Effects of Vitamin D and Calcium Supplementation on Falls: A Randomized Controlled Trial

Authors


  • The authors have no conflict of interest

Abstract

Specific receptors for vitamin D have been identified in human muscle tissue. Cross-sectional studies show that elderly persons with higher vitamin D serum levels have increased muscle strength and a lower number of falls. We hypothesized that vitamin D and calcium supplementation would improve musculoskeletal function and decrease falls. In a double-blind randomized controlled trial, we studied 122 elderly women (mean age, 85.3 years; range, 63–99 years) in long-stay geriatric care. Participants received 1200 mg calcium plus 800 IU cholecalciferol (Cal+D-group; n = 62) or 1200 mg calcium (Cal-group; n = 60) per day over a 12-week treatment period. The number of falls per person (0, 1, 2–5, 6–7, >7 falls) was compared between the treatment groups. In an intention to treat analysis, a Poisson regression model was used to compare falls after controlling for age, number of falls in a 6-week pretreatment period, and baseline 25-hydroxyvitamin D and 1,25-dihydroxyvitamin D serum concentrations. Among fallers in the treatment period, crude excessive fall rate (treatment − pretreatment falls) was compared between treatment groups. Change in musculoskeletal function (summed score of knee flexor and extensor strength, grip strength, and the timed up&go test) was measured as a secondary outcome. Among subjects in the Cal+D-group, there were significant increases in median serum 25-hydroxyvitamin D (+71%) and 1,25-dihydroxyvitamin D (+8%). Before treatment, mean observed number of falls per person per week was 0.059 in the Cal+D-group and 0.056 in the Cal-group. In the 12-week treatment period, mean number of falls per person per week was 0.034 in the Cal+D-group and 0.076 in the Cal-group. After adjustment, Cal+D-treatment accounted for a 49% reduction of falls (95% CI, 14–71%; p < 0.01) based on the fall categories stated above. Among fallers of the treatment period, the crude average number of excessive falls was significantly higher in the Cal-group (p = 0.045). Musculoskeletal function improved significantly in the Cal+D-group (p = 0.0094). A single intervention with vitamin D plus calcium over a 3-month period reduced the risk of falling by 49% compared with calcium alone. Over this short-term intervention, recurrent fallers seem to benefit most by the treatment. The impact of vitamin D on falls might be explained by the observed improvement in musculoskeletal function.

INTRODUCTION

HIP FRACTURES ARE the most serious and costly fractures among older persons.(1) About 90% of hip fractures involve falls. Fractures caused by falls occur in about 5% of elderly persons each year, 1–2% involving the hip.(2–4) Subjects with recurrent falls are the frailest and have the highest risk for sustaining a fracture because of an increased exposure.(5)

Chapuy et al.(6) and Dawson-Hughes et al.(7) have shown a significant reduction of nonvertebral fractures (32% and 58%, respectively) with long-term vitamin D and calcium supplementation (18 and 36 months) in institutionalized and community-dwelling elderly persons. Both groups of authors attributed the outcomes to the observed moderate increase in bone mineral density (BMD). However, an alternative explanation might be that these supplements affect factors directly related to fall risk. Epidemiologic, clinical, and laboratory evidence suggest a direct effect of vitamin D on muscle strength.(8–12) An association between low vitamin D levels and reversible myopathy has been documented in patients with osteomalacia(13–16) and uremia.(17–18)

Vitamin D, in addition to its effects on calcium homeostasis, binds to specific receptors on skeletal muscle for 1,25-dihydroxyvitamin D.(19–22) Sorensen et al.(23) observed an increase in the relative number and cross-sectional area of fast-twitch muscle fibers in biopsy specimens of elderly osteoporotic women who had been treated with 1-α-hydroxyvitamin D for 3 months.

We hypothesized that vitamin D and calcium supplementation would affect calcium homeostasis and increase muscle strength, which would reduce the risk of falling. If so, vitamin D would prevent fractures in two ways: by improving musculoskeletal function and by increasing calcium homeostasis.

