To be included in this review, an intervention was required to measure falls as a primary or secondary outcome. Thus, the 36 unique trials (39 articles) reviewed for KQ 1 are a subset of the 47 trials (51 articles) reviewed for KQ 2. Falling was assessed in a variety of ways in these intervention studies (e.g., number of fallers, fall rate, time to first fall, and number of frequent fallers). Number of fallers is the most consistent measure of falling across all studies with the remaining measures used selectively. To enhance comparability, we primarily discuss number of fallers (risk for falling) in the results below. We also reported fall rate if data were available.
KQ 1. Is There Direct Evidence That Primary Care Interventions Reduce Fall-Related Injury, Improve Quality of Life, Reduce Disability, or Reduce Mortality When Used Alone or in Combination to Reduce Falls in Community-Dwelling Older Adults?
Thirty-six of the 47 primary care interventions to prevent falling included in this review reported data on at least one health outcome.
Three studies67,75,76 pre-specified mortality as a health outcome and reported no reduction in mortality associated with the intervention. All-cause mortality was assessed in 26 studies as part of attrition (10 multifactorial assessment and management,62,69,77–84 10 single clinical treatment,66,67,85–92 and six exercise/physical therapy63,93–97). We found no evidence that primary care interventions had a significant impact on all-cause mortality after 3 to 36 months (Figure 2). The pooled relative risk for all-cause mortality was 0.90 (95% CI, 0.80 to 1.02) with low statistical heterogeneity (I2=0%). Results from the fixed effect model were similar. Data that would allow us to evaluate fall-related mortality were not available in the evaluated studies. The results do not rule out the possibility of a longer-term influence on all-cause mortality or fall-related mortality. Given that all-cause mortality was not identified as a health outcome in most studies that reported deaths, mortality is not discussed further as part of the health outcome results.
The evidence for fall-related fracture includes 16 unique trials assessing multifactorial assessment and management,69,75,76,82 hip protectors,66,85 correction of vision-related defects,87,90,91 vitamin D supplementation,67,86,89,98,99 and exercise or physical therapy interventions.96,100 Quality of life was reported in 12 unique trials assessing multifactorial assessment and management,79,81,84,101 correction of vision-related defects,87,91 vitamin D supplementation,102 clinical education/behavioral counseling,103 and exercise or physical therapy.96,100,104–106 Disability was reported in 13 unique trials assessing multifactorial assessment and management,75,77–81,84 correction of vision-related defects,87,91 and exercise or physical therapy.94,97,105–107 Results from these trials are discussed by intervention type. Given that few studies reported each of these health outcomes, the results are not robust. Furthermore, these results may reflect selective reporting and should be interpreted with caution.
Multifactorial Assessment and Management
Summary of findings. Of 14 multifactorial assessment and management interventions, 11 fair- to good-quality trials reported health outcomes identified for inclusion in this review. These studies included measures of fall-related fractures,69,75,76,82 quality of life,79,81,84,101 and disability (Table 2).75,77–81,84 All of these trials were conducted in high-risk populations; in seven trials participants were selected based solely on history of a fall75–77,79,80,82,84 and in three trials, based on presence of at least one of several fall-related risk factors (including history of a fall).69,78,81 While these studies provided no evidence that multifactorial assessment and management interventions improved quality of life, they provided limited evidence of reduced fall-related fractures and reduced disability.
The four fair-quality trials (n=1,282) that assessed fall-related fractures among older adults showed a nonsignificant reduced risk for fracture associated with multifactorial assessment and management.69,75,76,82 The pooled relative risk was 0.83 (95% CI, 0.61 to 1.14) with low heterogeneity (I2=0%). In one trial of a multifactorial intervention, Tinetti and colleagues69 reported fractures in four people out of 147 in the intervention group, compared with seven people out of 144 in the control group, during the 12-month followup period. We calculated a relative risk of 0.49 (95% CI, 0.09 to 2.58) for this trial. Hogan and colleagues reported fractures in three out of 75 in the intervention group, compared with five out of 77 in the control group, during the 12-month followup period. We calculated a relative risk of 0.62 (95% CI, 0.15 to 2.49) for this trial. Davison and colleagues76 reported fractures in six intervention subjects (4%) compared with 11 (7%) in controls during 12 months of followup (RR, 0.53 [95% CI, 0.20 to 1.39]). Spice and colleagues75 reported fractures in 40 of the subjects in the more intensive intervention group (19%) compared with 35 (22%) of the controls (OR, 0.90 [95% CI, 0.61 to 1.34]). These studies were not powered to evaluate a difference in fracture rate. These studies relied on self-reports of fall-related fractures using monthly calendars with followup phone calls to fallers to assess injuries resulting from the fall.
None of the four fair- to good-quality multifactorial assessment and management trials (n=914) in older adults reported a significant change in quality of life after 12 months of followup.79,81,84,101 Three trials assessed change in self-reported quality of life using the SF-36 instrument,79,81,101 and the fourth used the EuroQol instrument.84 Two of these studies reported blinding of outcome assessors.79,84
Seven fair-quality multifactorial assessment and management trials (n=3,237) evaluated effects on disability with mixed results.75,77–81,84 One trial (n=1,242) including older adults with one or more fall risk factors reported that a multifactorial assessment and management program was associated with a significant difference in the percentage of participants with worsened disability in the intervention group (15%) compared with controls (20%) after 12 months (p<0.05).78 Limitations in activities of daily life were assessed using the Medical Outcome Study physical function scale. Two studies reported slightly less disability, based on the Barthel Index, in the intervention group, compared with the control (mean difference in change ranged from 0.6 to 1.0 on a 100 point scale), representing a greater improvement in one study80 and less of a decline in the others.75,77 This difference, however, is not clinically meaningful. The remaining three studies reported no significant improvements in disability.79,81,84
Single Clinical Treatment
Eleven of the 16 clinical trials reported fall-related fracture, quality of life, or disability (Table 3). Interventions that included a single clinical treatment are diverse, and thus it is more relevant to describe the evidence for subgroups of similar interventions: vitamin D (with or without calcium),67,86,89,98,99,102 vision correction,87,90,91 and hip protectors.66,85 Eight of the 11 trials were restricted to women66,67,85–87,89,91,98 and all except one were conducted in high-risk populations.86
Summary of findings. Hip protectors. Hip protectors were not associated with a significant reduction in 24-month risk for fall-related fractures among high-risk women with an average age of 78–83 years (n=4,769) (Figure 3).66,85 One third (31%) to one half (53%) of women wore their hip protectors as intended.
Vision correction. Three of four vision-correction trials (n=1,161) among adults aged 78–81 years on average reported health outcomes.87,90,91 All studies were rated fair to good quality. The results for fall-related fractures were mixed and data were not pooled due to very high statistical heterogeneity. One of the trials of expedited cataract surgery was associated with a reduced risk for fractures,87 while the other was associated with a nonsignificant increased risk.91 The third trial treated vision deficiencies based on screening results and reported a nonsignificant increased risk for fall-related fractures (OR, 2.5 [95% CI, 0.5 to 12.5]).90 Additionally, no significant differences were observed in 6-month change in disability among 545 participants87,91 or quality of life among 239 participants.91
Vitamin D. Six of eight vitamin D trials67,86,89,98,99,102 reported health outcomes 6 to 36 months after the beginning of treatment. Trials included healthy ambulatory women and men aged 65 years or older;86,99 individuals with vitamin D deficiency and a history of falling,89 vitamin D deficiency without a history of falling,98 or a history of falling without vitamin D deficiency;102 and women at risk for hip fracture.67 No trials evaluated disability. All studies were rated as fair quality.
