The calcium and vitamin D controversy

Abstract

Areas of the world where vitamin D levels are low for months of the year and intakes of calcium are high have a high prevalence of osteoporosis and cardiovascular disease. This suggests a public health message of avoiding calcium supplements and increasing vitamin D intake. No message could be more welcome as vitamin D can be given as a bolus while calcium must be taken daily and may be poorly tolerated. This approach is based on no evidence from intervention studies. Randomized controlled trials (RCTs) suggest that vitamin D given with calcium elicits a small reduction in fracture risk and deaths. This has not been demonstrated for D given alone. The cardiovascular safety of calcium and vitamin D (CaD) supplements is difficult to ascertain due to weaknesses in RCT designs and adjudication that cannot be remedied by subanalysis. Moreover, no major new RCTs are in process to provide better evidence. It remains unclear that calcium from dietary sources has health advantages over supplements. Benefits may be confined to patients with poor nutritional intake and the small effects at societal levels may be derived from large effects in a small number of patients. This has been impossible to confirm given the limited information about baseline vitamin D and calcium status at entry into trials. Future intervention studies should carefully capture baseline characteristics as these may determine the strength of the response, and make more efficient use of randomization strategies allowing subsequent disassembly or subanalyses while maintaining balancing. Though large clinical RCTs currently evaluate the effects of higher vitamin D doses (equivalent to 50–83 µg/d) there is no current research effort regarding the calcium controversy. In the absence of such studies it is not possible to provide clinicians with evidence-based recommendations regarding the best use of CaD supplementation.

Introduction

To the clinician advising patients on skeletal health and nutrition, no subject is as full of controversy and strongly held opinions as calcium and vitamin D (CaD) supplements. Public opinion span from CaD having modest, if any, effects, be they benefits or harm, to supplements being either the elixir of long life or the certain path to ischaemic heart disease and death. There is a high prevalence of osteoporosis and cardiovascular disease in Northern Europe and North America, where vitamin D levels are low for several months of the year and dietary intake of calcium is twice that of South Asia. This may suggest that the best public health message would be to avoid calcium supplementation but make sure patients are vitamin D replete. No message could be more welcome to patients and their doctors as vitamin D supplements can be spaced and given as a bolus while calcium must be taken daily and often triggers discomfort or constipation. For readers accustomed to the rigour of RCTs for new pharmaceutical products it is important to know that by comparison, most CaD trials were conducted on very little funding with limited resources to document and adjudicate safety outcomes. Some studies also had major limitations in design and attempting to correct this by post-analysis has not always added more clarity. For some outcomes, the vitamin D controversy has been subject to so many meta-analyses and systematic reviews based on few trials that researchers felt a moratorium on meta-analysis might be the way forward.¹ The following is intended as a narrative review where the key focus is on the scientific disputes and the conflicts between studies. Cochrane reviews and systematic reviews have been cited for background context outside the areas of major dispute, purely to conserve space and limit the number of references. The aim in the following is to provide an interpretation of the state of the controversy and provide directions for research that could resolve the scientific disagreement.

Calcium

Maintenance of an adequate calcium level in the circulation in response to a variable and low supply of calcium from the environment was a crucial evolutionary challenge for early terrestrial vertebrates.² Modern mammals, including man, share a finely tuned parathyroid apparatus optimized for correcting incipient hypocalcaemia. Our mechanism for eliminating an excess of calcium through renal filtration comes with a risk of nephrolithiasis and nephrocalcinosis that is only partly offset by calcium-sensing receptor (CaSR)-mediated inhibition of arginine vasopressin(AVP)-driven water reabsorption. The dilute urine reduces the risk of precipitation of calcium salts but also results in polyuria and ultimately dehydration.³ Calcium absorption occurs chiefly in the ileum due to the slower transit time though active, saturable, strongly 1,25-dihydroxyvitamin D (1,25-(OH)_2D)-regulated calcium absorption is more pronounced in the proximal intestine.⁴ Nonsaturable, paracellular absorption is more prevalent in the distal intestine. Hence, calcium absorption is not tightly regulated. Curbing active calcium absorption in acute hypercalcaemia is a slow process due to the half-life of many hours of 1,25-(OH)_2D once formed. In women in their twenties, a large oral dose of 1200 mg calcium as calcium citrate results in an increase in S-Ca of 4% over placebo where calcium formate elicits an increase of 10% while calcium carbonate, the most widely used calcium supplement in trials, led to a placebo level change in serum calcium.⁵ In the elderly, the calcaemic response between calcium carbonate, calcium citrate and skimmed milk appears to be identical.⁶

