Skip to main content
Download PDF
  • Research Article
  • Open access
  • Published:

Vitamin D supplementation as a potential cause of U-shaped associations between vitamin D levels and negative health outcomes: a decision tree analysis for risk of frailty

BMC Geriatrics volume 17, Article number: 236 (2017) Cite this article

Abstract

Background

A recent controversy in vitamin D research is a “U-shaped association”, with elevated disease risks at both high and low 25-hydroxyvitamin D (25 (OH) D) levels.

Methods

This is a cross-sectional study of 238 male nursing home veterans in Hawaii. Classification and regression tree (CART) analysis identified groups based on 25 (OH) D and vitamin D supplementation for frailty risk. Characteristics were examined and compared across the groups using logistic regression and receiver operating characteristic (ROC) curve analyses.

Results

CART analysis identified three distinct groups: vitamin D supplement users (n = 86), non-users with low vitamin D (n = 55), and non-users with high vitamin D (n = 97). Supplement users were the most frail, but had high mean 25 (OH) D of 26.6 ng/mL, which was compatible with 27.1 ng/mL in non-users with high vitamin D, while mean 25 (OH) D of non-users with low vitamin D was 11.7 ng/mL. Supplement users and non-users with low vitamin D were significantly more likely to be frail (odds ratio (OR) = 9.90, 95% CI = 2.18–44.86, p = 0.003; OR = 4.28, 95% CI = 1.44–12.68, p = 0.009, respectively), compared with non-users with low vitamin D. ROC curve analysis showed the three groups significantly predicted frailty (area under the curve = 0.73), with sensitivity of 64.4% and specificity of 76.7%, while 25 (OH) D did not predict frailty.

Conclusions

In these nursing home veterans, vitamin D supplement users were the most frail but with high 25 (OH) D. This can potentially be a cause of U-shaped associations between vitamin D levels and negative health outcomes.

Peer Review reports

Background

Vitamin D deficiency is prevalent worldwide and has been recognized as a public health problem [1]. Low vitamin D has been extensively studied and shown to be associated with various negative health outcomes, such as osteoporosis, fracture, muscle weakness, falls, autoimmune diseases, and cardiovascular diseases [2]. Optimizing deficient vitamin D levels with supplementation can result in mostly favorable results, although definitive evidence from well-designed randomized controlled clinical trials is lacking [3]. In light of the accumulating evidence on the beneficial effects of vitamin D supplementation and its relatively safe side-effect profile, vitamin D supplementation has been recommended by multiple authorities and guidelines [4,5,6].

One of the recent controversies in vitamin D research is a “U-shaped association”, [7,8,9] in which both high and low levels of vitamin D are associated with elevated disease risks [7]. Regarding all-cause mortality, most observational population-based studies have shown that low vitamin D predicted premature death, which is supported by recent meta-analyses [10,11,12,13]. However, there were a considerable number of studies that showed U-shaped or reverse J-shaped associations between vitamin D and mortality [14,15,16,17,18,19,20]. One of the meta-analysis studies showed the possibility that mortality risk may become higher again at a serum 25-hydroxyvitamin D (25 (OH) D) level of 112.5 nmol/l or more [12]. The association between high vitamin D and high mortality appears counterintuitive and the underlying mechanisms are unknown. Criticisms of this paradoxical finding were that it was due to chance, [7] a lack of adequate adjustment, [8] an analytical bias due to serum vitamin D assays, [7] a small number of the highest serum vitamin D group, [18] or vitamin D intoxication [21].

Use of vitamin D supplementation may be a cause of increased disease risks at high vitamin D levels and can create U-shaped associations [22]. The Newcastle 85+ Study prospectively followed 775 men and women aged 85 or older over 6 years and examined all-cause mortality according to baseline 25 (OH) D levels [13]. Compared with the middle 25 (OH) D group, the highest and lowest 25 (OH) D groups had non-significantly increased mortality risks (adjusted hazard ratio [HR] = 1.25, 95% confidence interval [CI] = 0.97–1.63; HR = 1.10, 95% CI = 0.85–1.42, respectively). After excluding 150 vitamin D supplement users, the mortality risk for the highest 25 (OH) D group decreased by 16% (adjusted HR = 1.05, 95% CI = 0.73–1.53) whereas the mortality risk for the lowest 25 (OH) D group increased (adjusted HR = 1.22, 95% CI = 0.93–1.60) [13]. It is speculated that the supplement users were at high risk with high 25 (OH) D, likely due to the supplements, which led to the increased mortality risk of the highest 25 (OH) D group. Another meta-analysis study showed that the mortality risk for low 25 (OH) D was significantly lower among studies with prevalence of vitamin D supplement use of more than 10% compared with studies with less than 10% (p for meta-regression analysis <0.05), [11] which suggests that vitamin D supplementation attenuated inverse association between vitamin D and mortality, possibly by increasing 25 (OH) D of high-mortality-risk participants.