Five studies (n=4,252) tabulated the number of people who experienced a fall-related fracture67,86,89,98,99 and one study assessed quality of life.102 Risk for fall-related fractures was not reduced over 12 months among women aged 71–77 years on average (pooled RR, 0.85 [95% CI, 0.64 to 1.12]) (I2=0%) (Figure 4). The trial that evaluated improvements in quality of life found no significant improvement.102
Study details. Hip protectors. Two fair-quality trials66,85 (n=4,769) assessed the influence of hip protectors on overall risk for fracture among high-risk noninstitutionalized women over 24 months of followup, during which 415 women fell in one study 85 and there was a total of 1,437 falls in the other study (mean, 2.2 to 2.7 per person).66 Both interventions provided participants with semi-rigid shields sewn into modified underwear. One trial (n=600) provided intervention-group participants with a nurse to assist with fitting the protectors and encourage adherence. The nurse made three home visits followed by two telephone calls to intervention participants.66 The other intervention (n=4,169) mailed the hip protectors with an educational pamphlet.85 While adherence was higher in the trial including the nurse contact (53%)66 compared with the trial without contact (38%)85 at 6 months, neither was associated with a significant reduced risk for any fracture or hip fractures in intention-to-treat analyses. The pooled relative risk for experiencing a fall-related fracture during 24 months after initiation of the intervention was 0.89 (95% CI, 0.75 to 1.06) (I2=0%) (Figure 3).
Vision correction. Two trials evaluated expedited cataract surgery.87,91 A third trial evaluated single clinical treatment of vision problems identified through screening.90 All of the participants (n=1,161), women aged 78–81 years on average, were identified as high-risk populations by virtue of selection for frailty or for an age of 70 years or older and cataracts. Fall-related fractures were self-reported in all three trials and were assessed by monthly postcards, a telephone followup (if the postcard was not returned),90 or during a telephone interview or clinic visit every 3 months.87,91
The trials of expedited cataract surgery among women with unoperated cataracts87,91 (n=545) found mixed results. Harwood and colleagues87 reported a significant risk reduction for fall-related fracture (four persons in the intervention group compared with 12 in the control group [p=0.04]). Foss and colleagues,91 on the other hand, reported a nonsignificant increased risk in the intervention group compared with the control group. Expedited cataract surgery trials also reported no significant reduction in disability, as measured by the Barthel Index,87,91 or quality of life, as measured by the EuroQol.91 An intervention including an eye exam and treatment was associated with a significantly increased risk for sustaining a fall-related fracture in the intervention group, compared with the usual care control.90
Vitamin D. In a fair-quality study (n=148) restricted to women who were vitamin D deficient (25-hydroxycholecalciferol level<50 nmol/liter), a regimen of vitamin D and calcium supplement (400 IU vitamin D and 600 mg calcium daily) for 8 weeks was associated with a 4% risk for sustaining a fall-related fracture over 1 year, compared with 9% in the control group that received only the calcium supplement.98 This difference was not statistically significant, although the study may have been underpowered. A study (n=242) in healthy older adults that was also not powered to detect a significant reduction in the number of fractures reported 12 fractures in the calcium plus vitamin D group (800 IU of cholecalciferol per day) compared with 19 fractures in the control group that received calcium only (p=0.12).99 Another study (n=246) in healthy women reported a nonsignificant reduction in risk for nonvertebral fractures (RR, 0.60 [95% CI, 0.28 to 1.27]) for 0.25 μg of calcitriol twice per day compared with placebo.86 Another fair-quality study89 (n=302) restricted to women who were vitamin D deficient (25-hydroxycholecalciferol level<24.0 nmol/liter) and had fallen at least once during the previous year evaluated a higher-dose vitamin D and calcium supplement (1000 IU vitamin D2 and 250 IU calcium citrate tablets twice daily) compared with calcium supplementation alone for 1 year. One woman in each group experienced a fall-related fracture during the year of followup.89 A large study (n=3,314) among women with at least one risk factor for hip fracture evaluated a vitamin D plus calcium supplement (two tablets of 1000 mg of calcium and 800 IU of vitamin D3 daily) for 6 months.67 Women in the intervention group also received a brief education/counseling visit with a nurse that focused on reducing fracture risk and a pamphlet describing how to consume adequate calcium and vitamin D from dietary sources. The control group only received the pamphlet in the mail. After a median followup of 25 months, the risk for fall-related fractures was 4.8% in the intervention group and 5.0% in the control group.
Change in quality of life was assessed in an intervention comparing an intramuscular injection of 600,000 IU of ergocalciferol with a placebo injection among men and women (n=139) who had fallen during the previous 8 weeks.102 Dhesi and colleagues found no significant change in the intervention group between baseline and 6-month followup in any of the eight subscales of the SF-36 measure of health-related quality of life. Control participants, on the other hand, reported small but statistically significant improvements in two of the subscales (social functioning and role-emotional). The difference in changes between the two groups was not assessed.
The followup period for all of these trials was 6 to 36 months. As such, the absence of any influence on distal outcomes may reflect inadequate time to see an improvement or rare events, in the case of fractures.
Clinical Education/Behavioral Counseling
Summary of findings. One good-quality study (n=310) of a low-intensity behavioral counseling intervention (<26 contact hours) in community-dwelling older adults reported no significant difference in change in quality of life after 14 months103 (Table 4). The 7-week intervention included 2 hours per week of group sessions taught by an occupational therapist. The sessions were designed to assist older adults in developing strategies to reduce fall risk. The control group received two social visits with an occupational therapy student. Quality of life was assessed using the physical function and mental function composite scores of the SF-36. All women included in the trial had a history of at least one fall during the previous year or had a fear of falling.
Home Hazard Modification
No trials reported health outcomes related to home hazard modification interventions other than mortality (Table 5).
Exercise and Physical Therapy
Summary of findings. Of 17 exercise or physical therapy intervention trials, seven fair-quality trials (n=1,072) of physical activity in community-dwelling older adults assessed multiple health outcomes, including measures of fall-related fractures,96,100,107 quality of life,96,100,104–106 and disability94,97,105–107 (Table 7). Six of these trials were conducted in a high-risk population: three were identified based on gait and balance impairments,100,104,105 two were restricted based on chronic disease status (Parkinson’s disease)96 or recent stroke,94 and one was restricted to women aged 80 years or older.97 The remaining trial included an unselected population.107
In two trials (n=201), exercise interventions addressing muscle strengthening and balance did not significantly reduce risk for fall-related fractures in community-dwelling high-risk older adults 3 to 6 months after initiation of the intervention.96,100 The third trial included a Tai Chi intervention and did not report any fracture-risk data.107
Only one of the four studies evaluating change in quality of life found significant improvement after 3 to 6 months of followup.95,96,104–106 Ashburn and colleagues96 reported a significant difference of 5.7 units (95% CI, 0.47 to 11.0) in change in quality of life, as measured by the EuroQol instrument, after adjusting for baseline EuroQol, balance, functional reach, and disability. The remaining three trials assessed quality of life with the SF-36 instrument.95,104–106 Rubenstein and colleagues100 reported a nonsignificant improvement of 7 points in quality of life (physical function subscale) in the intervention group compared with the control group (p=0.08).
