Vitamin D Treatment for the Prevention of Falls in Older Adults: Systematic Review and Meta-analysis

Rita Rastogi Kalyani, MD, MHS; Brady Stein, MD, MHS; Ritu Valiyil, MD, MHS; Rebecca Manno, MD, MHS; Janet W. Maynard, MD, MHS; Deidra C. Crews, MD, ScM

J Am Geriatr Soc. 2010;58(7):1299-1310. 

Abstract and Introduction

Abstract

Objectives: To systematically review and quantitatively synthesize the effect of vitamin D therapy on fall prevention in older adults.
Design: Systematic review and meta-analysis.
Setting: MEDLINE, CINAHL, Web of Science, EMBASE, Cochrane Library, LILACS, bibliographies of selected articles, and previous systematic reviews through February 2009 were searched for eligible studies.
Participants: Older adults (aged ≥60) who participated in randomized controlled trials that both investigated the effectiveness of vitamin D therapy in the prevention of falls and used an explicit fall definition.
Measurements: Two authors independently extracted data, including study characteristics, quality assessment, and outcomes. The I2 statistic was used to assess heterogeneity in a random-effects model.
Results: Of 1,679 potentially relevant articles, 10 met inclusion criteria. In pooled analysis, vitamin D therapy (200–1,000 IU) resulted in 14% (relative risk (RR)=0.86, 95% confidence interval (CI)=0.79–0.93; I2=7%) fewer falls than calcium or placebo (number needed to treat =15). The following subgroups had significantly fewer falls: community-dwelling (aged <80), adjunctive calcium supplementation, no history of fractures or falls, duration longer than 6 months, cholecalciferol, and dose of 800 IU or greater. Meta-regression demonstrated no linear association between vitamin D dose or duration and treatment effect. Post hoc analysis including seven additional studies (17 total) without explicit fall definitions yielded smaller benefit (RR=0.92, 95% CI=0.87–0.98) and more heterogeneity (I2=36%) but found significant intergroup differences favoring adjunctive calcium over none (P=.001).
Conclusion: Vitamin D treatment effectively reduces the risk of falls in older adults. Future studies should investigate whether particular populations or treatment regimens may have greater benefit.
Falls are a common but serious cause of morbidity and mortality in older adults. Falls occur in up to 30% of community-dwelling adults and up to 50% of institutionalized older adults, resulting in nearly 16,000 deaths in 2006.[1–6] In addition to physical injury, falls can lead to a loss of independence and compromised emotional health.[7] Falls are costly, accounting for $19 billion in healthcare expenditures in 2000.[8,9] Thus, the implementation of an effective, inexpensive fall-prevention strategy is highly desirable.

Introduction

Falls result from a culmination of diverse risk factors acting in synergy with advancing age, disease states, and hazards in the environment. Although there are a multitude of recognized risk factors, certain risk factors portend a particularly high risk for future falls, including abnormalities in muscle strength, gait, and balance.[10,11] Recent studies have suggested that vitamin D supplementation is a safe, well-tolerated approach to improve muscle strength and function, leading to fewer falls.[12,13]

Older adults may be at greater risk for vitamin D deficiency for many reasons, including lower cutaneous synthesis of vitamin D,[14,15] so pharmacological supplementation of vitamin D is often required in older adults. The most commonly available forms are native vitamin D (cholecalciferol or vitamin D3, ergocalciferol or vitamin D2, 25-hydroxyvitamin D) and active vitamin D (calcitriol, alfacalcidol or 1,25-dihydroxyvitamin D).

Epidemiological studies support an association between vitamin D and falls.[12] Vitamin D may reduce the risk of falls in older adults through an improvement in muscle function and strength.[13] Muscle biopsies in patients who are deficient in vitamin D demonstrate atrophy of type II fibers, which are recruited first to prevent a fall.[16,17] Binding of the vitamin D receptor (VDR) in muscle affects transcription of genes that modulate calcium and phosphate uptake, phospholipid metabolism, and muscle cell proliferation and differentiation.[16,17] In VDR knockout mice, impaired motor coordination is observed.[16] Clinical vitamin D deficiency is associated with a proximal myopathy that improves with treatment.[16,17] Body sway improved more with vitamin D and calcium than with calcium alone in elderly ambulatory women.[18] Improvements in muscle strength, and subsequent gait stability, may explain the association between vitamin D therapy and fall prevention.

Although previous reviews have found insufficient evidence to conclude that vitamin D supplementation reduces the risk of falling[5,19–21] or greater benefit with active vitamin D,[22,23] one study[24] reported a significant 22% fewer falls with any type of vitamin D. In a recent meta-analysis of eight randomized controlled trials,[25] native vitamin D treatment reduced falls by 13%, although there was significant heterogeneity of studies. Only high-dose vitamin D (700–1,000 IU), achieved serum 25-hydroxyvitamin D concentrations greater than 24 ng/mL, and active vitamin D treatment significantly reduced fall risk by 19% to 23%. Most previous meta-analyses were underpowered or did not examine specific subgroups of patients in whom vitamin D may have differential effects. In recent years, additional randomized controlled trials using various fall ascertainment techniques have been published reporting the potential benefits of vitamin D in the prevention of falls; therefore, an updated meta-analysis is warranted.

The objective of the current study was to perform a comprehensive, updated systematic review and meta-analysis of randomized controlled trials that have evaluated the effectiveness of vitamin D therapy on fall prevention in older adults (mean age ≥60), adding to recent findings[25] by investigating if progressively higher doses of vitamin D are associated with greater benefits in fall risk. Whether beneficial effects of vitamin D treatment extend to hospitalized patients and whether particular patient factors (e.g., history of prior fracture or fall) or treatment regimens (e.g., adjunctive calcium use) contribute to the effect of vitamin D was also explored. Formal quality assessment of included studies was also undertaken to identify possible sources of systematic biases.

Methods

Data Sources and Selection

A standardized protocol was developed and followed for all steps of the review. Investigators searched MEDLINE, CENTRAL, EMBASE, CINAHL, Web of Science, and LILACS and hand-searched bibliographies of selected articles and previous systematic reviews to identify articles that potentially met the inclusion criteria. Medical subject headings, keywords, and truncated word vocabulary were used in the search strategies. The search strategy combined terms for the intervention of vitamin D (1,25-dihydroxycholecalciferol, 25-hydroxycholecalciferol, 1,25-dihydroxyvitamin D, 25-hydroxyvitamin D, vitamin D, ergocalciferol, cholecalciferol, calcitriol, hydroxycholecalciferol, dihydroxycholecalciferol, calcifediol, vitamin D2, vitamin D3, paricalcitol, vitamin D analogs and derivatives), outcome of falls (accidental falls, body sway, gait), and study design of randomized clinical trials using the Cochrane highly sensitive search strategy (MEDLINE only).[26] For smaller databases, to maximize the yield, the search strategy did not include syntax for study design. All electronic databases were accessed on February 9, 2009.