In order to further investigate how vitamin D supplementation affects associations between vitamin D status and health outcomes, we explored use of vitamin D supplementation and serum 25 (OH) D in relation to frailty, using a cohort of frail nursing home residents [23] among whom prevalence of both vitamin D supplement use and vitamin D deficiency are high. Our hypothesis was that those on a vitamin D supplement were highly frail even though serum vitamin D levels were elevated by supplements. They could be included in “high” vitamin D groups along with healthier vitamin D sufficient non-supplement users, leading to the seemingly paradoxically high risks of various health outcomes in those with high vitamin D.

Methods

Study design, setting, and population

This cross-sectional study was conducted at a Veterans Affairs nursing home in Honolulu, Hawaii, providing rehabilitation, skilled-nursing care, intermediate care, respite care, and hospice/palliative care for veterans. The study participants were all male veterans admitted to the study nursing home except for those admitted for hospice/palliative care, from 1st of January 2011 to 31st of December 2012. The data were anonymised and collected by a retrospective chart review. This study was approved by the Institutional Review Boards of Veterans Affairs Pacific Islands Health Care System. The study design, setting, and population have been described in detail elsewhere [24, 25].

Predictor variables - vitamin D supplementation and vitamin D level

Serum total 25 (OH) D was measured for all veterans on nursing home admission as a part of initial assessment. The information on vitamin D supplementation was obtained from transfer summaries, discharge summaries, outpatient clinic notes, or history taken from veterans or family. Various dosage, frequency, and duration were observed for vitamin D supplementation, from one multivitamin tab a day to 50,000 IU of ergocalciferol three times weekly. Use of vitamin D supplementation was calculated as a mean daily dosage of any vitamin D supplement over the previous 1 month and categorized into four groups: no supplement, 1–400 IU/day, 401–800 IU/day, or ≥801 IU/day.

Outcome variable - frailty

Frailty has been described as a state of reduced physiological reserve due to age related accumulation of multisystem impairments [26]. As people become frailer, they are more predisposed to increased risks of various adverse health outcomes, including falls, fractures, hospitalization, nursing home placement, disability, poor quality of life, and dementia [27,28,29,30,31,32,33,34]. Therefore frailty was considered as a good surrogate marker of biological aging [35]. In this study, frailty was measured by using a deficit accumulation model of the Frailty Index (FI) [35] FI is calculated from a variety of health deficits that usually include symptoms, signs, comorbidities, and disabilities that are biologically sensible, accumulate with age, do not saturate too early, and cover a range of systems [36]. Although FI does not require the same number of deficits or the same set of deficits, it is recommended to include at least 30 deficits [36]. We constructed FI based on 34 deficits including 12 chronic diseases, 9 psychological symptoms, 6 functional disabilities, 3 gait/fall-related problems, 2 cognitive symptoms, 1 obesity, and 1 pain symptom. (see the Additional file 1 for detail) [36, 37]. Although FI is a continuous score and not meant to be dichotomized, we used an empirical cut-point to define frailty as FI > =0.25 [38].

Covariates

Demographic data collected on admission were age, body mass index, ethnicity (White, Asian/Pacific Islander, or other), education, smoking, alcohol, place where veterans came from (home, acute care hospitals, or other nursing homes), season and reason for admission (rehabilitation, non-rehabilitation skilled-nursing care, intermediate care, or respite care).

