Systematic Review |
Systematic review of definitions and methods of measuring falls in randomised controlled fall prevention trials
1 Robert-Bosch Krankenhaus/Robert Bosch Stiftung, Stuttgart, Germany
2 Bethanien-Krankenhaus/Geriatrisches Zentrum a.d. Universität Heidelberg, Heidelberg, Germany
3 Warwick Emergency Care and Rehabilitation, Centre for Primary Health Care Studies, University of Warwick Medical School, University of Warwick, UK
4 Kadoorie Critical Care Research Centre, John Radcliffe Hospital, Oxford, UK
5 School of Nursing, Midwifery & Social Work, University of Manchester, Manchester, UK
Address correspondence to: K. Hauer, Bethanien-Krankenhaus/Geriatrisches Zentrum a.d. Universität Heidelberg, Heidelberg, Germany. Fax: (+49) 6221 319 1505. Email: khauer{at}bethanien-heidelberg.de
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionKey pointsReferences |
|---|
Objective: to review systematically the range of case definitions and methods used to measure falls in randomised controlled trials.
Design/methods: a Cochrane review of fall prevention interventions was used to identify fall definitions in published trials. Secondary searches of various databases were used to identify additional methodological or theoretical papers. Two independent reviewers undertook data extraction, with adjudication by a third reviewer in cases of disagreement.
Settings: community-dwelling and institutionalised older persons.
Results: 90 publications met the predefined inclusion criteria. Of these, 44 provided no definition of the term fall. In the remainder, there were substantial variations in the definition and methods of measuring falls. Reporting periods ranged from 1 week to 4 years with only 41% using prospective data collection methods.
Conclusion: the standard of reporting falls in published trials is poor and significantly impedes comparison between studies. The review has been used to inform an international consensus exercise to make recommendations for a core set of outcome measures for fall prevention trials.
Keywords: systematic review, methodology, fall-related outcomes, fall prevention trials, elderly
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionKey pointsReferences |
|---|
The prevention of falls in older people is a public health target in many countries around the world. A large and rapidly increasing number of randomised controlled trials of falls prevention have been published [1–90], which have contributed significantly to identifying effective and cost-effective interventions [91]. However, trials are a time-consuming and resource-intensive endeavour. It is essential that as much as possible is learnt from the many trials being undertaken, in the shortest possible time span. Systematic literature reviews and meta-analysis have proven effective in this respect. Although trials have a common target of fall prevention, the definition of a fall, the method of identifying when a fall has occurred, the details recorded, and the method of analysis appear to vary. This lack of standardisation represents a serious methodological pitfall for the evaluation and interpretation of prevention strategies and a more homogeneous methodology has been repeatedly requested [92, 93].
A first step to standardise the definition, measurement and analysis of falls is to summarise the methodology used in controlled studies. Therefore we conducted a systematic review of the measurement methods used in randomised controlled trials that report falls as a study endpoint.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionKey pointsReferences |
|---|
Search strategy
Literature searches were an exact replication of those used in the Cochrane Review of interventions to prevent falls in older people [91] (with permission). Studies published until January 2005 were included if they were randomised controlled trials recruiting in community or institutional settings.
Data sources
The Cochrane musculoskeletal group specialised register, Cochrane Central Register of controlled trials, MEDLINE, EMBASE, CINHAL, and reference lists of identified articles were searched. No language restrictions were applied. Further trials were identified by contact with researchers. More details of the search and methodology are provided elsewhere (http://www2.warwick.ac.uk/fac/med/healthcom/emergencycare/research/profane/addendum_to_consensus.pdf page 15).
