The NHANES is a series of cross-sectional studies designed to assess the health and nutritional status of adults and children in the United States [11]. NHANES uses a complex, stratified, multistage probability survey design representative of the civilian, noninstitutionalized United States population. Data are collected via household interviews, phone interviews, and physical examinations. Detailed information regarding NHANES is available elsewhere [11]. The survey was reviewed and approved by the NCHS Ethics Review Board, and participants provided written informed consent prior to participation. We selected nonpregnant adults aged ≥19 y from NHANES 2007–2018. Unweighted response rates among adults ranged from 34% (≥80 y) to 87% (16–19 y) depending on the survey cycle and age group [12]. Of the nonpregnant adults aged ≥19 y who attended the medical examination center (MEC) (n = 33,917), we excluded participants whose dietary recall status was not reliable, and participants with missing folic acid source information, leaving 31,128 participants for analyses (Figure 1).

The usual intake of dietary folate included folate from natural foods and folic acid from fortified foods. Total folate usual intake was defined as the sum of dietary folate and folic acid from supplements. We reported total folate intake in μg dietary folate equivalents (DFEs), defined as μg of natural food folate intake plus 1.7 times μg folic acid intake from fortified foods and/or supplements containing folic acid. For this analysis, we combined folic acid from CMF and ECGP into 1 group, as both represent folic acid from fortified staple foods, and USDA food codes used in the NHANES dietary interview data do not allow for sufficient granularity to separate CMF from ECGP. The 4 main folic acid sources (ECGP, CMF, RTE, and SUP) were divided into 4 mutually exclusive consumption groups:1) ECGP and CMF consumption only (ECGP/CMF only), 2) ECGP and CMF plus RTE (ECGP/CMF + RTE), 3) ECGP and CMF plus supplements containing folic acid (ECGP/CMF + SUP), and 4) folic acid from all available folic acid sources (ECGP/CMF + RTE + SUP). We modeled dietary folate usual intake from 2 24-hour dietary recalls in the NHANES dietary survey using the NCI method [13], accounting for sex, race, age, and weekday/weekend consumption. Dietary intake data for day 1 were collected in person in the MEC, and data for day 2 were obtained by telephone 3–10 d later. Dietary recall weights were computed using NHANES to adjust for differential allocation by day of the week for the dietary intake data collection and additional nonresponse not accounted for by MEC weights [11]. Dietary weights for day 1 of the recall were used for the usual intake calculation. Population distributions (e.g., median and IQR) of dietary usual intake were calculated by generating 300 pseudo people for each participant in the dataset [13].

We calculated the total intake of each nutrient as the sum of the diet (i.e., natural food folate, folic acid in fortified food) and dietary supplement using the “shrink then add” approach to avoid creating data that violate key assumptions of the shrinkage method [14]. The estimated supplement use was based on 30-d supplement use data collected via NHANES household interviews. A participant was classified as a consumer of supplements containing folic acid if he or she reported consuming any supplement with folic acid during the past 30 d. For each participant, the amount of folic acid was summed across all supplements consumed and divided by 30 to yield the average daily amount of supplemental nutrients. Vitamin B12’s usual intake was similarly calculated.

The estimated average requirement (EAR) is a daily nutrient intake value that is estimated to meet the requirement of half the healthy individuals in a group [15]. The EAR for folate is 320 μg daily of DFEs for adult men and women (≥ 19 y) [15]. The EAR for vitamin B12 for men and women aged ≥ 19 y is 2 μg daily [15]. The Institute of Medicine established the UL of 1000 μg/d folic acid for adults aged ≥19 y based on evidence that 5000 μg/d was the lowest observed adverse effect level for folic acid, which could mask the anemia of vitamin B12 deficiency and allow neurologic damage to progress when left untreated [15]. We estimated the prevalence of usual folic acid intake above the UL overall and by folic acid consumption groups.

Serum folate was calculated as the sum of individual folate forms, which were measured using microbiologic assays between 2007 and 2010 and by isotope-dilution high-performance liquid chromatography coupled to tandem MS (LC-MS/MS) between 2011 and 2018 [16–22]. Previous studies demonstrated excellent agreement between serum total folate measured using LC-MS/MS and microbiologic assays [23,24]. Whole-blood folate was measured using a microbiologic assay, whereas RBC folate was then calculated after subtracting the contribution of serum folate and normalizing the whole-blood folate data to the hematocrit [16–22]. One previous study showed that serum and RBC folate concentrations in the overall population showed minor fluctuations from 1999 to 2016 [25]. To increase statistical power, we combined data on folate biomarkers for the 6 NHANES cycles included—2007‒2018—except that we also compared biomarker concentrations before and after CMF fortification, which started in 2016. Among the 31,128 NHANES participants for this analysis, 3282 and 3169 participants had missing serum folate and RBC folate data, respectively. Serum and RBC folate deficiencies were defined as serum folate <7 nmol/L (~3 ng/mL) and RBC folate <305 nmol/L (~140 ng/mL), respectively [15]. High serum folate concentration thresholds were based on the distribution of this study population: 92 nmol/L (95th percentile), 77 nmol/L (90th percentile), and 60 nmol/L (80th percentile).