We tested this in a randomized double-blind trial comparing calcium plus vitamin D to calcium supplementation alone on the mean number of falls and recurrent falls per person. Secondary endpoints were change in musculoskeletal function and biochemical markers of bone metabolism.

MATERIALS AND METHODS

Study population

We took advantage of the fact that elderly persons, who are not able to live independently and awaiting placement, are cared for in long-stay geriatric care units in Switzerland. These individuals are at increased risk of vitamin D deficiency,(1,8–10) physical frailty, muscle weakness,(7) and falling.(5, 6) Fortification of food with vitamin D is not required in Switzerland.

We also timed the study to be done during the winter months (November 1999 and March 2000) in an attempt to accentuate as much as possible the differences in serum vitamin D hormone concentrations (25-hydroxyvitamin D, 1,25-dihydroxyvitamin D) between the treatment groups.(9)

Eligibility criteria were age 60 or older and the ability to walk 3 m with or without a walking aid. Exclusion criteria were primary hyperparathyroidism, hypocalcemia, hypercalciuria, renal insufficiency (creatinine > 117 μM), and fracture or stroke within the last 3 months. Also excluded were those who had received any treatment with hormone replacement therapy, calcitonin, fluoride, or bisphosphonates during the previous 24 months. Previous vitamin D supplementation was not an exclusion criterion. None of the women were having physical training at study entry, and no attempt was made to alter subject's diet, strength, or activity during the study. The study protocol was approved by the hospital ethics committee (University of Basel). Written informed consent was obtained from all subjects or from legal guardians.

Characteristics of the study population are given in Table 1.

Table Table 1.. Base-Line Characteristics Values in the Calcium Group (Cal-Group) and Calcium-Vitamin D Group (Cal + D-Group)
original image

Study design and endpoints

The double-blind randomized controlled trial had a 6-week pretreatment period and a 12-week treatment period. Number of falls per person during the treatment period was the primary outcome of the study. Secondary outcome measures were musculoskeletal function and bone remodeling (described below). At the start of the treatment period, subjects were randomized in groups of 4. The randomization was performed by an independent statistician.

Falls were recorded by the nurses on the inpatient units who had received training in the use of the fall protocol (date, time, circumstances, injuries). Falls were defined as “unintentionally coming to rest on the ground, floor, or other lower level.” Coming to rest against furniture or a wall was not counted as a fall.(24) Nurses completed the fall protocol if they observed or received a report of a fall.

Subjects were not followed into the nursing home because we expected varying numbers of falls because of the changed environment in the different nursing homes. In addition, the patient-nurse ratio in nursing homes was much lower compared with long-stay geriatric care in the acute hospital setting, which we expected to affect the report of falls.

After the 6-week pretreatment period, subjects were randomized to the two treatment protocols. Subjects randomly assigned to the vitamin D plus calcium group (Cal+D-group) received two tablets containing 600 mg of calcium carbonate and 400 IU of cholecalciferol per tablet. Subjects randomly assigned to the calcium group (Cal-group) received two tablets containing 600 mg of calcium carbonate per tablet. Tablets in both groups had an identical appearance and were administered twice per day with breakfast and dinner and swallowed in the presence of the study nurse to ensure compliance.

Patients, nurses, and all investigators were blinded to the treatment assignment throughout the study. Adverse reactions were reported to the physician in charge for the patient and to one research physician. The treatment allocation was kept in sealed envelopes to which the physician in charge of the patient had access to only in case of an emergency.

Patient descriptives

Comorbid conditions were evaluated by the Charlson Comorbidity Index.(25) Major comorbid conditions and previous nonvertebral fractures are stated in Table 2. Cognitive impairments were assessed by a Folstein Mini Mental Status.(26)

Table Table 2.. Major Baseline Comorbid Conditions
original image

Number of drugs was recorded as a score: score 1 represents less than four medications per day; score 2 represents four or more medications a day(27); benzodiazepine use and diuretic use is given separately.(28) A dietitian evaluated the mean calcium content of food consumed from each meal and drink during the baseline week. Overall diet was the same for all participants.