Statistical analysis

The classification and regression tree (CART) analysis [39] is a non-parametric classification technique that can deal with multiple predictors of both continuous and categorical data. It builds a decision tree by recursive partitioning to best explain the risk estimate of the dependent variable. This method has often been used for data mining and it was considered to be appropriate for the exploratory nature of this study. We used this method to split the cohort based on two variables: use of vitamin D supplementation and 25 (OH) D, into progressively smaller and more homogeneous subgroups with highest discriminative ability of identifying frailty risk. The minimum number of cases in a node was set at 23, 1/10 of the entire cohort, and the minimum change in improvement was set at 0.001. Ten-fold cross-validation was performed. The subgroups by the CART analysis were compared for the characteristics using a one-way analysis of variance (ANOVA) for continuous variables and a chi-square test for categorical variables. Correlation coefficient was examined using Spearman’s rho between 25 (OH) D and the FI in the entire cohort as well as supplement users and non-users. Relative likelihood of frailty of each CART group compared with the entire cohort was calculated. Univariate logistic regression models were used to examine risk of frailty for the CART groups and other characteristics. The CART groups were further examined for independent risk of frailty using a multivariate logistic regression model adjusted for variables which were significant in the univariate logistic regression models. Frailty risk discrimination by the CART analysis was assessed using the receiver operating characteristic (ROC) curve analysis and the area under the ROC curve (AUC), and was compared with that by 25 (OH) D as a continuous variable.

Among the supplement users, the characteristics were compared using a one-way ANOVA for continuous variables and a chi-square test for categorical variables according to the dosage of the supplement.

All statistical analyses were conducted using IBM SPSS Statistics (version 20, IBM Corporation, Armonk, NY, USA), and two-sided p value of <0.05 was considered statistically significant.

Results

There were a total of 302 male veterans admitted to the study nursing home during 2011 and 2012. Of those, 61 veterans who did not have 25 (OH) D measured within 7 days of admission and 3 veterans who missed more than 30% of deficits for constructing FI were excluded, leaving 238 veterans (78.8%) as a final sample. There were no significant differences in mean age, mean 25 (OH) D level, BMI, smoking, alcohol use and FI between those included (n = 238) and those excluded (n = 64). Those excluded had slightly but significantly longer duration of education compared with those included (13.7 v 13.0 years).

Decision tree analysis

Figure 1 displays a decision tree of the CART analysis. The cohort was initially divided into two groups: those on vitamin D supplement of any dosage (1–400 IU/day, 401–800 IU/day, or ≥801 IU/day, vitamin D supplement users, n = 86) and those without vitamin D supplement (non-users, n = 152). Non-users were further divided based on 25 (OH) D with a cut-off point of 17.5 ng/mL into two groups: non-users with low vitamin D level (n = 55) and non-users with high vitamin D level (n = 97). The cross-validation showed the same risk estimate. As a supplementary analysis the CART analysis was repeated using the FI as a continuous outcome variable, which showed very similar results (data not shown).

Fig. 1
figure 1

A decision tree by classification and regression tree analysis based on use of vitamin D supplement and 25-hydroxyvitamin D (25 (OH) D) among 238 male veterans in a nursing home. 25 (OH) D: 25-hydroxyvitamin D

Full size image

Cohort characteristics

Table 1 presents demographic characteristics of the entire cohort (N = 238). The mean age was 73.4 years. The mean Frailty Index (FI) was 0.39 and 87.4% (n = 208) had frailty (FI > =0.25). The mean 25 (OH) D was 23.4 ng/mL, ranging 4–53 ng/mL, and 37.4% (n = 89) had 25 (OH) D < 20 ng/mL.

Table 1 Characteristics according to three groups based on vitamin D level and supplement use
Full size table

The characteristics were compared across three groups by CART analysis: supplement users (n = 86, 36.1%), non-users with low vitamin D (n = 55, 23.1%), and non-users with high vitamin D (n = 97, 40.8%) (Table 1). The mean FI and prevalence of frailty were both significantly different across the groups, with supplement users being the most frail and non-users with high vitamin D being the least frail (mean FI: 0.43, 0.41, and 0.35, respectively. Frailty prevalence: 97.7, 90.9, and 76.3%, respectively. Both p < 0.001). Supplement users and non-users with high vitamin D had significantly higher 25 (OH) D, 26.6 and 27.1 ng/mL, respectively, than non-users with low vitamin D, 11.7 ng/mL. While FI and 25 (OH) D were not significantly correlated in the entire cohort and supplement users (r = −0.10 and −0.05, respectively), there was a significant inverse correlation between FI and 25 (OH) D in non-users (r = −0.19, p = 0.02).