Bone mineral density

There are two immediate reasons why one would not expect calcium supplements to boost bone formation in the adult skeleton. First, due to the tight regulation of serum calcium the concentration of calcium in the Extracellular Circulating Volume (ECV) remains close to constant even given extreme variation in intake. Second, mineralization of bone matrix is an active, not a passive, process. Reversal of a calcium-deficient diet or severe calcium malabsorption will lead to a gradual return of parathyroid hormone (PTH)-driven bone resorption to normal levels with the potential for a bone mineral density (BMD) increase by a mechanism paralleling that of anti-resorptive drugs. Calcium supplementation should have transient effects on BMD and the effect may well be limited to persons with inadequate intake or absorption of calcium. In a recent meta-analysis, Tai and colleagues identified 51 studies of BMD and calcium supplements, of which 13 studies used calcium in combination with vitamin D.⁷ A small increase in BMD of 1–2% was observed with calcium supplementation with most of the effect developing in the first year of supplementation with little addi-tional gains. There was no appreciable difference between calcium-alone trials and the CaD trials. Similarly, small transient effects were seen in studies of dietary intervention as an alternative to calcium supplements. A similar meta-analysis for vitamin D supplements suggested marginal BMD effects only, with marked heterogeneity and potential publication bias favouring positive studies.⁸

Fractures

In 2008, working on the assumption that individual patient characteristics could perhaps explain the huge difference between the conclusions of meta-analyses favouring higher vitamin D doses⁹ and meta-analyses favouring vitamin D combined with calcium,^10,11 we approached investigators who had published vitamin D and calcium fracture RCTs with >1000 participants and obtained patient-level data. Trials using a vitamin D bolus generally had recruited older participants while CaD trials had mainly targeted women in their sixties; the largest trial of this category being the Women’s Health Initiative (WHI) CaD study.¹² Patient-level analyses reported by our DIPART (vitamin D Individual Patient Analysis of Randomized Trials) collaboration indicated a modest reduction in hip and other fractures with low dose vitamin D (about 10 µg daily) given with a calcium supplement, while we found no risk reduction with vitamin D given alone irrespective of dose.¹³ This conclusion has been confirmed in Cochrane analyses, most recently in 2014.¹⁴

Cardiovascular concerns

In 2007, the WHI investigators reported on the cardiovascular outcomes on the large CaD trial embedded within the WHI study.¹⁵ In brief, more than 36,000 women with a mean age of 62 years were randomized to calcium carbonate with vitamin D or placebo. About half the participants used oestrogen therapy. The investigators found no differences between CaD and placebo in any of the >10 cardiovascular outcomes specified for analysis. A total of 801 patients were diagnosed with incident myocardial infarction (MI). The rate of MI was 3.2 per 1000 patient years with CaD intervention and 3.1 per 1000 patient years (py) in the control group (p = 0.50). The total incidence of MI, revascularization and death from cardiovascular causes was 7.2 per 1000 py in the CaD group compared with 6.6 in the control group (p = 0.10), an increase of 8% though nominally not statistically significant. A subsequent analysis by the Auckland group¹⁶ reasoned that outcomes could have been obscured by use of personal calcium supplements among participants in the CaD trial and performed a subanalysis on WHI data stratifying for this baseline characteristic. A total of 804 patients were classified as having had an MI and 753 a clinical MI (Figure 1). Overall, nine cardiovascular outcomes (unique or combined) were pursued, of which one was statistically significant, namely the combined risk of MI and revascularization, which occurred at a rate of 7.2 per 1000 py in the CaD group versus 6.2 in the placebo group (Figure 1). Some aspects of the findings are surprising. First, the incidence rates for these outcomes are lower, not higher, in the group that already used own calcium supplements, where readers would have expected a further increase in the cardiovascular risk when two supplements rather than one supplement were given. The increase in MI and revascularization by 16% attributed to CaD in the no-personal supplement arm is paralleled exactly by a 16% reduction in mortality in the personal supplement arm, where it would have been biologically more plausible to see an increased mortality, not a reduction.¹⁷ It is important to be aware that balancing of baseline risk factors by randomization is no longer maintained¹⁸ once the study becomes subdivided by preference for taking a calcium supplement, hence reducing the level of evidence to that of an observational study and reintroducing the issue of unmeasured confounders. Additional analyses have been performed¹⁹ to examine the potential interaction between key baseline variables (e.g. age, body mass index, hypertension, calcium intake, diabetes) and cardiovascular outcomes on the WHI dataset and found the association to be robust across such strata. This does not, however, eliminate concerns that a difference in base risk between the two arms may have been created by the break in randomized balancing in unmeasured confounders. A recent meta-analysis restricted to women with a mean cohort age of 50+ years was unable to confirm an excess risk of coronary heart disease with CaD [risk ratio (RR) 1.01, 95% confidence interval (CI), 0.95–1.08, p = 0.69], but this study too, does not provide weighty evidence against an association between calcium given alone and the risk of MI, (RR 1.37; 0.98–1.92, p = 0.07).²⁰ The major difference between this meta-analysis and the analysis by Bolland¹⁶ is whether the entire WHI cohort is included or only women not taking personal supplements.