Serum vitamin B12 was measured using an electrochemiluminescence immunoassay, and MMA was analyzed using LC-MS/MS [19,20]. MMA is one of the metabolites accumulating in cobalamin (vitamin B12)-deficient cells, and blood concentrations of MMA are functional markers of vitamin B12 status [6]. Vitamin B12 and MMA data were only available for the 2 NHANES cycles included in this analysis—2011–2014. Vitamin B12 deficiency and insufficiency were defined as serum vitamin B12 concentrations <148 pmol/L and <221 pmol/L, respectively [26].

We separated age into 3 categories: 19–39, 40–59, and ≥60 y. We divided race/ethnicity based on self-reports into 5 categories: Hispanic, non-Hispanic (NH) White, NH Black, NH Asian, and others. “Hispanic” includes respondents who self-identified as “Mexican American” and self-identified as “Hispanic” ethnicity. NH participants were then categorized based on their self-reported races. Categorization for “NH Asian” became available starting in 2011. Between 2007 and 2010, NH Asians were included in the “other” category, which accounted for 4% of the total study population. Given the change in the definition of “other” during our study period and the small number of participants in this category, we did not report results for “other” separately in the Results. We defined smoking status based on serum cotinine concentrations, measured using an isotope-dilution HPLC. We classified participants with cotinine levels above 10 ng/mL as smokers [27,28] and the rest as nonsmokers. BMI was calculated as weight in kilograms divided by height in meters squared, based on body measurements data collected in the MEC. We used BMI in this analysis as a categorical variable with 4 levels: underweight (<18.5 kg/m²), normal weight (18.5 to <25 kg/m²), overweight (25 to <30 kg/m²), and obese (≥30 kg/m²).

Usual intakes were modeled using SAS, version 9.4 (SAS Institute). All other analyses were conducted in R, version 4.0.2 (R Foundation), and its survey package, version 4.0, to account for the clustered design, probabilistic selection, and nonparticipation. To analyze demographic characteristics and estimate usual intake, we combined day 1 dietary recall weights from the 6 NHANES cycles included (2007–2018), e.g., by dividing the 2-y dietary weights by 6 for folate. For biomarker concentrations, all participants aged ≥1 y who attended the MEC examination were eligible for folate biomarker data before 2017. Starting in 2017, all examined female participants aged 12–49 y were eligible for folate biomarker data, whereas one-half of the other examined participants aged ≥1 y were eligible [22]. Due to this difference in the sampling of folate biomarkers in NHANES 2017–2018, we adjusted day 1 dietary recall weights to account for the probabilistic selection and nonparticipation in the folate subsample in analyses of folate biomarker concentrations by folic acid consumption group. Briefly, all 5092 persons in the folate subsample from NHANES 2017–2018 were stratified into 53 race/ethnicity-age-gender-specific groups. Using the day 1 dietary recall weight as the base weight, a new dietary folate subsample weight was calculated within each adjustment cell to account for selection probability and nonresponse to the laboratory tests [29]. Once poststratification was completed, the weights were raked by day of the week with 2 categories: weekdays defined as Monday to Thursday, and weekends as Friday, Saturday, and Sunday. This new dietary folate subsample weight was then combined with 2-y dietary weights from the other 5 NHANES cycles for analyzing serum and RBC folate biomarker data. For vitamin B12 and MMA, because their concentrations were only available for 2 NHANES cycles (2011–2014), analyses were conducted using a 4-y combined dietary recall weight for these 2 biomarkers. Combined weights were used to calculate adjusted geometric means and prevalence in chi-squared and Wald tests.

Weighted proportions and 95% CIs of United States adults in each of the 4 folic acid consumption groups were estimated in the overall study population and by demographic characteristics (i.e., sex, age category, and race/ethnicity). Differences in proportions were evaluated with the chi-square test. We used natural log transformations on the biomarker concentrations (serum and RBC folate, vitamin B12, and MMA) and estimated the geometric means via multiple linear regression, adjusting for sex, age category, race/ethnicity, current smoking status, and weight status category based on BMI. We assessed statistical differences in the adjusted geometric means across different folic acid consumption groups using the Wald test. Logistic regressions were used to estimate the prevalence of high serum folate, serum folate deficiency, RBC folate deficiency, and serum vitamin B12 deficiency and insufficiency by folic acid consumption group or vitamin B12 supplementation status, accounting for sociodemographic characteristics. Differences in the prevalence across folic acid consumption groups or demographic subgroups were evaluated with the chi-square test. All tests were 2-tailed with a significance level of 0.05.