Table 2 shows univariate and age-adjusted logistic regression models used to assess risk of frailty (FI ≥ 0.25). In unadjusted models, supplement users and non-users with low vitamin D were significantly more likely to be frail (Odds ratio [OR] = 13.05, 95% CI = 2.98–57.25, p = 0.001; OR = 3.11, 95% CI = 1.11–8.72, p = 0.03, respectively), with non-users with high vitamin D as a reference. In age-adjusted models, the risk of frailty for supplement users and non-users with low vitamin D remained significant after adjusting for age (OR = 9.90, 95% CI = 2.18–44.86, p = 0.003; OR = 4.28, 95% CI = 1.44–12.68, p = 0.009, respectively).

Table 2 Univariate and age-adjusted logistic regression models for frailty
Full size table

Figure 2 shows an ROC curve analysis showing that CART analysis groups accurately classified risk of frailty among the entire cohort, with AUC of 0.73 (95%CI = 0.65–0.82, p < 0.001). Being on a vitamin D supplement, compared with no use, has sensitivity of 64.4% and specificity of 76.7%. Conversely, 25 (OH) D was not a significant predictor of frailty (AUC = 0.56 95% CI = 0.46–0.66, p = 0.32). After removing supplement users in order to exclude effects of vitamin D supplementation, ROC curve analysis for 25 (OH) D was repeated only among non-users (n = 152). This repeated analysis showed that 25 (OH) D significantly predicted frailty among non-users, with AUC of 0.62 (95% CI = 0.52–0.73, p = 0.04) (Figure not shown).

Fig. 2
figure 2

Receiver operating characteristic (ROC) curve analyses for risk of frailty predicted by the classification and regression tree (CART) analysis groups (blue line) and 25-hydroxyvitamin D (25 (OH) D) level (green line). 25 (OH) D: Serum total 25-hydroxyvitamin D. CART: Classification and regression tree. Area under the ROC curve (AUC) = 0.73 (95% confidence interval (CI) = 0.65–0.82, p < 0.001) for CART analysis groups (blue line); AUC = 0.56 (95%CI = 0.46–0.66, p = 0.32) for 25 (OH) D (green line)

Full size image

Among 86 supplement users, the dosages of vitamin D supplementation were 1–400 IU/day (n = 27, 31.4%), 401–800 IU/day (n = 38, 44.2%), and ≥801 IU/day (n = 21, 25.3%). Table 3 compares the three groups and shows that higher dosage of vitamin D supplementation was significantly associated with higher 25 (OH) D (p for trend = 0.003) and higher body mass index (p = 0.05). Although statistical significance was not reached, those on higher dosage of supplementation were more likely from home and less likely from acute care hospitals (p = 0.06).

Table 3 Characteristics of vitamin D supplement users according to the dosage
Full size table

Discussion

In the current study, three distinct subgroups were successfully identified based on use of vitamin D supplementation and 25 (OH) D using the CART analysis. Taking vitamin D supplementation was the strongest frailty risk discriminative factor, and supplement users were found to have multiple distinguishable features. They were almost all frail, with the highest prevalence of 97.7% (84/86) among the three subgroups, but had high mean 25 (OH) D of 26.6 ng/mL, which was compatible with non-users with high vitamin D (27.1 ng/mL). Among supplement users, 25 (OH) D was not correlated with FI, while there was a significant correlation among non-users.

In light of these findings, we speculate that vitamin D supplement users are a group of highly frail older people with paradoxically high 25 (OH) D levels. This discrepancy between 25 (OH) D and frailty may result from the fact that vitamin D supplementation can quickly correct low 25 (OH) D [40]; however, it may take longer to see positive effects on mortality or other outcomes [11, 41]. One meta-analysis showed that vitamin D supplementation significantly decreased mortality only with follow-up longer than 3 years [41]. In addition, supplements increase 25 (OH) D more efficiently in those with lower baseline 25 (OH) D, [42] which will further exacerbate the discrepancy. Therefore, frail supplement users are likely to remain at high risk of negative health outcomes even with optimized vitamin D status by supplementation, which could potentially confound the profiles and outcomes of the vitamin D-sufficient group. In observational studies, true associations of 25 (OH) D with outcomes could be attenuated or reversed at high 25 (OH) D levels, creating falsely null or U-shaped associations unless controlled properly for the use of vitamin D supplements.