Study Selection

Two investigators independently reviewed the titles, abstracts, or full-text manuscripts of relevant articles identified through the literature search to determine whether they met eligibility criteria (Figure 1). A study was eligible for inclusion if it was a randomized, controlled trial; the mean age of study participants was 60 and older; the study compared vitamin D treatment with calcium therapy, placebo, or no treatment; the number of participants with one or more falls according to treatment arm was given; and an explicit fall definition was provided along with a description of how falls were ascertained. Falls were defined as "unintentionally coming to rest on the ground, floor, or other lower level."[27] Studies that did not have an explicit fall definition were included in a post hoc analysis. To examine potential subgroup effects, all eligible studies were included, regardless of dwelling (community, nursing institution, or hospital); participants' history of fracture or fall; vitamin D levels; adjunctive calcium therapy; or dose, type, and duration of vitamin D therapy. Studies that used only intramuscular vitamin D were excluded because this route is not as effective as oral.[28] Studies restricted to participants with significant neurological disabilities, such as Parkinson's disease or stroke with hemiplegia, in whom the independent effects of vitamin D on fall risk may be unclear, were also excluded. There were no restrictions on year or language, although at least one author had to be able to translate the study. Disagreements regarding inclusion and exclusion of studies were resolved according to group consensus and by referring to the original reports.

Figure 1.

 

Study selection flow chart according to QUORUM guidelines. RCT=randomized controlled trial

Data Extraction and Quality Assessment

Study investigators independently abstracted data in duplicate using a standardized form. The abstracted data were then entered into Microsoft Excel (Microsoft Corp, Redmond, WA), and a second reviewer verified them. Abstracted data included study design (date and location of study, sample size), patient characteristics (risk factors for falling, such as history of previous fracture or fall and comorbidities), study methodology (eligibility criteria, method of randomization, blinding), intervention (type, dose, duration of therapy), adverse events, and main results. Authors were not contacted directly for missing data. Methodological quality of included studies was investigated by collecting data on sources of systematic bias using published guidelines.[29] Collected data included information regarding sequence generation, allocation concealment, assessor blinding, incomplete outcome data, selective reporting, eligibility criteria, therapies, excluded patients, and reliability of fall ascertainment. Sources of funding were documented.

The primary outcome of the meta-analysis was the number of participants with one or more falls during follow-up. Studies in which only the relative risk of falling in each treatment arm was reported, but not the number of fallers, were included. Total number of falls was not analyzed because individuals with recurrent falls may have other significant risk factors for falling.

Data Synthesis and Analysis

A summary relative risk (RR) was calculated for the primary (dichotomous) outcome of number of participants with one or more falls during follow-up. The planned analysis was vitamin D arm (with or without calcium) versus comparator arm (placebo, calcium, or no treatment). For studies that tested multiple doses of vitamin D in separate intervention arms, a pooled effect estimate from all arms with vitamin D were compared with a pooled effect estimate from all arms without vitamin D for the main analysis.

The presence of heterogeneity was evaluated using the I2 statistic for pooled study-level data. An I2 statistic greater than 50% suggested moderate heterogeneity.[30] The a priori hypothesis was that there would be heterogeneity between and within studies, so a random effects model was used. The following a priori subgroup analyses were conducted to explore potential heterogeneity between studies: type, dose, and duration of vitamin D; use of adjunctive calcium in treatment arm; type of dwelling; history of fall or fracture in majority of participants; and mean baseline vitamin D level of 30 ng/mL or less (vitamin D insufficiency). For studies with arms that used multiple doses of vitamin D, each arm was compared with the control arm to derive a RR of falling for the relevant subgroup analyses. A meta-regression analysis examining the linear association between dose and duration of vitamin D treatment and the RR of falling was also performed. Sensitivity analysis was performed on studies with similar baseline characteristics between intervention and control groups. Publication bias was assessed using Begg's funnel plot and the Begg's and Egger's statistical tests.[31,32] All analyses were performed using Stata version 10.0 (StataCorp, College Station, TX).

Results

Search Results and Study Characteristics

A total of 1,679 unique titles and abstracts were retrieved from the search (Figure 1); 1,543 studies were excluded after title and abstract review because they did not satisfy the inclusion criteria. Subsequently, 136 articles underwent full text review, with 126 studies excluded. (Seven of these articles without an explicit fall definition were later examined in post hoc analysis.) Ultimately, 10 randomized controlled trials met inclusion criteria for primary analysis.

The mean ages of participants ranged from 71 to 92 (), and the majority were women. The included studies spanned eight countries; three were multicenter trials.[33–35] One study[36] included only hospitalized patients, four included institutionalized participants,[33,34,37,38] and five evaluated community-dwelling adults.[18,35,39–41] All studies in community-dwelling adults had a mean age younger than 80. Four studies specified that the majority of participants had a history of previous fracture or fall,[36–39] whereas four specified that most participants did not have history of fractures or falls.[18,34,35,40] In all studies in which baseline mean 25-hydroxyvitamin D levels were reported, the level in at least one treatment arm was less than 30 ng/mL.

Table 1.  Study Characteristics of Randomized Controlled Trials Included in Meta-Analysis

Study Therapy, Dose, Frequency N Age, Mean Follow-Up, Months How Fall Assessed Baseline/Change in 25-Hydroxy-Vitamin D, ng/mL* Previous Fracture, %* Previous Fall, %* Population
Studies included in primary analysis
   Bischoff et al.[37] I: Cholecalciferol 400 IU bid, calcium carbonate 600 mg bid 62 85 3 O 12/14 56 24 Institutionalized older adults awaiting nursing home placement
C: Calcium carbonate 600 mg bid 60 84 3 O 12/−0.2 52 23
   Bischoff-Ferrari et al.[41] I: Cholecalciferol 700 IU daily, calcium citrate 500 mg daily 219 71 36 Q, P M 33, F 28 Healthy, ambulatory, community dwelling
C: Placebo 226 71 36 Q, P M 33, F 25
   Broe et al.[38] I: Ergocalciferol 200 IU daily 26 92 5 O 18 69 Institutionalized, very old
I: Ergocalciferol 400 IU daily 25 88 5 O 21 68
I: Ergocalciferol 600 IU daily 25 89 5 O 17 64
I: Ergocalciferol 800 IU daily 23 89 5 O 21 65
C: Placebo+63% took multivitamin 25 86 5 O 21 44
   Burleigh et al.[36] I: Cholecalciferol 800 IU daily, calcium carbonate 1200 mg daily 100 82 1 O 9/1 26 53 Hospitalized, ill
C: Calcium carbonate 1,200 mg daily 103 84 1 O 10/0 26 48
   Dukas et al.[40] I: Alfacalcidol 1 μg daily 192 75 9 D 30 5 Ambulatory, community dwelling
C: Placebo 186 75 9 D 28 13
   Flicker et al.[34] I: Ergocalciferol 10,000 IU per week then 1,000 IU daily+calcium carbonate 600 mg daily 313 84 24 D Median<16 27 Institutionalized with vitamin D level between 25 and 90 ng/mL
C: Calcium carbonate 200 mg daily 312 83 24 D Median<16 24
   Graafmans et al.[33] I: Cholecalciferol 400 IU daily 177 83 7 D, Q Institutionalized
C: Placebo 177 83 7 D, Q
   Pfeifer et al.[18] I: Cholecalciferol 400 IU bid, calcium carbonate 600 mg bid 70 75 2 Q 26/40 0 Ambulatory, community dwelling, vitamin D<50 ng/mL
C: Calcium carbonate 600 mg bid 67 75 2 Q 26/18 0
   Pfeifer et al.[35] I: Cholecalciferol 400 IU bid, calcium carbonate 500 mg bid 122 76 12 D 22 0 Ambulatory, community dwelling, vitamin D<50 ng/mL
C: Calcium carbonate 500 mg bid 120 77 12 D 22 0
   Prince et al.[39] I: Ergocalciferol 1,000 IU daily, calcium citrate 500 mg bid 151 77 12 Q 18 60 Community dwelling, recruited from emergency department or nursing home, vitamin D<24 ng/mL
C: Calcium citrate 500 mg bid 151 77 12 Q 18 58
Studies included in post hoc analysis
   Gallagher et al.[42] I: Calcitriol 0.25 μg bid+calcium 500–1,000 mg daily 123 72 36 Q 31 28 Healthy, community-dwelling women; primary outcome bone mineral density
I: Estrogen+medroxyprogesterone+calcium 500–1,000 mg daily 121 72 36 Q 31 17
I: Calcitriol 0.25 μg bid+estrogen+medroxyprogesterone+calcium 500–1,000 mg daily 122 71 36 Q 32 16
C: Placebo+calcium 500–1,000 mg daily 123 71 36 Q 32 14
   Grant et al.[43] I: Ergocalciferol 800 IU daily 1,343 77 60 Q 100 Community-dwelling, recruited from fracture clinic or orthopedic ward; primary outcome fracture
I: Ergocalciferol 800 IU+calcium 1,000 mg daily 1,306 77 60 Q 100
I: Calcium 1,000 mg daily 1,311 78 60 Q 100
C: Placebo 1,332 77 60 Q 100
   Harwood et al.[44] I: Cholecalciferol 800 IU daily+calcium 1,000 mg daily 29 83 12 Q 12/8 100 Community dwelling, recruited<7 days after hip fracture surgery
C: None 25 81 12 Q 11/−0.1 100
   Law et al.[45] I: Ergocalciferol 1,100 IU daily 1,762 85 10 D Institutionalized, recruited from residential care homes
C: None 1,955 85 10 D
   Latham et al.[46] I: Ergocalciferol 300,000 IU single oral dose (1,600 IU daily) 108 80 6 D 15/9 44 57 Frail, hospitalized in acute care or rehabilitation facilities
C: Placebo 114 79 6 D 19/0 43 55
   Porthouse et al.[47] I: Cholecalciferol 800 IU daily+calcium 1,000 mg daily+leaflet 1,321 77 36 Q 59 34 Community-dwelling women, ≥1 risk factors for fracture; primary outcome fracture
C: Leaflet 1,993 77 36 Q 58 34
   Trivedi et al.[48] I: Cholecalciferol 100,000 IU every 4 months (800 IU daily) 1,027 75 48 Q Community-dwelling British doctors; primary outcome fracture and mortality
C: Placebo 1,011 75 48 Q