Table 2 shows the modeled usual daily intakes of natural food folate, folic acid, and total folate by folic acid consumption group and vitamin B12 intake by vitamin B12 supplementation status in the overall population and by demographic characteristics during 2007–2018. Natural food folate intakes were similar across the 4 folic acid consumption groups and among different demographic subgroups, with a median of 222 μg DFE/d (IQR, 175–276 μg DFE/d), which is below the EAR of 320 μg DFE daily. Among all the subgroups, NH Asian adults in the ECGP/CMF-only group had the highest folate intake from natural foods, with a median of 319 (IQR: 263–382) μg DFE/d. The median usual folic acid intake in the ECGP/CMF-only group was 134 μg/d, equivalent to 228 μg DFE/d, and this was comparable to the natural food intake in this group of 222 μg DFE/d. In the other 3 folic acid consumption groups, folic acid usual intakes were 313 (ECGP/CMF + RTE), 496 (ECGP/CMF + SUP), and 695 μg/d (ECGP/CMF RTE SUP) in the overall study population. Total folate, which is the sum of natural food folate and folic acid from fortified foods and supplements, increased with increasing folic acid intake sources or supplements containing folic acid among all demographic groups (i.e., ECGP/CMF only < ECGP/CMF + RTE < ECGP/CMF SUP < ECGP/CMF RTE SUP group). In the overall study population, 5.9% (95% CI: 5.7, 6.1) had a total folate intake below the EAR.

After voluntary CMF fortification, the median folic acid usual intake (μg/d) in the overall population was slightly lower (229 [IQR: 149, 403] during 2017–2018 compared with 237 [IQR: 152, 446] during 2007–2016), although it was slightly higher among Hispanic adults (211 [IQR: 148, 321] during 2017–2018 compared with 207 [IQR: 147, 304] during 2007–2016) (data not shown). After CMF fortification, the percentage of Hispanic adults with total folate intake below EAR remained steady at 5.5%, although it was slightly higher in the overall population at 6.1% (95% CI: 5.8, 6.5).

Folic acid and vitamin B12 supplement use were highly correlated. Among folic acid supplement nonusers, 95% did not take vitamin B12 supplements (95% CI: 94.6%, 95.7%). Among folic acid supplement users, 98% (95% CI: 97.5%, 98.5%) also took vitamin B12 supplements. The median usual vitamin B12 intake was 5.2 (IQR: 4.0, 6.7) μg/d among vitamin B12 supplement nonusers and 21.8 (IQR: 10.8, 53.8) μg/d among users (Table 2). The 25th percentiles of vitamin B12 intake for all demographic subgroups (ranging from 3.1 μg/d for NH Asian vitamin B12 supplement nonusers to 13.5 μg/d for age 60+ y vitamin B12 supplement users) were above the EAR of 2 μg daily.

Overall, 2.0% (95% CI: 1.7%, 2.3%) of adults consumed folic acid above UL during 2007‒2018. Figure 2 shows the cumulative distribution of usual intake for United States adults by folic acid source. For United States adults who did not consume supplements containing folic acid (i.e., in ECGP/CMF only and ECGP/CMF + RTE groups), 0% exceeded the UL. For supplement users, 3.9% exceeded the UL (95% CI: 3.4%, 4.5%) in ECGP/CMF + SUP group, and 10.9% exceeded the UL (95% CI: 9.2%, 12.7%) in ECGP/CMF + RTE + SUP group. Among people with usual intake above UL, ~69% were women, 80% were NH White; 36% were aged ≥60 y. After voluntary CMF fortification (from 2017 to 2018), the percentage of participants who consumed folic acid above the UL remained consistent (2.1% [95% CI: 1.6%, 2.5%] compared with 2.0% [95% CI: 1.7%, 2.3%] from 2007 to 2016). Among Hispanic adults, the percentage of participants who consumed folic acid above the UL remained the same at 1.3% before and after CMF fortification.