Data extraction
The definitions and method of recording falls were extracted by two independent reviewers (E.J., K.H.) using a database designed to support standardised extraction [94]. Disagreement was resolved by third party adjudication (S.L.). Falls definitions were analysed for clarity in the description of the event and for the sources of variation between definitions (such as the inclusion or exclusion of falls due to certain medical events). Methods of collecting and recording data were categorised into prospective, retrospective or record searching.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionKey pointsReferences |
|---|
Fall definitions
Ninety papers met the review inclusion criteria [1–90].Of these, 44 did not provide a definition of a fall. There was no single definition that stood out as a gold standard. The most frequent citations were the definitions of the Kellogg working group (n = 8) [95] and FICSIT collaboration (n = 9) [96]. However, some papers referencing these definitions had made amendments to or changed the original [23, 37, 46, 48, 57, 67, 68, 77, 80, 81, 88]. All other definitions (n = 28) differed from each other or were unreferenced.
Most definitions used a combination of topographical, biomechanical and behavioural components to describe a fall. A homogeneous component of definitions was that an individual must come to rest at a lower level, but the description of the level varied. Some studies counted a fall only if it resulted in body contact with the ground or floor. Other definitions also included furniture and wall contact. From a behavioural perspective, falls were defined variously as unintentional, inadvertent, involuntary, or accidental. Another significant and important source of variation was whether falls that were attributable to acute medical events such as syncope and seizures should be included or excluded. Falls resulting from environmental hazards or overwhelming external force and disease-related symptoms were also inconsistently included or excluded. Where studies felt a need to include or exclude falls of specific causes, there were between five and eight criteria in addition to the biomechanical and topographical criteria used to define the fall [29, 34, 95]. In contrast, some definitions explicitly included all falls related to diseases or unknown causes [27].
The potential for physical injury caused by a fall was another source of variation.
Only two of the 90 articles analysed the consequences of using different fall definitions on the outcome [4, 67]. Both reported that the variation in case definition influenced study results.
Methods of collecting falls data
There was considerable heterogeneity in reporting systems and the time period over which information was collected (Table 1). Three main methods of collecting falls data were found: (i) retrospective reporting systems using telephone interview, face-to-face interview or postal questionnaire (n = 24, 27%); (ii) prospective reporting systems using post-cards, calendars and diaries (n = 38, 42%); (iii) routine surveillance systems or abstraction from health care records (n = 16, 18%). In 12 of the papers (13%), no or insufficient information on data collection was given.
|
For retrospective studies, recall periods ranged from 1 week to 4 years. Prospective registration systems requested immediate return of the data, or return on specified time points ranging from 1 week to 6 months. The primary system was often backed up by secondary data capture mechanism. Only three articles [25, 85, 88] documented the adherence to the primary fall reporting system, indicating an insufficient response rate of 48–83%.
Methods of summarising data
Tables 2 and 3 detail the methods used to summarise falls data. The most frequently reported summary statistic was the number of participants sustaining a fall (i.e. number of fallers) (70%). The number of falls (i.e. incidence of falls) was reported in 61% of studies. Fall rates, expressed either as the number of falls per person or with an additional time denominator, were low (28%). Some papers further classified events or persons as recurrent fallers, and by the injury that resulted. Injury classifications were inconsistent. A few studies reported fall-free survival (i.e. time to a first fall) (18%) or fall rates adjusted by the level of physical activity using a questionnaire [50]. Other summaries included the number of falls in different locations (e.g. indoor and outdoor). Study endpoints included aspects such as the number of general practitioner (GP) visits or the number of fractures sustained without explicit attribution to a documented fall.
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionKey pointsReferences |
|---|
Fall definition
A first step in any epidemiological investigation is to develop a clear case definition. In the reviewed articles no single definition of fall was used as a designated standard. Although unsystematic and lacking explicit rationale, researchers have often tried to match their fall definition with the chosen intervention or target population. In half of the studies reviewed the notion of a fall is taken for granted and no definition is provided whatsoever. Whilst in everyday life it may seem self-evident what constitutes a fall, for research purposes this is not satisfactory and a clear definition of the target event is required.
In some studies when explicit definitions were used, extensive exclusions of what did not constitute a fall were also given, leading to highly selected study populations and fall observations [29]. Such falls would be difficult to classify retrospectively, and the results may be difficult to generalise.
