Strengths and limitations
The study takes advantage of a vitamin D fortification policy implemented in Denmark almost fifty years ago, and nests this “societal experiment” in a setting where the effect of conditions from conception (and before) for development of certain diseases can be followed among individuals from birth through adulthood. The study is also possible due to complete registration of every citizen in Denmark via a civil registration number. This number can be linked, on an individual level, to the Danish birth, patient and medical registries; social and ethnic registries; and clinical and other large databases. The comprehensive linkage of different datasets makes it possible to study the mediation, confounding or modification of effects of vitamin D for selected conditions through the development of sophisticated models in which individuals can be followed for periods of up to 50 years. Moreover, the large number of subjects that will be followed ensures that the effects of vitamin D exposure in relation to the timing of the seasons and gestational development can also be examined. Nevertheless, the proposed study design implies several uncertainties, including so-called ecological fallacies, addressed below.
Fortification supplied 15% of vitamin D intake on average, which may be considered low, particularly when the sun exposure is not ample. However, vitamin D fortification is especially beneficial when vitamin D status is low , e.g. when the previous 6 months of UVB radiation from sun exposure in Denmark was too weak to induce vitamin D conversion in the skin . The use of 50,000-100,000 individuals from adjacent birth years in the general population studies of events like fractures and diagnosis of T1D, a pivotal strength of this study, makes it possible to detect even minor effects of vitamin D.
Food disappearance statistics show that margarine intake decreased and low fat milk slightly increased during the period covered. Nevertheless, changes of this nature are not expected to influence results, because the intervention group and the control group will have been born in adjacent years. Moreover, any societal changes succeeding the initiation of mandatory fortification in 1961 or its termination in 1985 may influence the intervention and control groups similarly. Thus, we find it unlikely that changes in occurrence in the outcomes associated with the introduction and cessation of fortification will lead to spurious conclusions.
We do not collect any individual data on vitamin D intake. In contrary, the individuals in the database are unselected in relation to vitamin D exposure as well as to later disease occurrence. While assuming no ecological fallacies, this in fact may be considered a very strong feature of the present study. Neither confounding from external or internal factors, nor the lack of individual data on vitamin D intake, is a prerequisite for the study, as the influences of lifestyle differences (e.g. use of infant formula, individual diet intake, use of supplements) or societal changes (e.g. in breast feeding, fortified infant formulas, sun exposure) are independent of the vitamin D intervention.
We acknowledge, however, that the differences in prenatal exposure by any subsequent age imply a more simple difference in duration of exposure. Therefore, we will conduct robustness analyses where the age scale is adjusted so that the duration of exposure becomes the same and the difference then only whether the exposure was commenced/ended prenatal or postnatal.
Moreover, there might be regional and urban/rural differences in vitamin D intake. As information on individual’s place of residence is available from Statistics Denmark, it will be included into analyses.
As diet is usually different between families with higher socioeconomic status and those with lower status , maternal socioeconomic status may have affected margarine/milk intake and thus exposure to vitamin D. Information on socioeconomic position, education and ethnicity is available from Statistics Denmark and can be controlled for. Adjustments will also be made for maternal age, parity, gestational age and birth weight, which is information available from the birth and patient registries and Statistics Denmark (since 1973, 1977 & 1980, respectively).
Finally, the uncertainties related to lack of knowledge about real vitamin D status in mother and offspring may be eliminated by assessing the measurement of infant blood vitamin D levels in those developing the three disease entities and comparing to time and gender-matched controls to examine actual exposure differences at birth.
Unlike many other nutritional deficiencies, low vitamin D status is prevalent among industrialized populations, and, according to Danish health authorities, about half of all adults have vitamin D insufficiency (below 50 nmol/l) . Adequate exposure to UVB sunlight is paramount importance for supply of vitamin D, but oral intake, augmented by fortification and supplementation, is necessary to maintain baseline stores, particularly in winter when sunlight is limited [8, 33, 34]. Thus, as it is clear that vitamin D status of Danes needs to be improved, the Danish health authorities currently debate, whether to resume vitamin D fortification of food . The complete lack of knowledge on the general health effects of fortified foods is an important issue in this regard. Our results will significantly increase current understanding of the importance of early vitamin D intake for long-term health and the occurrence of diseases. The results will also provide solid quantitative justification whether to resume vitamin D fortification. From a public health perspective, the potential to prevent common chronic diseases via low-cost, simple and safe food fortification is an attractive option.
Additionally, we expect that the results of the study will open new research opportunities for other scientists and fields, e.g. human intervention studies to determine the level of vitamin D supplementation necessary to reduce susceptibility to diseases later in life and concomitant basic research exploring the biological mechanisms that may be involved. Because it is possible to obtain information in Denmark on individual health and social parameters via linkage to health registries using a unique personal identification number, many opportunities for collaboration will arise in relation to health aspects, e.g. growth and pubertal development, possibility related to vitamin D exposures in early life, perhaps even before conception, in addition to the outcomes studied. If and when our nationwide model is in operation for the three diseases studied, the approach used and the inherent methodologies applied can easily be adapted to study other health outcomes.