* If reported.
Primary outcome was falls unless otherwise indicated.
I=intervention; C=control; O=observed; D=diary; P=postcard; Q=questionnaire; bid=twice a day; M=male; F=female; IU=international units.

All intervention arms were compared with placebo or calcium. All studies were parallel group design and had two arms (intervention and control) except for one[38] that randomized patients to five treatment arms based on vitamin D dose (200–800 IU) and a placebo arm. Seven studies had intervention arms that included adjunctive calcium supplementation (500–1,200 mg);[18,34–37,39,41] although most of these also included calcium supplementation in the control arm, one used placebo in the control arm.[41] The remaining three trials compared vitamin D with placebo.[33,38,40] Three types of vitamin D (cholecalciferol, ergocalciferol, alfacalcidol) were used. Treatment duration ranged between 1 and 36 months, and native vitamin D dosage ranged between 200 and 1,000 IU. Adherence ranged from 86% to 98% for most studies in which this information was reported.[18,34,36,38] The follow-up rate was greater than 85% in studies that reported completion rates.[18,34,36,38–40] Two studies reported involvement with industry sponsorship.[18,35]

Falls were a primary outcome in all included studies and were ascertained according to direct observation in three studies,[36–38] questionnaire alone in two studies,[18,39] and fall diaries in three studies.[34,35,40] Two studies used a combination of methods, including questionnaire and postcard[41] and questionnaire and diary.[33]

Assessment of Methodological Quality

In general, methodological quality of included studies was good (). All studies had clearly defined eligibility criteria and therapies and reliable fall ascertainment. All studies were double-blind except for one,[33] which did not clearly mention the method of blinding and may have been subject to detection bias; in this study, a subgroup of participants was followed as part of a larger, observational study and randomized to vitamin D treatment. Sequence generation was adequately described in all studies except four.[18,33,35,41] In three of these studies,[18,33,35] there was insufficient information on allocation concealment, which may have made them vulnerable to selection bias. At least one of the following was absent or unclear in three studies[33,35,41]—incomplete outcome data addressed, similar rates of follow-up, and reasons for loss to follow-up—rendering these studies vulnerable to attrition bias. Reasons for exclusion were described in all studies except one.[33] Baseline characteristics were dissimilar between study arms in two studies because of differences in previous fracture rate[34] or anticoagulant use[40] and were unclear in two studies.[33,36]

Table 2.  Qualitative Analysis of Included Studies*

Study Adequate Sequence Generation Described Allocation Concealment Described Assessor Blinding Incomplete Outcome Data Addressed Free from Other Bias Eligibility Criteria Defined Excluded Patients Described (n and Reason) Rates of Follow-Up Similar Describe Reasons for Loss to Follow-Up Describe All Therapies Clearly Prospective Sample Size Justification Similar Baseline Characteristics Between Groups
Studies included in primary analysis
   Bischoff et al.[37] Y Y Y Y Y Y Y Y Y Y Y Y
   Bischoff-Ferrari et al.[41] U Y Y U Y Y Y U Y Y U Y
   Broe et al.[38] Y Y Y Y Y Y Y Y Y Y U Y
   Burleigh et al.[36] Y Y Y Y Y Y Y Y Y Y Y U
   Dukas et al.[40] Y Y Y Y Y Y Y Y Y Y Y N
   Flicker et al.[34] Y Y Y Y Y Y Y Y Y Y Y N
   Graafmans et al.[33] U U U U Y Y U U N Y U U
   Pfeifer et al.[18] U U Y Y U Y Y Y Y Y Y Y
   Pfeifer et al.[35] U U Y Y U Y Y U U Y Y Y
   Prince et al.[39] Y Y Y Y Y Y Y Y Y Y Y Y
Studies included in post hoc analysis
   Grant et al.[43] Y Y Y Y Y Y Y Y Y Y Y Y
   Gallagher et al.[42] Y Y Y Y Y Y Y Y Y Y Y Y
   Harwood et al.[44] Y Y N Y Y Y Y Y Y Y N Y
   Law et al.[45] Y Y N Y Y Y Y Y Y Y Y U
   Latham et al.[46] Y Y Y Y Y Y Y Y Y Y Y Y
   Porthouse et al.[47] Y U N Y U Y Y Y N U Y Y
   Trivedi et al.[48] U U Y Y Y Y N U N Y N Y

* Y=yes, N=no, U=unclear.

Statistical methods were described in all studies. Prospective sample size justification was not clearly stated in three studies,[33,38,41] whereas intention-to-treat analysis was clearly stated in all but one study.[33]

Evidence Synthesis for Primary Analysis The 10 studies included 2,932 participants (Figure 2). There was a statistically significant effect of vitamin D treatment on falls, with a pooled RR of 0.86 (95% confidence interval (CI)=0.79–0.93; I 2=7%; P=.38). From the pooled risk difference, the number needed to treat was 15, resulting in 68.1 (95% CI=35.1–98.5) out of 1,000 falls avoided by vitamin D treatment.

Figure 2.

