Extracting outcomes from narrative review.

The final set of articles (see Results section) was critically read and reviewed. Clinically related requirements were derived from the original conclusions and inferred from the text.

Focus group results analysis.

The focus group recordings were transcribed and then analyzed by a professional focus group content analyzer. The analysis was conducted in accordance with a standard content analysis method, in which words, sentences, or paragraphs relevant to the main questions were identified. These text units were then further analyzed to identify common themes, which were then condensed into statements reflecting the common ideas raised by the group. Finally, we interpreted these statements to extract clinically related requirements.

Participants.

Included participants were 18–80 years old, with chronic (i.e., no longer hospitalized due to injury) SCI. Inclusion criteria included presence of complete or incomplete SCI (ASIA score A, B, C, or D) with a neurological level of injury of C5 and above (level of spinal sever was on or above vertebra C5), and tetraplegia [30]. Overall, 23 individuals were recruited from the outpatient clinics at the Sheba Medical Center Rehabilitation Center. Their demographic and clinical information appear in Tables 1 and 2. They were screened over the phone and those who were eligible entered the study. The study protocol was approved by the Sheba Medical Center Institutional Review Board, and all participants signed informed written consent prior to entering the study.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Demographic data of the participants in the questionnaire study.

https://doi.org/10.1371/journal.pone.0218393.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Clinical data of the participants in the questionnaire study (N = 23).

https://doi.org/10.1371/journal.pone.0218393.t002

Construction of questionnaires.

We used parts of previously used questionnaires [7, 31–34] and added additional questions to create new questionnaires, tailored to the study design and to the special needs involved in interviewing patients with SCI.

The full questionnaire is provided in Supporting Information S1 Text. Briefly, it had three parts:

1. Items addressing demographic and clinical information
2. Items related to computer usage habits (e.g., daily duration of use), environments (e.g., home, office, etc.), and difficulties. To evaluate the most common purposes of computer use and other practical and emotional concerns, we used concurrent parts of the Matching Person and Technology questionnaire [33]. If the participants reported using an ATCC (e.g., a head-tracker [15]), they were asked to elaborate about their experience by answering the Quebec User Evaluation of Satisfaction with assistive Technology (QUEST) 2.0 questionnaire, which is widely used to assess assistive technology related satisfaction and is considered highly reliable and valid [35].
3. Open-ended questions targeting information on computer related needs, missing functions, and demands for improvement that participants raised with regard to the systems and/or assistive devices they were already using.

Procedure.

Interviews took place either at the hospital or at the homes of participants. The interviewer read the structured questionnaires and recorded participant responses. For open-ended questions, the interviewer recapped the participant responses in short, clear sentences. The questionnaires were available in Hebrew, Greek, and English (the latter is provided in Supporting Information S1 Text). All the participants with SCI were interviewed using the Hebrew form.

Questionnaire results analysis.

To analyze QUEST 2.0 results, average scores (which are in the form of a Likert scale from 1 to 5) were computed for the total questionnaire and for the ‘Device’ and ‘Services’ subscales. To assess the relative importance of different aspects of ATCCs, we used the results of the final part of the QUEST 2.0 questionnaire and an additional mini-questionnaire designed for this purpose (see Supporting Information S1 Text). For questionnaires using Likert scale-based responses, means and standard deviations (SD) were calculated. For descriptive questions (e.g., “Which operating system do you work with?”) and yes/no questions, frequencies were used. Clinically related requirements were then qualitatively derived from the final set of results.

In the Results sections, apart from demographic and clinical data, only the data that were used to draft clinically related requirements is reported. The rest of the data appears in Supporting Information S2 Text.

Loss of independence of people with high tetraplegia.

Persons with high SCI experience extreme loss of independence in almost all aspects of life: “The extent of loss for the families is huge. Loss on top of loss. Patients speak of the physical ability of walking, but also about the loss of ability to control their sphincters, the loss of sexual function, even not being able to scratch their own nose. The losses are incomprehensible; I have tetraplegia patients that have wanted to kill themselves.”

Significance of computer use for persons with SCI.

In particular, the following topics were emphasized: (1) After returning home from their daily dealings, the computer enables SCI patients to regain control and independently manage a part of their life; (2) The computer becomes an anchor for functional independence in these individuals; (3) The computer is important as a tool for maintaining communication and social interaction for SCI patients. The computer is a central tool in getting SCI patients back into social participation during the phase of mental acclimation to their disabilities.

