High-Dose Monthly Vitamin D for Prevention of Acute Respiratory Infection in Older Long-Term Care Residents: A Randomized Clinical Trial

Authors

  • Adit A. Ginde MD, MPH,

    Corresponding author
    1. Department of Emergency Medicine, School of Medicine, University of Colorado, Aurora, Colorado
    • Address correspondence to Adit A. Ginde, Department of Emergency Medicine, University of Colorado School of Medicine, 12401 E. 17th Avenue, B-215, Aurora, CO 80045. E-mail: adit.ginde@ucdenver.edu

    Search for more papers by this author
  • Patrick Blatchford PhD,

    1. Department of Biostatistics and Informatics, Colorado School of Public Health, Aurora, Colorado
    2. Geriatric Research, Education and Clinical Center, Eastern Colorado Department of Veterans Affairs, Denver, Colorado
    Search for more papers by this author
  • Keith Breese MA,

    1. Department of Emergency Medicine, School of Medicine, University of Colorado, Aurora, Colorado
    Search for more papers by this author
  • Lida Zarrabi MPH,

    1. Department of Emergency Medicine, School of Medicine, University of Colorado, Aurora, Colorado
    Search for more papers by this author
  • Sunny A. Linnebur PharmD,

    1. Department of Clinical Pharmacy, Skaggs School of Pharmacy and Pharmaceutical Sciences, University of Colorado, Aurora, Colorado
    2. Division of Geriatric Medicine, School of Medicine, University of Colorado, Aurora, Colorado
    Search for more papers by this author
  • Jeffrey I. Wallace MD, MPH,

    1. Division of Geriatric Medicine, School of Medicine, University of Colorado, Aurora, Colorado
    Search for more papers by this author
  • Robert S. Schwartz MD

    1. Geriatric Research, Education and Clinical Center, Eastern Colorado Department of Veterans Affairs, Denver, Colorado
    2. Division of Geriatric Medicine, School of Medicine, University of Colorado, Aurora, Colorado
    Search for more papers by this author

Abstract

Objectives

To determine the efficacy and safety of high-dose vitamin D supplementation for prevention of acute respiratory infection (ARI) in older long-term care residents.

Design

Randomized controlled trial investigating high-dose vs standard-dose vitamin D from 2010 to 2014.

Setting

Colorado long-term care facilities.

Participants

Long-term care residents aged 60 and older (n = 107).

Intervention

The high-dose group received monthly supplement of vitamin D3 100,000 IU, the standard-dose group received a monthly placebo (for participants taking 400–1,000 IU/d as part of usual care) or a monthly supplement of 12,000 IU of vitamin D3 (for participants taking <400 IU/d as part of usual care).

Measurements

The primary outcome was incidence of ARI during the 12-month intervention. Secondary outcomes were falls and fractures, 25-hydroxyvitamin D levels, hypercalcemia, and kidney stones.

Results

Participants (55 high dose, 52 standard dose) were randomized and included in the final analysis. The high-dose group had 0.67 ARIs per person-year and the standard-dose group had 1.11 (incidence rate ratio (IRR) = 0.60, 95% confidence interval (CI) = 0.38–0.94, P = .02). Falls were more common in the high-dose group (1.47 per person-year vs 0.63 in standard-dose group; IRR = 2.33, 95% CI = 1.49–3.63, P < .001). Fractures were uncommon and similar in both groups (high dose 0.10 vs standard dose 0.19 per person-year; P = .31). Mean trough 25-hydroxyvitamin D levels during the trial were 32. ng/mL in the high-dose group and 25.1 ng/mL in the standard-dose group. There was no hypercalcemia or kidney stones in either group.

Conclusion

Monthly high-dose vitamin D3 supplementation reduced the incidence of ARI in older long-term care residents but was associated with a higher rate of falls without an increase in fractures.

Introduction

Acute respiratory infection (ARI) is common in older adults and results in morbidity, healthcare usage, and functional decline.[1, 2] Older long-term care (LTC) residents are particularly vulnerable to ARI because their immune responses are impaired and their functional reserve is low.[3, 4] Strategies to reduce the incidence and severity of ARI in this population are limited, and effective vaccines are lacking for most common respiratory pathogens.