Strength and functional assessment

Overall musculoskeletal function was assessed by a summed score including the timed up&go test (TUG test), knee flexor strength, knee extensor strength, and grip strength. Assessments took place at the beginning of treatment period and after 12 weeks. Strength measures of hemiplegic subjects were performed on the normal extremity.

The TUG test, reported by Podsiadlo and Richardson, is a measure of functional mobility including muscle strength, gait speed, and balance.(29) The TUG test is measured is seconds, and the smallest value of three trials was documented.

Muscle strength was measured as knee extensor and flexor strength in kiloponds, with a validated, hand-held isometric method (model DPPH; Chatillon Inc., Greensboro, NC, USA). A nonelastic band was connected to a pull gauge with a continuos scale in kiloponds (1 kilopond = 10 Newton).(30) Grip strength was measured in the dominant hand with a Martin vigorimeter (Martin; Elmed, Addison, IL, USA), and the force was measured as bars (1 bar = 105 pascal = 1 Newton/m2).(31)

For strength measures, the best values out of three measurements recorded on each leg or hand were averaged for both hands and both legs. Strength and functional measures were not recorded if the person reported pain during measurements or was noncompliant. Encouragement in all measurements was standardized and performed by the same person (one physiotherapist trained in the treatment of elderly persons).

Methods of single musculoskeletal tests were validated in a random sample of 23 subjects and generally showed excellent reproducibility:

  • 1.knee flexor strength (hand-held gauge; isometric): 0.68 intrarater, 0.96 inter-rater reliability
  • 2.knee extensor strength (hand-held gauge; isometric): 0.84 intrarater, 0.89 inter-rater reliability
  • 3.grip strength (Martin Vigorimeter; dynamic): 0.94 intrarater, 0.94 inter-rater reliability
  • 4.TUG test: 0.88 intrarater; 0.95 inter-rater reliability

Laboratory studies

Fasting early morning venous blood was collected from resting subjects for measurement of serum calcium, phosphate, and albumin and total alkaline phosphatase. These were measured using automated serum chemical analysis. Blood calcium concentration was corrected for albumin with the following formula: corrected calcium = (measured calcium [mM] − 0.012) × (albumin [g/liter] − 39.5).(32)

Serum 25-hydroxyvitamin D, 1,25-dihydroxyvitamin D, and intact parathyroid hormone (iPTH) concentrations were measured by radioimmunoassay (Nichols, San Juan Capistrano, CA, USA): intra-assay variation was 5.1%, 5.0%, and, 1.8, respectively, and interassay variation was 7.9%, 10.8%, and 5.6%, respectively. Serum osteocalcin was measured by a radioimmunoassay (Nichols): intra-assay variation was 3.2% and interassay variation was 5.1%. Urinary N-telopeptide (NTX) and deoxypyridinolin (DPD) were measured by an ELISA (Osteomark, Seattle, WA, USA) from 2-h morning spot urine samples: intra-assay variation was 3.2% and 4.3%, respectively, and interassay variation was 3.0% and 3.1%, respectively. All samples were frozen at −80°C and analyzed by the same person who was blinded to treatment group in one batch.

Statistical analysis

All comparisons between treatment groups were based on an intention-to-treat analysis; each patient was analyzed as part of the group to which they were randomized.

For group comparison at baseline we used two-sample t-tests, Wilcoxon rank sum tests, χ2, and Fisher's exact tests.

Univariate tests for follow-up data (Wilcoxon rank sum tests) were used to evaluate median difference from baseline for laboratory investigations. Test results for all four musculoskeletal tests were transformed into z-scores and summed for an overall musculoskeletal function measure according to O'Brian.(33) This assumes that the measures are all indicators of the underlying construct of muscle strength.

A crude comparison of the mean number of excessive falls (number of falls during the treatment − falls during the pretreatment period) among subjects who fell during the treatment period was carried out by a two-sample t-test.