 

Forest plot comparing the risk of falling in groups treated with vitamin D and control groups in the primary (above) and post hoc (below) analyses. Squares represent the relative risk (RR) of falling in the treated groups versus those in control groups. The size of the square is proportional to the size of the trials, and error bars represent the 95% confidence intervals (CIs). The diamond shape represents the pooled RR, which was 0.86 (95% CI=0.79–0.93) for the 10 studies in primary analysis and 0.92 (95% CI=0.87–0.98) for the 17 studies in the post hoc analysis, which included seven additional studies that did not have explicit fall definitions. Pooled numbers in post hoc analysis do not include one study[47] that reported only a RR of falling. Relative weight (%) of each study in the pooled analysis is also indicated.

In subgroup analysis (), significantly fewer falls were found in the following subgroups: community-dwelling participants (aged <80)[18,35,39–41] (RR=0.79, 95% CI=0.69–0.92); majority of participants without history of fracture or fall[18,34,35,40] (RR=0.77, 95% CI=0.62–0.97); adjunctive calcium supplementation[18,34–37,39,41] (RR=0.83, 95% CI=0.75–0.92); and duration longer than 6 months[33–35,39–41] (RR=0.86, 95% CI=0.78–0.94). A dose of 800 IU or greater[18,34–39] (RR=0.80, 95% CI=0.70–0.91) was found to be more favorable than a dose less than 800 IU[33,38,41] (RR=1.01, 95% CI=0.85–1.20), but did not reach statistical significance (intergroup P=.06). Treatment with cholecalciferol[18,33,35–37,41] (RR=0.80, 95% CI=0.69–0.93) was more favorable than with ergocalciferol[34,38,39] (RR=0.89, 95% CI=0.80–1.00) but not statistically significantly (intergroup P=.34). Treatment with alfacalcidol[40] did not result in a statistically significantly lower risk of falls (RR=0.84, 95% CI=0.58–1.22).

Table 3.  Subgroup Analyses for Vitamin D Treatment and Prevention of Falls

Subgroup Studies, n Participants, n Relative Risk of Falling (95% Confidence Interval) I 2, % P Value*
Treatment Control
Studies included in primary analysis
   Adjunctive calcium therapy
      Yes[18,34–37,39,41] 7 1,037 1,039 0.83 (0.75–0.92) 22 .51
      No[33,38,40] 3 468 388 0.94 (0.77–1.15) 0
   Type of vitamin D
      Ergocalciferol[34,38,39] 3 512 488 0.89 (0.80–1.00) 0 Ref
      Cholecalciferol[18,33,35–37,41] 6 750 753 0.80 (0.69–0.93) 24 .34
      Alfacalcidol[40] 1 192 186 0.84 (0.58–1.22) .80
   Participant dwelling and age§
      Community-dwelling, aged <80[18,35,39–41] 5 754 750 0.79 (0.69–0.92) 29 .20
      Hospitalized or institutionalized, aged ≥80[33,34,36–38] 5 751 677 0.90 (0.80–1.01) 0
   History of fracture or fall (in majority of participants) if reported
      Yes[36–39] 4 412 339 0.84 (0.72–0.98) 0 .68
      No[18,34,35,40] 4 697 685 0.77 (0.62–0.97) 59
   Duration of vitamin D treatment
      ≤6 months[18,36–38] 4 331 255 0.79 (0.62–1.01) 0 .89
      >6 months[33–35,39–41] 6 1,174 1,172 0.86 (0.78–0.94) 21
   Dose of oral vitamin D, IU
      <800[33,38,41] 3 472 478 1.01 (0.85–1.20) 21 .06
      ≥800[18,34–39] 7 841 838 0.80 (0.70–0.91) 31
Studies included in post hoc analysis
   Adjunctive calcium therapy
      Yes[18,34–37,39,41–44,47] 11 3,938 4,622 0.86 (0.81–0.92) 0 .001
      No[33,38,40,43,45,46,48] 7 4,708 4,800 1.01 (0.96–1.07) 0
   Type of vitamin D
      Ergocalciferol[34,38,39,43,45,46] 6 5,082 5,200 0.99 (0.92–1.06) 23 Ref
      Cholecalciferol[18,33,35–37,41,44,47,48] 9 3,127 3,792 0.86 (0.77–0.96) 28 .08
      Alfacalcidol or calcitriol[40,42] 2 437 430 0.86 (0.75–1.00) 0 .16
   Participant dwelling and age§
      Community-dwelling, aged <80[18,35,39–44,47,48] 10 6,025 6,676 0.88 (0.81–0.96) 34 .10
      Hospitalized or institutionalized, aged ≥80[33,34,36–38,45,46] 7 2,621 2,746 1.00 (0.94–1.06) 0
   History of fracture or fall (in majority of participants) if reported
      Yes[36–39,43,44,46,47] 8 3,198 3,131 0.96 (0.88–1.04) 0% .11
      No[18,34,35,40,42] 5 942 929 0.82 (0.71–0.94) 46%
   Duration of vitamin D treatment
      ≤6 months[18,36–38,46] 5 439 369 0.93 (0.73–1.17) 37 .41
      >6 months[33–35,39–45,47,48] 12 8,207 9,053 0.92 (0.86–0.98) 42
   Dose of oral vitamin D, IU
      <800[33,38,41] 3 472 478 1.01 (0.85–1.20) 21 .35
      ≥800[18,34–39,43–48] 13 7,737 8.589 0.92 (0.84–1.00) 50

* P value indicates difference between subgroups.
Calcium therapy given in intervention and control arms, except for three studies[41,44,47] in which given only in intervention arm. For studies with multiple arms,[43] vitamin D alone arm was compared with placebo arm and vitamin D with calcium arm was compared with calcium arm.
Compared with ergocalciferol.
§ All studies in community-dwelling individuals had a mean age less than 80 except for one in post hoc analysis;[44] all studies in hospitalized or institutionalized individuals had mean age of 80 and older except for one in post hoc analysis.[46]

Meta-regression did not indicate any significant linear association between vitamin D dose (P=.13) or treatment duration (P=.38) and RR of falls. Furthermore, when the meta-regression was restricted to studies of a dose less than 800 IU, there was no significant linear association between dose and risk of falling (P=.35). The absence of a linear association raises the question as to whether an adequate minimal dose of vitamin D exists. Further analyses demonstrated that the pooled RR for doses of 400 IU and greater[18,33–39,41] (RR=0.86, 95% CI=0.77–0.96) was slightly less favorable but remained significant, compared to the pooled RR for doses of 800 IU and greater[18,34–39] (RR=0.80, 95% CI=0.70–0.91). In contrast, a significant lower fall risk was not observed at a dose of 200 IU[38] (RR=1.31, 95% CI=0.76–2.28). A minimum vitamin D dose of 400 IU may be needed to achieve significant benefits in fall reduction.

In sensitivity analysis, the effect estimate was examined using only studies with similar baseline characteristics[18,35,37–39,41] (RR=0.81, 95% CI=0.69–0.94, comparable with the overall RR when all 10 studies were included).

There was no evidence of significant publication bias according to the Begg's and Egger's tests for the 10 studies included in the primary analysis.

In the included studies, adverse events were minimal or not reported. Hypercalcemia was reported to occur in 0% to 3% of participants in four studies[36,37,39,40] but was not addressed in the remaining studies. Hypercalciuria was not described in any of the included studies.

Post Hoc Analysis

A post hoc analysis was conducted including all studies that had falls as an outcome, regardless of whether the definition of falls was explicitly given in the article, to test the robustness of this inclusion criterion. The post hoc analysis included 17 studies (10 with and 7 without explicit fall definitions); a summary of the seven additional studies is provided in .

