is driven separately with internal feedback of the position of the fingers so that when tips opposition is selected the fingers are driven by the controller so that the tips will meet. Like the Touch Bionics hand, the thumb rotation is passive, but it has two set positions for the thumb (tips and lateral opposition) and an internal switch in the thumb selects lateral opposition without an additional user input (in the i-limb hand, the lateral position has to be selected with the EMG channel as well as with passive rotation of the thumb by external forces). The relative positions of opposition of the tips of thumb and index and third fingers is set in the clinic. If the fingers are set to provide a three-jaw chuck, when two-jaw chuck was chosen, the tip of the index finger was at roughly 45° to the thumb. Common to both of these first two hands, the BeBionic has undergone a number of changes to the design, including strengthening parts and changing firmware. The hand tested here represents a Mark 2 mechanism, with some of the strengthening features of the Mark 3 and the firmware of that generation of hand.


The Michelangelo is of a different design, it has a single large brushless motor in the palm, with a short crank to the base of the fingers and thumb. The position of a cam on the motor dictates if the thumb opposes the tip or side of index finger (a second small motor shifts the thumb orientation to assist in this change). The digits have no articulation except at the base, but they are flexibly linked to each other so that in a limited degree, the hand can conform around an object. The fingers also abduct as they close so that once the hand has closed sufficiently, the fingers close together and become a ridged group. In the Lateral grip, the fingers are driven together to form a single base to oppose the closing of the thumb. Compared with the other two hands, this is a far firmer platform. At the time of writing, the hand was rated at a closing force of 80 N. All Michelangelo hands are also mounted on a sprung wrist flexion wrist, which is settable at five rigid positions from flexion to extension, or has some compliance. This feature alone allows the users to use it to extend the wrist passively against a horizontal surface, so the hand is parallel with the surface and is better able to pick up small objects (such as from the floor). The wrist is similar to the Motion Control Flexi-wrist tested in the previous study18 in that it is flexible and sprung. Although details of the mechanisms may be different, the overall flexibility is likely to be similar, and the previous results suggest that this flexibility alone would have an impact on the functional range measured in this test.


One problem with all multifunction hands is how to control the additional options made available with the independent digits. The manufacturers have created a wide range of different options. These options are then set at the clinic, so that at any time, the range of grips available to an individual is more limited. Selection of options is important and is chosen to fit with the user's needs and abilities. All three designs use coding of the myoelectric channel to select the options. Co-contraction of the two channels is used as one switch. If the digits are extended fully, the user can place the Touch Bionics hand in a state ready for a series of pulses (two or three), whereas the BeBionic hand waits until the extend signal has exceeded a set time threshold before making a transition. The BeBionic also detects the position of the thumb to switch into Lateral grip, using context as a clue. It also possesses a switch in the middle of the dorsal surface of the hand, which can be customized to the needs of the user. For this study, the grips were chosen to allow the SHAP objects to be manipulated in the grips selected in the SHAP protocol with the most rapid and easiest transitions to be selected by the subject to maximize the score. Table 2 details the grips and transitions. More recent switch options with RFiD tags and other external switches were discounted as they require the environment to be prepared for the user, which may occur in daily life, but is not relevant to this study.

Table 2:
Grips and transitions used in this study for each of the SHAP tasks



The original study focused on the performance of the older generation of myoelectric hands that typified most devices before the launch of the Touch Bionics i-limb hand.17 This study measured the functional capabilities of the individual device, attempting to remove the influence of the operator as much as possible; hence, it used a single user with repeated measures. The relative scores emphasize the differences between mechanisms and control schemes. This protocol was repeated in this study.

The subject was the author who, while able bodied, has irregularly used myoelectric control over the preceding 25 years. He has operated a variety of hands using a custom socket over his forearm to retain the device. He also conducted the original study on the older generation of hands.15,16 SHAP is designed to test the functional capabilities of the subject and hand. By the use of a single subject, intersubject variations could be reduced substantially. The use of the experimenter as subject allowed many more repetitions of the test to be performed than would be practical or ethical with a cohort of limb-absent users.


