Predicting Vitamin D Deficiency Accurately Using a Simple Questionnaire
Laboratory tests usually offer the best insights into nutritional status for the micronutrients. Unfortunately, laboratory tests tend to be expensive, take time to administer, and carry a slight risk of injury. In 2010 for example, Medicare, the US government’s national social insurance program for people aged over 65, paid over $220 million for vitamin D tests and $75 million for vitamin B12 tests, according to the Office of Inspector General. The actual cost of the vitamin D test to insurers varied considerably but was generally in the range of $30-$50 per test. Questionnaires, on the other hand, can be administered remotely, do not involve invasive procedures (with perhaps the exception of a paper cut), and can be evaluated very easily.
A recent publication by Sohl and associates looked at whether it was feasible to assess vitamin D deficiency through simple patient characteristics. They used data from 1106 subjects participating in a longitudinal study on aging conducted in the Netherlands. 46% of participants had low-normal vitamin D levels of less than 50 nmol/l and 17% were deficient with levels under 30 nmol/l. They used a statistical technique that identified various patient characteristics that had an influence on vitamin D levels, such as age, season, physical activity and use of dietary supplements. Ultimately, the authors assigned a score to each patient characteristics to give a score that predicted whether patients had low-normal or deficient vitamin status.
For each year of age over 65, the chance of vitamin D deficiency increased by 2 on a scale from 6 to 204. The other 9 factors found to increase the chance of vitamin D deficiency were smoking, alcohol consumption less than two drinks per day, winter or spring season, a lack of vitamin D supplementation, no cycling, not being able to use public transport, no gardening, not able to remember the year, and use of medications.
The model that assessed low-normal levels of vitamin D also used age as a predictor, increasing by 1 on a scale of 2 to 97 for each year of age above 65. A further 12 predictors influenced the model; additional to the predictors for vitamin D deficiency female sex, obesity, no physical activity, poor appetite, and living alone, while limitations in transport use and the ability to remember the year were no longer good predictors.
Around 86% of people who are vitamin D deficient according to the model will actually be vitamin D deficient. Additionally, 16% of people who are found to be sufficient will actually be deficient.
This short list of questions can be used to advise patients in the general population about their likely risk of vitamin D deficiency and need for supplementation. The list of criteria is appropriate to the Dutch population for which cycling is very prevalent and no vitamin D can be synthesized in the skin in the winter time. For other populations in other locations, the questionnaire would have to be adjusted to reflect likely sun exposure, seasonal effects and normal outdoor activities.
Evelien Sohl, Martijn W Heymans, Renate T de Jongh, Martin den Heijer, Marjolein Visser, Thomas Merlijn, Paul Lips, and Natasja M van Schoor. Prediction of vitamin D deficiency by simple patient characteristics. Am J Clin Nutr 2014 ajcn.076430; First published online February 19, 2014. doi:10.3945/ajcn.113.076430
Department of Health and Human Services. Office of Inspector General. Comparing Lab Test Payment Rates: Medicare Could Achieve Substantial Savings. June 2013. https://oig.hhs.gov/oei/reports/oei-07-11-00010.pdf