No significant reduction in disability was observed among older adults followed for 4 to 6 months (n=708).94,97,105–107 The followup period for all of these trials was short. As such, the absence of any influence on distal outcomes may reflect inadequate time to see an improvement or rare events, in the case of fractures. Additional details about the interventions are provided below in KQ 2.
KQ 2. Do Primary Care Interventions Used Alone or in Combination in Community-Dwelling Older Adults Reduce Risk for or Rate of Falls/Fallers?
Multifactorial Assessment and Management Interventions
Summary of findings. In seven arms (n=3,195) of comprehensive multifactorial assessment and management interventions among older adults,69,75,76,78,80,82,108 the interventions were associated with reduced risk for falling compared with usual care 12 months after entry. The pooled estimate of relative risk was 0.75 (95% CI, 0.58 to 0.99), with high heterogeneity (I2=86.4%) (Figure 5). Among the nine noncomprehensive multifactorial clinical assessment intervention arms,71,75,77,79,81,83,84,108 the risk for falling was not reduced (RR, 1.04 [95% CI, 0.98 to 1.10]) (Figure 6). The statistical heterogeneity of these trials was low (I2=0%).
Study details. Fourteen trials of multifactorial assessment and management interventions62,69,71,75–84,101 measured fall outcomes. These trials (n=5,570) randomized community-dwelling adults aged 65 years or older (Table 6). The average age of participants was 75 years or older, except for one study in which the average age was 72.5 years.78 The majority of participants were women (percentage ranged from 40% to 77%) and none of the studies reported a substantial proportion of nonwhite or Hispanic participants. Only four of the trials were conducted in the United States,69,78,83,101 while the majority were conducted in other countries, including the United Kingdom,75–77,80 the Netherlands,71,84 Australia,62,81 and Canada.82 Most trials identified participants through primary care practices or insurance rolls, while four identified community-dwelling participants presenting to the emergency department for a fall-related event.76,77,80,84 One study recruited directly from the community.83 Seven trials recruited participants who had a history of falling,75–77,79,80,82,84 and four trials recruited a high-risk population by screening for multiple possible risk factors.62,69,71,78,101 The remaining two trials were conducted in unselected populations.81,83 In the eleven studies that reported a history of falls during the 12-month period preceding the trial initiation, the percentage of individuals experiencing a fall ranged from 33% to 100%.69,71,75–80,82,84,101 Most studies excluded older adults with evidence of cognitive impairment or physical disability.
Components of multifactorial assessment and management interventions are described in Table 8. The primary fall risk factors identifiable during a clinical evaluation were generally evaluated as part of the multifactorial assessments;109 the majority of assessments included visual acuity, gait and balance, medication use, and home environment. We evaluated a total of 16 different active treatment arms in 14 trials that ranged from comprehensive (multifactorial assessment and provision of medical and social care) 69,75,76,78,80,82,108 to noncomprehensive (multifactorial assessment and referral or limited management).62,71,75,79,81,83,84,101 Control groups primarily received usual care, although two trials (described in three studies) provided social visits designed to mirror time and attention provided to the intervention group.69,82,110 These may at least partially explain the reduced effect size reported in one of these trials.82
We conducted meta-analyses of all 14 trials reporting risk for falling. The reduced risk for falling was not statistically significant (RR, 0.90 [95% CI, 0.80 to 1.02]), but the statistical heterogeneity was substantial (I2=79.6%). After removing one study that appeared to be an outlier,80 the heterogeneity was reduced to 44.9% and the relative risk estimate was attenuated to 0.96 (95% CI, 0.90 to 1.04). Publication bias was assessed using the funnel plot and Egger’s regression test; no important publication bias was detected. Of the study-level factors evaluated in a meta-regression, only comprehensiveness (comprehensive vs. noncomprehensive) explained a significant amount of the heterogeneity in the effect estimate (−0.30 [SE, 0.11]) (p=0.009).
Given the significant results in the meta-regression for comprehensiveness, we conducted meta-analyses stratified by the comprehensiveness of the intervention (Figures 5 and 6). The pooled relative risk was 0.75 among the seven comprehensive trial arms (95% CI, 0.58 to 0.99). The statistical heterogeneity, however, was substantial (I2=86.4%). This analysis was restricted to the most comprehensive arm of the trials with two intervention arms.62,75 After removing an outlier,80 the heterogeneity was substantially reduced to 44.4% and the relative risk remained significant (RR, 0.90 [95% CI, 0.82 to 0.99]). The remaining heterogeneity was not explained in meta-regressions using study-level variables. Among the seven noncomprehensive trial arms, the risk for falling was not reduced (RR, 1.04 [95% CI, 0.98 to 1.10]). The statistical heterogeneity was low (I2=0%).
Among the comprehensive interventions, one study also reported a significant reduction in fall rate. Tinetti and colleagues reported 31% reduction in fall rate in the intervention group compared with the control group (incidence rate ratio, 0.69 [95% CI, 0.52 to 0.90]).69 Only one noncomprehensive trial measured fall rate and reported no reduction in the rate of falling.79
Study design and quality. We rated two79,83 trials as good quality and the remaining trials as fair quality (see Appendix B Table 3 for quality criteria). Most trials did not report whether treatment allocation was blinded or whether those conducting followup assessments were blind to the treatment condition. The majority had retention rates between 70% and 90%. While recent trials used prospective methods to assess falls, older studies 78,101 assessed falls retrospectively after 12 months. Most trials reported the percentage of fallers based on the number available for analysis rather than the number randomized.
Overall, the results of these trials may not be generalizable to nonwhite or Hispanic older adults, or older adults with cognitive limitation or physical disability. Two studies relied on self-referral from community-based or health-care based advertising,82,83 and thus may have enrolled participants who were more motivated to participate in interventions. The 12-month risk for falling in the control group ranged from 37%78 to 79%82 following study initiation.82 Overall, the risk for falling in the included participants was higher than that of the average community-dwelling adult aged 65 years and older, of whom approximately one third fall every year.16 Thus, these findings may be most relevant to a high-risk population. Additionally, the majority of these studies were conducted outside of the United States and may vary from “usual care” in the United States.
Single Clinical Treatment
Sixteen trials evaluating single clinical treatment strategies to reduce falling are described by subgroups: vitamin D (with or without calcium),67,86,89,92,98,99,102,111 vision correction,61,87,90,91 hip protectors,66,85 medication withdrawal,112 and nutritional supplementation88 (Table 9).
