Automatic sensors

Automatic fall detector sensors operate either directly or indirectly. The main components of most direct sensors are an accelerometer which detects the body's motion during a fall, and a microcontroller that sends a wireless signal to generate an alarm when a fall is detected. Direct sensor devices are worn either round the waist or on the wrist and manufacturers claim that they are unobtrusive and of low cost for the user (Campo and Grangereau Reference Campo and Grangereau2008; Miskelly Reference Miskelly2001; Prado-Velasco, Rio-Cidoncha and Ortiz-Martin Reference Prado-Velasco, Rio-Cidoncha and Ortiz-Martin2008). The differences between the various devices are their added features that aid the accelerometer in detecting falls and decrease the number of false alarms. Some examples of these added features include a tilt meter to determine the wearer's orientation after acceleration (Doughty, Lewis and McIntosh Reference Doughty, Lewis and McIntosh2000), a meter to determine whether or not an impact has taken place after acceleration (Noury et al. Reference Noury, Fleury, Rumeau, Bourke, Laighin, Raille and Lundy2007), a motion detector to determine the wearer's state of consciousness after acceleration (Bourke and Lyons Reference Bourke and Lyons2008), and a GPS tracking device in order to locate the wearer and enable a rapid response (Prado-Velasco, Rio-Cidoncha and Ortiz-Martin Reference Prado-Velasco, Rio-Cidoncha and Ortiz-Martin2008). Emerging data from pilot studies into the use of automatic sensor devices have revealed good rates of sensitivity to the detection of falls in experimental settings (Bourke and Lyons Reference Bourke and Lyons2008; Kangas et al. Reference Kangas, Konttila, Lindgren, Winblad and Jamsa2008; Lindemann et al. Reference Lindemann, Hock, Stuber, Keck and Becker2005), but no device completely eliminates the possibility of a false alarm.

Indirect sensors include motion detection devices such as bed occupancy sensors, floor pressure sensors, or inactivity sensors that use passive infra-red technology to detect when a person is not moving. Indirect sensors can be programmed to raise alarms when patterns of activity are recorded that deviate from a person's normal pattern of movements (Campo and Chan Reference Campo and Chan2002). For example, a bed occupancy sensor would need to be programmed so that an alarm is not raised during a nightly trip to the bathroom, but sensitive enough to raise an alarm if the sensor registers that the user was out of bed for a prolonged period of time. The use of indirect movement sensors may be effective in monitoring that a person has fallen and raising an alarm accordingly, but such sensors do not, of course, operate specifically as fall detection devices. In practice, there will be many other reasons why an alert is raised that have nothing to do with a person falling. In some cases, an alert will be raised when the person has not come to harm, but has simply deviated from a typical daily or nightly routine. As such, the likelihood of false alarms being raised with movement or inactivity sensors is likely to be significantly higher than with the use of direct sensors.

Video monitoring

Video monitoring involves real-time remote surveillance of the user while in the home setting, and may be considered to be a specific type of indirect automatic sensor. Older systems rely on video and audio communication using either a monitor or television and a touch-tone telephone (Miskelly Reference Miskelly2001). Some systems can relay the images to a secure website, and some cameras have the ability to pan, tilt, zoom and scan the accommodation. They may allow a remote viewer to monitor the user at any time, or alert the viewer when an alarm, such as a fire alarm, has been set off. Newer video systems are exploring less obtrusive real-time monitoring methods. These systems detect falls from the user's lying posture using scene analysis, or they detect abrupt movements using vector analysis software. Vector analysis involves detecting variations in a user's movements which are then sorted according to their direction and/or amplitude (Noury et al. Reference Noury, Fleury, Rumeau, Bourke, Laighin, Raille and Lundy2007). By focusing on the differences between successive images from the user's body positions and only transmitting the distinct variations that are consistent with a fall, these systems avoid relying on constant remote surveillance to detect when a fall has taken place. Video cameras using scene or vector analysis may also reduce the invasion of the user's privacy by using low image resolution, filtering the image into shape outline and motion vectors, or transmitting full images only once a fall is detected.