The main adjusted analysis used the Poisson regression to compare the number of falls in the two treatment groups. The predefined response categories of the primary outcome (0, 1, 2–5, 6–7, >7 falls) were grouped as shown in the Table 3 to reduce the possibility that one person contributed a disproportional number of falls and to obtain more conservative results. In addition to treatment, the model included baseline covariates that reached a significance level of p < 0.1 when compared between the two arms. In this way we included the most likely confounders. Adjustment for individual observation time with treatment was forced into the model because of the expected moderately high drop-out rate. We also show the raw data result without conservative inference based on the absolute mean number of falls.

Table Table 3.. Absolute Distribution of Falls
original image

The planned study population was 120 subjects based on the power calculation and an expected 30% drop-out rate because of discharge to nursing homes, but also because of death, hip fracture, intercurrent illness, and withdrawal of consent. The number of subjects was chosen to have sufficient power to detect a difference in the number of falls. Based on the Poisson distribution, a reduction of 43% can be detected. A standard method for comparing Poisson distributions shows that a reduction of 43% or more in falls provides 90% power to detect a significant difference between the groups as long as the groups each contain 44 subjects.(34)

Intrarater and inter-rater reliability for the four musculoskeletal tests were assessed on data from a random sample of 23 patients using Spearman's rank correlation coefficient.

A 5% significance level was maintained throughout these analysis, and all tests were two-sided.

Data were analyzed by SPSS computer program (Version 9.0) and SAS computer program (Version 7.0; SAS Institute Inc., Cary, NC, USA).

RESULTS

Baseline

From a population of 130 eligible elderly institutionalized women in two hospitals with long-stay geriatric care units (Geriatric University Hospital and Felix Platter Spital Basel, Basel, Switzerland), 124 agreed to participate in the study. Of these 124 subjects, 122 were still hospitalized after a 6-week pretreatment period (mean age 85.3 ± 6.6 years). Of the 122 subjects randomized at baseline of the treatment period, 62 subjects were in the Cal+D-group and 60 subjects were in the Cal-group. All subjects received standard intervention as allocated.

Patient characteristics, number of falls (Tables 1 and 2), laboratory measures (Table 4), and musculoskeletal function (Table 5) did not differ between groups at baseline. Mean dietary calcium intake was low in all subjects, 600–700 mg/day.

Table Table 4.. Serum Biochemical Values in the Two Treatment Groups at Baseline and After 3 Months Follow-Up
original image
Table Table 5.. Musculoskeletal Function at Baseline and Percentage Difference From Baseline After 3 Months of Follow-Up in the Two Treatment Groups
original image

Vitamin D deficiency at study entry was highly prevalent by any definition(35–37): 50% of women had 25-hydroxyvitamin D serum concentrations below 12 ng/ml (30 nmol/liter), 90% below 31 ng/ml (78 nmol/liter), and 95% below 40 ng/ml (100 nmol/liter). Seventeen percent of subjects met the criteria for secondary hyperparathyroidism (2°HPT), namely increased iPTH levels (iPTH ≥ 55 pg/ml) and normal serum calcium levels.

There was a significant difference in baseline 25-hydroxyvitamin D between subjects taking vitamin D supplements (n = 23) at study entry and those without (median in nanograms per milliliter plus interquartile range [IQR]: with vitamin D = 29.0 [15.8–38.0]; without = 11.1 [9.0–15.7]; p < 0.0001). Also, baseline serum iPTH differed significantly between these groups (median in picograms per milliliter [IQR]: with vitamin D = 20.9 [14.2–32.6]; without = 36.5 [26.3–48.5]; p < 0.0001).

Follow-up

Falls

Of 42 falls in pretreatment period, 22 (15 persons) occurred in the Cal+D-group and 20 (14 persons) in the Cal-group. Of 80 falls in treatment period, 25 (14 persons) were in Cal+D-group and 55 (18 persons) in Cal-group (Table 3).

In the adjusted analysis, including all subjects, the following predictors entered the screening multiple regression model: number of falls and fallers in the 6-week pretreatment period, age, height, weight, body mass index (BMI), use of a walking aid (no walking aid, cane, walker), previous vitamin D use (yes, no), Charlson Comorbidity index, number of medications at baseline (<4; ≥4), length of stay in the institution at study entry, length of observation during treatment period, muscle strength and function (baseline grip strength, knee extensor and flexor strength, TUG test), and baseline 25-hydroxyvitamin D, 1,25-dihydroxyvitamin D, iPTH, and albumin serum concentrations.

