Discussion Questions

• ■
  

  What are the overall aims of the study?

  
  
• ■
  

  How important is this question?

  
  
• ■
  

  What is the NHANES dataset?

  
  
• ■
  

  Can you describe this complicated analysis in simpler terms?

  
  
• ■
  

  What information is in the tables?

  
  
• ■
  

  What are the study’s strengths and weaknesses?

  
  
• ■
  

  What do we learn about vitamin D supplementation?

Introduction

This month, the Journal Club discussed a timely article on vitamin D levels in women of child-bearing age. Experts have been reassessing optimal vitamin D intake; specifically, whether more is better. Currently, the Food and Nutrition Board at the Institute of Medicine of the National Academies is examining dietary reference intakes for vitamin D and calcium and is due to release a report by the end of summer 2010. In addition, the American Academy of Pediatrics has raised the recommended pediatric daily intake of vitamin D, and because maternal levels influence fetal bone formation, the organization has suggested that pregnant women be screened for vitamin D deficiency.

George A. Macones, MD, MSCE, Associate Editor

Background

Macones: What are the overall aims of the study? How important is this question?

Goetzinger: While it is known that vitamin D plays a critical role in bone mineralization, emerging evidence suggests that it may play an even more crucial role in the general health of both children and adults. Vitamin D insufficiency has been associated with many adverse health outcomes, including childhood respiratory infection, type 1 diabetes, cardiovascular disease, and even overall adult mortality.

The primary aims of this study were to evaluate a representative sample of pregnant and nonpregnant women of childbearing age in the United States for vitamin D insufficiency and to determine the role of vitamin D supplementation in the prevention of vitamin D insufficiency. The potential implications of this study are of great importance to obstetricians and gynecologists as they may result in practice-changing recommendations for preconception and antepartum vitamin D supplementation. They may even influence patient counseling about long-term maternal and child health.

Study Design

Macones: What is the NHANES dataset? I have seen several papers in the obstetric/gynecologic literature that use these data, so it would be great if you could explain exactly what it is.

Harper: The NHANES (National Health and Nutrition Examination Survey) is a population-based survey that has been administered continuously since 1999 by the National Center for Health Statistics, a division of the Centers for Disease Control and Prevention. The survey, which includes about 5000 people per year, includes samples from 15 different counties across the United States with the goal of achieving a nationally representative sample. These data have been used to create national standards, growth curves, and recommendations regarding diet and nutrition.

Macones: How are data collected in NHANES?

Harper: As the survey team, which consists of a physician, medical technicians, and dietary and health interviewers, travels to new locations, households in the study area receive notification of the survey. Interviews are conducted in the participant’s home to gather information on health status, disease history, activity, and diet. The next step is a physical examination by a physician at the NHANES mobile examination center. Subjects also undergo measurement of blood pressure, height, and weight, bone densitometry, a dental examination, and vision, hearing, and breathing tests. Urine and blood samples are collected. Transportation is provided to and from the mobile facility to remove barriers to participation, and subjects are compensated for their time.

Macones: One thing to keep in mind is that NHANES is a publicly-accessible dataset. So in essence, anyone could have accessed these data and done this study.

Macones: This seems like a complicated analysis. Can you break this down for us?

Allsworth: Absolutely. Dr. Harper has spoken about the general characteristics of the survey, but the element most relevant to appropriate statistical analysis is the survey’s complex design. Basically, to assemble a cohort of participants that is small, yet representative of the US population, the survey samples participants in multiple stages. The first stage is to select the counties from which participants will be recruited and then to select households within those counties.

Once households are selected, they are contacted by interviewers who collect essential demographic information from residents to determine eligibility before final selection of individual NHANES’ participants. To account for this process, NHANES releases variables that allow researchers to account for the stratification/clustering of individuals, as well as differential probabilities of selection (sampling weights). The sampling weights reflect the unequal probabilities of selection, nonresponse adjustments, and adjustments to independent population controls. Statistical consideration of the features of study design is mandatory to produce accurate estimates and standard errors for these estimates.

Further, as a continuous survey, NHANES is currently completed in 2-year waves with the weights in each of these surveys calculated to represent the entire US population. This analysis combines 3 waves of data and therefore must adjust sampling weights so variance estimates are based on a single nationally-representative sample. Once the complex survey design was accounted for in the analysis, Ginde et al studied the sample using standard statistical methods, such as multivariable linear and logistic regression.