Summary of findings. Vitamin D. Among seven fair-quality trials (n=5,216) in adults aged 71–77 years on average, vitamin D with or without calcium was associated with a reduced risk for falling during 6 to 36 months of followup (pooled RR, 0.83 [95% CI, 0.75 to 0.91]) (I2=14.6%).67,86,89,92,98,99,102,111 Only one of the individual trials,99 however, reported a statistically significant reduction in fall risk (Figure 7).
Vision correction. Among four fair- to good-quality trials (n=1,437) in adults aged 76–81 years on average, risk for falling during 12 to 18 months of followup was not reduced as a result of vision correction (improving acuity or correcting cataracts with surgery).61,87,90,91
Hip protectors. In two trials (n=4,769), the effect of hip protectors on falling risk was mixed among high-risk women aged 78–83 years on average, with a significant protective effect in one trial85 and no effect in the second.66 Adherence to the hip protectors was low in both studies.
Medication withdrawal. One fair-quality study (n=48) among adults with an average age of 75 years reported no reduction in fall rate associated with medication withdrawal (with or without exercise).112 An additional three multifactorial assessment and management interventions also included medication assessment and management and were effective in preventing fallers.78,80,82
Protein supplementation. In one fair-quality study (n=50) among frail adults with an average age of 78 years, nutritional supplements and home visits over 12 weeks were significantly associated with a reduced risk for falling (zero falls in the intervention group vs. five falls in the control group at 3 months) (p<0.05).88
Study details. Vitamin D. We evaluated eight trials (n=5,216) of vitamin D supplementation conducted in community-dwelling older adults aged 71–77 years on average (Table 9). These studies included more women than men. Four of the trials’ participants were 100% female and the remaining four’s participants ranged from 51% to 80% female. Six studies did not report race or ethnicity; the two that did consisted primarily of nonHispanic white participants (77% to 97%).102,111 Four trials were conducted in populations defined as high risk by virtue of recent falls and/or vitamin D deficiency, and the remaining four studies used populations that were unselected except for age (≥65 years).
The overall vitamin D intervention dosages ranged from 22,400 IU over 8 weeks98 to 766,500 IU over 36 months.111 Delivery was variable across studies, including intramuscular injection, daily oral dose, and mega oral dose every 4 months. Two of the studies evaluated ergocalciferol (vitamin D2)89,102 and the remaining studies evaluated cholecalciferol (vitamin D3). Five trials included calcium supplements with the vitamin D. The control groups ranged from no intervention to placebo to calcium supplements only.
Among adults aged 71–77 years on average (n=5,216), vitamin D with or without calcium was associated with a reduced risk for falling during 6 to 36 months of followup (pooled RR, 0.83 [95% CI, 0.75 to 0.91]) (I2=14.6%). However, only one of the individual trials reported statistically significant reductions (Table 9, Figure 7).99 Only one study evaluated differences in fall rate and reported a significantly lower rate of falls in the group receiving calcitriol compared with the group receiving placebo (0.27 falls/year in intervention group vs. 0.43 falls/year in placebo group [p=0.0015]). The majority of these trials were not sufficiently powered to observe a significant reduction in risk for falling. Excluding one large study,67 these individual studies randomized between 139 to 445 participants. Two of the trials conducted analyses of the cumulative development of the number of subjects with no falls.99,111 These analyses suggest that the minimum required time to observe an effect on falls outcomes appears to be 12 months and that between group differences may be maintained but probably do not increase after 12 months.
All studies were rated as fair quality. Only three trials assessed falls prospectively using a diary or questionnaire;86,99,102 the remaining trials assessed falls retrospectively with periods of recall ranging from 6 weeks89 to 12 months.98
Vision correction. Four trials (n=1,437) in adults aged 76–81 years on average evaluated the effect of vision correction (after screening for visual impairment) on risk for falling. Three trials were conducted in populations deemed high-risk because of frailty or having uncorrected cataracts (with or without a history of falling). The fourth study had a population that was unselected except for age (>70 years). Two studies included only women and evaluated expedited cataract surgery compared with routine wait controls.87,91 One of the trials of expedited cataract surgery evaluated first cataract surgery,87 while the other evaluated second cataract surgery.91 The other two studies primarily included women (60%–68%) and evaluated vision screening and referral/treatment.61,90
None of the trials significantly reduced the risk for falling.61,87,90,91 The trial evaluating first cataract surgery, however, reported a significantly reduced fall rate among the intervention group participants compared with the controls. Rate of falling was reduced by 34% in the operated group (incidence rate ratio, 0.66 [95% CI, 0.45 to 0.96]).87 One of the trials that evaluated the effect of visual treatment impairment based on screening results reported a statistically significant 30% increased risk for falling in the intervention group compared with the control group (95% CI, 14 to 50).90 This result is described further in KQ 3 (harms). Studies were rated as fair- to good-quality.
Hip protectors. In one large, fair-quality study (n=4,169), hip protectors were associated with a significant reduction in 12-month risk for sustaining a fall among high-risk women with an average age of 78 years (28% in the intervention group vs. 38% in the control group [p<0.001]).85 These results may be conservative, as the 6-month adherence rate among the intervention group was 38%. A second smaller, fair-quality study reported no significant effect of hip protectors on the total number of falls or on being a frequent faller.66 This study employed a study nurse to make routine home visits to encourage use of the hip protectors among intervention group participants. While the adherence was higher in this study than in others, it was still low (57% after 12 months) (Table 9). Additional study details were provided in KQ 1.
Medication withdrawal. One fair-quality study (n=48) evaluated the influence of medication withdrawal (with and without exercise) on fall outcomes among adults with an average age of 75 years and taking psychotropic medications, but did not report fall risk.112 The number of falls per group was lower in the intervention group (17 falls out of 48 people) compared with the control group (29 falls out of 22 people). The fall rate, however, was not significantly lower among the intervention group compared with the groups that did not receive medication withdrawal (difference of 0.64 falls/person-year [95% CI, −0.07 to 1.35]) (Table 9). These results compare the rate of falling among the trial arms that received medication withdrawal (medication withdrawal alone and medication withdrawal plus exercise) to the rate of falling in the trial arms that did not receive medication withdrawal (exercise plus regular medications with no alteration and a control group that received only regular medications). An additional four multifactorial assessment and management interventions also included medication assessment and withdrawal.75,78,80,82 All four trials were associated with a significantly reduced risk for falling. The pooled estimate for all five studies together was noninformative because of high statistical heterogeneity (I2=91.9%). These studies have been described previously, thus they are not discussed further in this section.
Protein supplementation. One small study (n=50) evaluated the effect of protein supplementation among frail adults with an average age of 79 years.88 The majority of participants were female and had less than high school education. All were frail and at nutritional risk, defined as involuntary weight loss of greater than 5% body weight in previous month, greater than 7.5% in 3 months, or greater than 10% in 6 months, and a body mass index of less than 24.
The intervention group received 235 mL of liquid supplement twice a day and home visits once per week. The control group received weekly home visits providing encouragement and dietary suggestions. The intervention was associated with a reduced risk for falling (zero falls in the intervention group vs. five falls in the control group at 3 months [p<0.05]).