Autonomy

Respect for autonomy, interpreted here as the right of each individual to control her own life and be free from unwanted interference, is a central principle of health and social care ethics. The fact that a person is older, and more vulnerable to falling, does not of itself diminish the importance of this principle. This principle has implications for three main aspects of the use of fall detectors: consent for their use; control of their use; and the nature of the response. It might also be seen as the principle that underlies the importance of privacy which will be discussed in the next section.

If the user has the capacity to decide whether, and how, a fall detector should be used then such a detector should be used only with her valid consent. Most liberal societies insist on this principle of consent in the setting of health and social care, and allow adults with capacity to refuse beneficial, and even life-saving treatment. Valid consent involves three elements: the person has the capacity to make the relevant decision, is appropriately informed, and is able to make a free, voluntary decision. The issue of informing users about the use of fall detectors may not be straightforward in practice. Consider a woman with early mild dementia who is at risk from falls and aware of this risk. Her family suggest that a video monitoring system is put in place, a camera in each room of her flat. The system will send real-time images to a secure website accessible only to a few specified family members. The woman, let us suppose, is capable of understanding this and agrees to the video surveillance. She appears both to have the capacity to consent and to have been fully informed. One camera surveys her bedroom. Because of the dementia she readily forgets that some of her family can see exactly what she is doing. In the privacy of her bedroom she behaves in ways that she would not want her family to observe. Was her consent to have the system installed valid? What exactly is the necessary information for valid consent? If, at the time of giving consent, she is not able to imagine all the situations that might be observed, does she have the capacity to consent?

Even if the user gives valid consent for a monitoring system to be set up, the principle of autonomy would require service providers or carers to respect any change of mind or decision to interrupt the monitoring. Some devices have to be worn around the waist or upper chest; they are not discreet and could be stigmatising or distressing for the user. It is reasonable to believe that a person who originally agreed to wear such a device may change her mind as there is some evidence that wearing electronic devices can cause older adults to feel embarrassed or fearful (Robinson et al. Reference Robinson, Hutchings, Corner, Finch, Hughes, Brittain and Bond2007). Clearly the ability for users to change their mind might decrease the value of the monitoring system and might increase the risk of harm, but considerations of autonomy emphasise that there would need to be very good reasons to justify overriding personal control of the use of the device. Indeed, the principle of autonomy would challenge arguments about the use of non-removable devices in scenarios where the user was judged to be at very high risk of falling. Much of the literature, however, interprets non-compliance as a problem – a result of inadequate device design or systemic failings, for example – rather than as the valid expression of the user's free choice (Alwan et al. Reference Alwan, Rajendran, Kell, Mack, Dalal, Wolfe and Felder2006; Levine and Tideiksaar Reference Levine and Tideiksaar1995; Nelson et al. Reference Nelson, Powell-Cope, Gavin-Dreschnack, Quigley, Bulat, Baptiste, Applegarth and Friedman2004). This failure to ensure that the principle of autonomy lies at the heart of all decisions relating to the use of fall detectors could be explained by the claim that telecare only promotes forms of self-management that concord with compliance to medical instruction and standardised care regimes (Schermer Reference Schermer2009).

In the interests of respecting their autonomy, users, if they have the capacity, must agree not only to the nature of the monitoring but also the responses that will be initiated should a fall be detected. This would require information to be provided in relation to a number of uncertain, and potentially difficult, questions: Who is told about a fall? Who has access to their home if they do fall? What might be the longer-term consequences of people knowing about falls? In relation to the last question, for example, data from alarms that are raised by automatic sensors or through video monitoring could lead family members to become so concerned that they take steps to move their relative into a care home. As a corollary to the decision about the nature of the response invoked once an alarm is raised, service providers would need to give careful consideration to ensure that users were fully informed about potential implications of the detection, alert, and response process being proposed.

We have emphasised some ways in which fall detection devices might interfere with autonomy. These devices may also promote users' autonomy, for example through increasing independence for longer periods of time, enabling users to continue directing their own life plan to the fullest extent, and without unnecessary interference from external services. Given that the evidence suggests that most people with care needs would like to receive care in their own home setting rather than moving to a residential care or nursing home (Help the Aged 2008), the use of fall detectors is likely to respect autonomy, assuming that the challenges of fully informing the user in the process of gaining consent can be overcome.

