Multiple imputation to evaluate the impact of an assay change in national surveys

Population RBC folate concentrations can be accurately estimated from measured whole blood folate, measured hemoglobin, and predicted serum folate-cross-sectional data from the NHANES 1988-2010

BACKGROUND

RBC folate (RBF) is an indicator of folate status and risk of neural-tube defects. It is calculated from whole blood folate (WBF), serum folate (SFOL), and hematocrit (Hct). SFOL and/or Hct are sometimes unavailable; hemoglobin (Hb) is generally available in surveys.

OBJECTIVES

We assessed the ability of different RBF approximations to generate population data in women aged 12-49 y.

METHODS

Using SFOL, RBF, Hct, Hb, and mean corpuscular Hb content (MCHC) from prefortification (1988-1994) and postfortification (1999-2006, 2007-2010) NHANES we applied 6 approaches: #1) assume SFOL = 0; #2) impute SFOL (population median); #3) impute Hct (population median); #4) estimate Hct (Hb/MCHC); #5) assume SFOL = 0 and estimate Hct; and #6) predict SFOL (from WBF) and estimate Hct. For each approach, we calculated the paired percentage difference to the "true" RBF and estimated various statistics.

RESULTS

For 2007-2010 (unweighted data), the median relative difference from "true" RBF was lowest for approaches #2 (-0.74%), #4 (-0.96%), and #6 (-1.15%), intermediate for #3 (-3.36%), and highest for #5 (4.96%) and #1 (5.78%). The 95% agreement limits were smallest for approach #1 (2.33%, 13.0%) and largest for #3 (-20.8%, 11.3%). Approach #2 showed concentration-dependence (negative compared with positive differences at low compared with high RBF). Using weighted data, we found similar patterns across approaches for mean relative differences by demographic subgroup for all 3 time periods.

CONCLUSIONS

We obtained the best agreement between estimated and "true" RBF when we predicted SFOL using a regression equation obtained from a subset of samples (approach #6). Alternatively, the consistent overestimation of RBF when assuming SFOL = 0 (∼6%) could be addressed by adjusting the data (approach #5). Similar observations for pre- and postfortification periods suggest applicability to low and high folate status situations, but should be confirmed elsewhere. To estimate RBF, at least WBF and Hb are needed.

Folate status in the US population 20 y after the introduction of folic acid fortification

BACKGROUND

Enriched cereal-grain products have been fortified in the United States for >20 y to improve folate status in women of reproductive age and reduce the risk of folic acid-responsive neural tube birth defects (NTDs).

OBJECTIVES

Our objectives were to assess postfortification changes in folate status in the overall US population and in women aged 12-49 y and to characterize recent folate status by demographic group and use of folic acid-containing supplements.

METHODS

We examined cross-sectional serum and RBC folate data from the NHANES 1999-2016.

RESULTS

Serum folate geometric means increased from 2007-2010 to 2011-2016 in persons aged ≥1 y (38.7 compared with 40.6 nmol/L) and in women (35.3 compared with 37.0 nmol/L), whereas RBC folate showed no significant change. Younger age groups, men, and Hispanic persons showed increased serum and RBC folate concentrations, whereas non-Hispanic black persons and supplement nonusers showed increased serum folate concentrations. The folate insufficiency prevalence (RBC folate <748 nmol/L; NTD risk) in women decreased from 2007-2010 (23.2%) to 2011-2016 (18.6%) overall and in some subgroups (e.g., women aged 20-39 y, Hispanic and non-Hispanic black women, and supplement nonusers). After covariate adjustment, RBC folate was significantly lower in all age groups (by ∼10-20%) compared with persons aged ≥60 y and in Hispanic (by 8.2%), non-Hispanic Asian (by 12.1%), and non-Hispanic black (by 20.5%) compared with non-Hispanic white women (2011-2016). The 90th percentile for serum (∼70 nmol/L) and RBC (∼1800 nmol/L) folate in supplement nonusers aged ≥60 y was similar to the geometric mean in users (2011-2014).

CONCLUSIONS

Blood folate concentrations in the US population overall and in women have not decreased recently, and folate insufficiency rates are ∼20%. Continued monitoring of all age groups is advisable given the high folate status particularly in older supplement users.

Vitamin D status in the United States, 2011-2014

BACKGROUND

Vitamin D is important for bone health; in 2014 it was the fifth most commonly ordered laboratory test among Medicare Part B payments.

OBJECTIVES

The aim of this study was to describe vitamin D status in the US population in 2011-2014 and trends from 2003 to 2014.

METHODS

We used serum 25-hydroxyvitamin D data from NHANES 2011-2014 (n = 16,180), and estimated the prevalence at risk of deficiency (<30 nmol/L) or prevalence at risk of inadequacy (30-49 nmol/L) by age, sex, race and Hispanic origin, and dietary intake of vitamin D. We also present trends between 2003 and 2014.

RESULTS

In 2011-2014, the percentage aged ≥1 y at risk of vitamin D deficiency or inadequacy was 5.0% (95% CI: 4.1%, 6.2%) and 18.3% (95% CI: 16.2%, 20.6%). The prevalence of at risk of deficiency was lowest among children aged 1-5 y (0.5%; 95% CI: 0.3%, 1.1%), peaked among adults aged 20-39 y (7.6%; 95% CI: 6.0%, 9.6%), and fell to 2.9% (95% CI: 2.0%, 4.0%) among adults aged ≥60 y; the prevalence of at risk of inadequacy was similar. The prevalence of at risk of deficiency was higher among non-Hispanic black (17.5%; 95% CI: 15.2%, 20.0%) than among non-Hispanic Asian (7.6%; 95% CI: 5.9%, 9.9%), non-Hispanic white (2.1%; 95% CI: 1.5%, 2.7%), and Hispanic (5.9%; 95% CI: 4.4%, 7.8%) persons; the prevalence of at risk of inadequacy was similar. Persons with higher vitamin D dietary intake or who used supplements had lower prevalences of at risk of deficiency or inadequacy. From 2003 to 2014 there was no change in the risk of vitamin D deficiency; the risk of inadequacy declined from 21.0% (95% CI: 17.9%, 24.5%) to 17.7% (95% CI: 16.0%, 19.7%).

CONCLUSION

The prevalence of at risk of vitamin D deficiency in the United States remained stable from 2003 to 2014; at risk of inadequacy declined. Differences in vitamin D status by race and Hispanic origin warrant additional investigation.