The subject was fit with a splint over the left forearm, extending from the fingers to the olecranon. Figure 1 shows examples of the splint used with two of the hands under test. The prosthesis was positioned over the dorsal surface of the arm and displaced 15 cm from it.15,16 Ottobock myoelectrode amplifiers/processors were retained over flexor and extensors, in the same manner as a conventional self-suspending forearm socket. The splint was manufactured by the prosthetics team at (Atlantic Clinic for Upper Limb Prosthetics, Institute of Bioengineering, University of New Brunswick, Canada), using conventional materials and techniques. The hand was retained at the end of a beam with a quick disconnect wrist unit; the socket housed the batteries. Additional modifications were required to mount the Michelangelo at the same position. Generally, the hand-eye line was very similar to that of a routine fitting. The left hand was chosen because it was the subject's nondominant hand. This was considered appropriate, as the majority of prosthesis users with a single side loss use their prosthetic device in a nondominant manner. SHAP requires the hand under test to be the leading hand, thus a genuine user performing SHAP would be using their nondominant hand, much as the hand was used in this experiment. This is similar to the previous studies.17,18

Figure 1:
The subject performing SHAP tasks using the splint and the Touch Bionics and Ottobock multifunction hands.


Similar to the earlier trials, preparation included the subject performing general activities with the hands so the basic learning of its use had passed. Compared with the conventional prostheses, the control formats were more distinct, so the subject was given longer to familiarize himself with the controller.17 Only when he was satisfied he had mastered the controller was the next stage of the trial begun. Mastery was gauged by achieving switch control to a different hand state on the first attempt greater than 90% of the time. The subject was seated at a table; when the hands were resting on the table, the subject's elbows were at a 90° angle. The form-board with timer was placed directly in front of the subject, approximately 10 cm from the edge of the table. Each task commenced with the hand under test starting the timer, the subject then performed the action and then turn the timer off with the same hand. Unless a serious error occurred (e.g., timer failed to be turned off), the first time was used.

The test was performed twice a day for 20 successive working days. This was longer than the preceding tests, again to ensure that any learning effects were marginalized. The next 20 days (40 runs) were divided into four epochs of equal length. Any learning and accommodation would be observed from the difference of the last two epochs. The result used was the mean of final 5 days (ten runs) (a summary of the protocol is shown in Figure 2).

Figure 2:
Experimental design. Data were taken over 4 epochs of 10 trials each after an extended period of familiarization.



A Touch Bionics i-Pulse hand, a Steeper BeBionic V2a hand, and an Ottobock Michelangelo hand, largest sizes, left-sided, were tested. For grip types and controller switch options, see Figure 3.

Figure 3:
Control options chosen for the different hands under test. Transitions between grip states are shown. All hands start in full tips (in the center) and move to the other grips via the commands indicated.


With only one subject, there is only one score per condition, the conditions (different hands) where grouped together for the analysis. The sDoF hand that had the highest score in the earlier study,17 (Motion Control) was added to all subsequent comparisons. The Overall SHAP scores of each of the hands were used. Learning effects were checked using an unpaired Student's t-test between the penultimate and the final epochs. A between subject analysis of variance (ANOVA) was used to compare the different Overall scores. The independent variables were the hands, the dependent variables were the Overall scores from final epoch (40 scores). In all cases, alpha was set at 0.05. SHAP creates score for individual grips, which are used to calculate the Overall score for the six grips. These scores were normalized against the Overall score by dividing the individual grip score by the Overall score, to show the scores relative to the Overall performance.19 The scores per category were also ranked, and unpaired t-tests for those that showed distinct differences were applied. This follows the same analysis performed in the preceding studies.



Figure 4 shows the scores for the Overall rating scores (and the standard deviations) for the 10 runs of the final epoch. The ANOVA of all of the Overall scores showed that they were not all the same population (4 hand/controllers, 10 scores per hand, df = 39, P > 0.05, F-Stat = 76.8, dfnum = 8, dfdenom = 81). The comparison of the penultimate to final epochs showed no significant changes in overall scores (df = 19, P > 0.05); hence, any learning had ceased, and the scores used were seen to be stable.