Figure 1.

Post-hoc analysis of WHI calcium vitamin D intervention trial, divided into categories according to baseline use of personal calcium supplement. Numbers are events per 10,000 person years. The randomization procedure was not designed to ensure balancing of patients according to this criterion. As categories were based on individual choice of the study subject with no subsequent balancing procedure possible this should be viewed as an observational analysis done within what was originally designed as a RCT. Note the lower rate of MI and revascularization events in users of personal calcium supplements irrespective of study CaD intervention or no intervention, and the apparent opposite effects of CaD supplementation on cardiovascular events and mortality in the group not already using supplements. Numbers from the Bolland analysis.¹⁶

CaD, calcium and vitamin D; MI, myocardial infarction; RCT, randomized controlled trial; WHI, Women’s Health Initiative.

Following the Bolland meta-analyses^16,21 there has naturally been a drive towards improving calcium intake through the diet as this is perceived as being more natural and safer than supplements. This may not be evidence-based. Dairy products are sources of not only calcium but also phosphate and energy in the form of lactose and milk fat. NHANES III data suggest a sex-discrepant effect of serum calcium, with higher serum calcium levels in women and low serum calcium in men being linked to Ischaemic Heart Disease mortality. However, there was no association between daily calcium intake from diet and supplements and cardiovascular death.²² In the US NIH (National Institutes of Health) AARP study of 388,000 men and women there was no association between dietary calcium and cardiovascular disease (CVD) death, while supplements were associated with elevated risk in men but not in women.²³ In some studies association with cardiovascular morbidity and mortality have opposite directions. In an observational study of 23,980 German adults,²⁴ there was a nonsignificant trend towards fewer MI events in patients in the highest quartile of dietary calcium intake [hazard ratio (HR) = 0.92 (0.66–1.27)] yet also a trend towards higher CVD mortality [1.30 (0.87–1.94)]. Interestingly, calcium intakes up to 1350 mg/day have been associated with decreased CVD morbidity and mortality (Table 1) in a recent analysis from the Melbourne Collaborative Cohort Study of 41,500 men and women,²⁵ while two Swedish observational cohort studies found increasing all-cause mortality at HR = 1.15 (1.13–1.17) in women and 1.03 (1.01–1.04) in men for each additional glass of milk.²⁶ A new meta-analysis of 22 observational studies revealed no significant association between calcium intake and all-cause or CVD mortality, but it was noted that studies with <10 years of observation time tended to show a risk reduction with higher calcium intake while studies of longer duration generally pointed towards increased risks.²⁷ It is possible that this may be due to greater weight gain over time in persons with a high intake of dairy products or the discrepancy may be influenced by period effects (e.g. food and exercise habits as well as more effective cardiovascular treatment).

Table 1.

Comparison of absolute calcium intake and risk of cardiovascular outcomes between the Melbourne²⁵ and Uppsala²⁶ observational cohort studies.

	Uppsala (61,400 women)		Melbourne (41,514 women and men)
	Range	Total mortality	Median	Total mortality
Q1	<600	1.38 (1.27–1.51)	641	1
Q2	600–699	1	785	0.93 (0.84–1.04)
Q3	1000–1399	1.00 (0.96–1.04)	899	0.88 (0.79–0.99)
Q4	>1400	1.40 (1.17–1.67)	1076	0.86 (0.76–0.98)
	Range	Cardiovascular disease deaths	Median	Cardiovascular disease deaths
Q1	<600	1.63 (1.42–1.87)	641	1
Q2	600–699	1	785	0.92 (0.72–1.17)
Q3	1000–1399	1.01 (0.94–1.09)	899	0.91 (0.71–1.19)
Q4	>1400	1.49 (1.09–2.02)	1076	0.83 (0.62–1.12)
	Range	Stroke deaths	Median	Stroke deaths
Q1	<600	1.50 (1.14–1.97)	641	1
Q2	600–699	1	785	0.74 (0.46–1.20)
Q3	1000–1399	1.02 (0.89–1.17)	899	0.73 (0.44–1.23)
Q4	>1400	0.73 (0.33–1.65)	1076	0.61 (0.34–1.09)