Vitamin D supplements have been used more and more commonly in general populations. According to NHANES, the use of dietary vitamin D supplements has increased among both men and women in most age groups from 1988 to 1994 to 2003–2006 [43]. This increase may be explained by several reasons. With guidelines and authorities advocating the importance of treating low vitamin D, clinicians more often check patients’ serum vitamin D levels and prescribe vitamin D supplement for low vitamin D. Increasing sun protection for skin cancer prevention can also contribute to the increase in vitamin D supplementation. Due to growing media exposure regarding risks of low vitamin D levels, the general public has more awareness of the importance of vitamin D than ever. Furthermore, vitamin D supplements are inexpensive and readily available over the counter with almost no serious adverse effects, which further lowers the threshold for people to start the supplements. In this context, future vitamin D research should take into account the possibility of high prevalence of vitamin D supplementation in various populations, which may have significant confounding effects.

This study has some limitations and its findings must be interpreted with caution. First, it included a small number of only male veterans in a nursing home at a single facility in Hawaii, and the findings may not be entirely generalizable to other populations or to women. Second, some important information related to vitamin D status is missing, such as dietary vitamin D intake, sunlight exposure, or reasons for vitamin D supplementation. Third, we did not have information on what assay was used to measure 25 (OH) D given that 25 (OH) D levels vary according to the assay used [44]. Fourth, due to the small number of participants, we included those who had available data for at least 70% of the deficits, instead of 80%, which is typically required for calculation of the FI [36]. Fifth, the multiple imputation could have been conducted for missing value of 25 (OH) D. Lastly, the cross-sectional study design hinders assessing prospective associations and interactions among 25 (OH) D, vitamin D supplementation, and health outcomes, as vitamin D supplementation may improve supplement users’ overall health status and decrease the risks over years.

Conclusion

Three distinct groups were categorized from the study: 1) vitamin D supplement users, 2) non-users with low vitamin D, and 3) non-users with high vitamin D levels. This study shows that use of vitamin D supplements can potentially be a cause of paradoxical U-shaped associations between vitamin D levels and negative health outcomes, by creating a unique group of participants who are the most frail but have high 25 (OH) D levels. This highlights the importance of identifying vitamin D supplement users and ideally obtaining information on the dosage and duration of the supplementation in order to better examine the true association between vitamin D and health outcomes, including frailty status, by controlling for vitamin D supplementation effects. Future studies can further examine and clarify these effects.

Abbreviations

25 (OH) D:

25-hydroxyvitamin D

ANOVA:

Analysis of variance

CART:

Classification and regression tree

FI:

Frailty Index

ROC:

Receiver operating characteristic

References

  1. Wahl DA, Cooper C, Ebeling PR, Eggersdorfer M, Hilger J, Hoffmann K, Josse R, Kanis JA, Mithal A, Pierroz DD, et al. A global representation of vitamin D status in healthy populations. Arch Osteoporos. 2012;7:155–72.

    Article  CAS  PubMed  Google Scholar 

  2. Holick MF. Vitamin D deficiency. N Engl J Med. 2007;357(3):266–81.

    Article  CAS  PubMed  Google Scholar 

  3. Wimalawansa SJ. Non-musculoskeletal benefits of vitamin D. J Steroid Biochem Mol Biol. 2016;

  4. Ross AC, Manson JE, Abrams SA, Aloia JF, Brannon PM, Clinton SK, Durazo-Arvizu RA, Gallagher JC, Gallo RL, Jones G, et al. The 2011 report on dietary reference intakes for calcium and vitamin D from the Institute of Medicine: what clinicians need to know. J Clin Endocrinol Metab. 2011;96(1):53–8.

    Article  CAS  PubMed  Google Scholar 

  5. Holick MF, Binkley NC, Bischoff-Ferrari HA, Gordon CM, Hanley DA, Heaney RP, Murad MH, Weaver CM. Evaluation, treatment, and prevention of vitamin D deficiency: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab. 2011;96(7):1911–30.