Micronutrient deficiencies are common in LTC residents[5, 6] and can exacerbate age-related changes in immune function.[7, 8] Multivitamin supplementation appears to improve indices of immune function,[9] but the clinical effect on ARI and other infections has been modest.[10, 11] Vitamin D has an important role in many aspects of immune function, particularly innate immunity.[12] Older adults are at high risk of vitamin D deficiency,[13] and epidemiological studies demonstrate a consistent association between vitamin D deficiency and ARI,[14, 15] but clinical trials of vitamin D supplementation for ARI prevention have been mixed, with metaanalyses suggesting modest benefit but substantial heterogeneity.[15, 16] The use of low doses for short durations in predominantly healthy populations have limited previous trials.

Therefore, a randomized, controlled trial was conducted to evaluate the efficacy and safety of high-dose vitamin D3 supplementation for 12 months to reduce the incidence and severity of ARI in older LTC residents.

Methods

Study Design

This study was a double-blind, parallel-group, randomized, controlled Phase II trial comparing oral high-dose with standard-dose vitamin D3 supplementation administered monthly for 12 months. The trial was conducted from June 2010 to January 2014. The Colorado Multiple Institutional Review Board approved the protocol, which an independent Data Safety and Monitoring Board (DSMB) supervised. Written informed consent was obtained from each participant or his or her legally authorized representative. The trial was registered at Clinicaltrials.gov NCT01102374 (see Online Supplement for full protocol).

Eligibility Criteria

Older residents (aged ≥ 60) from 25 selected Colorado LTC facilities (skilled nursing or assisted living facilities) were eligible for participation. Exclusion criteria were terminal illness; anticipated facility discharge within 12 months; inability to take whole or crushed tablets; active cancer, except squamous or basal cell carcinoma; underweight (body mass index <18 kg/m2); taking immunosuppressive medications; renal failure (estimated glomerular filtration rate (eGFR)<15 mL/min per 1.73 m2); taking more than 1,000 IU per day of vitamin D; personal (or strong family) history of kidney stones; history of sarcoidosis or other granulomatous disorder; baseline hypercalcemia (albumin-adjusted calcium >10.5 mg/dL); baseline serum 25-hydroxyvitamin D (25(OH)D) level of 40 ng/mL or greater (to convert to nmol/L, multiple by 2.496); inability of participant or legally authorized representative to speak and understand English and no available interpreter; and inability to provide informed consent and no available healthcare proxy.

Intervention

Participants were randomized to one of two vitamin D dose groups: high dose (equivalent to 3,000–4,000 IU/d) or standard dose (equivalent to 400–1,000 IU/d). A participant's total vitamin D dose included any study drug supplementation plus vitamin D taken as part of usual care (0–1,000 IU/d). Because the facilities were not formally engaged in the research, the study team administered the study drug monthly and did not interfere with usual clinical care. Based on institutional review board and DSMB recommendations, all trial participants received at least 400 IU per day equivalent of vitamin D to meet the Institute of Medicine Dietary Reference Intake (including typical dietary intake).

Participants already taking 400 to 1,000 IU per day of vitamin D as usual care were randomized to one of two study drugs (oral 100,000 IU vitamin D3 monthly or matched oral placebo monthly) while continuing their usual vitamin D regimen. Participants taking less than 400 IU per day of vitamin D supplementation as usual care were also randomized to one of two study drugs (oral 100,000 IU vitamin D3 monthly or oral 12,000 IU vitamin D3 monthly) while continuing any usual vitamin D regimen. The 12,000 IU per month supplementation ensured at least 400 IU per day.

Pencol Compounding Pharmacy (Denver, CO) provided study drugs and performed routine testing of study drug content throughout the trial to verify study drug content. Vitamin D and placebo were identical in size, weight, color, smell, texture, and taste. If necessary, the capsule was opened and sprinkled on food for administration. The research pharmacy provided study medication to the research staff in numbered blister packs, and the sequence was concealed until interventions were assigned. The protocol and study drugs were also under the oversight of the Food and Drug Administration as an Investigational New Drug.

Randomization

The research pharmacy (Veterans Affairs Eastern Colorado Research Pharmacy, Denver, CO) performed one-to-one randomization in permuted block sizes of four to eight stratified according to site and baseline vitamin D supplementation (<400 IU/d or 400–1,000 IU/d). Study personnel, outcome assessors, study participants, and treating clinicians were blinded to study group assignment, allocation sequence, and baseline 25(OH)D level (the study team was informed of only dichotomous data on eligibility as <40 or ≥40 ng/mL). Unblinding was not required for any participant during the course of the trial.