Privacy

All fall detectors compromise the users' privacy, although this occurs in different ways and to different extents. Video monitoring, where the viewer has the ability to see exactly what the user is doing, can be a highly significant invasion of privacy, as in the example of the woman with mild dementia discussed above. The extent of this invasion may depend on which rooms contain a camera. It seems intuitively likely to us that most people would prefer cameras not to be placed in private settings such as bathrooms or bedrooms, although these may be rooms in which the risk of falling is most significant. This issue could be exacerbated further if data about third parties are collected by video cameras or a motion-detection device such as a bed occupancy sensor. Such data might include visual recordings of users and their homes, or it might include the collection of other forms of personal data, such as information about partners' daily activities and routines.

Even if users lack the capacity to consent to, or refuse, the use of monitoring, they still have strong privacy interests and these should be violated only with very good reason. Further empirical evidence is required to flesh out the nature and scope of user and public concerns about privacy in relation to the use of fall detectors. A corollary of this concern is that in those situations where considerations of risk reduction justify the use of monitoring, privacy should be compromised to the least possible extent. Here, the use of certain technologies would appear to be more morally defensible than the use of other technologies. Cameras that use intelligent software to detect falls without providing the viewer with a continuous video of the user will reduce the invasion of privacy, as will automatic sensors. However, personal data are still produced by these devices, such as the location of the user in the home (Noury et al. Reference Noury, Fleury, Rumeau, Bourke, Laighin, Raille and Lundy2007). Even the information that the user has fallen impacts negatively on privacy interests and needs to be justified.

The ‘added features’ of fall detectors can also change the nature and extent of the invasion of privacy. GPS tracking systems used as part of a fall detection system will provide the viewer with information about the user's movements outside the home, unless the system is designed to operate within a limited range. A system that only provides this information after a fall is detected would reduce, although not eliminate, such an invasion of privacy. Finally, privacy concerns would be raised if the data collected by monitoring systems were used for purposes other than that for which they have been designed, or to which the user had consented. Concerns have been raised in empirical studies that commercial companies could acquire such data to direct marketing strategies to sell aids, adaptations or other devices, or that governmental agencies might use data about a person's functional abilities to make changes to a person's benefit payments or eligibility for publicly-funded social-care support (Percival and Hanson Reference Percival and Hanson2006; Percival, Hanson and Osipovic Reference Percival, Hanson, Osipovic, Loader, Hardey and Keeble2009).

Benefit

Reviews of the literature have demonstrated that the best form of intervention for falls in assisted living and nursing home settings is an individualised system catered specifically to the user (Tilly and Reed Reference Tilly and Reed2006), within a broad interagency strategy focused on reducing the frequency of the person falling (Martin Reference Martin2009). The main purpose of fall detectors is to improve safety. There is some evidence that they decrease users' fears about falling (Brownsell and Hawley Reference Brownsell and Hawley2004a), although this is far from conclusive (Brownsell and Hawley Reference Brownsell and Hawley2004b; Laviolette and Hanson Reference Laviolette and Hanson2007). Evidence from qualitative studies suggesting that electronic devices worn on the body do not make people feel safer, but rather more vulnerable – either because they are constant reminders of the dangers that they face (Brownsell and Hawley Reference Brownsell and Hawley2004a), because they feel embarrassed, or because they feel that they are more likely to be the victim of theft (Robinson et al. Reference Robinson, Hutchings, Corner, Finch, Hughes, Brittain and Bond2007). However, if fall detectors do reduce the fear of falling, they will likely impact positively on the user's quality of life, not only because of the positive effect of making people feel less afraid, but also because of the likelihood that the person will engage in a more active lifestyle, be involved in an increased range of activities, and be better integrated in the community. Benefits will also be realised if, as they are intended to do, fall detectors enable people to enjoy greater freedom, and remain in their own homes for longer than it would otherwise be possible or safe to do so. Given the evidence of two recent surveys (Landau et al. Reference Landau, Werner, Auslander, Shoval and Heinik2009; Rialle et al. Reference Rialle, Ollivet, Guigui and Herve2008), an interview study (White, Montgomery and McShane Reference White, Montgomery and McShane2010), and a focus group study (Landau et al. Reference Landau, Auslander, Werner, Shoval and Heinik2010) of caregivers' attitudes towards different kinds of telecare devices, it is apparent that fall detectors may also benefit others by providing reassurance to carers and reducing the economic or psychological burden of direct caring.