Table 1.  Study Characteristics of Randomized Controlled Trials Included in Meta-Analysis

Study Therapy, Dose, Frequency N Age, Mean Follow-Up, Months How Fall Assessed Baseline/Change in 25-Hydroxy-Vitamin D, ng/mL* Previous Fracture, %* Previous Fall, %* Population
Studies included in primary analysis
   Bischoff et al.[37] I: Cholecalciferol 400 IU bid, calcium carbonate 600 mg bid 62 85 3 O 12/14 56 24 Institutionalized older adults awaiting nursing home placement
C: Calcium carbonate 600 mg bid 60 84 3 O 12/−0.2 52 23
   Bischoff-Ferrari et al.[41] I: Cholecalciferol 700 IU daily, calcium citrate 500 mg daily 219 71 36 Q, P M 33, F 28 Healthy, ambulatory, community dwelling
C: Placebo 226 71 36 Q, P M 33, F 25
   Broe et al.[38] I: Ergocalciferol 200 IU daily 26 92 5 O 18 69 Institutionalized, very old
I: Ergocalciferol 400 IU daily 25 88 5 O 21 68
I: Ergocalciferol 600 IU daily 25 89 5 O 17 64
I: Ergocalciferol 800 IU daily 23 89 5 O 21 65
C: Placebo+63% took multivitamin 25 86 5 O 21 44
   Burleigh et al.[36] I: Cholecalciferol 800 IU daily, calcium carbonate 1200 mg daily 100 82 1 O 9/1 26 53 Hospitalized, ill
C: Calcium carbonate 1,200 mg daily 103 84 1 O 10/0 26 48
   Dukas et al.[40] I: Alfacalcidol 1 μg daily 192 75 9 D 30 5 Ambulatory, community dwelling
C: Placebo 186 75 9 D 28 13
   Flicker et al.[34] I: Ergocalciferol 10,000 IU per week then 1,000 IU daily+calcium carbonate 600 mg daily 313 84 24 D Median<16 27 Institutionalized with vitamin D level between 25 and 90 ng/mL
C: Calcium carbonate 200 mg daily 312 83 24 D Median<16 24
   Graafmans et al.[33] I: Cholecalciferol 400 IU daily 177 83 7 D, Q Institutionalized
C: Placebo 177 83 7 D, Q
   Pfeifer et al.[18] I: Cholecalciferol 400 IU bid, calcium carbonate 600 mg bid 70 75 2 Q 26/40 0 Ambulatory, community dwelling, vitamin D<50 ng/mL
C: Calcium carbonate 600 mg bid 67 75 2 Q 26/18 0
   Pfeifer et al.[35] I: Cholecalciferol 400 IU bid, calcium carbonate 500 mg bid 122 76 12 D 22 0 Ambulatory, community dwelling, vitamin D<50 ng/mL
C: Calcium carbonate 500 mg bid 120 77 12 D 22 0
   Prince et al.[39] I: Ergocalciferol 1,000 IU daily, calcium citrate 500 mg bid 151 77 12 Q 18 60 Community dwelling, recruited from emergency department or nursing home, vitamin D<24 ng/mL
C: Calcium citrate 500 mg bid 151 77 12 Q 18 58
Studies included in post hoc analysis
   Gallagher et al.[42] I: Calcitriol 0.25 μg bid+calcium 500–1,000 mg daily 123 72 36 Q 31 28 Healthy, community-dwelling women; primary outcome bone mineral density
I: Estrogen+medroxyprogesterone+calcium 500–1,000 mg daily 121 72 36 Q 31 17
I: Calcitriol 0.25 μg bid+estrogen+medroxyprogesterone+calcium 500–1,000 mg daily 122 71 36 Q 32 16
C: Placebo+calcium 500–1,000 mg daily 123 71 36 Q 32 14
   Grant et al.[43] I: Ergocalciferol 800 IU daily 1,343 77 60 Q 100 Community-dwelling, recruited from fracture clinic or orthopedic ward; primary outcome fracture
I: Ergocalciferol 800 IU+calcium 1,000 mg daily 1,306 77 60 Q 100
I: Calcium 1,000 mg daily 1,311 78 60 Q 100
C: Placebo 1,332 77 60 Q 100
   Harwood et al.[44] I: Cholecalciferol 800 IU daily+calcium 1,000 mg daily 29 83 12 Q 12/8 100 Community dwelling, recruited<7 days after hip fracture surgery
C: None 25 81 12 Q 11/−0.1 100
   Law et al.[45] I: Ergocalciferol 1,100 IU daily 1,762 85 10 D Institutionalized, recruited from residential care homes
C: None 1,955 85 10 D
   Latham et al.[46] I: Ergocalciferol 300,000 IU single oral dose (1,600 IU daily) 108 80 6 D 15/9 44 57 Frail, hospitalized in acute care or rehabilitation facilities
C: Placebo 114 79 6 D 19/0 43 55
   Porthouse et al.[47] I: Cholecalciferol 800 IU daily+calcium 1,000 mg daily+leaflet 1,321 77 36 Q 59 34 Community-dwelling women, ≥1 risk factors for fracture; primary outcome fracture
C: Leaflet 1,993 77 36 Q 58 34
   Trivedi et al.[48] I: Cholecalciferol 100,000 IU every 4 months (800 IU daily) 1,027 75 48 Q Community-dwelling British doctors; primary outcome fracture and mortality
C: Placebo 1,011 75 48 Q

* If reported.
Primary outcome was falls unless otherwise indicated.
I=intervention; C=control; O=observed; D=diary; P=postcard; Q=questionnaire; bid=twice a day; M=male; F=female; IU=international units.

In contrast to studies that were included in the primary analysis, a few studies included participants at high risk for falls, such as individuals who were frail[46] or who had history of fracture.[43,44] Vitamin D interventions were similar except for one study that used calcitriol, estrogen, and medroxyprogesterone in the treatment arms.[42] The vitamin D regimen varied in these studies from 800 to 1,100 IU daily to 100,000 IU every 4 months. The studies tended to be of longer duration (6–60 months) than those in the primary analysis. Falls were not a primary outcome in all studies.[42,43,47,48]

These studies generally had good methodological quality, similar to those in the primary analysis (), with a few exceptions. In three studies, placebo was not given, so neither participant nor investigator was blinded.[44,45,47] One study[47] may have been subject to attrition bias because reasons for losses to follow-up were not completely described. In this study, falls were self-reported at 6-month intervals, which is not as reliable as questionnaires given every 6 weeks.[39] One study had involvement with industry sponsors,[47] and one study did not provide a flow diagram describing why patients were excluded.[48]

Table 2.  Qualitative Analysis of Included Studies*

Study Adequate Sequence Generation Described Allocation Concealment Described Assessor Blinding Incomplete Outcome Data Addressed Free from Other Bias Eligibility Criteria Defined Excluded Patients Described (n and Reason) Rates of Follow-Up Similar Describe Reasons for Loss to Follow-Up Describe All Therapies Clearly Prospective Sample Size Justification Similar Baseline Characteristics Between Groups
Studies included in primary analysis
   Bischoff et al.[37] Y Y Y Y Y Y Y Y Y Y Y Y
   Bischoff-Ferrari et al.[41] U Y Y U Y Y Y U Y Y U Y
   Broe et al.[38] Y Y Y Y Y Y Y Y Y Y U Y
   Burleigh et al.[36] Y Y Y Y Y Y Y Y Y Y Y U
   Dukas et al.[40] Y Y Y Y Y Y Y Y Y Y Y N
   Flicker et al.[34] Y Y Y Y Y Y Y Y Y Y Y N
   Graafmans et al.[33] U U U U Y Y U U N Y U U
   Pfeifer et al.[18] U U Y Y U Y Y Y Y Y Y Y
   Pfeifer et al.[35] U U Y Y U Y Y U U Y Y Y
   Prince et al.[39] Y Y Y Y Y Y Y Y Y Y Y Y
Studies included in post hoc analysis
   Grant et al.[43] Y Y Y Y Y Y Y Y Y Y Y Y
   Gallagher et al.[42] Y Y Y Y Y Y Y Y Y Y Y Y
   Harwood et al.[44] Y Y N Y Y Y Y Y Y Y N Y
   Law et al.[45] Y Y N Y Y Y Y Y Y Y Y U
   Latham et al.[46] Y Y Y Y Y Y Y Y Y Y Y Y
   Porthouse et al.[47] Y U N Y U Y Y Y N U Y Y
   Trivedi et al.[48] U U Y Y Y Y N U N Y N Y

* Y=yes, N=no, U=unclear.