Outcomes

The primary outcome was total number of incident ARIs during 12 months of follow-up. Upper (common colds, sinusitis, pharyngitis, otitis media) and lower (acute bronchitis, influenza, pneumonia) ARIs that required medical attention (nurse or physician assessment, new prescribed treatment) were measured using a chart review method validated in the LTC setting,[17] with additional active surveillance during monthly study visits.

The infection-related secondary outcomes included severity of ARIs as measured according to emergency department visits or hospitalization for ARIs, time to first ARI, and incidence of other infections (urinary tract, skin or soft tissue, other) during the 12-month follow-up period. Efficacy of the intervention was also assessed according to change in 25(OH)D levels from baseline at 3, 7, and 11 months. Collection of samples for trough 25(OH)D levels occurred just before the next monthly dose, and analysis was performed at the University of Washington using the liquid chromatography—tandem mass spectrometry method and accounting for the C-3 epimer of 25(OH)D. Cryopreserved samples were batch tested after trial participation to avoid unblinding and reduce measurement variability.

The primary safety outcome was incident hypercalcemia, defined as albumin-adjusted serum calcium greater than 10.5 mg/dL, measured at 3, 7, and 11 months. Incident falls, fractures, kidney stones, all-cause hospitalizations, and all-cause death were also measured using chart review. At monthly medication administration visits, research assistants queried participants and clinical staff for new adverse events during and classified them using the Medical Dictionary for Regulatory Activities (MedDRA) hierarchy.

Data Collection

Baseline data collected from chart review and interviews with participants, legally authorized representatives, and clinical providers included demographic characteristics; facility length of stay; comorbid conditions; advance directives; vaccination status (seasonal and H1N1 influenza, pneumococcal); smoking history; body mass index (BMI); physical activity; and current medications, including vitamin D and calcium supplementation. Outcome and adverse event data were collected at monthly medication administration visits, with expanded data collection, chart review, and blood draws at 3, 7, and 11 months.

Statistical Analysis

The primary intention-to-treat analysis measured the effect of randomized treatment group (high- vs standard-dose vitamin D) on the number of incident ARIs observed during follow-up. The treatment effect was estimated using Poisson regression as performed using the PROC GENMOD procedure using SAS version 9.3 (SAS Institute, Inc., Cary, NC), expressed as incidence rate ratios. Some participants had less than 12 months of observation time for reasons such as drop out, death, and loss to follow-up. The logarithm of participant time active in the study was included in the model as an offset to account for the different observation times. Secondary analyses of count data were treated in a manner similar to analyses of the primary outcome of ARI. All outcomes were prespecified in the study protocol. Prespecified subgroup analyses included age, sex, residence, comorbidities, baseline vitamin D supplementation, BMI, renal function, and completers (defined as ≥11/12 possible doses of study medication).

The sample size calculation assumed a control group rate of 1.4 ARIs per person-year and an estimated 35% lower rate in the high-dose vitamin D intervention group. It was anticipated that 80% of the total possible follow-up time would be obtained (because of censoring). With a two-sided type I error rate of 0.05 and 80% power, a total sample size of 200 randomized participants was planned for (100 per group). In consultation with the DSMB and the National Institutes of Health (NIH), recruitment for the trial ended at 107 randomized participants because of the lack of available local participants and insufficient resources to expand recruitment beyond the local geographic area.

Results

Participants

Of the 1,055 LTC residents screened, 107 were eligible and randomized (55 high dose, 52 standard dose). The primary reason for exclusion was clinician discretion based on their knowledge of exclusion criteria and study intervention (Figure 1). The intention-to-treat analysis included all 107 randomized participants.

Figure 1.

Consolidated Standards of Reporting Trials diagram for flow of participants in trial.

Baseline participant characteristics are presented in (Table 1). There were modest differences in some baseline characteristics, with the high-dose group having a higher mean BMI and rates of chronic obstructive pulmonary disease (COPD) and diabetes mellitus but lower rates of current smoking, asthma, coronary artery disease, and dementia. Baseline vitamin D supplementation and serum 25(OH)D levels were similar between the two groups.

Table 1. Baseline Characteristics of Randomized Study Participants
CharacteristicHigh Dose, n = 55Standard Dose, n = 52
  1. SD = standard deviation.