    Article  CAS  PubMed  Google Scholar 

  6. Mayor S. Public Health England recommends vitamin D supplements in autumn and winter. BMJ. 2016;354:i4061.

    Article  PubMed  Google Scholar 

  7. Grant WB, Karras SN, Bischoff-Ferrari HA, Annweiler C, Boucher BJ, Juzeniene A, Garland CF, Holick MF. Do studies reporting ‘U’-shaped serum 25-hydroxyvitamin D-health outcome relationships reflect adverse effects? Dermatoendocrinol. 2016;8(1):e1187349.

    Article  PubMed  PubMed Central  Google Scholar 

  8. Grant WB. Critique of the U-shaped serum 25-hydroxyvitamin D level-disease response relation. Dermatoendocrinol. 2009;1(6):289–93.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  9. Davis CD. Vitamin D and health: can too much be harmful? Am J Lifestyle Med. 2009;3(5):407–8.

    Article  Google Scholar 

  10. Garland CF, Kim JJ, Mohr SB, Gorham ED, Grant WB, Giovannucci EL, Baggerly L, Hofflich H, Ramsdell JW, Zeng K, et al. Meta-analysis of all-cause mortality according to serum 25-hydroxyvitamin D. Am J Public Health. 2014;104(8):e43–50.

    Article  PubMed  PubMed Central  Google Scholar 

  11. Chowdhury R, Kunutsor S, Vitezova A, Oliver-Williams C, Chowdhury S, Kiefte-de-Jong JC, Khan H, Baena CP, Prabhakaran D, Hoshen MB, et al. Vitamin D and risk of cause specific death: systematic review and meta-analysis of observational cohort and randomised intervention studies. BMJ. 2014;348:g1903.

    Article  PubMed  PubMed Central  Google Scholar 

  12. Zittermann A, Iodice S, Pilz S, Grant WB, Bagnardi V, Gandini S. Vitamin D deficiency and mortality risk in the general population: a meta-analysis of prospective cohort studies. Am J Clin Nutr. 2012;95(1):91–100.

    Article  CAS  PubMed  Google Scholar 

  13. Granic A, Aspray T, Hill T, Davies K, Collerton J, Martin-Ruiz C, von Zglinicki T, Kirkwood TB, Mathers JC, Jagger C. 25-hydroxyvitamin D and increased all-cause mortality in very old women: the Newcastle 85+ study. J Intern Med. 2015;277(4):456–67.

    Article  CAS  PubMed  Google Scholar 

  14. Melamed ML, Michos ED, Post W, Astor B. 25-hydroxyvitamin D levels and the risk of mortality in the general population. Arch Intern Med. 2008;168(15):1629–37.

    Article  PubMed  PubMed Central  Google Scholar 

  15. Visser M, Deeg DJ, Puts MT, Seidell JC, Lips P. Low serum concentrations of 25-hydroxyvitamin D in older persons and the risk of nursing home admission. Am J Clin Nutr. 2006;84(3):616–22. quiz 71-2

    CAS  PubMed  Google Scholar 

  16. Jia X, Aucott LS, McNeill G. Nutritional status and subsequent all-cause mortality in men and women aged 75 years or over living in the community. Br J Nutr. 2007;98(3):593–9.

    Article  CAS  PubMed  Google Scholar 

  17. Michaelsson K, Baron JA, Snellman G, Gedeborg R, Byberg L, Sundstrom J, Berglund L, Arnlov J, Hellman P, Blomhoff R, et al. Plasma vitamin D and mortality in older men: a community-based prospective cohort study. Am J Clin Nutr. 2010;92(4):841–8.

    Article  CAS  PubMed  Google Scholar 

  18. Sempos CT, Durazo-Arvizu RA, Dawson-Hughes B, Yetley EA, Looker AC, Schleicher RL, Cao G, Burt V, Kramer H, Bailey RL, et al. Is there a reverse J-shaped association between 25-hydroxyvitamin D and all-cause mortality? Results from the U.S. nationally representative NHANES. J Clin Endocrinol Metab. 2013;98(7):3001–9.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  19. Dror Y, Giveon SM, Hoshen M, Feldhamer I, Balicer RD, Feldman BS. Vitamin D levels for preventing acute coronary syndrome and mortality: evidence of a nonlinear association. J Clin Endocrinol Metab. 2013;98(5):2160–7.