This study was rated fair quality. Outcome assessors were blinded to intervention status and there was no attrition from assessment during the 3-month followup. However, participants in the control group were slightly more likely to report having a good appetite at baseline (suggesting potentially better health) and falls were assessed retrospectively at 6 and 12 weeks.
Exercise and Physical Therapy
Summary of findings. Of the 17 fair- to good-quality trials (n=3,716) of exercise/physical therapy in community-dwelling older adults, 13 were consistent with a reduced risk for falling. The majority of differences, however, were not statistically significant. Pooled data suggest that those participating in an exercise/physical therapy intervention were less likely to fall during followup (RR, 0.86 [95% CI, 0.80 to 0.92) (I2=5.4%) (Figure 8). There was an indication that exercise had a greater effect on fall risk in the trials that were conducted in higher-risk populations (Figures 9 and 10).
Study details. These trials (n=3,985) randomized community-dwelling older adults to exercise/physical therapy interventions (Table 10). The majority of participants were aged 75 years or older on average. In five trials, the average age was 80 or older.63,93,95,113 Women were more common than men in the majority of trials; most trials consisted of between 60% and 100% female participants. One study was restricted to men,100 and three trials were evenly split or had more men.94,96,105 The majority of trials did not report race or ethnicity. The trials reporting race or ethnicity data were primarily nonHispanic white (≥90%). Ten of the physical activity trials were conducted in high-risk populations. Of these 10 trials, three were identified based on gait and balance impairments,100,104,105 one based on falls during the past year or one or more risk factors for falling,93,114 one based on use of psychotropic medication,112 and one based on visual impairment; two were restricted based on chronic disease status (Parkinson’s disease plus at least one fall in the previous year)96 or recent stroke,94 and one included only people with a recent hospitalization or period of bed rest.107 The remaining studies were conducted in unselected populations.61,95,97,107,108,115,116 The percentage of participants who reported falling during the 12 months preceding the intervention (excluding trials that selected participants based on history of a fall at baseline) ranged from 10% to 64%.
Trial characteristics are listed in Table 11. Exercise/physical therapy trials included a variety of components that can be summarized into three major categories: gait, balance, or functional training; strength or resistance exercise; and general exercise (including walking, cycling, aerobic activity, and endurance exercise). All but one trial included gait, balance, and/or functional training. The majority of trials included at least two exercise components, about half of which were primarily group-based exercises conducted in the community. The remaining trials were individual-based and generally conducted at home. Five studies were conducted in the United States. The remaining trials were conducted in the United Kingdom, Finland, Australia, and New Zealand.
We evaluated a total of 20 treatment arms in 17 unique trials. Three trials evaluated 12-month interventions.63,95,104 The duration of the remaining trials ranged from 6 to 26 weeks (median, 12.5 weeks). Treatment intensity (estimated in hours of contact) ranged from 2 to 80 hours. Control groups varied from no treatment/usual care to social visits or educational information unrelated to falls. One Tai Chi intervention matched the number of contact hours in the intervention group with stretching and relaxation classes.116
The majority of trials were consistent with a beneficial effect on falling or fall rate compared with controls. Most of these differences, however, were not statistically significant. Across all trials, the 12-month risk for falls in the intervention group compared with the control group ranged from 39% lower116 to 34% higher.94 Meta-analysis of all 15 trials reporting fall risk was conducted (Figure 8). Among physical activity interventions, the pooled relative risk estimate for reduction in percentage of fallers comparing the intervention to the control group was 0.86 (95% CI, 0.80 to 0.92), with minimal heterogeneity (I2=5.4%). We detected no evidence of publication bias based on Egger’s test and a visual inspection of Begg’s funnel plot. Two intervention arms of a trial intended to improve balance (Tai Chi or platform training) were combined in our meta-analysis.107 Results were unchanged when the more intense of the two arms was included. A parallel analysis was conducted on the subset of trials that reported fall rate, which resulted in a very similar pattern of results (RR, 0.76 [95% CI, 0.67 to 0.89]) (I2=55.4%).
Three of the trials had intervention arms that included exercise plus another intervention type.61,63,112 One trial included an intervention arm that evaluated exercise plus home hazard modification in addition to an exercise-only arm.63 One trial included an intervention arm that evaluated exercise plus treatment of visual impairment and an intervention arm that evaluated exercise plus home hazard modification in addition to an exercise-only arm.61 A third trial evaluated exercise (exercise alone and exercise plus medication withdrawal), comparing these groups to a group that did not receive exercise (a medication withdrawal group plus the control group). We conducted an additional meta-analysis including these combination intervention arms and the results were similar, although the statistical heterogeneity was slightly greater (I2=12.4%).
Meta-regression identified no study-level variables that explained significant within-study variability. Visual analyses examining pattern of results by intervention intensity, number of intervention components, type of intervention (group vs. not), and risk status of participants (selected vs. unselected and percentage fallers in previous year) confirmed that there was no pattern in intervention effectiveness by these variables. Differences in results, however, were observed when the observed fall risk among the control group after randomization (>35% vs. <35%) was used as a measure of risk status. This stratification roughly represents the groups above and below the average fall risk for people aged 65 years or older. Meta-analyses stratified by fall risk in the control group support the qualitative observation. The pooled risk from 10 trials with more than 35% of the control group falling during the intervention was similar to the overall results (RR, 0.84 [95% CI, 0.78 to 0.91]) (I2=1.1%) (Figure 10). However, the pooled relative risk among the five trials with 35% or less of the control group falling was not significant (0.98 [95% CI, 0.82 to 1.62]) (I2=0%) (Figure 9).
Study design and quality. One trial was rated good quality,115 while the remaining trails were rated fair quality (see Appendix B Table 3 for quality criteria). It is not possible to blind participants to their group assignment in physical activity trials, and most of the trials did not report whether those collecting followup assessments of falls were blinded to intervention status. The majority of the interventions did not provide an attention control to address a possible Hawthorne effect.117 Many studies reported retention of 80% or more, although retention in five trials was below 80%.93,108,112,113,116 Three studies appeared to have differential attrition,97,100,105 but these differences were not tested statistically. All of the studies relied on self-reported falls. Participants in most trials were provided with a calendar or diary to record falls and queried at least monthly by study staff to report falls over the past month throughout the study period; five trials assessed falls retrospectively with periods of recall ranging from 2 weeks100 to 3 months.94 Trials conducted in unselected older adults tended to limit the population to healthier older adults, resulting in a large percentage of screened older adults not randomized for the intervention.95,107,116 Similarly, the high-risk populations are also very unique (e.g., stroke patients94 or extremely frail105 or recently hospitalized).113 As such, results may not generalize to other selected populations, such as those selected based on history of falls.
