Practice Parameter: Assessing patients in a neurology practice for risk of falls (an evidence-based review)

Types of falls considered in this review.

Falls are generally defined as sudden, unintentional, and unexpected events that result in a person's coming to rest on the ground or at a lower level. Usually excluded from studies of the medical risks for falls are those that result from overwhelming environmental hazards (e.g., icy walkways) or unusual activities or events (e.g., playing sports or being shoved) that would place any person at high risk. While falls frequently follow loss of consciousness due to seizures or syncope, managing the risk of falling due to these conditions is distinct from that for most other falls and is usually considered in separate publications. We have limited our review to studies that address falls occurring without prior loss of consciousness and to studies where falls are analyzed as a principal outcome, rather than fall-related injuries.

Search strategy.

We searched the National Library of Medicine MEDLINE database for articles indexed under the search term (medical subject heading) “accidental falls” (or its subcategories) and under either 1) at least one of the terms “screening” or “functional testing” or “clinical evaluation,” or 2) “nervous system diseases” (or its subcategories, which include specific diagnoses) and “epidemiologic methods” (or its subcategories). The search was limited to English-language articles published between January 1980 and January 2005. Key articles were also identified from comprehensive recent reviews of risk factors for falls found from this search and a search of the Cochrane Library.

Of the 193 potentially relevant citations retrieved by this search, 86 articles met our criteria for relevance: 1) they measured non-syncopal falls as an outcome, and 2) they addressed specific neurologic risk factors or screening tools that could be easily applied in a clinical setting without special equipment. The majority of articles described the experience of seniors living in the community. Most commonly, the articles described prospective cohort studies of incident falls; exceptions (i.e., retrospective cohort or case-control studies) are noted in table e-1 on the Neurology® Web site at www.neurology.org. These reports were analyzed in full by at least two of the authors and rated according to the AAN criteria for determining quality of evidence relating to prognosis or prediction of outcomes (appendix e-2). A few articles received discordant ratings, resolved by consensus after discussions between the reviewers. Information pertaining to 1) the nature of the risk factor and measurement of risk or 2) the screening test, its intended use, and its sensitivity, specificity, and predictive value, was extracted for articles rated as Class III or higher. Where at least two Class III—or one or more Class I or II—articles pertaining to a single risk factor or screening test were found, these were included in this practice parameter and in the evidence table available online (table e-1). Conclusions and recommendations were made according to the AAN criteria (appendix e-3) for translating the quality of prognostic evidence to recommendations. Because of space limitations, evidence supporting Level C recommendations was described only in appendix e-4.

Risk of falls determined from history of falls.

Many follow-up studies we reviewed identified a previous history of falls among their participants, enabling calculations of the risk of falls and an analysis of the sensitivity and specificity of this predictor. Most commonly, the period described for the history of falls was 1 year (range: 3 months to 2 years). Some studies distinguished reports of single and multiple earlier falls, with higher risks of future falls associated with the latter.

Five Class I studies examined the risk of future falls among older adults with a history of recent falls. These found relative risks (RRs) of 2.4 (95% CI 1.9 to 2.9),⁹ 2.6 (no CI given),¹⁰ and 2.5 (1.9 to 3.4)⁶ and ORs of 3.7 (1.3 to 10.0)¹¹ and 2.4 (1.3 to 4.4).¹² Among 10 Class II studies, ORs were 3.0 (1.3 to 7.3),¹³ 16.0 (4.4 to 58.0),¹⁴ 5.0 (1.2 to 20.9),¹⁵ 2.0 (1.2 to 3.5),¹⁶ 3.5 (1.2 to 9.6),¹⁷ 2.4 (1.3 to 4.4),¹² 5.4 (1.8 to 15.7),¹⁸ 3.1 (2.2 to 4.4),¹⁹ 12.5 (3.0 to 52.2),²⁰ and 4.0 (1.3 to 12.1).²¹ Other Class II follow-up studies reported RRs of 1.9 (1.3 to 2.9) and 1.8 (1.3 to 2.4),²² and 2.0 (1.3 to 3.1).²³ Class III studies yielded ORs of 4.2 (1.6 to 10.5)²⁴ and 9.0 (4.3 to 19.1),²⁵ and an RR of 1.8 (1.5 to 2.2).²⁶

One Class II study reports that a history of falls has a sensitivity of 0.86 and specificity 0.86 as a predictor of future falls.¹³ These measures can be calculated from the data in six other prospective Class I and II studies, which yield a range of sensitivity from 37% to 76% and a range of specificity from 63% to 91%.^{6,9,16,18,20,24} Pooled data from these six studies yield a combined sensitivity of 50% and specificity of 80%, with an RR of 2.4 (95% CI 2.1 to 2.8) and an OR of 4.2 (3.3 to 5.3). Given a history of falls, the pooled absolute risk of falling during follow-up was 55%.