Age, mean ± SD80 ± 1082 ± 10
Female, n (%)33 (60.0)29 (55.8)
Non-Hispanic white, n (%)48 (87.3)48 (92.3)
Facility length of stay, months, mean ± SD24 ± 2428 ± 34
Skilled nursing facility, n (%)15 (27.3)13 (25.0)
Required surrogate for consent, n (%)19 (34.5)25 (48.1)
Do-not-hospitalize order, n (%)1 (1.8)3 (5.8)
Body mass index, kg/m2, mean ± SD28.1 ± 6.826.0 ± 5.4
Smoking history, n (%)
Current6 (10.9)9 (17.3)
Former20 (36.4)11 (21.2)
Never28 (50.9)31 (59.6)
Comorbidities, n (%)
Asthma1 (1.8)5 (9.6)
Chronic obstructive pulmonary disease17 (30.9)14 (26.9)
Congestive heart failure12 (21.8)15 (28.8)
Coronary artery disease8 (14.5)14 (26.9)
Diabetes mellitus21 (38.2)14 (26.9)
Dementia16 (29.1)25 (48.1)
Depression30 (54.5)28 (53.8)
History of cancer7 (12.7)5 (9.6)
Osteoporosis2 (3.6)4 (7.7)
Documented influenza vaccination in past 12 months, n (%)30 (54.5)32 (61.5)
Outdoor physical activity in past month, n (%)
None24 (43.6)23 (44.2)
At least monthly4 (7.3)4 (7.7)
At least weekly19 (34.5)15 (28.8)
Daily8 (14.5)10 (19.2)
Vitamin D supplementation dose, IU/d
Mean ± SD226 ± 279232 ± 304
400–1,000, n (%)23 (41.8)20 (38.5)
Serum 25-hydroxyvitamin D, ng/mL
Mean ± SD23.0 ± 8.423.0 ± 9.9
<20, n (%)18 (32.7)19 (36.5)
Serum albumin-adjusted calcium, mg/dL, mean ± SD9.1 ± 0.39.1 ± 0.4
Serum phosphorus, mg/dL, mean ± SD3.5 ± 0.73.5 ± 0.7
Estimated glomerular filtration rate, mL/min per 1.73 m2, mean ± SD69.7 ± 23.270.2 ± 30.0

Vitamin D Intervention

The number of monthly study drug doses received during the 12 months was similar between the two treatment groups (high-dose group: median 11 (interquartile range (IQR) 8–12); standard-dose group: median 11 (IQR 6–12)). Mean 25(OH)D levels increased in both groups (Figure 2). Mean trough 25(OH)D levels in the high-dose group remained greater than the target 30 ng/mL throughout the trial and were significantly higher than in the standard-dose group at every time point (P < .001).

Figure 2.

Change in 25-hydroxyvitamin D levels in two randomized groups: high-dose vitamin D (solid line), standard-dose vitamin D (dotted line).

Primary Outcome

The incidence of ARI was lower in the high-dose group than the standard-dose group (incidence rate ratio (IRR) = 0.60, 95% confidence interval (CI) = 0.38–0.94, P = .02). There were 17 (31%) participants in the high-dose group and 24 (46%) in the standard-dose group who had at least one ARI (P = .10). Time to first ARI analysis (Figure S1) demonstrated an effect size similar to that of the primary analysis (hazard ratio = 0.59, 95% CI = 0.32–1.09, P = .09).

Secondary Outcomes

The high-dose group had a lower incidence of upper ARI (IRR = 0.52, 95% CI = 0.31–0.90, P = .02) and skin and soft tissue infections (IRR = 0.32, 95% CI = 0.13–0.80, P = .02) than the standard-dose group (Table 2). There were no differences in incidence of lower ARI, urinary tract infection, or other infections or hospitalizations for ARI.

Table 2. Incidence of Efficacy and Safety Outcomes According to Vitamin D Dosage Group
Outcome MeasureWith ≥1 EventsIncidence Rate Ratio (95% Confidence Interval)P-Value
High Dose, n = 55aStandard Dose, n = 52b
EventsIncidenceEventsIncidence
  1. a

    47.8 person-years.

  2. b

    42.8 person-years.

  3. c

    Primary outcome.