In the final Poisson regression vitamin D plus calcium treatment accounted for a 49% reduction in falls (estimate: −0.68; 95% CI, 14–71%; p = 0.01), after adjustment for age, number of falls in the pretreatment period, being a faller in the pretreatment period, baseline1,25-dihydroxyvitamin D and 25-hydroxyvitamin D, and observation time during treatment. Significant predictors other than treatment are listed in order of importance: being a faller in the pretreatment period, number of falls in the pretreatment period, and age.

Without conservative statistical interference (using the absolute number of falls as the primary outcome), vitamin D plus calcium treatment accounted for a 62% reduction in falls (estimate: −0.96; 95% CI, 37–77%; p < 0.0002) after adjustment for the same covariates. In addition to the above listed covariates, low baseline 1,25-dihydroxyvitamin D was a significant predictor of fall risk. Figure 1 illustrates adjusted Poisson probabilities of having zero, one, or multiple falls for subjects in both treatment groups.

Figure FIG. 1..

Adjusted probabilities and SEs for having zero, one, or multiple falls for subjects in both treatment groups. SEs are calculated by taking 1 SD above and below the mean rate of falls and calculating the resulting Poisson probabilities. Adjustments have been performed for length of observation in the treatment period, previous falls in pretreatment period, a person who fell in the pretreatment period, age, and baseline 25-hydroxyvitamin D and 1,25-dihydroxyvitamin D.

The crude t-test comparing the mean number of excessive falls among fallers of the treatment period was lower in the Cal+D-group (p = 0.045), suggesting a decrease in recurrent falls with the Cal+D supplementation. The number of fallers did not differ between the treatment arms (RR = 0.7; 95% CI, 0.3–1.5).

Musculoskeletal tests

In 62 women with complete data set for all musculoskeletal tests at baseline and follow-up, musculoskeletal function improved significantly in the Cal+D-group compared with the Cal-group by comparing changes over time (p = 0.0094; Table 5).

Biochemical changes

Among subjects in the Cal+D-group, there were marked increases in median serum 25-hydroxyvitamin D (+71%) and 1,25-dihydroxyvitamin D (+8%), and decreases in serum iPTH (−29%), alkaline phosphatase (−7%), urinary DPD (−30%), and urinary N-telopeptides (−26%). No significant differences between groups were found for serum calcium (corrected for albumin), serum phosphate, serum creatinine, and serum osteocalcin (see Table 4). Compared with baseline, significant decreases in serum osteocalcin (p < 0.0001) were found in both groups. We found a small (−4%) insignificant decline in 25-hydroxyvitamin D serum concentrations in the control group over the winter, which may be explained by the low baseline vitamin D status in most of the subjects measured in November.

Side effects

Two subject reported increased constipation in the Cal+D-group, which did not lead to discontinuation of treatment. None of the women developed hypercalcemia, as documented in the laboratory investigations after treatment. One person discontinued medication in the Ca-group independent of side effects.

Subjects lost to follow-up

Of the 122 women enrolled, 89 completed the study. Mean observation time of subjects during total study period (scheduled: 126 days) was not significantly different between treatment groups (113 ± 22.3 days in the Cal+D-group; 119.2 ± 16.6 days in the Cal-group; p = 0.25).

The causes of dropout during the study were similar in both treatment groups (31% in Cal+D-group and 25% in Cal-group). The primary reason was being discharged to a nursing home (11 subjects of Cal-group and 12 subjects of Cal+D-group). Other causes were balanced between groups (five deaths, two intercurrent illnesses, and three hip fractures).

Of 89 subjects who completed the study, 27 had missing or incomplete musculoskeletal assessment at baseline or follow-up (13 subjects refused to join the measurements, 4 reported pain during measurements, and 10 had decreased compliance). There were no significant differences in baseline strength and functional measures or other baseline characteristics between subjects with complete and subjects with incomplete or missing data at follow-up.