In most of the reviewed articles, falls resulting from acute medical events and/or external force were explicitly excluded. Falls resulting from external force, such as a collision, occur in younger adults or vigorous elderly persons and are perceived as an accident as opposed to the effect of motor deficits that could be reversed by intervention. This logic appears flawed. Excluding falls because they are not believed to be amenable to the intervention should not be necessary. A randomised design should distribute such confounders equally into each arm of the trial. Defining away the ‘unavoidable’ falls permits trial sizes to be smaller as the effect size is bigger, but at the expense of introducing definitional artefact and potential for observer or researcher bias. Overall the review reveals that, in general, it was a subjective decision by researchers as to which medical conditions or environmental hazards were included.
One example of the impact of the chosen fall definition is the intervention study of Wolf et al. [67]. In this case the effect of a Tai Chi Chuan intervention was studied. The choice of the fall definition including ‘near falls’ led to a significant result whereas the exclusion of near falls would have led to a less favourable result. This demonstrates the need to standardise the topographical component of the fall definition.
The wording ‘involuntary’, ‘unintentional’, ‘unexpected’, ‘inadvertent’, ‘unplanned’, or ‘sudden’ describes an external perspective not always experienced or verbalised by fallers. People who fall may use different wording, e.g. stumbling, slipping or tripping [98]. This clearly is an understudied area.
No publication contained information on how participants or proxies were instructed or trained. It remained unclear whether the documentation of falls used in studies was based on the given definition and only a few studies described an active process of an expert confirmation of fall reports as a mandatory second step [4, 77, 81, 88]. Buchner et al. [4] reported that only a few falls which had been reported by participants were rejected by a blinded review committee.
A comprehensive, non-exclusive fall definition is to be preferred and is recommended for future research. Definitions need to be simple, and understood reliably by lay people, who document their own falls. This could be supplemented by a further subclassification of falls by time, location, activity, etc., when further documented by staff. This is in concordance with a recently published consensus statement that recommends a fall to be defined as ‘an unexpected event in which the participant comes to rest on the ground, floor, or lower level’. Including the lay perspective participants should be asked: ‘. . . have you had any fall including a slip or trip in which you lost your balance and landed on the floor or ground or lower level?’ [98].
Fall reports
The method used to report falls also remains problematic and highly variable.
Only one study investigated adherence to the reporting system, documenting poor results [30]. There are some conflicting reports on the accuracy of different data collection methods [98–101]. Routine health care records are limited by their quality and availability in different settings. In the community they are of little use since they record less than 20% of the fall events reported by patients [100]. Prospective registration systems are superior, but the issue of under- or over-reporting still remains unsettled [102]. The recall period is a substantial source of variation in the number of falls reported. Fall diaries can lead to a substantial increase in reported falls compared to report without a diary—a classic Hawthorne effect [103, 104]. The validity of reporting systems was not referenced in any study and only some studies gave references of a former use in previous intervention studies or common data sets.
Three systems of prospective reporting have been used in the reviewed articles: calendar, diary and postcard. The accuracy of each reporting system is difficult to determine [101]. More work is needed to establish methodological rigour. Even the difference between prospective and retrospective systems is difficult to clarify since both approaches were frequently used in the same study. Prospective systems often have a back-up retrospective recall (e.g. by a telephone call), thus introducing retrospective recall error into the data [105], sufficient to introduce significant deviation in outcome [102].
Fall prevention studies need a long follow-up [91, 99, 106] for sufficient events to occur, and to ensure longer term effects of interventions can be detected. However, to maximise accuracy the recall period over which participants report the absence or presence of a fall event must be short. The implication is that studies will require intensive follow-up over long periods. The cited consensus expert meeting recommended prospective daily recording and adequate surveillance of documentation and ascertainment of details of falls at least once a month [98].
