Fall Incidence as the Primary Outcome in Multiple Sclerosis Falls-Prevention Trials: Recommendation from the International MS Falls Prevention Research Network

When developing a recommendation for a specific primary outcome for a falls-prevention program, a range of options should be considered. These include fall risk, the number of people falling, total number of falls in a given period, rate of falls, and injurious falls. In the MS population, where relapses and fatigue are an issue, it may be useful to also capture the number of falls per day and to examine the pattern over a calendar month.

To date, a limited number of falls-prevention and fall risk reduction programs for people with MS have been investigated (Table 1).12–17 Owing in part to the varied nature of investigations, several different fall-related outcomes have been used. These include balance, dynamic mobility, balance confidence, adoption of falls-prevention strategies, composite measures of fall risk (ie, physiological profile assessment [PPA] of fall risk), retrospective recall of falls, and prospective falls diaries. The outcome measures used have a range of strengths and weaknesses. For example, the adoption by people with MS of falls-prevention strategies is a potentially positive outcome, and interventions have been shown to improve relevant knowledge and skills among people with MS.15 However, people who use fall risk reduction strategies still fall frequently,18 suggesting that the adoption of fall risk reduction strategies should not be the sole outcome for falls-prevention studies.

Three meta-analyses19–21 have focused on risk factors for falls in people with MS, and all have concluded that balance deficits, progressive MS, and mobility aid use are associated with and predict falls status. Despite the commonality of conclusions, the ability of balance and mobility measures to predict future falls continues to be unclear. To date, the limited data from prospective studies have been inconsistent, and no studies have presented analyses of individual measures that can be considered excellent in predicting which MS patients will fall. For example, in the largest prospective study to date, the best model to explain falls included the PPA, Expanded Disability Status Scale (EDSS), and Ashworth scale, and had only fair to good predictive ability (sensitivity 70%, specificity 69%), maintaining the possibility of a significant number of false-negative values. Given the limited number of studies as well as lack of consensus in the findings, at this time it is not possible to definitively propose a measure of fall risk for people with MS that can be used as an outcome measure in falls-prevention studies. Additionally, there is evidence from the research on older adults that reducing fall risk does not necessarily lead to a reduction in the number of falls.8,22 Therefore, the IMSFPRN proposes that the primary outcome should be a reduction in falls rather than fall risk.

In the MS literature to date, a number of definitions of a fall have been used, and these are presented in Table 2.23–31 In a systematic review of the literature on falls in older people, Hauer et al.32 found that the lack of definition in some studies and the range of definitions in others precluded comparison between studies. This review led to the definition proposed by ProFaNE,10 which recommends that a fall be defined as “an unexpected event in which the participants come to rest on the ground, floor, or lower level.” This definition has been used verbatim by several MS studies to date, and variations of it by others. The advantage of this definition is that it is straightforward and minimizes subjective judgment by the researcher and/or participants. Consequently, the IMSFPRN proposes to adopt this definition.

One possible successful outcome of a falls-prevention intervention is a reduction in the number of people falling; a second is a reduction in the number of falls that each person has. Therefore, the IMSFPRN is interested not only in whether someone fell during a defined period, but also in how many times they fell. Within the MS population, an individual's number of falls is very important given the high percentage of recurrent fallers.21 In analyzing falls data, the absolute numbers for each variable may be recorded, which subsequently may be used to classify each participant's fall status (for example, as a “faller” or “frequent faller”). Lamb et al.10 recommend that both the number of falls and the number of fallers be presented, as well as the number of frequent fallers. However, they fail to specify the number of falls a person must have in order to be defined as a “frequent faller.” Studies of people with MS to date have defined fall status using a number of different metrics. For example, in prospective studies Gunn et al.33 used the criterion of having two or more falls to be classified as a faller, Nilsagård et al.23 considered a faller as anyone having one or more falls, while Hoang et al.34 used the criterion of three or more falls to define frequent fallers. Interestingly, overall the results of these investigations highlight similar fall risk factors. As it only takes one fall to cause injury or related consequences, we propose that a faller be defined as anyone who records one or more falls during the observation period. We also propose that a multiple or frequent faller be defined as anyone reporting two or more falls during that period.

