W8 - Nutritional Biomarkers Study (NBS)
This page provides study documentation for Core Study W8. For description of the specimen results, see
Specimen Results Description (open to public). Data sets of the specimen results are included in the existing WHI datasets located on the
WHI Data on this site (sign in and a completed Data Distribution Agreement are required; see details on the Data site.
Investigator Names and Contact Information
Ross Prentice, PhD; rprentic@whi.org
Introduction/Intent
The principal DM trial results contrasted intervention and control groups in respect to breast cancer, colorectal cancer, CHD, and a range of other outcomes. These contrasts did not rely on nutrient and food consumption data for their validity, but the interpretation of the results depended on the assessment of the magnitude of the difference in dietary habits between DM intervention and control groups, and on the C-I differences for specific nutrients. Aside from blood concentrations for selected vitamins in the 6% CT subsample (4.3% in the DM cohort), we have only dietary self-report data from which to "estimate" individual consumption of specific nutrients. In particular FFQs are available at baseline, 1 year, and approximately every third year thereafter for DM women. The measurement properties of FFQs, as well as food records and recalls, in populations like the WHI DM cohort are substantially unknown, due to the typical absence of adequate objective marker substudies in nutritional epidemiologic studies to date. Data from the early 2000s on this topic (e.g., from NCI's OPEN Study, and from substudies in the European Prospective Investigation of Cancer) suggested systematic biases in nutritional self-report data of a magnitude that could greatly affect both the validity of observational studies and the interpretation of intervention trials in the nutritional epidemiology areas. For example, measurement properties may depend on such study subject characteristics as body mass, age, ethnicity, and social-desirability factors, and may depend on DM intervention assignment. Study results may be greatly affected if factors related to measurement biases are also related to the clinical outcomes of interest.
A doubly-labeled water (DLW) measure of short-term energy consumption, and a urinary nitrogen measure of protein consumption, are so-called "recovery" biomarkers for which a simple measurement model evidently holds. The blood-based biomarkers for various other nutrients are "concentration" biomarkers which respond to dietary intake, but for which a more complex measurement model is expected to be needed. Under this substudy the recovery biomarkers will allow a calibration of FFQ energy and protein assessments. A sample size of 500 will be sufficient to accommodate selected sources of systematic bias in this calibration procedure. Though some further methodology development is needed, the intention is to combine the recovery biomarkers with nutrient-specific concentration markers to calibrate FFQ consumption estimates for various other nutrients. The biomarker data on the subsample will then be used in a fundamental way, in conjunction with FFQ data on the entire cohort, in explanatory analyses in the DM trial. (This same basic approach will also be used to strengthen nutritional epidemiology analyses in the OS, or nutritional analyses in the combined CT/OS.)
A Nutrient Biomarker (sub)Study involving 250 DM intervention and 250 comparison women was proposed for implementation in early to mid-2004 and completion by late 2004 to early 2005. The additional data and specimens obtained in this substudy, though not able to address all aspects of DM dietary adherence, and of DM trial interpretation, demonstrates a "cutting-edge" attempt by WHI to do the best job in these areas that technology of the time allows.
This Core study was funded and implemented accordingly through the WHI. The final sample size was 544 with 268 DM intervention and 276 DM comparison women.
This sample size is perhaps a little on the small side for WHI purposes, though large by doubly-labeled water standards, and as large as seems advisable given DLW costs. Simulation studies at the CCC indicate that reliable corrected estimates of nutrient-specific Hazard Ratios will be able to be obtained with this sample size, and some DMC members are exploring the possibility of a later companion study in the OS using external funds (see AS 218). The NBS timing, while the DM intervention was still active, was needed to determine whether the dietary reporting habits of DM intervention and comparison women differ, and to separately calibrate FFQ estimates in the two groups for energy and other nutrients. The repeat application of the protocol in a subset of the women was necessary to assess biomarker variation, an integral element of the calibration procedure. One hundred women allows for sufficient precision in the estimation of measurement error variances. In actuality, 111 women participated in the repeat application of the protocol. Study subjects were selected from 12 geographically dispersed Clinical Centers to provide a suitable distribution of baseline BMI, race/ethnicity, and age so that the calibration procedure can reliably accommodate these factors.
The overall objective of this study was to collect objective (i.e., blood and urine) measures of dietary intake to use in regression calibration models in the Women's Health Initiative that will correct the random and systematic bias of dietary self-report. The calibrated nutrient intake data will then be used for a broad range of nutrient-disease analyses from the Women's Health Initiative.
The specific aims were:
- To determine the association of self-reported measures of dietary intake from a food frequency questionnaire (FFQ) with actual intake, with the help of objective measures (biomarkers) of macronutrient intake (energy, fat, protein).
- To determine the association of self-reported measures of dietary intake from a food frequency questionnaire (FFQ) with actual intake, with the help of biomarkers of micronutrient intake (tocopherols, retinol, folate, carotenoids, B-vitamins, selenium, potassium. sodium).
Results/Findings
See Publications: 464 (methods), 624 (cancer), 646 (CVD), 831 (protein and frailty), 941 (diabetes), 945 (protein and impaired renal function). Several additional papers have been published from W8, the titles and PubMed abstract links may be found by searching the WHI Bibliography. WHI publications by study lists published WHI papers that have been generated by ancillary studies. A complete list of WHI papers is available in the Bibliography section of this website.
WHI publications by study lists published WHI papers that have been generated by ancillary studies. A complete list of WHI papers is available in the
Bibliography section of this website.