Fall documentation
The way the data were summarised differed across articles. Although statistical analysis was not the target of this review we observed that the choice of summarising the primary outcome corresponded to different statistical procedures. This exacerbates the problems of evaluating and comparing different interventions for fall prevention [93]. For example, time to event data are perhaps best approached using Cox’s proportional hazard modelling, whilst binomial event data are analysed using logistic regression. Defining falls in terms of resource use was common, but is not recommended for international comparison, where access and provision of services may vary substantially.
For better comparability of study results, a core set of fall outcome measures including number of falls, fallers, fall rate and time to first fall (as a safety measure) to summarise fall data along with a standardised statistical analysis indicating the absolute risk difference between groups will improve reporting standards for future intervention trials [98].
Limitations of the study
This study has some limitations since only randomised controlled studies were included. We are aware that by such pre-selection available information may have been neglected, especially in terms of descriptive epidemiological studies. However, such an approach covers most relevant, high-quality intervention studies and documents the current understanding and methodological practice in this research field.
To conclude, in this systematic review a substantial heterogeneity in fall definitions, the way falls are documented and analysed was identified. We need to standardise methods in fall prevention if advances are to be made. Based on this review, recommendations for the use and standardisation of these and additional methodological aspects related to fall prevention have been developed in an expert consensus conference by the ProFaNE group [98].
|
Key points |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionKey pointsReferences |
|---|
- A large variation in parameters was found indicating a substantial lack of standardisation in the use and application of terminology and documentation methods concerning falls.
- A comprehensive, non-exclusive fall definition is recommended for future research. Definitions need to be simple, and understood reliably by lay people, who document their own falls.
- Prospective daily recording of falls, surveillance of documentation and ascertainment of details of falls are recommended.
- For better comparability of study results, a core set of fall outcome measures is recommended.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionKey pointsReferences |
|---|
The authors wish to thank Lesley Gillespie from the Cochrane Working group of fall prevention for cooperation in identifying previously published fall prevention trials. The authors are participants of a thematic network (PROFANE-Network) which is funded by the European Commission (QLRT-2001-02705) (Key Action #6; the ageing population and their disabilities; Quality of Life and Management of Living Resources Program). The content of the manuscript does not represent the opinion of the European Community and the Community is not responsible for any use that might be made of the information presented in the text.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionKey pointsReferences |
|---|
-
PLEASE NOTE: The very long list of references supporting this review has meant that only the most important are listed here and are represented by bold type throughout the text. References 1–90 are available on the journal website (http://www.ageing.oxfordjournals.org/).
- Gillespie LD, Gillespie WJ, Robertson MC, Lamb SE, Cumming RG, Rowe BH. Interventions for preventing falls in elderly people (Cochrane Review). Cochrane Library 2003, Issue 4. Oxford: Updated Software.
- Lavery LL, Studenski SA. Tai Chi, falls, and the heritage of JAGS. J Am Geriatr Soc 2003; 51: 1804–5.[Medline]
- Cumming RC, Kelsey JL, Nevitt MC. Methodological issues in the study of frequent and recurrent health problems. Ann Epidemiol 1990; 1: 49–56.[Medline]
- Haywood KL, Hargreaves J, White R, Lamb SE. Reviewing measures of outcome: reliability of data extraction. J Eval Clin Pract 2004; 10: 329–37.[Medline]
- Kellogg International Work Group on the Prevention of Falls by the Elderly. The Prevention of Falls in Later Life. Dan Med Bull 1987; 34: 1–24.[Medline]
- Buchner DM, Hornbrook MC, Kutner NG et al. Development of the Common Data Base for FICSIT Trials. J Am Geriatr Soc 1993; 41: 297–308.[Web of Science][Medline]
- Ballinger C, Payne S. The construction of the risk of falling among and by older people. Ageing Soc 2002; 22: 305–24.[CrossRef]
- Lamb SE, Jorstad-Stein EC, Hauer K, Becker C, Prevention of Falls Network Europe and Outcomes Consensus Group. Development of a common outcome data set for fall injury prevention trials: the Prevention of Falls Network Europe consensus. J Am Geriatr Soc 2005; 53(9): 1618–22.[CrossRef][Web of Science][Medline]
- Kanten DN, Mulrow CD, Gerety MB, Lichtenstein MJ, Aguilar C, Cornell JE. Falls: an examination of three reporting methods in nursing homes. J Am Geriatr Soc1993; 41: 662–66.[Web of Science][Medline]
- Haga H, Yasumura S, Niino N, Ueno H, Oshima M, Higuchi Y. An examination of two reporting methods of falls among the elderly living in the community. Nippon Koshu Eisei Zasshi 1996; 43: 983–88.[Medline]
- Hale WA, Delaney MJ, Cable T. Accuracy of patient recall and chart documentation of falls. J Am Board Fam Pract 1993; 6: 239–42.[Medline]
- Fujimoto K, Kondo H, Okanda K et al. A comparison between three methods to investigate falls among elderly living in the community. Nippon Koshu Elisei Zasshi 2000; 47: 430–9.