DISCUSSION

Falls are a hallmark of the frail elderly.(38) Functional decline is often heralded by the onset of gait instability, falls, and the resulting trauma.(39) This seemingly inseparable relationship of falls to worsening health status has led to pessimism on the part of physicians and patients alike when faced with falling, especially recurrent falling. The present study adds to a growing body of literature,(40–44) which should encourage the search for treatable causes, by showing that a simple, inexpensive, and well-tolerated intervention, such as vitamin D and calcium, may help to reduce the burden of falling in the elderly. To our knowledge, this is the first randomized controlled trial studying the effects of vitamin D and calcium on falls.

The results of the study indicate that vitamin D and calcium supplementation reduced the number of falls per person by 49%, improved musculoskeletal function, increased vitamin D status (both 25-hydroxyvitamin D and 1,25-dihydroxyvitamin D), and decreased parathyroid hormone secretion and bone resorption (as assessed by urinary NTX and DPD) within 3 months of treatment in elderly women with vitamin D deficiency. We did not see a significant difference in the number of fallers between the treatment groups, which might be caused by a lack of sample size, because the study was not powered to see a difference in fallers. However, the direction of the effect was in favor of the Cal+D-treatment with a 30% lower risk to be a faller.

Falls prevention by risk factor reduction is a major concern in the care of the elderly. A number of approaches have been shown. Multifactorial approaches, such as medical and occupational therapy assessment or adjustment in medications, behavioral instructions, and exercise programs, as demonstrated in the PROFET trials(40) and FICSIT trials,(41) as well as single intervention strategies, such as Tai Chi balance training(42) and exercise,(43) reduced falls by 25–50%. However, they are expensive, take considerable time to implement, and are strongly dependent on the compliance of subjects.

This is the first study that documents a reduction in falls among frail elderly women with a single medication and within 3 months of treatment. Previous studies which have examined the relationship between vitamin D and muscle strength in elderly persons usually demonstrated a beneficial effect on muscle strength or balance. In a prospective study, Soerensen et al.(23) documented an increase in cross-sectional area and number of fast-twitch muscle fibers with 1-α-hydroxyvitamin D treatment (1 μg/day) within 3 months in osteoporotic elderly women. This was accompanied by a clinical improvement of muscle function (time to dress). In ambulatory elderly women, Pfeifer et al.(44) found a decrease in body sway after a 2-month treatment with vitamin D (800 IU/day) and calcium (1200 mg/day), as well as a reduction of falls within a 1-year uncontrolled follow-up. Cross-sectional studies described low 25-hydroxyvitamin D levels in elderly fallers(45) and increased muscle strength in ambulatory elderly with higher 1,25-dihydroxyvitamin D levels.(46)

On the other hand, in one controlled study in ambulatory elderly patients, 1,25-dihydroxyvitamin D administration (0.5 μg/day) did not improve leg extension strength.(47) However, 1,25-dihydroxyvitamin D serum concentrations were not as low as in our subjects and did not rise significantly in the treatment group (baseline: 35.6 pg/ml; after 6 months: 39.7 pg/ml). Alternatively, cholecalciferol may be the more effective agent for muscle strength improvement.

A possible explanation for the effect of vitamin D on muscle strength is that vitamin D acts directly on muscle tissue in addition to the well-described effect on calcium homeostasis.(19–22) The effects of vitamin D on muscle tissue might be mediated by de novo protein synthesis, affecting muscle cell growth.(48) Because this effect on muscle tissue seems to result in clinical improvement after short-term intervention,(11–16,23) it is of major clinical interest for fall and fracture prevention in elderly people, especially if compared with the long-term intervention needed to enhance BMD.(1, 2) In fact, Glerup et al.(49) showed that in vitamin D-deficient subjects, severely impaired muscle function may be present even before biochemical signs of bone disease develop. Another support for a positive interaction between vitamin D and strength is given by Geusens et al.,(50) who found a major association of an allelic variant of the vitamin D receptor genotype with muscle strength in elderly nonobese women, suggesting a genetic predisposition of strength decline may be linked to the vitamin D receptor.