  4. ARI = acute respiratory infection; NA = not applicable.

Efficacy
ARIc32 (17)0.6748 (24)1.120.60 (0.38–0.94).02
Upper21 (13)0.4436 (20)0.840.52 (0.31–0.90).02
Lower11 (7)0.2312 (9)0.280.82 (0.36–1.86).64
Hospitalization for ARI3 (2)0.066 (5)0.140.45 (0.11–1.79).26
Skin or soft tissue infection6 (6)0.1317 (11)0.400.32 (0.13–0.80).02
Urinary tract infection38 (19)0.8022 (14)0.511.55 (0.92–2.62).10
Other infection27 (16)0.5726 (18)0.610.93 (0.54–1.60).80
Safety
Falls70 (20)1.4727 (15)0.632.33 (1.49–3.63)<.001
Fractures5 (4)0.108 (8)0.190.56 (0.18–1.71).31
All-cause hospitalizations33 (22)0.6933 (24)0.770.90 (0.55–1.45).66
Death90.19130.300.62 (0.27–1.45).27
Incident kidney stone0000NA
Incident hypercalcemia0000NA
25-hydroxyvitamin D level ≥80 ng/mL0000NA

No prespecified vitamin D-related safety outcomes were observed in either group (hypercalcemia, kidney stones, hypervitaminosis D). The overall proportion of participants with all-cause hospitalizations (46% high dose, 43% standard dose) and death (22% high dose, 21%, standard dose) was high in both groups of LTC residents but not different between groups.

The high-dose group had a higher incidence of falls (IRR = 2.33, 95% CI = 1.49–3.63, P < .001). At least one fall during the follow-up period was recorded for 20 (36%) participants in the high-dose group and 15 (29%) in the standard-dose group (P = .41). Similarly, the time-to–first fall analysis (Figure S2) demonstrated no marked difference between groups (hazard ratio = 1.34, 95% CI = 0.68–2.59, P = .41). Thus, participants with multiple recorded falls appeared to drive the overall difference in fall incidence.

The overall incidence of fractures was low and did not differ between the groups (IRR = 0.56, 95% CI = 0.18–1.71, P = .31), with one or more fractures in four (7%) high-dose and eight (15%) standard-dose group participants (P = .18).

Adverse Events

There were no significant between-group differences in the recorded adverse events overall or according to MedDRA group (Table S1).

Subgroup Analyses

Prespecified subgroup analyses of the primary ARI outcome is displayed in (Figure 3) along with post hoc subgroup analyses for the secondary fall outcome. The following subgroups had a significantly lower incidence of ARI in the high-dose than the standard-dose group: baseline vitamin D supplementation less than 400 IU, baseline 25(OH)D level 20 ng/mL or greater, 11 or more study medication doses received, aged 80 and older, dementia, and eGFR of 60 mL/min per 1.73 m2 or greater. The following subgroups had a greater observed incidence of falls in the high-dose than the standard-dose group: baseline vitamin D supplementation of 400 IU and greater, 11 or more study medication doses received, younger than 80 years, dementia, and BMI of 25 kg/m2 or greater.

Figure 3.

Subgroup analyses of acute respiratory infection and fall outcomes: (A) acute respiratory infection, (B) falls. 25(OH)D = 25-hydroxyvitamin D; COPD = chronic obstructive pulmonary disease; eGFR = estimated glomerular filtrate rate.

Discussion

In this double-blind, Phase II, randomized, controlled trial, older LTC residents receiving monthly high-dose vitamin D supplementation had a 40% lower incidence of ARI during 12 months of follow-up than those receiving standard-dose vitamin D. There were no observed differences between the high-dose and standard-dose groups in hypercalcemia, hypervitaminosis D, kidney stones, hospitalizations, death, or fractures, although there was a markedly higher incidence of falls in participants in the high-dose vitamin D group driven by participants with multiple falls.

To the authors’ knowledge, this is the first trial to evaluate high-dose vitamin D supplementation for prevention of ARI in older LTC residents. Secondary analyses of two randomized controlled trials suggest that 800 IU per day vitamin D supplementation had modest benefit in preventing ARI in older adults.[18, 19] It has been reported that 60,000 IU monthly of vitamin D reduced antibiotic use in community-dwelling adults aged 70 and older,[20] but that was a post hoc analysis of trials designed for other purposes. A recent randomized trial found that 96,000 IU of vitamin D every 2 months in sheltered care residents in the United Kingdom did not influence the incidence of ARI,[21] although these participants were younger and healthier than the current study population of older LTC residents and received half the dose used in the current trial.