Figure 4:
Comparison of the Overall scores of the three multifunction hands compared with the highest scoring sDoF hand from the earlier study.15 Grips where there is a statistical difference between scores are indicated with a star.

An unpaired t-test was applied to the scores from different hand from the final epoch that were clearly distinct from each other; the Motion Control hand had a significantly higher score than the OttoBock (P > 0.05), which was distinct from the other two hands (P > 0.05, Bonferroni correction applied).


Table 3 outlines those grip scores that were statistically different. For all analyses, the P value was very much less than 0.05 (with a Bonferroni correction applied).

Table 3:
Significant differences between grip scores for the hands tested


Similar to previous studies,17–19 the individual grip scores were normalized against the Overall scores. This method points up the grips that are the most and least effective from each hand, Figure 5.

Figure 5:
Normalized individual grip scores of the three multifunction hands compared with the highest scoring sDoF hand from the earlier study.15 A score of 1.0 indicates that the score is the same as the Overall; below indicates a poorer performance; and greater than 1.0, better performance.



What follows are a series of observations from the subject on his experience of using each of the hands:

Adaptive grip: One benefit of an adaptive grip is that it was possible to open the jar with the lid held by the prosthesis despite the fact that the grip force in the BeBionic and the i-Pulse was not particularly great. This was because the digits were able to form around the object and spread the load more evenly compared with simply gripping the object firmly in a power grip. This also means that the majority of grip forms are derived from the starting pose of the full Tips grip, so that as few of the grips as possible need to incorporate transition times into those grips (9 of the 26 grips needed grip selection).

Predictable Grip: The thumb mechanism in the BeBionic hand is set to achieve a three-jaw chuck in tips opposition or a two-jaw chuck. The position is set in the clinic. This is not a disadvantage compared with the infinite number of positions the i-limb, set by the user. This is because to position the i-limb thumb in the correct place requires greater attention. Once the BeBionic user has selected the grip, the thumb will be in a predictable place; it is always in that position. If the thumb is set for a three-jaw chuck, when the two-finger pinch is selected, the index finger opposes slightly laterally, but even here, users can confidently adapt their grip strategy to grip successfully because they know that the thumb will always be in the same place. The same cannot be said for the ProDigit thumb, which needs to be observed closely to know where it is exactly. Hence, the uncertainty can slow down the user and increase the chances of the grip not being used or the score increased.

Similarly, when switching into the Lateral grip, the uncertainty of the position of the Touch Bionics hand makes it slower; the newer active rotation of the thumb may change this only if the resulting position is consistent.

Michelangelo: The fingers closing together creates a rigid base in the Ottobock hand, this is in either Tip prehension or Lateral grip, making a more predictable tool. Its inability to achieve a two-jaw chuck makes handling small objects less easy. The approach of the thumb in the Michelangelo is more anthropomorphic as it adducts as it flexes, this can be visually pleasing. It makes the task of lining it up with a small object different to the habits of years with the simpler hands. Similarly, the Michelangelo may go from neutral to grasp shape by extending the thumb to rotate it into the correct position, even when the user myo-command is exclusively to flex.

This confounds expectations built up over many years of using conventional hands. It is probable that users of many years experience will notice these differences and need time to accommodate the new motions.

The earlier study17 showed that control format can have an impact on the measured function. By comparison, it is not possible to control the hands with the same command structure. So although some transitions seem quicker and should result in higher scores, it is not possible to demonstrate this unequivocally. Therefore the comparisons are based on the overall systems.