Risk of falls due to neurologic conditions determined from history and examination.

Most studies of the risk of falls have focused on older populations and enrolled modest numbers of participants. Accordingly, available data address these risks only for relatively common neurologic conditions, especially those affecting predominantly older populations. Diagnoses for which an increased risk of falling has been documented include stroke, Parkinson disease, dementia, and peripheral neuropathy. Impairments of neurologic function that carry an increased risk of falling, irrespective of diagnosis, include disorders of balance, disorders of gait, lower extremity weakness or sensory loss, and loss of vision.

Stroke.

Three Class I studies found significantly greater risk of falling among persons with a past history of stroke, yielding RRs of 1.9 (95% CI 1.5 to 2.5),⁹ 2.3 (1.2 to 4.3),²⁷ and 2.4 (1.3 to 4.5).¹⁰ Another Class I study yielded an insignificant RR of 1.7 (0.8 to 3.4).²⁸ With a history of stroke, the absolute risk of falling during follow-up was 34%, using data pooled from three of these studies.^9,27,28 One Class III study demonstrated that patients with stroke have an increased risk of falls among persons undergoing rehabilitation, with an OR of 4.0 (2.5 to 6.5).²⁹ For predicting risk of falls among hospital patients with stroke receiving acute care or rehabilitation, one Class III study identified cognitive impairment (RR 1.9, CI 1.2 to 3.1) and confusion (RR 1.5, CI 1.0 to 2.3),³⁰ and another Class III study identified impairment in activities of daily living (OR 2.9, CI 1.4 to 6.0).³¹

Parkinson disease.

One Class I study estimated an increased risk of falls (OR 9.5, CI 1.8 to 50.1) among seniors with Parkinson disease¹²; this association was supported by two Class III studies yielding an OR of 2.5 (1.5 to 4.1)³² and a RR of 2.2 (1.5 to 3.1).²⁶ One Class II study reported the absolute risk of falls among persons with this condition as 68% during the follow-up period.²¹ When the occurrence of falls was compared among persons with more severe Parkinson disease and unaffected controls, an OR > 100 (3.1 to 585) was obtained (Class II)¹⁵; when those with more severe Parkinson disease were compared with those less severely affected, an RR of 1.8 (1.1 to 2.9) was found (Class II).³³ Other Class II or III studies of persons with Parkinson disease revealed that those with postural instability were at much greater risk of falls than those without instability (OR 8.9, CI 3.1 to 26.1),³⁴ and those with absent arm swinging during walking were also at greater risk (OR 4.3, CI 1.3 to 13.7).²¹

Dementias and cognitive impairment.

Among 12 studies of fall risk in persons with dementia or cognitive impairment, definitions of these conditions varied. Most commonly, diagnoses were based on findings from the standardized Mini-Mental State Examination,³⁵ or criteria of the American Psychiatric Association's Diagnostic and Statistical Manual of Mental Disorders, 3rd edition, revised³⁶ (table e-1). Two Class I studies of community-dwelling seniors found an increased risk of falls among those with cognitive impairment, yielding an RR of 2.3 (1.5 to 3.5)⁶ and an OR of 6.4 (2.9 to 13.2).¹¹ Six Class II studies, representing both community-dwelling and institutionalized older populations, indicated increased risks of falls in the presence of dementia (ORs of 4.8, CI 1.7 to 13.6³⁷; 1.9, CI 1.2 to 2.9³⁸; and 6.7, CI 1.1 to 42.5²¹) or cognitive impairment (ORs of 15.2, CI 1.5 to 149.3³⁹ and 1.4, CI 1.1 to 1.8⁴⁰ and an RR of 5.0, CI 1.5 to 16.6⁴¹). One Class II study found a higher risk of falls among moderately demented persons than in those who were mildly demented (OR 2.5, CI 1.0 to 6.2).⁴² Four Class III studies also provide evidence that dementia or cognitive impairment increases the risk of falling among institutionalized seniors, yielding ORs of 1.9 (1.2 to 3.1)²⁵ and 2.0 (1.0 to 4.0)⁴³ and an RR of 1.7 (1.3 to 2.3).²⁶ An analysis of pooled data from five of these studies indicates an absolute risk of falling of 47% among patients with dementia during study follow-up.^{6,37,40,42,43}

Peripheral neuropathy.