The time frame in which falls occur is important and adds to the information on the type and severity of the problem; for example, three falls in 1 month is a potentially different problem than three falls in 1 year. In order to address this, and the issue of defining frequent fallers, the computation of “fall rate” is recommended in the literature on the elderly.10 This is calculated as the number of falls per person per year and has been used as the primary outcome in a number of Cochrane reviews of the literature on the elderly and stroke-related falls.9,35,36 Falls per person per year can be computed from any time period of diary recording, allowing comparison across studies with different reporting periods. We therefore propose that MS falls-prevention intervention studies use the number of falls per person per year as the primary outcome, with the number of falls, number of fallers, and frequent fallers as additional outcome variables. The length of time for recording falls will be considered in a later section.

Several studies of falls in people with MS have highlighted discrepancies between reported/recorded fall prevalence using retrospective recall and fall incidence using prospective falls diaries. In a study of 76 community-dwelling people with MS in Sweden,23 the reported/recorded retrospective fall prevalence was 36%, while the reported/recorded prospective fall incidence for the same cohort was 63%. Similarly, in a UK sample of 148 people with MS,33 57% of the sample retrospectively reported falls, while 70.3% recorded falls prospectively using diaries. These studies are further supported by the findings of Dibble et al.,37 who reported that 63% of their sample underestimated the number of falls they had recorded during a 12-month period. These are not unexpected findings considering the high prevalence of cognitive dysfunction in people with MS.38 Therefore, the IMSFPRN recommends that future research with the aim of reducing falls in people with MS should prospectively record falls using falls diaries.

Falls diaries can provide valuable information regarding the occurrence of falls. If designed correctly, they also can provide data on various other aspects of the falls, including the context (time of day, activity at the time, location, etc.), the causes (legs giving way, slip, trip, loss of balance, etc.), and consequences of the fall (injury, activity curtailment, care, etc.). When incorporating a falls diary into a falls intervention, care should be taken to balance the burden on the participant with maximizing data collection.

In order to inform the design of falls diaries, we reviewed those diaries used by four MS research groups to date.4,33,34,39 All used daily reporting with different formats for recording falls. Three used a 1-month-view calendar with space to write the number of falls beneath each date,4,34,39 while the other group used a format with one page per day.33 For all four groups, diaries were presented to the participants in a pack or binder at the start of the trial, and the investigators suggested that these constituted an important visual reminder to record falls. Diaries were generally returned by mail on a monthly basis, although one group used a biweekly time frame for returns.33 One group using monthly returns used biweekly phone reminders to optimize the accuracy of falls recording,39 and all followed up by phone or e-mail if the diaries were not returned within 7 to 10 days of the expected date.

The reporting period and frequency of returning diaries is a balance between burden and inconvenience to participants and loss of accuracy of data due to memory problems. For instance, if monthly returns are used, the participants may have forgotten to record for several weeks before it is noticed that a diary is missing, and recall over many weeks to subsequently complete the diary may result in inaccuracies. Biweekly returns may be more accurate in terms of numbers of falls but result in increased postal charges and place extra demands on the participant. Owing to the prevalence of cognitive dysfunction in people with MS, we suggest that biweekly phone reminders be used, and that diaries be returned on a monthly basis, with a further phone reminder if the diary is not received within 7 days of the expected date. A sample falls diary is included in Supplementary Appendix 1, which is published in the online version of this article at ijmsc.org. While this diary is based on the best available information at this time, it does require validation.

Given the young age of people with MS who fall,21 and their potential familiarity with technology, the use of other forms of reporting and reminders such as text alerts, mobile phone applications, or other fall-sensing devices should be evaluated. To date few studies have been published evaluating these technological approaches in people with MS.

It is not clear for what time frame the falls diaries should be kept, such as 3 months, 6 months, or 12 months. The reporting period needs to be long enough to capture falls before intervention and a change in falls after intervention, but short enough to avoid excessive interruption of everyday life. Fall rates in people with MS have been shown to be significantly higher than those observed in older people.40 For example, in one study,41 fall rate for the total sample (N = 148) was estimated at 18.4 falls per person per year. When calculated for fallers only (70% of the total sample, n = 104), the fall rate equated to 26.2 falls per person per year (or 2.15 falls per person per month). Based on these data, it is suggested that 3 months is long enough to capture falls frequency before intervention. However, the reporting period should also consider the mechanism of the intervention; for example, interventions targeting behavior change may require a longer time to have an observable impact. Longer-term follow-up after the post-intervention phase should also be considered in order to evaluate the long-term effect of the intervention on fall rate.

