The study aimed at describing the nursing staffs' opinion of what causes falls, documentation and report related to fall-risk and fall incidents, interventions and their experiences and reactions associated with working with older people with dementia. The discussion of the results is presented according to these four areas.
Causes of falls are, according to the respondents, most often related to the individual's condition. Four causes stand out as the most frequent: forgetfulness related to physical impairment, impaired mobility, anxiety, and inability to call for help. These results are in line with other studies [14, 15, 34, 35]. Impaired short-time memory makes people with dementia forget their physical impairment, and falls can occur when they are getting up from the bed or a chair [17, 29, 36].
Those who can get up from a chair, but cannot stand upright or sit down unaided are particularly exposed to fall-risk [16, 17].
Anxiety was reported to "often" cause falls among the respondents. This results are in line with other studies showing that anxiousness and confusion are symptoms that evidently precede falls [22, 34].
Hyperactive, restless and paranoid behaviour may characterise some persons with dementia, and it is reported that residents who have the highest risk of falls are wanderers . As opposed to this the nursing staff in this study did not explicitly report wanderers as a risk group.
Environmental factors were not reported as a frequent cause of falls by the respondents. For comparison, studies suggest that external, physical factors are estimated to precipitate about 8% of falls in residential care facilities , while others have found that 14,5% of falls are associated with environmental factors . A Swedish study indicates that well-planned furnishing and the use of colour to achieve a plain, clearly-defined environment reduce fall incidence in residential care facilities . An explanation to the fact that the respondents do not emphasise this issue may be that the facilities in the nursing homes in question are modern, surveyable and well customised for persons with dementia.
Disturbing behaviour of co-residents "sometimes" precipated falls according to the nursing staff in question. Previous studies do not point out mistreatment or disturbances from other residents as a significant precipitant to falls , but if these kinds of environmental factors are associated with anxiety and anger they could increase the odds of inducing falls, according to French et al. .
The respondents seem to emphasise staff-to-resident ratio as an indirect cause of falls to a higher degree than how the staff plan and organise their work. These factors probably interact. According to a study by Pellfolk et al.  there is no difference between fallers and non-fallers regarding staff-to-patient ratio. However, the factors communication, management policies and teamwork are considered to be of importance for successful fall management . Lack of staff influences the staff's priorities in daily work and may force them to organise in ways that are efficient regarding the basic tasks in the care unit. To find time to discuss systematic routines of fall prevention and implement these in the care units may be a challenge in a busy working day.
The majority of the respondents in this study reported having sufficient knowledge about fall prevention. However, there were differences between the compared groups regarding their reports on what causes falls. As previously described, respondents with longer experience are significantly more aware of the condition of the floors as directly or indirectly causes of falls. According to Lundin-Olsson et al.  nursing staff's assessment of fall-risk seems to be fairly accurate in predicting falls in residential care facilities. This may indicate that the attention to environmental risk-factor evolves with years of experience, rather than educational level.
The two most frequent fall-preventing interventions reported by the respondents were conversation and closeness and assistance with personal hygiene. Impaired cognition and ADL function are evidently risk-factors . It also seems that the majority of falls occur in the residents' own rooms, unwitnessed . Thus, the nursing staff's choice of interventions seem appropriate.
Comparisons between the sub-groups showed significant differences regarding what they preferred as adequate interventions. In general, staff with more experience in health services chose assistance with personal hygiene as an intervention to a significantly higher degree than less experienced staff. RNs chose conversation and closeness as an intervention to a significantly higher degree than the ENs. The results do not describe whether the choice of interventions were based on evidence or the staff's personal experiences, but in this case education seems to make a difference in how fall prevention is carried out.
Although nearly all respondents reported that they had sufficient competence in fall prevention, the findings did not describe a systematic strategy with multi-factorial and individually-targeted risk assessment and intervention, which is recommended to reduce fall rate in an institution [1, 3, 14, 40]. However, the multiple experience-based and single-targeted interventions conducted by the nursing staff may be important knowledge, as a basis for the development of more feasible routines to prevent falls.