Although not a primary focus of this trial, falls are also an important cause of morbidity in older adults. There is growing interest in the role of vitamin D in fall prevention, relating to muscle function and balance. An American Geriatrics Society consensus statement recommends 1,000 to 4,000 IU daily of vitamin D supplementation for fall prevention in older adults.[22] Recruitment for the current study was challenging because of clinician use of high-dose vitamin D for usual care of many older LTC residents, although the incidence of falls was 2.3 times as high in participants in the high-dose group. The mechanism of this finding requires further investigation, including exploration of the hypothesis that high-dose vitamin D leads to greater mobility, resulting in greater exposure to falls. Two prior trials of high intermittent doses of vitamin D (500,000 IU annually, 60,000 IU monthly) in older community-dwelling adults also reported higher fall rates in the intervention group.[23, 24] Although higher fracture rates were not found in the current study or in these two prior trials, higher fall rates from these three trials raise the question of the potential safety of high intermittent doses of vitamin D for older adults. This dosing paradigm was used to reduce the need to deliver daily study doses to these LTC residents. Whether daily high-dose vitamin D supplementation prevents (or increases) falls in older adults remains unknown and is the focus of an ongoing Phase III trial (NCT02166333).

Several other secondary and subgroup analyses could have important clinical implications meriting further investigation. The incidence of skin and soft tissue infections was lower in the high-dose group and is an underexplored infectious outcome in vitamin D trials. High-dose vitamin D may be particularly effective in ARI prevention for individuals with COPD, which has important clinical implications because of ARI-associated disease exacerbation. Several recent clinical trials support the potential for high-dose vitamin D to improve COPD outcomes, particularly in individuals with severe disease and severe vitamin D deficiency.[25-27].

The results of this trial should be interpreted in the context of several limitations. The sample size was modest and did not reach goal recruitment, which affects trial power and precision, although outcome data did not influence the decision to end trial recruitment, limiting the potential for bias and Type 1 error. Modest differences in some baseline characteristics may have influenced results, although subgroup analyses suggest that this was unlikely. Intermittent bolus dosing was selected largely for feasibility, and the regimen is not the normal way that humans acquire vitamin D from ultraviolet B radiation or dietary sources (in small, consistent doses). Baseline fall history was not measured or adjusted for, so imbalance in the number of frequent fallers in the high-dose group may have driven the higher rate of total falls in this group. Nevertheless, others have observed higher falls with high intermittent doses of vitamin D, so it is recommended that equivalent daily doses be tested in future trials for ARI prevention in this population. Because of ethical concerns to ensure at least standard amounts of vitamin D for the control group in this vulnerable population, there was no true placebo group. This may have reduced the effect size for some outcomes of interest. In addition, some missed doses of study medication (due to logistical issues and hospitalizations) may have also led to regression to the mean. The observed rise in 25(OH)D levels was not as robust as anticipated, which may have mitigated potential benefits (as well as risks) of high-dose supplementation. Completers (those receiving ≥11 of 12 possible doses of study medication) appeared to have more-robust findings of ARI prevention and fall increase than noncompleters, providing further causal evidence of these results.

In summary, monthly high-dose vitamin D supplementation reduced the incidence of ARIs but increased falls, without an increase in fractures, in older LTC residents. If these results are confirmed in a larger trial, high-dose vitamin D, ideally using daily dosing to minimize fall risk, has the potential for substantial public health benefit through ARI prevention for the large and growing population of LTC residents.

Acknowledgments

We thank the LTC facilities, residents, and families who participated in this trial and the DSMB members for their careful review and oversight of this trial.

This trial was supported by the Beeson Career Development Award (NIH, National Institute on Aging Grant K23AG040708), NIH, National Center for Advancing Translational Sciences Colorado Clinical and Translational Science Awards Grant UL1TR001082, and the American Geriatrics Society Jahnigen Career Development Scholars Award. Dr. Blatchford and Dr. Schwartz were supported by the Eastern Colorado Veterans Affairs Geriatric Research, Education and Clinical Center. Contents are the authors’ sole responsibility and do not necessarily represent official NIH or Department of Veterans Affairs views.

Conflict of Interest: None.

Author Contributions: Dr. Ginde had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Ginde, Schwartz. Acquisition of data: Ginde, Breese, Zarrabi. Statistical analysis: Blatchford. Interpretation of data: Ginde, Blatchford, Breese, Zarrabi, Linnebur, Wallace, Schwartz. Drafting of the manuscript: Ginde. Critical revision of manuscript for important intellectual content: Ginde, Blatchford, Breese, Zarrabi, Linnebur, Wallace, Schwartz.

Sponsor's Role: The sponsors had no role in the design, methods, subject recruitment, data collection, analysis, interpretation, or presentation of the study.

Ancillary