Table 1.
Selected Limitations to be Considered Related to Study Design in Vitamin D and Infection.
| Observational Studies |
| Independent variable: vitamin D status (serum 25(OH)D concentration) |
| Confounding variables: associated with both 25(OH)D and the outcome |
| Sample size must be adequate to adjust for known confounding variables |
| Seasonal variation of 25(OH)D and respiratory illnesses |
| Those with chronic illness have less sunlight exposure to produce 25(OH)D |
| Obesity is associated with both lower 25(OH)D and adverse outcomes |
| 25(OH)D may be inversely related to inflammatory markers in severe illness |
| Racial groups with dark skin may have lower 25(OH)D and different outcomes than Caucasian whites |
| Vitamin D fortified foods increase 25(OH)D, but other nutrients in food may be related to outcomes |
| 25(OH)D level is related to genes involved in vitamin D transport and metabolism, which could be linked to other genes affecting disease outcomes |
| Laboratory variation in 25(OH)D measurements and methodology requires standardization |
| Selection bias: Those with 25(OH)D measurements available were selected for study. They likely differ from those who did not have 25(OH)D measured. |
| Healthy user bias: Those who take vitamin D may be healthier than those who do not. |
| Post-hoc analysis: A pre-specified hypothesis is needed to correctly apply significance testing. Analyses of multiple outcomes, subgroups, and 25(OH)D cut points can lead to erroneous conclusions (statistical type 1 error). |
| Publication and reporting bias: Journals are more likely to publish studies that show potential benefit of an intervention than studies with negative results. |
| Preprint server publications are not peer-reviewed and results should be considered preliminary. |
| Randomized Controlled Trials |
| Independent variable: vitamin D supplementation (dose of vitamin D) |
| Control group may also take vitamin D, potentially attenuating any observed benefit |
| Inadequate number of persons with vitamin D deficiency (or vitamin D deficient subjects excluded) |
| Dose and duration of vitamin D may be related to benefit |
| Formulation of vitamin D may be related to benefit (e.g., cholecalciferol vs. calcifediol) |
| Daily vs. bolus dosing of vitamin D may have different metabolic effects |
| Timing of vitamin D administration in relation to illness onset, stage of disease, or illness severity |
| Interaction of vitamin D with other treatments for disease (e.g., corticosteroids) |