Home Hazard Modification
Summary of findings. Three fair-quality trials (n=2,348) in community-dwelling adults aged 76–84 years on average assessed the influence of home hazard assessment and modification on risk for falling. All three trials were consistent with a reduced risk, ranging from 7% to 41%, when comparing the intervention group with the control group. This risk reduction, however, was only statistically significant in one study (n=196) conducted among high-risk older adults.63 In a meta-analysis, the pooled relative risk was 0.81 (95% CI, 0.63 to 1.04), with significant heterogeneity (I2=70.4%) (Figure 11). An additional five multifactorial assessment and management trials included home hazard assessment and modification as a component of the overall intervention.71,75–77,84 Adding these trials to the meta-analysis attenuated the risk reduction and did not change the significance (RR, 0.90 [95% CI, 0.77 to 1.06]) (I2=54.5%). As these three trials have been discussed previously, they will not be discussed further in this category.
Study details. Three studies (n=2,718) randomized community-dwelling older adults to home hazard modification interventions (Table 12). All participants were aged 70 years or older. Trials primarily included women, with the proportion of women ranging from 52% to 70%. No trials reported race/ethnicity or socioeconomic status of the included participants. One study was restricted to a high-risk population, defined as people with visual impairment.63
Trial characteristics are listed in Table 12. All three trials evaluated an assessment of the home and modification of any identified hazards (e.g., adding nonslip tape to rugs and steps) and/or provision of free safety devices (e.g., grab bars). Two interventions also provided intervention participants with behavioral counseling.63,70 Occupational therapists conducted the assessment and oversaw modifications in only one trial,63 and the remaining two trials used trained assessors/research nurses. We evaluated a total of six treatment arms. In addition to evaluating home hazard modification separately, two of the trials evaluated home hazard assessment in combination with physical activity61,63 or vision assessment and correction.61 Two trials provided the control participants with social visits,63,70 while the third provided nothing to the control group.61
Only one home-hazard modification trial (n=196) conducted in New Zealand reported a significant beneficial effect on risk for falling compared with controls.63 Modification of home hazards and behavioral counseling among vision-impaired adults with an average age of 84 years was associated with a significant reduction in the percentage that fell at least once during the 12-month intervention (36% vs. 61% [p<0.05]).63 Notably, the process of assessment and modification was guided by the Canadian Model of Occupational Performance and conducted by occupational therapists, whereas the assessment was conducted by research nurses in the remaining two trials and modification was conducted by participants themselves and the same nurses70 or by a city maintenance worker.61 The other two home-hazard modification trials that did not demonstrate significant results were conducted in Australia among unselected adults with an average age of 76 years.
Two of these trials also evaluated intervention arms that combined home hazard modification with exercise61,63 or vision correction.61 The results from the combination arms were generally similar to the results in the home hazard modification only arm. When these combination intervention arms were added to the home hazard modification only arms, the results of the meta-analysis were unchanged.
Study quality. The three studies were rated as fair quality. Falls were recorded prospectively in all three studies. Only one study reported blinded assessment of outcomes.61 All three studies reported attrition less than 10%, although attrition was differential in one study.70 Compliance with the intervention ranged from 83%70 to 90%.63 One study also reported crossover, such that 16% of the control group also made home safety modifications.70 Two of these studies were conducted before 2005,61,70 and all were conducted outside of the United States. One of the studies conducted in an unselected population was restricted to healthier older adults.61
Clinical Education and Behavioral Counseling
Summary of findings. Only one good-quality trial (n=310) evaluated the influence of an intervention that was primarily educational.103 This study reported no evidence of a reduced risk for falling associated with behavioral counseling. We conducted a meta-analysis including all twelve trials providing minimal education or counseling in conjunction with other interventions.62,69–71,75,77,78,80,83,101,103,112 One of these trials included two intervention arms (exercise and medication assessment/withdrawal) and shared a single control group.112 To address the lack of independence, two meta-analyses were conducted: the first with the physical activity arm and the second with the medication assessment/withdrawal arm. The pooled estimates from these meta-analyses were consistent with reduced risk for falling, but were not informative because of high statistical heterogeneity (I2=84.0% and 80.1%, respectively). These additional nine studies have been described previously and thus are not discussed further in this section.
Study details. A multifaceted community-based learning program was conducted among adults selected to be higher risk for falling (Table 13). The majority of participants were female (74%) and 65% had fallen at least once during the previous year. Those who had not fallen during the past year but were afraid of falling were also included. The average age of participants was 78 years; race/ethnicity and socioeconomic status were not reported. The intervention group received seven weekly group sessions of 2 hours duration and a single booster session held 3 months after the final group session. All sessions were conducted by an occupational therapist. The control group received two social visits conducted by an occupational therapist. Among the intervention group, 52% fell during the 14-month followup, compared with 58% among the control group. This difference was not statistically significant.
Attrition was low (<10%). Falls were reported monthly using diaries; 90% of intervention participants attended at least five educational sessions. Homebound individuals and those with cognitive problems were excluded.
KQ 1a. Do These Interventions Reduce Injury, Improve Quality of Life, Reduce Disability, or Reduce Mortality in Older Adults Specifically Identified as High Risk for Falls?
KQ 2a. Do These Interventions Reduce Falls in Older Adults Specifically Identified as High Risk for Falls?
A variety of methods were used to select higher-risk populations in the interventions included in this review (Table 14). The evidence for improved health outcomes other than falling among high-risk populations was insufficient to evaluate, and thus is not further discussed. Meta-regression analyses were conducted to evaluate the association between study characteristics related to risk status and estimates of the effect of the intervention on risk for falling. Risk status characteristics included in meta-regressions included history of a fall at baseline (percent), average age (≥80 years vs. <80 years), percent female subjects, high-risk selection (history of falls, gait and balance limitations, other, none), and fall risk in the control group during followup. When possible, we conducted the meta-regression by type of intervention.
In meta-regression analyses with all included studies, none of the risk status characteristics explained a significant amount of the intervention effect. Evidence related to risk for falling within each intervention category is described below.
Multifactorial Assessment and Management
There was sufficient variability to evaluate the influence of age, history of falls, and high-risk selection on the effect of multifactorial assessment and management programs on fall risk via meta-regression. None of these risk characteristics could explain the significant between-study variability in effect sizes.
Twelve of the 14 trials were conducted in populations selected for falls risk. One trial, conducted entirely among women, reported no significant reduction of fall risk overall. Post-hoc analyses were restricted to those participants with more than two falls during the 3 months prior to the study, and this subgroup analysis supported a significant difference between the intervention and control groups (p=0.046).82 Another study that conducted a similar post-hoc analysis reported no differences in fall risk between the intervention and control groups by history of one or more falls during the past 3 months at baseline.83 A third study also found no significant difference in post-hoc analyses on fall outcomes among participants at higher fall risk: people with history of two or more falls during the previous year, people with mobility impairments, and people with limitations in activities of daily living.84
Single Clinical Treatment
Two of the Vitamin D trials in unselected populations conducted post-hoc analyses to evaluate fall outcomes separately for high-risk populations: people with low dietary calcium intake or low serum 25-hydroxyvitamin D levels92,111 and women.111 The effect of vitamin D alone on risk for falling was larger among participants with less than 512 mg/d calcium intake.92 The study’s authors, however, do not provide the statistical significance of the interaction term. In the second study, fall reduction was not enhanced among individuals with low serum 25-hydroxyvitamin D levels (p=0.71) or in women (p=0.25).111
One of the four vision-correction trials conducted a subgroup analysis of those with a history of falls during the past year at baseline and found no significant differences in effect.90
Exercise and Physical Therapy
Ten of the 17 trials were conducted in populations selected to be high risk; none identified a high-risk population based solely on history of falls. There was sufficient variability to evaluate the influence of age, history of falls, and high-risk selection (selected vs. unselected) on the effect of exercise and physical therapy programs on fall risk through meta-regression. None of these risk characteristics explained a significant amount of between-study variability in effect sizes.