Computer use habits.

(1) No particular type of computer platform (e.g., laptop versus desktop) or brand stood out as preferable to participants with SCI; (2) Sitting was the preferred body position for computer use: "Most computer use is done sitting down, when you lie down it limits the movement; most functions are performed whilst sitting in a wheelchair."

Difficulties and barriers faced by people with SCI when operating computers.

The group acknowledged that operation of computers was a challenge in itself: “Everything that seems trivial to us, for them is a huge effort; if I think of a person that works with a head curser and has to lean back and move his head, and these are the only muscles that work, this is quite a big effort for him. He might be limited in terms of range of movement; these movements might cause him pain.”

Specialized computer-use needs of people with SCI.

(1) Securing privacy: “I have a patient who, when she wants to check the status of her bank account, her caregiver is automatically exposed to her code. It’s not just the caregiver; there is something very public about it.” (2) Identification of gentle muscle movements: “(for computer use)…there is a need for accuracy using the muscles, the gentle movements are much more vulnerable than the gross muscles, and they need to make delicate movements to operate the computer.”

Based on these themes and on additional discussion points, we identified, for example, that a proposed assistive device to be operated by gaze and brain waves should “be able to ‘translate’ thought about an action into its required stages.” Additional requirements are presented in Tables 3 and 4.

Clinical and demographic information.

The questionnaire study was completed by 23 participants with tetraplegia (16 males, 69%). Differences in gender proportions indicate that the sample is representative of the general SCI population, as 80% of people who have survived an SCI are men [50]. The mean age (and SD) was 49.22±15.8 years (range: 22–74). Additional details about employment and marital status, number of children, and employment status are provided in Table 1. Eighteen (77%) of the participants with tetraplegia had C4 and C5 spinal injuries, while the rest had higher injuries. The most common cause of injury was trauma (91.3%), of which the most common was traffic accidents (n = 11, 47%), followed by falls (n = 4, 17%).

Since ATCC can be used in bed, we asked the participant to report about the number of hours spent in bed per day. Since people with SCI report sleeping seven hours per day [51], we can deduct that people with SCI spend almost four extra hours not sleeping, in bed. Thus, although the questionnaires indicated that the preferred operating position is while sitting in a wheelchair, we did acknowledge that the assistive device should be designed to be operated in different positions. Interestingly, this requirement was extracted from the focus group and from the literature (See Table 1). Additional clinical data is provided in Table 2.

Social life and computer-use habits.

The mini-questionnaire results revealed that most of the SCI participants reported a decline in their social life (60%), engagement in leisure activities (78%), and having regular mobility (73%). The platforms most owned by the participants were personal smartphones (69%) and laptop computers (65%). However, the platforms most used were desktop PCs (34.8%) and smartphones (30.4%).

The following results are from participants who reported owing a desktop or a laptop PC (18 participants, 78.3%). These participants reported using the computer for a mean of 5±3.4 hours daily (min 1, max 12) and having 20.5±9.1 years of computer use experience (min 10, max 40). The most commonly noted uses were ‘communication’ (82%) and ‘work and productivity’ (82%). The most popular computer applications were ‘internet browser’ (72%) and ‘word processor’ (72%), but when asked to indicate the most important one, all participants indicated the ‘Internet browser’ (100%). In addition, the vast majority of participants reported using Windows OS (94%) as their operating system. Participants also indicated that they used the computer mostly at home (88%), while sitting on a motorized wheelchair (61%). The most common aspects of life to which computers were reported to contribute were education (72%), interpersonal interactions and relationships (54%), and work and employment status/potential (54%). The most commonly reported locations of pain and fatigue caused by current operation method were the neck (55%) and the shoulders (38%).

When asked to indicate the most common difficulties they encountered while using computers, participants most often noted the following: typing on the keyboard (70%), pressing two buttons at once (60%), and zooming and panning (50%).

Results of the mini-questionnaires taken from the QUEST 2.0 revealed that the most commonly noted aspects of the ATCCs were comfort (77.8%), durability (66.7%), and effectiveness (66.7%). Results of another mini-questionnaire showed that the most important technical aspects were independent operation (85%), ease of use (71%), noninvasiveness (57%), and productivity (57%).

The clinically related requirements extracted from the questionnaires study are presented in Tables 3 and 4.

Comparison to previously extracted clinically related requirements of new assistive devices.