The causes, context, and consequences of falls are also frequently included in falls diaries, with researchers either asking about the most recent fall as an example or asking for details of each fall that occurs. The accuracy of self-reports of causes of falls is not known. As some participants have up to 63 falls in a 3-month period,41 the recording of causes and context for each fall may be overly burdensome. Once again, the balance between the data set and participant burden comes into play. While extensive detailed information on these factors might be important in studies intended to gather a wide range of information to inform intervention development and report predictors of falls, it may be less important in intervention studies where rates of falling are the primary outcome.

Of the falls diaries we reviewed, all but one recorded the context (location, time of day, activity being performed at the time, etc.) and the perceived cause (slip, trip, knee gave way, etc.) of falls. A change in falls context may be a positive outcome for an intervention trial. A person with activity curtailment who rarely left the house and whose pre-intervention falls were all indoors might consider it more positive if their falls happened outside of the home during social or recreational activities, suggesting a possible reduction of their activity curtailment. Therefore, while it is not a priority, the IMSFPRN recommends that the context be considered for recording. The context may allow for individualization of any intervention.

Similarly, a change in the cause of falls might also indicate improvement; for example, a person who falls because of their knees giving way may not experience that after intervention if quadriceps-strengthening exercises are a key component of the intervention. It is probably unreasonable to burden participants with recording these for each fall, but it may be feasible to record the perceived cause of falls for two falls in each month.

The consequences of falls, in terms of injuries or health-care use, are an element of falls that is essential to record. Estimates of injurious falls for people with MS vary. The literature clearly indicates that falls in people with MS are more injurious than those in people without MS. For instance, the ProFaNE group10 proposes that injurious falls be recorded as the “number of radiologically confirmed peripheral fracture events per person year.” Several studies have suggested that the fracture rate among people with MS is considerably greater than in the general population. A Danish study42 comparing people with MS to the National Hospital Discharge register found incidence rate ratios of 3.36 (tibia), 6.66 (femur), and 3.20 (hip) for fractures in people with MS. A study of 721 veterans with MS reported that 2.8% of them had an injurious fall coded in their medical record, compared with 1.5% of veterans without MS.1 A British study comparing people with MS to the UK General Practice Register43 found a threefold increased risk of hip fracture, confirming that fractures due to falls are important to record and report. However, falls of a much lesser severity can also result in considerable costs to the person and to the health-care system. It appears that these less injurious falls may be more common than severe injuries. For instance, in a phone survey of the North American Research Committee on Multiple Sclerosis database,44 23% reported seeking medical attention for a fall, while another survey of community-dwelling people with MS revealed that 58% of falls were found to be medically injurious.2 Additionally, given that many falls are not reported to health-care professionals in both the MS2 and elderly45 populations, recording medical attention for falls may not be an accurate representation of injurious falls. For that reason, the IMSFPRN proposes that falls diaries contain all self-reported injuries in addition to health-care use information. We also recommend that specific training in the completion of the falls diary for all study participants be implemented at the start of the trial, with the biweekly phone calls used to verify that information is being recorded accurately.

The IMSFPRN recommends that a reduction in falls should be the primary outcome of falls-prevention programs. Consistent with recommendations from ProFaNE,10 a fall should be defined as “an unexpected event in which the participants come to rest on the ground, floor, or lower level.” At this time, the IMSFPRN recommends using prospective falls diaries to record the number of falls each day. Participants should be reminded by phone every 2 weeks to complete the diaries, which should be returned on a monthly basis with a further reminder if not received within 7 days of the expected date. The length of reporting may depend on the nature of the intervention but should be at least 3 months. Falls data should be presented as the rate of falls per person per unit time, in addition to the total number of falls and the number of fallers. Information on the consequences of each fall—that is, self-reported injury and health-care use—should also be recorded and reported. The IMSFPRN recommends that information on the causes and context of falls should be captured for the first two falls in each month.