Of 17 studies, which included 18,068 individuals, the RR of falling in the treatment compared with the control group was 0.92 (95% CI=0.87–0.98) with an I2 of 36% (P=.07) (Figure 2). For one study,[47] only the RR estimate for falling was provided.

Given that post hoc studies had quality measures similar to those of the studies in primary analysis, aside from the exceptions listed above, exploratory subgroup analyses were performed (). Significant intergroup differences were found favoring adjunctive calcium therapy[18,34–37,39,41–44,47] versus none[33,38,40,43,45,46,48] (P=.001), resulting in a 14% lower fall risk and heterogeneity that completely resolved (I2=0%) in stratified analysis. In post hoc analyses, treatment with cholecalciferol[18,33,35–37,41,44,47,48] (RR=0.86, 95% CI=0.77–0.96) was more favorable than ergocalciferol[34,38,39,43,45,46] (RR=0.99, 95% CI=0.92–1.06), but did not reach statistical significance (intergroup P=.08).

Table 3.  Subgroup Analyses for Vitamin D Treatment and Prevention of Falls

Subgroup Studies, n Participants, n Relative Risk of Falling (95% Confidence Interval) I 2, % P Value*
Treatment Control
Studies included in primary analysis
   Adjunctive calcium therapy
      Yes[18,34–37,39,41] 7 1,037 1,039 0.83 (0.75–0.92) 22 .51
      No[33,38,40] 3 468 388 0.94 (0.77–1.15) 0
   Type of vitamin D
      Ergocalciferol[34,38,39] 3 512 488 0.89 (0.80–1.00) 0 Ref
      Cholecalciferol[18,33,35–37,41] 6 750 753 0.80 (0.69–0.93) 24 .34
      Alfacalcidol[40] 1 192 186 0.84 (0.58–1.22) .80
   Participant dwelling and age§
      Community-dwelling, aged <80[18,35,39–41] 5 754 750 0.79 (0.69–0.92) 29 .20
      Hospitalized or institutionalized, aged ≥80[33,34,36–38] 5 751 677 0.90 (0.80–1.01) 0
   History of fracture or fall (in majority of participants) if reported
      Yes[36–39] 4 412 339 0.84 (0.72–0.98) 0 .68
      No[18,34,35,40] 4 697 685 0.77 (0.62–0.97) 59
   Duration of vitamin D treatment
      ≤6 months[18,36–38] 4 331 255 0.79 (0.62–1.01) 0 .89
      >6 months[33–35,39–41] 6 1,174 1,172 0.86 (0.78–0.94) 21
   Dose of oral vitamin D, IU
      <800[33,38,41] 3 472 478 1.01 (0.85–1.20) 21 .06
      ≥800[18,34–39] 7 841 838 0.80 (0.70–0.91) 31
Studies included in post hoc analysis
   Adjunctive calcium therapy
      Yes[18,34–37,39,41–44,47] 11 3,938 4,622 0.86 (0.81–0.92) 0 .001
      No[33,38,40,43,45,46,48] 7 4,708 4,800 1.01 (0.96–1.07) 0
   Type of vitamin D
      Ergocalciferol[34,38,39,43,45,46] 6 5,082 5,200 0.99 (0.92–1.06) 23 Ref
      Cholecalciferol[18,33,35–37,41,44,47,48] 9 3,127 3,792 0.86 (0.77–0.96) 28 .08
      Alfacalcidol or calcitriol[40,42] 2 437 430 0.86 (0.75–1.00) 0 .16
   Participant dwelling and age§
      Community-dwelling, aged <80[18,35,39–44,47,48] 10 6,025 6,676 0.88 (0.81–0.96) 34 .10
      Hospitalized or institutionalized, aged ≥80[33,34,36–38,45,46] 7 2,621 2,746 1.00 (0.94–1.06) 0
   History of fracture or fall (in majority of participants) if reported
      Yes[36–39,43,44,46,47] 8 3,198 3,131 0.96 (0.88–1.04) 0% .11
      No[18,34,35,40,42] 5 942 929 0.82 (0.71–0.94) 46%
   Duration of vitamin D treatment
      ≤6 months[18,36–38,46] 5 439 369 0.93 (0.73–1.17) 37 .41
      >6 months[33–35,39–45,47,48] 12 8,207 9,053 0.92 (0.86–0.98) 42
   Dose of oral vitamin D, IU
      <800[33,38,41] 3 472 478 1.01 (0.85–1.20) 21 .35
      ≥800[18,34–39,43–48] 13 7,737 8.589 0.92 (0.84–1.00) 50

* P value indicates difference between subgroups.
Calcium therapy given in intervention and control arms, except for three studies[41,44,47] in which given only in intervention arm. For studies with multiple arms,[43] vitamin D alone arm was compared with placebo arm and vitamin D with calcium arm was compared with calcium arm.
Compared with ergocalciferol.
§ All studies in community-dwelling individuals had a mean age less than 80 except for one in post hoc analysis;[44] all studies in hospitalized or institutionalized individuals had mean age of 80 and older except for one in post hoc analysis.[46]

Discussion

This systematic review and meta-analysis demonstrates that there is a protective effect of vitamin D supplementation on fall prevention in community-dwelling and institutionalized older adults. An overall RR of 0.86 (95% CI=0.79–0.93) suggested a 14% lower risk of falls. The effect of vitamin D on fall reduction was significant in several subgroups of individuals: community-dwelling participants with a mean age younger than 80, adjunctive calcium therapy, no history of fracture or fall, duration longer than 6 months, dose of 800 IU or greater, and cholecalciferol therapy, although no evidence was found of a linear association between higher doses of vitamin D or longer duration of vitamin D therapy and treatment effect. When trials without explicit fall definition were included in the post hoc analysis, the overall RR was smaller but remained significant (0.92, 95% CI=0.87–0.98), although the heterogeneity was substantial. In the post hoc analysis, significant intergroup differences favoring adjunctive calcium therapy were found with heterogeneity that completely resolved.

The inference from this work is similar to that of an earlier meta-analysis[24] reporting a corrected pooled odds ratio of 0.78 (95% CI=0.64–0.92) in five studies and 13% lower odds of falling in sensitivity analysis of 10 studies. The magnitude of the effect reported here is only slightly less in comparison. Because the fall outcome was common in the current study, it is likely that the odds ratio would overestimate the effect estimate, and it was decided to analyze the data using relative risks.