How to ensure a qualified and co-ordinated nursing staff may be an interesting subject of research. So far there is no strong evidence that educational programs for nursing staff have any significant impact on reducing fall rate among people with dementia [3, 14]. Implementation of models of interventions have proved to be difficult, and it is suggested that the most successful strategy appears to be changing attitudes of nurses in order to increase the focus on fall prevention .
Routines of documentation and report was an area where the respondents had considerable variation in perceptions. The disagreement on this issue may be understood as lack of routines or inadequate following-up of existing routines. The impact of multi-factorial risk-assessment and interventions seems to presuppose a comprehensive co-ordinated group of health workers . The respondents in this study discussed fall incidents on a regular basis, but only 60% register fall-risk in the individual care plans. As pointed out in different studies the first step in the prevention of falls in nursing care is systematic, individual risk assessment [14, 42]. This presupposes adequate communication between the nursing staff within a care unit, and cooperation between different professions and units in the nursing home. To obtain this, a sufficient number of qualified staff who are organised with feasible routines of documentation and reports are required. Lack of routines of documentation and report related to fall-risk assessment may cause an inattentive nursing staff. A consequence of this may be that fall-preventing interventions are not planned or conducted before it is too late.
Nursing staffs' experiences and reactions when attending fall situations is a subject that has been given little attention in other studies. In the present study ENs and RNs only seldom experienced being present at fall situations or preventing falls. When falls occurred, the majority of the respondents seldom felt stress, unease or guilt related to attending fall situations, but RNs feel stress to a significantly higher degree than ENs. These results oppose some studies which describe nursing staffs' experience of stress related to being responsible for the residents' safety [27, 29]. However, when one third of the respondents "sometimes" or "often" felt unease when they experienced these situations, the findings should be of interest for further exploration. Rush et al.  describes acute nurses' experience with resident falls, and found that this created considerable stress for nurses which prompted them to use a range of coping strategies. Knowing that the residents are safe has the potential to resolve the tension between patient safety and independence 
Although the nursing staff in this study considered that prioritising the residents' freedom of movement "sometimes" or "often" cause falls, they prioritised autonomy and freedom as often as they chose a strategy of control. One type of protection and control is physical restraint. It is reported that nursing staff often use restraints to protect these residents, although the effectiveness of these interventions has not been proved . Previous studies have found that there is a comprehensive on-going assessment by the nursing staff to balance the residents' autonomy-versus-control in order to minimise fall-risk . It is argued that in this context there are some essential values and norms that should be observed in an ethical evaluation of physical restraint [13, 45].
The RNs to a significantly higher degree prioritised the residents integrity and autonomy as a risk factor. These differences may be explained by the fact that years of experience is important for developing skills in clinical observation and assessment, while nurse education aims at developing a higher degree of ethical reflection, evidence-based knowledge and critical thinking.
The four nursing homes were quite similar with regards to staff-resident ratio, buildings and other enviromental factors. In total the residents in these care units for persons with dementia represented a wide variation related to physical and cognitive impairment. Thus the findings represent a representative cross-sectional picture of the nursing staffs' challenges related to caring for people with dementia at risk of falling.
The response rate was 66%, which is acceptable for a survey . Study limitations are related to a small sample and the sample distribution (ENs: n = 40, RNs: n = 14). Further studies in larger scale are required.
The questionnaire was developed according to previous research on the subject . Hence the use of sedatives is not listed as a possible cause of falls, but is related to the nursing staffs' interventions. On the other hand, side effects of sedatives, such as dizziness and gait or balance problems, are listed as options in the questionnaire.
The questions about checking remedies had a low response rate (27%) and were insufficiently filled in by some of the respondents. Whether this is related to the formulation of the questions or a general doubt regarding these routines is difficult to know. However, the findings indicate divergent opinions within the nursing staff that may call for action with regards to routines in the care units in question.
To validate the questionnaire the research team and head nurses participated in the formulation, and pilot tests were conducted. Still, further testing of the instrument is required for validity and reliability.