Mortality

The largest CaD intervention study to address all-cause mortality was the WHI CaD study,^12,28 which is described in more detail above. There was no increase in mortality among participants randomized to CaD (HR = 0.88; 95% CI, 0.75–1.03). The relative importance of calcium versus vitamin D on mortality was assessed at patient level in the collaborative DIPART study of trials with a total of 70,528 patients, which included the WHI study; vitamin D alone did not affect mortality, but risk of death was reduced if vitamin D was given with calcium (HR = 0.91; 95% CI, 0.84–0.98).²⁹ Though the most recent Cochrane analysis reports reduced mortality with vitamin D3 in their conclusions, subanalyses were only nominally significant for D3 given with calcium, not if given alone.³⁰ Other meta-analyses have reported a small but nominally significant risk reduction with vitamin D with or without calcium (HR = 0.96; 95% CI, 0.93–1.00, p = 0.04),³¹ or a reduction of the same size that was not statistically significant.¹⁴

Renal stones

Like cardiovascular risk, the risk of nephrolithiasis with CaD supplements has also been surrounded by confusion. For example, if analysed by the intention-to-treat principle, the WHI CaD trial showed an increase in renal stone events of the same magnitude as the number of fractures avoided.

Meta-analyses report an increased risk of renal stones with the combination of vitamin D and calcium though not for vitamin D supplementation itself.³²

Current status of the field

Both the benefits and harms of CaD supplementations appear to be very modest indeed in the general population (Table 2). It is reasonable to speculate that benefits may be confined to patients with poor nutritional intake or uptake of these two important nutrients and that the small effects seen at societal levels are derived from large effects in a small number of subjects. This has been very difficult to confirm, however, given the paucity of high quality information about baseline vitamin D and calcium nutritional status at entry into the intervention trials. Because CaD supplements are cheap, limited funding has been available to thoroughly evaluate the harm and benefits compared with what would have been provided had they been novel drugs for which patents could be obtained. There is a risk of over- or misinterpreting association studies where vitamin D and calcium supplements may be used either by patients in poor health to perhaps improve daily activities and quality of life or in persons who are very health aware and who make a range of life choices associated with lower morbidity and mortality. It is clear, of course, to most clinicians and scientists that the effects, good or bad, of supplementation can only truly be determined by monitoring the outcomes of intervening in one group and leaving a comparable group without intervention. Observation is human nature and has served us well in some fields but not in all. We know that film stars may prefer a certain brand of instant coffee or skin product, yet there is no evidence that stardom follows from choosing the same brands.

Table 2.

Author’s interpretation of current evidence base from controlled intervention studies.

	Calcium alone	Vitamin D alone	CaD combined
BMD	Small gain	No effect	Small gain
Fractures	No effect	Conflicting evidence	Small risk reduction in the elderly
Cardiovascular events	Conflicting evidence	Conflicting evidence	Conflicting evidence
All-cause mortality	No effect	No effect	Small risk reduction in the elderly

BMD, bone mineral density; CaD, calcium and vitamin D.

Directions for future research

First, in the design of future intervention studies it would be well worth incorporating, not only careful scrutiny of baseline patient characteristics as they may determine the strength of the response, but also the efficient use of randomization strategies that allow subsequent disassembly or subanalyses while maintaining balancing on key clinical characteristics such as calcium intake, prior fractures and vitamin D serum levels.

Second, the bulk of ongoing research focuses on higher dose vitamin D and not on calcium supplements. More than 700 trials of vitamin D that are currently recruiting or planning to recruit patients are registered at clinicaltrials.gov at the time of writing with an additional 240 ongoing trials registered with the European trials register. The largest ongoing trials are summarized in Table 3. By contrast, a search of these trials registers found no planned or ongoing vitamin D trials with calcium outside the specific areas of hyperparathyroidism, colon cancer prevention and prevention of renal stones in conditions linked to citrate deficiency. Hence, despite the combination of CaD having shown the strongest anti-fracture and survival benefits but also the strongest suspicion of renal and cardiovascular potential for harm, there is currently a complete lack of clinical trial activity to resolve this central controversy and inform clinical guidance to patients and public health strategies.

Table 3.

Notable ongoing large vitamin D RCTs.

Study and reference	Inclusion started year	Duration, years	Population	Intervention	Daily dose equivalent	Outcomes
VITAL33	2010	5	US Age 50+ (men) 55+ (women) n = 25,874	Factorial: Cholecalciferol 2000 IU/d Fish oil placebo	50 µg	Cancer and cardiovascular disease.
VIDA34	2011	3.3	NZ Age 50–84 years n = 5,110	Cholecalciferol 100,000 IU/month	83 µg	(1) Cardiovascular disease (2) Acute respiratory infection, falls, nonvertebral fractures
D-Health35	2014	5(+5)	Australia Age 65–84 n = 21,315	Cholecalciferol 60,000 IU/month placebo	50 µg	(1) Mortality (2) Cancer

NZ, New Zealand; RCT, randomized controlled trial; US, United States.