Apart from vitamin D, hyperparathyroidism is considered to play a potential role in both the peripheral neuropathy and proximal myopathy.(51) However, we found no evidence in our study that iPTH is related to musculoskeletal function or falls. This may be because of the small proportion of subjects with 2°HPT (17%). Alternatively, the observed changes in parathyroid status may not have been as large as those changes in vitamin D status.

Some limitations of the study require mention. One is the drop-out rate, which was expected and largely because of transfers to nursing homes during the course of the study. However, the adjusted analysis controlled for individual observation time. To address the issue of multiple falling, we controlled for the number of women who fell in the pretreatment period and number of falls in the pretreatment period. Furthermore, the response categories were collapsed to obtain more conservative results (the raw results showed a 62% reduction of number of falls per person instead of the 49% observed in the conservative approach). The model provided 90% power to detect a 43% difference in the mean number of falls per person comparing the groups each containing 44 subjects. Generalizability is an issue because the patients were white institutionalized elderly women with low vitamin D serum concentrations from Switzerland, where vitamin D fortification is not mandatory. Among institutionalized elderly in the United States, 25-hydroxyvitamin D serum concentrations have been reported to be higher than in Europe.(8) However, fracture prevention by vitamin D and calcium supplementation showed similar beneficial effects in vitamin D-deficient European institutionalized elderly(6) and in less deficient ambulatory elderly in the United States.(7)

A strength of the study is the consistency of all findings in terms of fall risk, musculoskeletal function, and bone metabolism, with significant benefits in the Cal+D-group compared with the Cal-group. The effect size of the reduction of number of falls by vitamin D and calcium supplementation is large (49%) and is supported by the observed significant improvement in musculoskeletal function. In addition, biological mechanisms that explain the relationship between vitamin D deficiency and muscle weakness seem plausible in the light of the discussed literature.(11–21,44–46,49,50)

Much emphasis has been placed on fractures as the most serious consequence of falling in elderly persons. Important studies have been made in the past decade in osteoporosis treatment with vitamin D and calcium, leading to effective fracture reduction when given chronically.(1, 2) The direct effects of vitamin D on musculoskeletal function may offer a way to reduce not only fractures, but the many distresses associated with recurrent falling. These findings suggest that vitamin D and calcium supplementation may assume even more prominence as a prevention strategy in older people.

In conclusion, our results show that vitamin D and calcium supplementation within a 3-month application reduces falls by 49% in frail elderly women with vitamin D deficiency. Recurrent fallers seem to benefit most by the Ca+D supplementation. The beneficial effect could be modulated by the observed improvement in musculoskeletal function. In addition, supplementation of vitamin D and calcium significantly improved bone metabolism, with a 29% decrease in serum iPTH, a 71% rise in 25-hydroxyvitamin D, and a 30% decrease in bone resorption, measured by urinary DPD. Vitamin D and calcium were superior to calcium supplementation alone in regard to fall prevention, musculoskeletal function, and bone metabolism.

Acknowledgements

The authors are indebted to John Orav (Harvard University, Boston, MA, USA) for his statistical advice and to Lyn Singer-Lindpaintner, MD (Dept. of Geriatrics, University Basel, Basel, Switzerland) and Matthew H Liang, MD, MPH (Division of Rheumatology, Immunology and Allergy, Brigham and Women's Hospital, Boston, MA, USA) for their insights and assistance in the preparation of this manuscript. We are grateful to Beat A Michel, MD (Department of Rheumatology, University Zurich, Zurich, Switzerland) for his support and encouragement in the preparation of the study protocol. Thanks are especially due to the nurses of the Geriatric Departments for their great commitment in recording the falls of our patients. This study was supported by Strathmann AG; Germany; International Foundation for the Promotion of Nutrition Research and Nutrition Education (ISFE); Swiss Orthopedic Society; and Swiss Foundation for Nutrition Research (SFEFS).

Ancillary