Papers defining clinically related requirements for the design of new assistive devices are sparse. Medola et al. [1] and Caltenco et al. [7] administrated questionnaires to individuals with SCI to assess their experiences and problems and reported results similar to ours, but without spelling out specific requirements.

Harmo et al. [24] used a similar, two-way method that included questionnaires, interviews, and a literature review to assess the needs and problems of elderly and disabled people in relation to home automation and service robots. However, they did not specifically indicate, or discuss, which conclusions were based on which method. In accordance with our results, communication and productively needs and physiological problems such as reduced fine motor skills also emerged as key factors in their analyses.

Lenker et al. [52] reviewed papers on consumer-identified assistive technology related outcome variables and conducted a number of focus groups with end users to assess these outcomes. Unlike the distinct requirement definitions presented in the current study, their outcomes were less specific, probably due to their inclusion of diverse end user cohorts (e.g., visual impairments and mobility impairments) and their reference to non-specific technologies.

Finally, Bright & Coventry [23] derived mobility related automated assistive technology requirements for older adults from interviews of care-givers and technology ‘tea parties’ (qualitative measures). Their analyses yielded socio-emotional and psychological design guidelines for future assistive devices. As in the present study, their conclusions were specific to a certain cohort. Interestingly, the qualitative methods in both our study and theirs revealed the requirement that the device should "look as normal as possible."

Evaluation of the requirement extraction methods.

Very few papers discuss the advantages of the different methods used to produce user requirements for assistive or medical devices, and no papers were found that discuss their disadvantages. Garmer et al. [53], for example, identified the strengths of usability tests and focus groups, but did not mention any of their weaknesses. We argue that an effective discussion of the advantages of the combined method must also address the disadvantages of the different methods. Thus, we describe the advantages and limitations of each extraction method in general, and specifically with respect to the present study.

In general, the literature review provided us with existing knowledge and with an idea of the general framework of prior efforts similar to ours. It has the advantage of being based on the most up-to-date published knowledge from clinical trials, systematic reviews, and meta-analyses, among other studies. Conversely, it is limited to published topics, it is likely that standard publication cycle times prevent the most recent technologies from appearing. Furthermore, the narrative review is not as comprehensive as a systematic review.

The focus group method produces firsthand opinions of experienced health professionals. Its advantages are the provision of comprehensive clinical perspectives based on the professional knowledge and experience of personnel who work closely with patients with SCI. As such, it can provide insights concerning end users that would not be acquired otherwise. For example, in the current study, the focus group raised the possibility that computer use could be a coping strategy for users with severe depression. One limitation of the focus group method is its dependence on the background of participants. In our study, the psychologist, social worker, and to some extent the nurse, referred mostly to emotional, social, and motivational aspects, while for other medical professionals, physical aspects of SCI were the baseline for discussion. Furthermore, the requirements extracted from this method may be theoretical and general. For example, in the present study, the moderator explicitly asked group members to try to suggest every solution for SCI that they can imagine, without limiting themselves. In her experience, limiting the group in advance was not recommended and should be done only during the discussions. Nonetheless, some of the stated requirements stemmed from "wishful thinking" that could greatly improve SCI computer use but might not be feasible for an ATCC. With respect to the current study specifically, the single meeting and small number of pre-selected members might have limited the scope of extracted requirements. We would also like to acknowledge the potential added value of focus groups of potential end users (e.g., Lenker et al. [52]), which were not included in the present study.

Questionnaires enable a direct, structured investigation with potential end users. Because information is extracted from the actual targeted end users, this method has the advantage of providing an authentic picture of their current use habits, difficulties, and needs. This is especially important in the field of computer use, in which habits, difficulties, and needs can change relatively fast and findings can become irrelevant fairly quickly. On the other hand, questionnaire-driven requirements can represent somewhat "narrow" thinking, since the questions themselves imposes limits on potential responses. This limitation is alleviated to some extent by the inclusion of open-ended questions, though responses to these questions lack standardization. Beyond this, the generalizability of the questionnaires used in the present study was limited, since they were not tested for reliability and validity. The convenience sampling method also limited the representativeness of outcomes. Finally, when considering data from SCI participant questionnaires, requirements should be derived in relation to upper limb deficits. For example, the needs of a patient with residual hand movements may differ from those of a completely immobile individual. However, due to the relatively small number of participants with C2 and C3 injury level (Table 2), we chose to generalize the conclusions to upper limbs deficits.