There have been two smaller meta-analyses that have shown statistically significantly lower fall risk with vitamin D treatment. A meta-analysis of two trials[23] showed an even more protective effect of active vitamin D (calcitriol, alfacalcidol), with a 34% lower odds of falling in vitamin D–treated groups than in controls. Another meta-analysis of 11 studies[22] showed an overall protective effect of 8%, but this was mostly due to a statistically significantly lower risk of falling in users of active vitamin D (RR=0.79, 95% CI=064–0.96) than in those taking native vitamin D (RR=0.94, 95% CI=0.87–1.01). In comparison, it was found that active vitamin D may have greater benefits than ergocalciferol but benefits similar to those of cholecalciferol.

A recent meta-analysis of eight randomized controlled studies[25] found an overall 13% lower risk (RR=0.87, 95% CI=0.77–0.99) with significant heterogeneity (Q test: P=.05). A similar overall risk reduction of 14% in 10 randomized controlled trials is reported in the current study but with a relatively more precise estimate (RR=0.86, 95% CI=0.79–0.93) and minimal heterogeneity (I2=7%; P=0.38). Greater benefit with higher-dose vitamin D was similarly found when studies were dichotomized as high or low dose, although in the post hoc analysis, the difference became less significant. In the meta-regression analysis containing fewer assumptions, no evidence was found of a linear association between dosage or duration and treatment effect. In addition, studies investigating hospitalized patients were not excluded. Although hospitalized patients could have comorbid conditions that predispose them to falls, the rate of falling in controls was similar in hospitalized patients[36] (fall rate 44%) and community-dwelling and institutionalized persons[18,33–35,37–41] (fall rates ranging between 28% and 59%). Vitamin D treatment reduced the risk of falling by 18% (RR=0.82, 95% CI=0.59–1.16) in hospitalized patients,[36] within the range of RRs reported by other studies included in this meta-analysis. Furthermore, the rigorous quality assessment allowed more high-quality studies that had been dismissed to be identified. As a result, it was possible to perform subgroup analyses that other meta-analyses may have been underpowered to conduct or did not investigate.

There are limitations to this systematic review and meta-analysis. It was not possible to investigate whether the effect of treatment with vitamin D on fall prevention would also apply to populations that were not deficient in vitamin D at baseline because all included studies had populations with vitamin D levels that were 30 ng/mL or less. The ascertainment of falls varied between studies, which may have contributed to inaccuracies in outcomes reporting. Although how many falls subsequently led to fracture was not examined, the distinction between falls that are injurious and noninjurious may be important. Although self-reported adherence was high in most studies in which it was reported (86–98%), a more-objective marker might have been changes in endogenous vitamin D levels before and after treatment, although few studies explicitly provided this information,[18,36,37] and vitamin D changes were highly variable (0–40 ng/mL). Calibration of vitamin D across different assays and laboratories is problematic, with mean between laboratory variation ranging from 2.9 to 5.2 ng/mL;[49] without proper standardization, such information may not be reliable. Last, subgroup analyses are observational by nature and may be subject to confounding according to study-level characteristics; these exploratory analyses require confirmation in randomized clinical trials comparing benefits in particular patient populations or with particular treatment regimens.

The post hoc analysis of 17 studies resulted in six times as many individuals in the analysis, yet the treatment effect remained significant although smaller in magnitude, despite including studies without an explicit fall definition. It was decided to report this as a post hoc analysis for several reasons. First, it was important to retain the initial eligibility criteria formulated in the protocol. Second, the addition of studies without an adequate fall definition increased I 2 considerably, suggesting that these studies may be more heterogeneous. Third, falls can be described differently and result in inconsistencies if not explicitly defined. Last, the post hoc analysis might have underestimated the true effect of vitamin D by including some studies that assessed falls as a secondary outcome and that were not double-blinded, although given that the quality assessment suggested that studies in the post hoc analysis were otherwise similar to those in primary analysis, exploratory subgroup analysis was performed to better discern possible differences. A limitation of any pooled analysis is that results will be weighted more toward larger studies (with smaller standard errors), although in analyses with many participants, such as the current one, the relative influence of each study may be less.

This systematic review and meta-analysis has provided a comprehensive update of prior reviews and identified two additional randomized controlled trials[36,47] not previously examined. A comprehensive quality assessment of all studies included in the meta-analysis has also been provided, allowing possible sources of bias to be identified. Significant intergroup differences favoring participants given vitamin D with calcium over those given vitamin D alone were found among the 17 studies in the post hoc analysis, which is a novel finding. Adequate calcium supplementation is necessary for optimal vitamin D action and may explain the greater benefits found with this regimen.

In summary, vitamin D supplementation is an effective strategy for reducing falls in older adults and should probably be incorporated into the clinical practice of providers caring for older adults, especially those at risk for falling. Although the effect appears to be modest, possibly because of inadequate dosing, vitamin D is inexpensive and well tolerated; a slight reduction in falls with vitamin D supplementation might lead to a significant decrease in the costs associated with fall morbidity and mortality.

Nevertheless, there are several outstanding questions. Research has not determined the optimum dose of vitamin D needed to reduce falls. More studies are needed to investigate the sustainability of beneficial effects conferred by vitamin D treatment on falls. Also, the utility of vitamin D treatment in patients who are not deficient in vitamin D is not clear. Studies that are better powered to detect differences in fall reductions according to subgroups may provide further insight into selected populations that could most benefit from vitamin D. Last, there is a paucity of studies evaluating the effects of vitamin D on fall prevention in hospitalized patients. In the future, investigations on these topics may facilitate the development of effective clinical guidelines regarding the administration of vitamin D for the prevention of falls in older adults.

References

  1. Campbell AJ, Borrie MJ, Spears GF et al. Circumstances and consequences of falls experienced by a community population 70 years and over during a prospective study. Age Ageing 1990;19:136–141.

  2. Blake AJ, Morgan K, Bendall MJ et al. Falls by elderly people at home: Prevalence and associated factors. Age Ageing 1988;17:365–372.

  3. Prudham D, Evans JG. Factors associated with falls in the elderly: A community study. Age Ageing 1981;10:141–146.

  4. Tinetti ME, Williams CS. The effect of falls and fall injuries on functioning in community-dwelling older persons. J Gerontol A Biol Sci Med Sci 1998;53A: M112–M119.

  5. Gillespie LD, Gillespie WJ, Robertson MC et al. Interventions for preventing falls in elderly people. Cochrane Database Syst Rev 2003; 4: CD000340.

  6. Self-reported falls and fall-related injuries among persons _65 yearsFUnited States, 2006. MMWR Morb Mortal Wkly Rep 2008;57:225–229.

  7. Vellas BJ, Wayne SJ, Romer LJ et al. Fear of falling and restriction of mobility in elderly fallers. Age Ageing 1997;26:189–193.

  8. Rizzo JA, Friedkin R, Williams CS et al. Health care utilization and costs in a Medicare population by fall status. Med Care 1998;36:1174–1188.

  9. Stevens JA, Corso PS, Finkelstein EA et al. The costs of fatal and non-fatal falls among older adults. Inj Prev 2006;12:290–295.

  10. Ganz DA, Bao Y, Shekelle PG et al. Will my patient fall? JAMA 2007;297:77– 86.

  11. Tinnetti ME. Clinical practice: Preventing falls in elderly persons. N Engl J Med 2003;348:42–49.

  12. Stein MS,Wark JD, Scherer SC et al. Falls relate to vitamin D and parathyroid hormone in an Australian nursing home and hostel. J Am Geriatr Soc 1999;47:1195–1201.