    Article  CAS  PubMed  Google Scholar 

  20. Durup D, Jorgensen HL, Christensen J, Schwarz P, Heegaard AM, Lind B. A reverse J-shaped association of all-cause mortality with serum 25-hydroxyvitamin D in general practice: the CopD study. J Clin Endocrinol Metab. 2012;97(8):2644–52.

    Article  CAS  PubMed  Google Scholar 

  21. Zittermann A, Prokop S, Gummert JF, Borgermann J. Safety issues of vitamin D supplementation. Anti Cancer Agents Med Chem. 2013;13(1):4–10.

    Article  CAS  Google Scholar 

  22. Korgavkar K, Xiong M, Weinstock MA. Review: higher vitamin D status and supplementation may be associated with risks. Eur J Dermatol. 2014;24(4):428–34.

    CAS  PubMed  Google Scholar 

  23. Kojima G. Prevalence of frailty in nursing homes: a systematic review and meta-analysis. J Am Med Dir Assoc. 2015;16(11):940–5.

    Article  PubMed  Google Scholar 

  24. Kojima G, Tamai A, Masaki K, Gatchell G, Epure J, China C, Ross GW, Petrovitch H, Tanabe M. Prevalence of vitamin D deficiency and association with functional status in newly admitted male veteran nursing home residents. J Am Geriatr Soc. 2013;61(11):1953–7.

    Article  PubMed  Google Scholar 

  25. Kojima G, Tanabe M, Masaki K, Ross GW, Tamai A, Petrovitch H, Gatchell G, Fujii D. Association between depressive symptoms and vitamin D deficiency in recently admitted nursing home residents. J Am Geriatr Soc. 2016;64(4):893–5.

    Article  PubMed  Google Scholar 

  26. Clegg A, Young J, Iliffe S, Rikkert MO, Rockwood K. Frailty in elderly people. Lancet. 2013;381(9868):752–62.

    Article  PubMed  Google Scholar 

  27. Kojima G. Frailty as a predictor of future falls among community-dwelling older people: a systematic review and meta-analysis. J Am Med Dir Assoc. 2015;16(12):1027–33.

    Article  PubMed  Google Scholar 

  28. Kojima G. Frailty as a predictor of fractures among community-dwelling older people: a systematic review and meta-analysis. Bone. 2016;90:116–22.

    Article  PubMed  Google Scholar 

  29. Kojima G. Frailty significantly increases the risk of fractures among middle-aged and older people. Evid Based Nurs. 2017;20(4):119–20.

  30. Kojima G. Frailty as a predictor of hospitalisation among community-dwelling older people: a systematic review and meta-analysis. J Epidemiol Community Health. 2016;70(7):722–9.

    Article  PubMed  Google Scholar 

  31. Kojima G. Frailty as a predictor of nursing home placement among community-dwelling older adults: a systematic review and meta-analysis. J Geriatr Phys Ther. 2016;

  32. Kojima G, Taniguchi Y, Iliffe S, Walters K. Frailty as a predictor of Alzheimer disease, vascular dementia, and all dementia among community-dwelling older people: a systematic review and meta-analysis. J Am Med Dir Assoc. 2016;17(10):881–8.

    Article  PubMed  Google Scholar 

  33. Kojima G. Frailty as a predictor of disabilities among community-dwelling older people: a systematic review and meta-analysis. Disabil Rehabil. 2016:1–12.

  34. Kojima G, Iliffe S, Jivraj S, Walters K. Association between frailty and quality of life among community-dwelling older people: a systematic review and meta-analysis. J Epidemiol Community Health. 2016;70(7):716–21.

    Article  PubMed  Google Scholar 

  35. Mitnitski AB, Mogilner AJ, Rockwood K. Accumulation of deficits as a proxy measure of aging. ScientificWorldJournal. 2001;1:323–36.

    Article  CAS  PubMed  Google Scholar 

  36. Searle SD, Mitnitski A, Gahbauer EA, Gill TM, Rockwood K. A standard procedure for creating a frailty index. BMC Geriatr. 2008;8:24.