To perform any task, it is necessary to place the hand in the correct position and orientation relative to the target so the digits to close on optimal position on the object to hold it securely during the task. For some tasks, this is achieved by placing the object in the palm and closing the hand around it (Cylinder); in other tasks, it is critical for the grasping surfaces to be parallel to the surface on the tab on the zipper or the faces of the coin. Natural fingers have greater flexibility with the palps on the digit tips. These conform around the object in a similar manner to a liquid. When mechanical compliance is supplied by the glove’s rubber, it tends to oppose the grasp and eject the object, rather than admit it, making holding an object between the finger tips harder to achieve.

The two multi-axis, multifunction hands (BeBionic and i-limb) have the advantage of the fingers forming around the object, both through curling and through the differential driving of the digits. The Michelangelo achieves some of this with the compliant links between the fingers. This should make the approach of the hand less critical, and some circumstantial evidence for this can be observed in Bertel's study,14 where the approach paths of the conventional prostheses to the target were very limited compared with natural hands. So it should be observable that the approach paths for any of these new hands should be less restricted. It remains for future studies to confirm this.


The study shows that the hands have different performances, with the simplest hand having the highest performance. The other hands were able to achieve some of the same grips generally used by unimpaired subjects; the sDoF hands were set in a Tips grip for all tasks.

Although one multifunction hand had a better Overall score than the other hands, the differences between individual grip scores was more complex.


In Figure 5, the bars are laid out with the grips in the same order as in the previous study,17 the order reflecting the best to worst.

There is a considerable variation in the normalized scores for each grip across the different prostheses, showing different grips have different relative performance for each model. These differences are likely to be due to a combination of the design of the hand structure and control format used (these factors cannot be teased out using this methodology), but earlier work suggest both are factors.17

One consistent finding with different hand designs is that the Tips grip has a lower relative score than other grasps.17 It is worth considering what causes this lowered functional score. One possibility is that it is a weakness of the test, always producing a low Tips score. However, when SHAP is used with other impairments (such as missing digits or distal radial fractures19), this distinction is not seen, reducing the probability that this is an artifact of the test. If instead it is caused by a functional weakness of all prosthetic hands, what factors might cause this? In the tests of other hand impairments, the Tips grips score better than the Overall score, while prostheses scored lower. By comparison, when impaired wrists are tested, they reduce the normalized score by no more than 0.03 (data taken from 17), much smaller than the prosthetic hands' impact on the score. In this case, the natural fingers are very flexible and sensate. Tips grips are used in precision activities (zipper, coins, buttons), which require greater attention to perform the task, and they provide the flexibility that the impaired wrists lose. Thus, it seems likely that the reduction in the Tips score is not an artifact, but reveals an important limitation in the performance of the hand. Further work to confirm this assertion is required, but the fact that natural hands, although physically impaired but possessing an intact sensory system, can perform the tasks more swiftly than a person with a prosthesis suggests this is a contributory factor. These tasks require greater concentration and have sensory feedback from the contact points of the fingers to the object and the positions of the fingers relative to each other. Thus the question remains to be answered is, “What is the missing information?” What is missing is the lack of the force/contact sense or proprioception in the prosthetic hands.

The minimum level of sensory information needed to perform any particular task can be estimated. For example, to be sure that the finger has grasped the zipper securely requires that it is held in a small area of full opposition between finger and thumb tip (this was guessed from the position of the hand relative to the zip). With a zipper tag of 4 mm across, it must be felt to be central on the fingertip, giving a resolution of better than 2 mm laterally. Many feedback sensors tested previously have had a much lower resolution than this,25,26 although some recent developments have produced compact sensors with a higher resolution.27 Only a prosthesis with this resolution and a satisfactory means to feed this information to the user are likely to have an improved Tips score.


SHAP was designed to measure aspects of the function and activity domains in the World Health Organization's International Classification of Function.15,16 It cannot begin to measure factors in the participation domain, but, anecdotally, the multifunction hands have been received well by many users.