- Peel N. Validating recall of falls by older people. Acc Anal Prev 2000; 31: 371–2.
- Mayo E. The social problems of an industrial civilization. London: Routledge & Kegan Paul, 1949.
- Lachenbruch PA, Reinsch S, MacRae PG, Tobis JS. Adjusting for recall bias with proportional hazard model. Methods Int Med 1991; 30: 108–10.
- Cumming SR, Nevitt MC, Kidd S. Forgetting falls. The limited accuracy of recall of falls in the elderly. J Am Geriatr Soc 1988; 36: 613–6.[Web of Science][Medline]
- Hufford MR, Shiffman S. Assessment methods for patient-reported outcomes. Dis Manag Health Outcome 2003; 11: 77–86.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  T. P Haines, T. Russell, S. G Brauer, S. Erwin, P. Lane, S. Urry, J. Jasiewicz, and P. Condie Effectiveness of a video-based exercise programme to reduce falls and improve health-related quality of life among older adults discharged from hospital: a pilot randomized controlled trial Clinical Rehabilitation, November 1, 2009; 23(11): 973 - 985. [Abstract] [PDF]
|
|
|
|
 |
|
 E. Th. Petridou, E. G. Manti, A. G. Ntinapogias, E. Negri, and K. Szczerbinska What Works Better for Community-Dwelling Older People at Risk to Fall?: A Meta-Analysis of Multifactorial Versus Physical Exercise-Alone Interventions J Aging Health, August 1, 2009; 21(5): 713 - 729. [Abstract] [PDF]
|
|
|
|
 |
|
 R. Boyd and J. A. Stevens Falls and fear of falling: burden, beliefs and behaviours Age Ageing, July 1, 2009; 38(4): 423 - 428. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. G. Donaldson, B. Sobolev, W. L. Cook, P. A. Janssen, and K. M. Khan Analysis of recurrent events: a systematic review of randomised controlled trials of interventions to prevent falls Age Ageing, March 1, 2009; 38(2): 151 - 155. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S Gates, J D Fisher, M W Cooke, Y H Carter, and S E Lamb Multifactorial assessment and targeted intervention for preventing falls and injuries among older people in community and emergency care settings: systematic review and meta-analysis BMJ, January 19, 2008; 336(7636): 130 - 133. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Hartikainen, E. Lonnroos, and K. Louhivuori Medication as a Risk Factor for Falls: Critical Systematic Review J. Gerontol. A Biol. Sci. Med. Sci., October 1, 2007; 62(10): 1172 - 1181. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. Jansson Research on the prevention of fall injuries still makes prediction for practice difficult Age Ageing, May 1, 2007; 36(3): 351 - 352. [Full Text] [PDF]
|
|
|
|
|
|
D. A. Skelton, K. Hauer, and S. Lamb Re: 'Falls Definition Validation' Age Ageing, January 1, 2007; 36(1): 111 - 112. [Full Text] [PDF]
|
|
|
|
|
|
J. Dickens, M. Jones, and A. Johansen Falls definition--reliability of patients' own reports Age Ageing, July 1, 2006; 35(4): 450 - 451. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||