  13. Young A, Brenton P, Edwards RHT. Analysis of muscle weakness in osteomalacia. Clin Sci Mol Med 1978;54:31P.

  14. Holick MF. The role of vitamin D for bone health and fracture prevention. Curr Osteopor Rep 2006;4:96–102.

  15. Holick MF. Vitamin D deficiency. N Engl J Med 2007;357:266–281.

  16. Sorensen OH, Lund B, Saltin B et al. Myopathy in bone loss of ageing: Improvement by treatment with 1 alpha-hydroxycholecalciferol and calcium. Clin Sci (London) 1979;56:157–161.

  17. Ceglia L. Vitamin D and skeletal muscle tissue and function. Mol Aspects Med 2008;29:407–414.

  18. Pfeifer M, Begerow B, Minne HWet al. Effects of a short-term vitamin D and calcium supplementation on body sway and secondary hyperparathyroidism in elderly women. J Bone Miner Res 2000;15:1113–1118.

  19. Latham NK, Anderson CS, Reid IR. Effects of vitamin D supplementation on strength, physical performance, and falls in older persons: A systematic review. J Am Geriatr Soc 2003;51:1219–1226.

  20. Oliver D, Connelly JB, Victor CR et al. Strategies to prevent falls and fractures in hospitals and care homes and effect of cognitive impairment: Systematic review and meta-analyses. BMJ 2007;334:82.

  21. Jackson C, Gaugris S, Sen SS et al. The effect of cholecalciferol (vitamin D3) on the risk of fall and fracture: A meta-analysis. QJM 2007;100:185–192.

  22. Richy F, Dukas L, Schacht E. Differential effects of D-hormone analogs and native vitamin D on the risk of falls: A comparative meta-analysis. Calcif Tissue Int 2008;82:102–107.

  23. O'Donnell S, Moher D, Thomas K et al. Systematic review of the benefits and harms of calcitriol and alfacalcidol for fractures and falls. J Bone MinerMetab 2008;26:531–542.

  24. Bischoff-Ferrari HA, Dawson-Hughes B,Willett WC et al. Effect of vitamin D on falls: A meta-analysis. JAMA 2004;291:1999–2006.

  25. Bischoff-Ferrari HA, Dawson-Hughes B, Staehelin HB et al. Fall prevention with supplemental and active forms of vitamin D: Meta-analysis of randomised controlled trials. BMJ 2009;339:b3692.

  26. Robinson KA, Dickerson K. Development of a highly sensitive search strategy for the retrieval of reports of controlled trials using PubMed. Int J Epidemiol 2002;31:150–153.

  27. Buchner DM, Cress ME, Wagner EH et al. The Seattle FICSIT/MoveIt study: The effect of exercise on gait and balance in older adults. J Am Geriatr Soc 1993;41:321–325.

  28. Romagnoli E, Mascia ML, Cipriani C et al. Short and long-term variations in serum calciotropic hormones after a single very large dose of ergocalciferol (vitamin D2) or cholecalciferol (vitamin D3) in the elderly. J Clin Endocrinol Metab 2008;93:3015–3020.

  29. Higgins J, Altman DG. Assessing risk of bias in included studies. In: Higgins J, Green S, editors. Cochrane Handbook for Systematic Reviews of Interventions. Oxford, UK: The Cochrane Collaboration, 2008, pp 187–242.

  30. Higgins JP, Thompson SG, Deeks JJ et al. Measuring inconsistency in metaanalyses. BMJ 2003;327:557–560.

  31. Begg CB, Mazumdar M. Operating characteristics of a rank correlation test for publication bias. Biometrics 1994;50:1088–1101.

  32. Egger M, Smith GD, Altman DG. Systematic Reviews in Health Care: Meta-Analysis in Context. London: BMJ Publishing Group, 2001.

  33. Graafmans WC, Ooms ME, Hofstee HM et al. Falls in the elderly: A prospective study of risk factors and risk profiles. Am J Epidemiol 1996;143: 1129–1136.

  34. Flicker L, MacInnis RJ, Stein MS et al. Should older people in residential care receive vitamin D to prevent falls? Results of a randomized trial. J Am Geriatr Soc 2005;53:1881–1888.

  35. Pfeifer M, Begerow B, Minne HWet al. Effects of a long-term vitamin D and calcium supplementation on falls and parameters of muscle function in community-dwelling older individuals. Osteoporos Int 2009;20:315–322.

  36. Burleigh E, McColl J, Potter J. Does vitamin D stop inpatients falling? A randomised controlled trial. Age Ageing 2007;36:507–513.

  37. Bischoff HA, Stähelin HB, Dick W et al. Effects of vitamin D and calcium supplementation on falls: A randomized controlled trial. J Bone Miner Res 2003;18:343–351.

  38. Broe KE, Chen TC,Weinberg J et al. A higher dose of vitamin d reduces the risk of falls in nursing home residents: A randomized, multiple-dose study. J Am Geriatr Soc 2007;55:234–239.

  39. Prince RL, Austin N, Devine A et al. Effects of ergocalciferol added to calcium on the risk of falls in elderly high-risk women. Arch Intern Med 2008;168:103–108.

  40. Dukas L, Bischoff HA, Lindpaintner LS et al. Alfacalcidol reduces the number of fallers in a community-dwelling elderly populationwith aminimum calcium intake of more than 500 mg daily. J Am Geriatr Soc 2004;52:230–236.

  41. Bischoff-Ferrari HA, Orav EJ, Dawson-Hughes B. Effect of cholecalciferol plus calcium on falling in ambulatory older men and women: A 3-year randomized controlled trial. Arch Intern Med 2006;166:424–430.

  42. Gallagher JC, Fowler SE,Detter JR et al. Combination treatmentwith estrogen and calcitriol in the prevention of age-related bone loss. J Clin Endocrinol Metab 2001;86:3618–3628.

  43. Grant AM, Avenell A, Campbell MK et al. Oral vitamin D3 and calcium for secondary prevention of low-trauma fractures in elderly people: A randomised placebo-controlled trial. Lancet 2005;365:1621–1628.

  44. Harwood RH, Sahota O, Gaynor K et al. The Nottingham Neck of Femur (NONOF) Study. A randomised, controlled comparison of different calcium and vitamin D supplementation regimens in elderly women after hip fracture: The Nottingham Neck of Femur (NONOF) Study. Age Ageing 2004;33: 45–51.

  45. Law M, Withers H, Morris J et al. Vitamin D supplementation and the prevention of fractures and falls: Results of a randomised trial in elderly people in residential accommodation. Age Ageing 2006;35:482–486.

  46. Latham NK, Anderson CS, Lee A et al. A randomized, controlled trial of quadriceps resistance exercise and vitamin D in frail older people: The Frailty Interventions Trial in Elderly Subjects (FITNESS). J Am Geriatr Soc 2003;51: 291–299.

  47. Porthouse J, Cockayne S, King C et al. Randomised controlled trial of calcium and supplementation with cholecalciferol (vitamin D3) for prevention of fractures in primary care. BMJ 2005;330:1003.

  48. Trivedi DP, Doll R, Khaw KT. Effect of four monthly oral vitamin D3 (cholecalciferol) supplementation on fractures and mortality in men and women living in the community: Randomised double blind controlled trial. BMJ 2003;326:469.

  49. Binkley N, Krueger D, Gemar D et al. Correlation among 25-hydroxy-vitamin D assays. J Clin Endocrinol Metab 2008;93:1804–1808.