Differences in results, however, were found when the observed fall risk among the control group after randomization (>35% vs. <35%) was used as a measure of risk status. This stratification roughly represents the groups above and below the average fall risk among older adults. Meta-analyses stratified by fall risk in the control group support the qualitative observation. The pooled risk from 10 studies conducted among high-risk populations was 0.84 (95% CI, 0.78 to 0.91) (I2=1.1%), while the pooled relative risk was no longer significant (RR, 0.98 [95% CI, 0.82 to 1.17]) (I2=0%) among the five studies conducted in a low-risk population (Figure 9).
None of the physical-activity trials conducted post-hoc subgroup analyses to evaluate the influence of high-risk characteristics.
Home Hazard Modification
The only home-hazard modification trial conducted among a selected population—older adults aged 75 years or older with reduced distance visual acuity—reported a significant reduction in fall risk. None of the trials with unselected populations conducted subgroup analyses to evaluate specific risk factors.
Clinical Education/Behavioral Counseling
The single trial solely evaluating a clinical education/behavioral counseling intervention was conducted in a high-risk population and did not evaluate subgroups.
KQ 2b. Are There Positive Outcomes Other Than Reduced Falls and Related Morbidity and Mortality That Result From Primary Care Falls Interventions?
The main outcomes assessed by included trials were fear of falling or falls efficacy67,69,71,76,80,85,87,91,103,114,116,118 and balance, gait, and mobility measures.75,83,94,96,114,116,118 Additional positive outcomes assessed in trials, but not reported as part of this report, are described in Appendix H Table 2.
Multifactorial Assessment and Management
Five trials evaluated other positive outcomes, including falls efficacy69,71,76 and neuromuscular performance measures.75,83 Three multifactorial assessment and management trials (n=930) reported a significant difference in 12-month change in fear of falling using the falls efficacy scale favoring the intervention group.69,71 The differences in the mean change score ranged from 1.4 to 7.5 and were consistent with a decline in falls efficacy in the control group, compared with no change in the intervention group. One moderate-intensity trial (n=453) reported significantly different 12-month improvements in the timed Get Up and Go test and the Berg balance scale in intervention participants compared with controls.83 This trial provided all intervention group participants with a referral to a physical activity intervention. Another trial reported a significantly greater percentage of intervention participants who completed the timed Get Up and Go test in less than 30 seconds (82%) compared with the control group (72%).75
Single Clinical Treatment
Two vision correction trials (n=545) reported a significant difference in 6-month change in falls efficacy favoring the intervention group.87,91 The differences in the mean change score were consistent with reduced falls efficacy in the control group, accompanied by stable or slightly increased falls efficacy in the intervention group (differences ranged from 3.6 to 5.4).
One of the two trials of hip protectors evaluated differences in fear of falling between the intervention and control groups at 12 months and reported significantly less fear of falling in the intervention group (p=0.003).85
Exercise and Physical Therapy
Four trials (n=1,000) evaluated other positive outcomes, including the timed Get Up and Go test,96 functional reach,96,116 Berg balance scale,96,104,114,116 timed walk,94,104 and falls efficacy.114,116 One study reported significantly better falls efficacy, functional reach, and balance in the intervention group compared with the control groups at 12 months (p<0.05). All the measures were balanced at baseline.116 Only one other trial reported significantly different mean change in functional reach after 6 months (adjusted difference in change, 5.7 [95% CI, 0.47 to 11.0]), although no differences were found in the timed Get Up and Go test or the Berg balance score.96 The remaining studies reported no significantly different changes in these positive outcomes.
Home Hazard Modification
None of the studies evaluated other positive outcomes.
Clinical Education/Behavioral Counseling
The community-based group behavioral counseling intervention was not associated with a significantly greater improvement in falls efficacy when comparing the intervention group with the control group over the 14-month followup (mean difference, 1.74 [95% CI, −6.1 to 2.7]).
KQ 3. What Are the Adverse Effects Associated With Interventions to Reduce Falls?
In addition to reviewing the trials included in KQs 1 and 2, we conducted additional searches for evidence addressing significant clinical harms (i.e., intervention-related events requiring medical services) for each intervention type. Due to both practical limitations and the availability of recent AHRQ-funded evidence reports, we did not systematically review the evidence of harms of vitamin D supplementation, vision screening, or early vision correction in older adults. These topics are briefly addressed in the Discussion section. For KQ 3, we included harms reported in 44 trials from KQs 1 and 2 (Appendix C Tables 1–5), two additional trials on exercise interventions,119 and one systematic review on protein and energy supplementation in older adults120(Appendix D Tables 1 and 2). We found no observational studies examining clinically significant harms of fall prevention interventions.
Summary of findings. Based on the meta-analyses conducted for KQ 2, there was limited evidence of paradoxical effects of the falls prevention interventions resulting in an increased number of fallers, fall-related fractures, or increased rate of falls. A few physical activity interventions94,100,114 and multifactorial assessment and management interventions62,71,75,79 reported increased falls in the intervention group, but only one of these was statistically significant.79 In addition, there does not appear to be an increase in all-cause mortality or disability or decrease in self-reported quality of life with falls prevention interventions (Figure 2). We found no evidence to suggest serious harms of hip protectors, medication withdrawal, liquid protein-energy supplementation, vitamin D supplementation, clinical education and counseling, home hazard modification, or exercise and physical therapy interventions. In one trial (n=312) conducted in New Zealand, a group randomized to receive a nurse-conducted multifactorial clinical assessment plus referral intervention had more fallers than the control group.79 Based on one fair-quality trial (n=616) in Australia, vision assessment and correction in frail older adults may have increased fallers.90 Overall, trials were primarily designed to evaluate the intervention’s efficacy or effectiveness, and therefore did not report adverse outcomes other than falls outcomes.
Minor adverse outcomes associated with specific interventions include: increased fall-related outpatient visits after comprehensive falls assessment, self-reported musculoskeletal complaints (but not outpatient visits or hospitalizations) with exercise interventions, increased outpatient visits for abnormal heart rhythm with exercise intervention, minor local skin irritation or infection with use of hip protectors, gastrointestinal side effects with liquid protein-energy supplementation, and transient or asymptomatic hypercalcemia and hypercalciuria with vitamin D supplementation.