Higher serum and RBC folate concentrations were associated with the consumption of additional folic acid sources or supplements containing folic acid (P < 0.05) in the overall population and all demographic groups. For example, participants consuming RTE and/or supplements with folic acid had significantly higher RBC folate concentrations (nmol/L) (ECGP/CMF only: 976 [95% CI: 963, 989], ECGP/CMF + RTE: 1130 [95% CI: 1110, 1150], ECGP/CMF + SUP: 1307 [95% CI: 1283, 1332], ECGP/CMF + RTE + SUP: 1467 [95% CI: 1439, 1495]; Table 3). Among the different race and ethnicity groups, NH White adults had the highest serum and RBC folate concentrations, whereas NH Black adults had the lowest concentrations, followed by Hispanic adults. After voluntary CMF fortification starting in 2016, RBC folate concentration (nmol/L) in Hispanic adults was higher (1042 (95% CI: 1004, 1083) during 2017‒2018 compared with 995 [95% CI: 978, 1012] during 2007–2016) whereas serum folate concentration (nmol/L) was slightly lower (32.6 [95% CI: 30.8, 34.4] during 2017‒2018 compared with 34.4 [95% CI: 33.7, 35.1] during 2007–2016) (data not shown). In comparison, RBC folate concentrations (nmol/L) in the overall study population remained consistent (1117 [95% CI: 1079, 1157] during 2017–2018 compared with 1115 [95% CI: 1098, 1132] during 2007–2016) and serum folate concentration (nmol/L) was slightly lower (35.0 [95% CI: 33.0, 37.1] during 2017–2018 compared with 38.0 [95% CI: 37.3, 38.6] during 2007–2016) (data not shown).

Serum vitamin B12 concentrations (pmol/L) were significantly higher among vitamin B12 supplement users (482, 95% CI: 468, 496) than nonusers (347, 95% CI: 341, 353), which is consistent across age, sex, and race/ethnicity groups (Table 4). Among vitamin B12 supplement users, serum vitamin B12 concentration increased with age, though among nonusers, the youngest age group (19–39 y) had the highest vitamin B12 concentration. Correspondingly, MMA concentrations were lower among vitamin B12 supplement users than nonusers in the overall study population and all demographic groups (Table 4; not statistically significant for adults aged 19–39 y, NH Asian adults). In addition, MMA concentrations increased significantly with age in the overall study population as well as across vitamin B12 supplement use strata.

Percentages of adults with high serum folate increased in the order of ECGP/CMF only, ECGP/CMF + RTE, ECGP/CMF + SUP, and ECGP/CMF + RTE + ECGP consumption groups, regardless of thresholds used (Table 5). Chi-square tests showed that these percentage differences across folic acid consumption groups were statistically significant. Supplement users (in the ECGP/CMF + SUP and ECGP/CMF + RTE + SUP groups) had higher percentages of high serum folate than nonusers. This increase in percentages of adults with high serum folate between supplement users and nonusers remained true across demographic groups (i.e., by sex, age category, and race/ethnicity). In the overall study population, there was a significantly higher proportion of females than males with high serum folate, and the proportion with high serum folate increased with age. Folate deficiency was rare in the study population—mean serum folate deficiency was ≤0.7% and mean RBC folate deficiency was ≤ 1.2% in the 4 folic acid groups (Table 5). For the overall population, there was no difference between males and females in the proportion of serum or RBC folate deficiency, though proportions differed significantly for different age groups and race and ethnicity groups. For the overall population, folic acid consumption groups did not have a significant association with the proportion of RBC folate deficiency but had a significant association with the proportion of serum folate deficiency.

The mean vitamin B12 deficiency was 3.6% in the overall population and 4.4% and 2.3% between vitamin B12 supplement nonusers and users, respectively (Table 6). Except for those aged 40–59 y and NH Black adults, vitamin B12 supplement users had a significantly lower proportion with vitamin B12 deficiency than nonusers in the overall study population as well as across demographic groups. Likewise, vitamin B12 supplement users had a significantly lower percentage of vitamin B12 insufficiency (4.5 to 8.2% among vitamin B12 supplement users across demographic groups compared with 12.4 to 18.5% among nonusers).

Our study has several strengths. First, we used available data between 2007 and 2018 from a large nationally representative sample of United States adults, with oversampling of certain population subgroups by age and race/ethnicity. As a result, we had a large sample size of over 30,000 participants, which in most scenarios provided stable estimates in subgroups that account for only a small percentage of the population. For example, we reported data for NH Asian adults as its own category, which became available in NHANES starting in 2011. Second, folate and vitamin B12 biomarkers were used to evaluate dose-response across folic acid consumption groups or vitamin B12 supplement use status. Third, we applied the updated NCI method to estimate the usual daily intake of folate and vitamin B12. The NCI method allows the estimation of usual intake distributions of daily and episodically consumed components for populations and sub-populations and the prediction of individual intakes, which allows for the incorporation of individual and time-dependent covariates [13]. Nutrient usual intakes from natural and fortified foods and supplements were combined to estimate total usual intakes using the preferred “shrink then add” method.

Potential limitations include the inability to examine temporal associations due to the cross-sectional design of the survey. Dietary data were self-reported and collected at the time of the medical examination or shortly thereafter. Supplement data were collected during household interviews, in which participants were asked about their use of dietary supplements during the past 30 d. Thus, to estimate the total intake, we combined data obtained from 2 different instruments. We did not analyze the usual folic acid intake from CMF alone, as USDA food codes used in the NHANES dietary interview data did not allow for the separation of CMF from ECGP.