One Class I study²⁸ yielded an RR of 3.9 (2.1 to 7.0) with an absolute risk of falling of 55% during an average follow-up time of nearly 6 months. Two Class III studies^44,45 also found an increased risk of falls among persons with peripheral neuropathy, yielding ORs of 6.3 (1.4 to >100)⁴⁴ and 17 (2.5 to 100).⁴⁵

Disorders of gait and balance.

Ten studies associated various signs or symptoms of gait or balance abnormalities with increased risks of falling. The definitions of such abnormalities varied among studies in substance and amount of detail (table e-1). Some addressed abilities to perform standing tasks (e.g., feet together or semi-tandem, response to limited push); most addressed the evenness, steadiness, or pace of normal walking; and some included special walking tasks (e.g., varying pace, turning, or tandem gait). The populations studied included older adults residing in a variety of settings—communities, housing for seniors, or nursing homes—depending on the study. Only three studies focused on specific neurologic diagnoses, i.e., stroke^17,46 or Parkinson disease.³⁴

In two Class I studies, one reported an OR of 2.7 (1.1 to 6.2)¹² and the other reported RRs between 1.7 (1.1 to 2.7) and 2.5 (1.7 to 3.8), depending on the number of abnormalities observed.⁶ Four Class II studies yielded ORs of 3.3 and 7.0,³⁹ 6.0 (1.7 to 21.6),¹⁷ 8.1 (2.0 to 32.8),⁴⁷ and 1.9 (1.4 to 2.7),⁴⁰ and an RR of (1.4 to 4.5).²³ Two Class III studies yielded an OR of 2.8 (2.0 to 3.9)³² and an RR of 1.4 (1.2 to 1.7).²⁶ For self-reported difficulties with balance or walking, two Class II studies found ORs of 3.0 (1.7 to 5.0)⁴⁸ and 6.2 (1.8 to 21.5).¹⁷

Lower extremity weakness or sensory loss.

Among studies, the criteria for determining lower extremity weakness or sensory loss also varied. One Class I study reported an RR for falls of 2.4 (1.7 to 3.2) among seniors with lower extremity disability manifest by “problems with strength, sensation, or balance.”⁶ A Class II study reported ORs of 2.2 (1.1 to 4.7) among stroke survivors with lower extremity motor impairment and 3.1 (1.5 to 6.8) among those with combined lower extremity motor and sensory impairments.⁴⁶ One Class III study reported an OR of 1.8 (1.1 to 3.2) for seniors with lower extremity sensory loss and an OR of 4.1 (2.2 to 7.4) for those with hip flexion weakness.⁴⁹

Use of assistive devices.

One Class I study reported an RR for falls of 2.5 (1.4 to 4.4) among seniors who used a walker or cane.²⁸ Two Class III studies reported sensitivities of 59% and 23% with specificities of 76% and 100% for this criterion.^50,51

Vision loss.

One Class I study reported an RR for falls of 1.7 (1.2 to 2.3) among seniors with vision loss (≥20% diminished near acuity measured by Jaeger's test type),⁶ and a Class I study found an increased risk of falls among seniors with impaired near contrast sensitivity, with a hazard ratio of 1.7.⁵²

Among Class III studies, for persons with corrected visual acuity less than 20/30, one indicated an RR for falling of 2.1 (C.I. 1.3 to 3.4),⁵³ while another, using a corrected visual acuity of 20/60 as the cutoff value, found a sensitivity of 76% and specificity of 57%.⁵⁴ A third Class III study reported an OR of 1.8 (1.5 to 2.1) for the risk of falling among seniors with impaired vision.⁵⁵ Other Class III studies yielded ORs of 2.9 (1.8 to 4.7) for adults with nuclear cataracts⁵⁶ and 3.2 (1.5 to 6.8) for seniors who were blind.⁵⁷ For self-reported impaired eyesight, a Class II study found an OR of 2.6 (1.7 to 4.1).⁴⁸

Risk of falling determined from other screening assessments of mobility, gait, and balance.

Clinical context.

Interventions to reduce identified fall risks are beyond the scope of this parameter. However, other evidence-based guidelines for the management of these risks have been developed that may be consulted,^7,8,64 as well as guidelines for the treatment of underlying disorders where possible.