Page 121 - JSOM Spring 2021

retinol) decreases bone mass, decreases rates of bone forma- fractures) as separate data points. In Table 1, a RR<1 indi-
tion, increases osteoblast formation (that removes bone), in- cated that vitamin A decreased risk of fracture, while a RR>1
duces hypercalcemia (presumably mobilized from bone), and indicated vitamin A increased fracture risk. The meta-analy-
induces fragile bones and bone fractures. These actions may be ses generally indicate little difference in total fracture risk be-
mediated by direct action of vitamin A to upregulate proteins tween high versus low intake of total vitamin A or retinol. For
involved in bone resorption and/or through effects on vitamin β-carotene, the meta-analysis involving the fewest number of
D or calcium-regulating hormones. studies suggests little difference in total or any fracture risk
51
between high and low intake, but a later meta-analysis involv-
ing a greater number of studies indicated that consumption
53
Vitamin A and Risk of Bone Fractures
of high levels of β-carotene likely had a protective effect on
While animal studies suggest excessive vitamin A can produce both total fracture risk and hip fracture risk. The two me-
bone fractures through the mechanisms described above, what ta-analyses 50,51 that separated total fracture risk from hip frac-
is the evidence in humans? There have been four meta-analyses ture risk suggested that high levels of total vitamin A or retinol
examining studies that have explored associations between di- increased hip fracture risk, while high levels of β-carotene had
etary intake of various forms of vitamin A and risk of fractures no statistically significant effect.
in humans. 50–53 Meta-analysis is a technique that combines data
from studies that have a common outcome and can be analyzed It is not totally clear why high total vitamin A or high reti-
statistically. In this case, the meta-analyses analyzed data from nol might increase hip fracture risk but not total fracture risk.
studies that had investigated the influence of high and low di- Zhang et al. suggested fracture risk at high levels of total
51
etary intake of vitamin A on risk of bone fractures. All four vitamin A or retinol may be associated with bone mineral den-
meta-analyses extracted (or calculated) the risk of fractures sity and may differ by bone site. For example, one study found
at the lowest level of vitamin A consumption and compared a positive association between bone mineral density and reti-
that with the risk of fractures at the highest level of vitamin A nol intake at a number of bone sites (lumbar vertebrae, Ward’s
consumption. The authors extracted (or calculated) from vari- triangle, trochanter) but not at the femoral neck. This sug-
55
ous studies the risk ratios (RRs) and 95% confidence intervals gested to Zhang et al. that the effect of vitamin A may differ
51
(95% CIs) of fracture statistically adjusted for the largest num- at the femoral neck and that this may be mediated through
ber of covariates (e.g., age, sex, body mass index). Most of the bone mineral density. It is also possible that the geometry of
meta-analyses examined studies that had determined vitamin A the hip bone (femoral neck) makes it more susceptible to frac-
content from food-frequency questionnaires. Food-frequency ture at high levels of vitamin A. Further studies that examine
56
questionnaires attempt to measure long-term dietary intake by fractures at specific bone sites are necessary to resolve the pos-
asking for information on participants’ usual average intake sible site specificity of the association between bone fractures
over an extended period. Nutrients (including vitamin A and and vitamin A.
its analogs) are calculated from the questionnaire responses. 54
The major problem with all meta-analyses is that they depend
The earliest meta-analysis was conducted by Wu et al., who on studies using food-frequency questionnaires. On these ques-
50
examined the association between risk of total fracture or hip tionnaires, information is self-reported and suffers the usual
fractures and dietary intake of various forms of vitamin A. weaknesses associated with this method, including recall bias,
They included eight prospective studies with a total of 283,930 social desirability, errors in self-observation, and inadequate
participants. Six studies examined only hip fractures, two pro- recall. 57,58 In addition, studies included in the meta-analyses
vided total fractures, and one study provided both (n = 3,693 differ in the vitamin A levels that were considered high and
hip fractures and 3,101 total fractures). The second meta- low. This does not provide a level at which vitamin A might be
analysis by Zhang et al. was very similar to that of Wu et al. adequate versus that which might increase fracture risk.
51
50
Zhang et al. examined the association between risk of total
51
fracture or hip fractures and dietary intake of various forms of In summary, there is a suggestion that high β-carotene intake
vitamin A. One study covered in the Wu et al. meta-analysis might actually reduce fracture risk. For total vitamin A and
50
53
was excluded, but a new study was included. Thus, there were retinol, there was little effect on total fracture risk, but when
eight prospective studies including a total of 307,093 partici- hip fracture risk was examined separately, there was increased
pants; four studies involved hip fractures, three total fractures, risk from high vitamin A intake. More studies are needed that
and two both (n = 3,593 hip fractures and 18,968 total frac- separate the various forms of vitamin A and different types of
tures). A meta-analysis by Zhou et al. examined only studies fractures.
52
involving total dietary vitamin A intake and total fracture risk.
They analyzed nine studies involving 302,821 participants and Conclusion
39,901 total fractures. Charkos et al. looked exclusively at
53
dietary β-carotene intake and fracture risk. The meta-analysis Adequate levels of vitamin A are necessary for vision, gene
included nine studies, six involving hip fractures and three in- expression, immune function, reproduction, and growth.
volving any fracture, with a total of 190,545 participants. The Meta-analysis of observational studies suggest dietary con-
number of fracture cases was not reported. sumption of large amounts of vitamin A in the form of β-
carotene may reduce the risk of fractures. Meta-analyses that
Table 1 shows the results of all four meta-analyses. When have specifically examined hip fractures have shown that to-
there are more data points than studies, this is usually be- tal vitamin A or retinol consumption increases the risk of hip
cause the RRs for men and women were reported separately fractures. Until more information is available, it is advisable
in some of the reviewed studies, and these were entered into to consume vitamin A primarily from plant sources (where
the meta-analyses as separate data points. In some cases, the vitamin A is in the form of carotenoids such as β-carotene)
investigators entered different fracture types (hip versus total and avoid excessive consumption from dietary supplements,

Vitamin A and Fractures | 117