    Article  PubMed  PubMed Central  Google Scholar 

  37. Kojima G, Tanabe M. Frailty is highly prevalent and associated with vitamin D deficiency in male nursing home residents. J Am Geriatr Soc. 2016;64(9):e33–5.

    Article  PubMed  Google Scholar 

  38. Rockwood K, Andrew M, Mitnitski A. A comparison of two approaches to measuring frailty in elderly people. J Gerontol A Biol Sci Med Sci. 2007;62(7):738–43.

    Article  PubMed  Google Scholar 

  39. Breiman L, Friedman J, Stone CJ, Olshen RA. Classification and regression trees. USA: CRC press; 1984.

  40. Przybelski R, Agrawal S, Krueger D, Engelke JA, Walbrun F, Binkley N. Rapid correction of low vitamin D status in nursing home residents. Osteoporos Int. 2008;19(11):1621–8.

    Article  CAS  PubMed  Google Scholar 

  41. Zheng Y, Zhu J, Zhou M, Cui L, Yao W, Liu Y. Meta-analysis of long-term vitamin D supplementation on overall mortality. PLoS One. 2013;8(12):e82109.

    Article  PubMed  PubMed Central  Google Scholar 

  42. Shab-Bidar S, Bours S, Geusens PP, Kessels AG, van den Bergh JP. Serum 25 (OH) D response to vitamin D3 supplementation: a meta-regression analysis. Nutrition. 2014;30(9):975–85.

    Article  CAS  PubMed  Google Scholar 

  43. Gahche J, Bailey R, Burt V, Hughes J, Yetley E, Dwyer J, Picciano MF, McDowell M, Sempos C. Dietary supplement use among U.S. adults has increased since NHANES III (1988-1994). NCHS Data Brief. 2011;(61):1–8.

  44. Enko D, Fridrich L, Rezanka E, Stolba R, Ernst J, Wendler I, Fabian D, Hauptlorenz S, Halwachs-Baumann G. 25-hydroxy-vitamin D status: limitations in comparison and clinical interpretation of serum-levels across different assay methods. Clin Lab. 2014;60(9):1541–50.

    CAS  PubMed  Google Scholar 

Download references

Acknowledgements

This material is the result of work supported with resources and the use of facilities of the Veterans Affairs Pacific Islands Health Care System, Honolulu, Hawaii. The contents do not necessarily represent the views of the Department of Veterans Affairs or the United States Government.

Funding

None.

Availability of data and materials

The data that support the findings of this study are not publicly available.

Author information

Authors and Affiliations

  1. Department of Primary Care and Population Health, University College London (Royal Free Campus), Rowland Hill Street, London, NW3 2PF, UK

    Gotaro Kojima & Steve Iliffe

  2. Veterans Affairs Pacific Islands Health Care System, Honolulu, Hawaii, USA

    Marianne Tanabe

Authors
  1. Gotaro Kojima
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Steve Iliffe
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Marianne Tanabe
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

GK contributed to the conception and design, collected the data, conducted statistical analyses, contributed to interpretation of the data, drafted the manuscript, revised the manuscript critically for important intellectual content, read and approved the final manuscript. SI contributed to the conception and design, contributed to interpretation of the data, revised the manuscript critically for important intellectual content, read and approved the final manuscript. MT contributed to interpretation of the data, revised the manuscript critically for important intellectual content, read and approved the final manuscript.

Corresponding author

Correspondence to Gotaro Kojima.

Ethics declarations

Ethics approval and consent to participate

This study was approved by the Institutional Review Boards of Veterans Affairs Pacific Islands Health Care System. Consent for participate was not required as the data were anonymised and collected by a retrospective chart review.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Additional file

Additional file 1:

List of 34 deficits for constructing Frailty Index. (DOC 90 kb)

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Kojima, G., Iliffe, S. & Tanabe, M. Vitamin D supplementation as a potential cause of U-shaped associations between vitamin D levels and negative health outcomes: a decision tree analysis for risk of frailty. BMC Geriatr 17, 236 (2017). https://doi.org/10.1186/s12877-017-0631-0

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12877-017-0631-0

Keywords

BMC Geriatrics

ISSN: 1471-2318

Contact us