Re: "the vitamin D hypothesis revisited: race-based disparities in birth outcomes in the United States and ultraviolet light availability".

Abstract

Thayer (1) set out to test the hypothesis that variation in ultraviolet (UV) light–induced vitamin D synthesis is an important contributor to birth outcome inequalities between blacks and whites. The author used US birth certificate data from 2007 to calculate state-specific differences in rates of low birth weight and preterm birth between non-Hispanic black and non-Hispanic white mothers, and she related this state-level disparity measure to the UV index, an estimate of statewide ambient UV light exposure (2). Thayer found larger black-white gaps in poor birth outcomes in states with greater UV index scores, and she concluded that UV radiation–induced vitamin D status is not an important contributor to racial/ethnic disparities. We agree that analyses are needed that evaluate the contribution of safe and inexpensive interventions (e.g., moderate increases in sunlight exposure or vitamin D intake) in reducing birth outcome inequalities. However, these ecological data are not suitable for this task. The UV index used in this analysis is not an accurate measure of statewide ambient UV light exposure. Of the states with available data, state-level UV index scores were calculated as the mean of daily UV index scores in 2007 for only 1 city in 43 states and for 2 or 3 cities in the remaining 5 states. This approach ignores variability in statewide spatial factors known to affect ground-level UV radiation, such as urbanicity, altitude, amount of ozone, and cloud cover (3). It also overlooks population mobility, including seasonal migrations (which can dramatically alter UV light exposure (4)) and residence in early life (which may be a critical window for UV light and/or vitamin D exposure (5, 6)). Statewide UV index scores are also poor proxies for personal UV light exposure because they disregard key individual characteristics, including time spent outdoors (duration and time of day), sun-protective behaviors, and variation in skin pigmentation (5). Even when data on ambient and personal UV light exposures are rigorously collected and representative of an individual's true UV light exposure, research has shown that solar radiation plays a very minor role in determining individual vitamin D status (7, 8)—possibly because sun avoidance is so prevalent in the United States (4, 8). It is also challenging to make accurate inferences using ecological low birth weight and preterm birth data. Low birth weight is widely recognized as a heterogeneous outcome that does not reliably reflect perinatal or long-term population health (9). Although preterm birth rates more accurately reflect infant risk, there is substantial state-level variation in preterm births that occur as a result of differences in sociodemographic and medical/obstetrical risk factors; availability and accessibility of quality health care; reporting, accuracy, and completeness of birth data; different definitions of fetal death and livebirth; and provider practice variation (10). Without sufficient information on these numerous confounding factors, between-state comparisons of preterm birth are questionable, at best (10). Ecological data can be of tremendous use for some research questions, but the accuracy of the results is determined by the scale of the ecological exposure, the quality of the ecological data, and the analytical methods. In assessing the relationship between UV light–induced vitamin D synthesis and birth outcome disparities, Thayer relies on crude state-level measures of both the exposure and outcome; uses what is likely too large a spatial scale (i.e., states); and relies on aspatial analytical methods that ignore regional correlations. Better ecological data sources and analytical methods exist to address this research question. Moreover, large, diverse pregnancy cohorts with banked biospecimens for vitamin D assay and rich confounder data are widely available and may more accurately estimate the nature of the relationship between vitamin D and birth outcome inequalities.