Study details. We found 12 fair- to good-quality trials (n=5,099) that evaluated the effectiveness of multifactorial assessment interventions.62,69,71,75,77–84 Overall, there was no evidence for clinically significant harms. One good-quality trial (n=312) in New Zealand evaluating a nurse-conducted multifactorial clinical assessment with referral had a slightly higher proportion of fallers and frequent fallers at 12 months (i.e., 2 or more falls) in the intervention group compared with the control group (68.4% vs. 62.4% [p=0.040] and 44.5% vs. 34.4% [p=0.067], respectively).79 Only three of the 12 trials explicitly reported on additional adverse effects.69,77,84 In one fair-quality trial (n=348) evaluating a nurse-conducted multifactorial assessment and referral in the United Kingdom, persons in the intervention group had more fall-related outpatient visits to their general practitioner than persons receiving usual care.77 In another fair-quality trial (n=301) evaluating a comprehensive multifactorial assessment with management, persons in the intervention group self-reported more musculoskeletal symptoms, which were probably related to the exercise program according to the study investigators. However, there was no increase in falls, hospitalizations, or deaths.69
There were 15 fair- to good-quality trials (n=12,133) evaluating different single clinical treatments. Two fair-quality trials (n=4,769) evaluated hip protectors in older community-dwelling adults.66,85 In these two trials, there was no statistically significant increase in falls or frequent fallers. In one trial, the investigators reported that 5% of persons in the intervention group had minor local complications, including skin irritation or infection due to wearing hip protectors.66 In one fair-quality trial (n=93) of persons receiving medication management and medication withdrawal to prevent falls, no adverse effects were reported.112
Four fair- and good-quality trials (n=1,437) included interventions to correct vision to prevent falls.61,87,90,91 In one fair-quality trial (n=616) in Australia, frail older adults received a comprehensive eye exam with subsequent treatment of vision problems. Approximately 44% of participants received some sort of vision-related intervention. Persons in the intervention group, compared with those in the control group, had a higher proportion of fallers (65.0% vs. 49.8% [p=0.0001]) and frequent fallers (37.9% vs. 30.6% [p=0.003]).90 There was also a nonsignificant trend in fall-related fractures. The trial investigators hypothesized that corrected vision may have increased the level of activity of these frail older adults, thereby increasing their risk for falls. In the two trials (n=545) evaluating expedited cataract surgery, complication rates from the cataract surgery were reported for the intervention group (cataract surgery at approximately 4 weeks) but not for the control group (surgery at approximately 12 months). Complications included iris damage, anterior vitrectomy performed, and posterior capsular opacification noted 6 months afterward. Harms for vision assessment and early treatment are addressed in the discussion.
Only one fair-quality trial (n=50) evaluated liquid protein and energy supplementation in frail older adults.88 This trial did not report any adverse effects. One good-quality systematic review designed to evaluate the effectiveness of oral protein and energy supplementation for older adults found 18 trials that reported adverse effects.120 Ten of the 18 trials found some problems with tolerance and gastrointestinal side effects (e.g., nausea, vomiting, and diarrhea). Most of these trials, however, were conducted in hospitals or nursing homes. Only two trials were conducted among community-dwelling adults: one trial in persons with diabetic foot ulcers and one trial in undernourished persons recently discharged from the hospital. Most of these trials did not report methods for assessing harms or for conducting comparisons with the control group.
Based on eight fair-quality trials (n=5,216) evaluating vitamin D supplementation for the prevention of falls in older adults, there is no evidence of an increase in falls or fallers or other significant clinical harms. Most of the trials, however, did not report adverse effects.67,89,92,98,99,102,111 Three trials (n=926) reported transient and asymptomatic hypercalciuria or hypercalcemia in the intervention group but no differences in side effects or significant harms, such as incident kidney stones, cancer, ischemic heart disease, or stroke.86,89,92 Harms of vitamin D supplementation are addressed in the Discussion section.
In one good-quality trial (n=310) evaluating primarily clinical education and behavioral counseling to prevent falls, there was no increase in falls or fallers. However, no additional adverse effects were reported.103
Based on the three fair-quality trials (n=2,348) that included home hazard modification interventions, there was no evidence of increased falls or fallers.61,63,70 None of these trials reported additional adverse events.
There is no evidence of an increase in falls or fallers due to exercising, based on 17 fair- to good-quality trials (n=3,985) examining exercise and physical therapy interventions to prevent falls61,63,93–97,100,104,105,107,108,112–116 plus two additional fair-quality trials (n=496) identified in our search for harms.119,121 Few of these trials report additional adverse effects. Two trials (n=312) reported one fall while exercising as instructed, although there was no increased number of fallers in the intervention group overall.95,119 One fair-quality trial (n=424) that explicitly evaluated adverse effects found that persons in the exercise group had more physician visits for abnormal heart rhythm compared with those in the control group (20.2% vs. 11.4% [p=0.016]), but not for other problems, including syncope, shortness of breath, or musculoskeletal complaints.121 There were also no statistically significant differences between the two groups for serious harms, including clinically significant abnormal laboratory or other diagnostic testing, hospitalization, or life-threatening event.
KQ 4. How Are Community-Dwelling Older Adults Identified for Primary Care Interventions to Prevent Falls?
Summary of findings. The intervention studies that focused on higher-risk community-dwelling older adults considered different, noncomparable sets of self-reported or measured risk factors to identify those at risk for falling. These risk factor assessment approaches included from one to six different risk factors (some with multiple measures for a single risk factor, such as balance). Few studies used the same set of risk factors or used a formal battery (risk assessment tool). To complicate matters further, the measured risk factors (e.g., balance, gait speed, mobility) used several different measures across studies and many studies used measures that would not be feasible for use in routine primary care. Other reviewers have noted similar challenges in addressing the falls epidemiology and intervention literature.122
Study details. Most of the fall prevention interventions selected participants considered high risk for reasons in addition to age. Nine of the trials, however, defined high risk only as an age of 70 years or older,61,67,70,92,99 75 years or older,63,81,95 or 80 years or older97 (Table 14). Five trials78,83,108,111,115 were unselected for falls risk. Of the 32 interventions that selected participants based on risk factors for falls, seven trials 66,85,87,89,91,97,98 included only women, although female gender was not the only selection criteria.
Twelve of the 32 interventions that selected participants based on other risk factors for falls defined high risk according to fall history. A history of at least one fall in the last 2 to 12 months was required for participation in seven studies75,79,82,89,100,102,103 and a history of more serious falls (e.g., falls leading to hospitalization or urgent/emergency/specialty health care services, multiple falls) was required in another seven studies.66,71,76,77,80,84,93
Even within these studies, few used the same method to assess fall history. Several studies used fall history as one of several risk factors assessed to qualify participants for interventions.66,71,79,89,93,100,102,103,114 In two of these studies, people who had fallen during the past year represented less than half of the selected participants.71,93
Eight studies included participants at high risk as defined by measures of gait or balance impairment or mobility limitation.62,69,71,93,100,104,105,114 While some of these studies used similar measures to define gait, balance, or mobility impairment, none were precisely the same, and most would not be feasible in the primary care setting.
Agency for Healthcare Research and Quality (US), Rockville (MD)
Michael YL, Lin JS, Whitlock EP, et al. Interventions to Prevent Falls in Older Adults: An Updated Systematic Review [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2010 Dec. (Evidence Syntheses, No. 80.) 3, Results.