SHAP relates the grip form most commonly applied by unimpaired subjects when performing the task while it does not penalize a subject for using a different grasp if they choose. This is because it was designed to measure the person’s function, irrespective of impairment.20 From the hands described, the Motion Control could only perform one prehensile grasp; the Ottobock, two; and the other hands many more (precise number dependent on definition). An objective observer would suggest that they could perform five of the six SHAP-defined grasps (not Extension), but the definition of grasp depends on the requirement of the study (from greater than 10 in Cutkosky28 to three in MacKenzie29 or two basic forms in Napier.30). The definitions of the grasp used as listed in Table 2 reflect the final grip form, once the flexibility of the mechanism have positioned them, the controllers do not make the distinction between some grasps.

Not withstanding this distinction, it was still easier or more secure to perform other grasps while using the hands. For example, manufacturers and clinicians have reported anecdotally that users of multifunction hands use the Lateral grip more often than the unimpaired population. Personal experience would suggest that is because this is the most stable grip any of the multifunction hands can impart: the thumb opposes a larger mostly flat platform on the side of the fingers, something not necessary with the natural hand. This is especially true of the Michelangelo because the fingers are driven to adduct together to achieve this grip, forming an extremely stable surface.

A reading of these results are that the simpler, dependable sDoF hands have a greater functional score than any of the much more expensive multifunction hands. This assessment is based entirely on the different hands' ability to perform the task set for them. SHAP cannot capture the quality of the movement or the appearance of the hand. It simply records the hand's ability to perform the task.

As such, the Motion Control hand outperforms the more expensive and sophisticated hands. This is a one-dimensional analysis of the problem.

If appearance and the dynamic cosmesis of the devices is incorporated, then the other hands are likely to score far better. Even then, the other important dimension, price, is not included.

More anthropomorphic-looking hands introduce an aspect of embodiment that is missing from the single-axis hands. Clinically, the recommendations are clear: if the potential user's requirements are fast, efficient grip function, then the simpler hands are at least as useful, but any user needs to try the devices for herself or himself. If the potential user has a need for a greater aesthetic considerations, then the newer hands indeed have a wider range of performance and look more lifelike. As always it depends on the individual. This study aims to provide clinicians with greater understanding of the choices available to them.


Using a validated procedure to measure hand function, the more complex multiarticulated hands were tested and they did not show improved functional performance compared with the simpler prosthetic designs. Each device requires more actions to trigger the different grips to respond to the range of objects and tasks. The factors that affect the Overall score include the control format and the design of the hand, as it was not possible to program each of the hands with the same control formats; thus it was not possible to separate the different factors. All three hands were more anthropomorphic in action and appearance than the earlier hands, but this did not result in greater function than the simpler fixed geometry hands.