There are, however, a number of possible disadvantages associated with the use of fall detectors. First, it has been claimed that remote monitoring and detection technology will lead to a reduction in human interactions and an increase in social isolation (Lowe Reference Lowe2009; Marshall Reference Marshall2000) as, ironically given the aim of telecare, people become institutionalised in their own homes (Eccles Reference Eccles2010). For example, a care-giver may decrease the number of visits because of the ability to monitor remotely, and the delivery of care at distance would be detached from the value of the interpersonal act of caring itself. The empirical evidence of the impact of technological devices and telecare on social isolation, user–provider care relationships and interaction is conflicting. On the one hand, this intuition appears to be substantiated by research (Tang, Gann and Curry Reference Tang, Gann and Curry2000), with concerns also being raised by service providers that telecare will be used to justify cost-cutting and redundancies in professional care services (Percival and Hanson Reference Percival and Hanson2006; Percival, Hanson and Osipovic Reference Percival, Hanson, Osipovic, Loader, Hardey and Keeble2009). On the other hand, a recent study in the Netherlands of nursing practices has shown that new technologies can facilitate social and affective relationships between care providers and care recipients (Pols Reference Pols2010; Pols and Moser Reference Pols and Moser2009).

Secondly, fall detectors may lead to increased, rather than decreased, risks from falls either because a user might be encouraged to engage in behaviours that increase the likelihood of falling (Brownsell and Hawley Reference Brownsell, Hawley, Wootton, Dimmick and Kvedar2006), or because too much reliance is placed upon them. This problem would be made worse by the possibility that the user switches off the device. Thirdly, instead of decreasing the chance that a person may be moved into a care home, a fall detector may increase this chance through making relatives aware how frequently the person falls. This might, or might not, be in the person's best interests. For example, a user who falls may alarm relatives unduly and even if a care home were to provide a safer environment it might nevertheless be in the person's overall best interests to remain in their own home. This claim cannot, however, be substantiated empirically as there is currently limited evidence of relatives' attitudes towards the use of fall detectors or telecare more broadly. Finally, there are significant concerns about the likelihood of false positives and false negatives, which have been documented in empirical studies of direct sensors (Brownsell and Hawley Reference Brownsell and Hawley2004a), and raised as serious concerns by both users and care providers (Brownsell and Hawley Reference Brownsell and Hawley2004b).

Use of resources

The remote monitoring devices and responses that follow the detection of a fall require resources of time and money, regardless of whether the fall detector is integrated within a family carer or professional system of support. Inevitably, the use of such resources raises the question of whether, in a particular instance, using a detector is a good or a poor use of resources. There may also be an issue of whether the use of the detector is fair on the person, or people, who take the brunt of the burden to respond. The mere possibility of using a detector may have a deleterious effect on a carer, relative or service. For example, a daughter who lives near her frail father may visit him every day. When she goes on holiday another family member or friend takes over the visits. New technology, by providing a means by which the daughter can continue to check on her father, may deprive her of a much-needed break from her responsibility, or alternatively fundamentally change the way she perceives that responsibility. Again, empirical research with relatives is required to substantiate this claim, and assess the impact of resource-use concerns associated with the act of responding.

State resources and private insurance have to be used efficiently. Even when it is in the best interests of the user for such resources to be used, it does not necessarily follow that this will be the best use of those resources. Every time social- or health-care services respond to the detection of a fall, there are opportunity costs associated with this action. The response is at the expense of something else, helping another patient, for example. False positives will be particularly wasteful of resources. In some circumstances these services may choose whether or not to take on the role of monitoring or responding, but this will not always be the case. Relatives may draw more heavily upon primary care services as a result of making use of the detector, for example making an appointment for their father to be checked by the general practitioner every time a fall is detected. Without the detection system many of these falls are likely to occur without the family member's knowledge. Again, such appointments have an opportunity cost, and might not be judged to be a good use of the doctor's time.