1. Popov B. The bio-electrically controlled prosthesis. J Bone Joint Surg Br 1965;47:421–424.
2. Sollerman C, Ejeskär A. Sollerman hand function test. A standardised method and its use in tetraplegic patients. Scand J Plast Reconstr Surg Hand Surg 1995;29(2):167–176.
3. Atkins D, Heard DCY, Donovan DH. Epidemiologic overview of individuals with upper-limb loss and their reported research priorities. J Prosthet Orthot 1996;8(1):2–11.
4. Kyberd PJ, Wartenberg C, Sandsjö L, et al. Survey of upper extremity prosthesis users in Sweden and the United Kingdom. J Prosthet Orthot 2007;19(2):55–62.
5. Putti V. Historical prostheses. 1925. J Hand Surg Br 2005;30(3):310–325.
6. Borchardt M, Hartmann K, Leymann, Radike, Schlesinger, and Schwiening. Ersatzglieder und Arbeitshilfen für Kriegsbeschädigte und Unfallverletzte. Germany: Springer Verglag; 1919.
7. Nightingale JM. Microprocessor control of an artificial arm. J Microcomputer Appl 1985;8:167–173.
8. Almström C. An electronic control system for a prosthetic hand with six degrees of freedom. Technical Report 1:77, Research laboratory of Medical electronics, Chalmers University of Technology, 1977.
9. Almström C, Herberts P, Körner L. Experience with Swedish multifunctional prosthetic hands controlled by pattern recognition of multiple myoelectric signals. Int Orthop 1981;5:15–21.
10. Gow D, Douglas W, Geggie C, et al. The development of the Edinburgh Modular Arm System. Proc Inst Mech Eng H 2001;215:291–298.
11. Ross M. Development of a quantitative test for prosthetic function using motion analysis and Activities of Daily Living. dissertation, Masters, Department of Mechanical Engineering, University of New Brunswick, 2005.
12. MacPhee B. Examining the prosthetic function and body behaviour of prosthesis users performing Activities of Daily Living. Masters dissertation, Department of Mechanical Engineering, University of New Brunswick, 2007.
13. Zinck A. The investigation of compensatory movements in prosthesis users and the design of a novel wrist. Masters dissertation, Department of Mechanical Engineering, University of New Brunswick, 2008.
14. Bertels T, Schmalz T, Ludwigs E. Objectifying of functional advantages offered by wrist flexion. J Prosthet Orthot 2009;21(2):74–78.
15. Hill W, Stavdahl Ø, Hermansson LN, et al. Functional outcomes in the WHO-ICF model: establishment of the upper limb prosthetic outcome measures group. J Prosthet Orthot 2009;21(2):115–119, April.
16. World Health Organization. ICF International Classification of Functioning, Disability and Health. 2001.
17. Kyberd PJ. The influence of control format and hand design in single axis myoelectric hands: assessment of functionality of prosthetic hands using the Southampton Hand Assessment Procedure. Prosthet Orthot Int 2011;35(3):285–293, September.
18. Kyberd PJ. The influence of passive wrist joints on the functionality of prosthetic hands. Prosthet Orthot Int 2012;36(1):33–38, March.
19. Kyberd PJ, Murgia A, Gasson M, et al. Case studies to demonstrate the range of application of the Southampton Hand Assessment Procedure. Br J Occup Ther 2009;72(5):212–218.
20. Light CM, Chappell PH, Kyberd PJ. Establishing a standardized clinical assessment tool of pathologic and prosthetic hand function. Arch Phys Med Rehabil 2002;83:776–783.
21. Metcalf CD, Woodward H, Wright V, et al. Changes in hand function with age and normative unimpaired scores when measured with the Southampton Hand Assessment Procedure. Br J Hand Ther 2008;13(3):79–83.
22. Vasluian E, Bongers R, Reinders-Messelink H, et al. Learning effects of repetitive administration of the Southampton Hand Assessment Procedure in novice prosthetic users. J Rehabil Med 2014;46(8):788–797.
23. Vasluian E, Bongers RM, Reinders-Messelink HA, et al. Preliminary study of the Southampton Hand Assessment Procedure for Children and its reliability. BMC Musculoskelet Disord 2014;15:199.
24. Light CM. An Intelligent Hand Prosthesis and Evaluation of Pathological and Prosthetic Hand Function. PhD thesis, Electrical Engineering Department, University of Southampton, 2000.
25. Salehi S, Cabibihan JJ, Ge SS. Artificial skin ridges enhance local tactile shape discrimination. Sensors (Basel) 2011;11:8626–8642.
26. Matulevich B, Loeb GE, Fishel JA. Utility of contact detection reflexes in prosthetic hand control. In: 2013 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). 4741–4746, Tokyo, Japan, 3–7 November 2013.
27. Winstone B, Griffiths G, Melhuish C, et al. Tactip - tactile fingertip device, challenges in reduction of size to ready for robot hand integration. In: 2012 I.E. International Conference on Robotics and Biomimetics (ROBIO). 160–166, 11–14 December 2012.
28. Cutkosky MR. On Grasp Choice, Grasp Models, and the Design of Hands for Manufacturing Tasks, Robotics and Automation. IEEE Trans Power Electron 1989;5(3):269–279.
29. MacKenzie CL, Iberall T. The Grasping Hand. North Holland; 1994.
30. Napier JR. The prehensile movements of the human hand. J Bone Joint Surg Br 1956;38-B(4):902–913.

assessment; function; SHAP; multifunction prosthetic hands

Copyright © 2017 American Academy of Orthotists and Prosthetists