1. Introduction
Suboptimal iron status negatively impacts cognitive function in women of reproductive age, representing a significant health problem in light of the prevalence of iron deficiency. National surveys within North America report rates of iron deficiency in women aged 20–49 years of 15% (United States) [
1], 19%–27% (Mexico) [
2], and 9% (Canada) [
3]. An even higher rate of iron deficiency among female university students in the United States (30%–50%) [
4,
5,
6,
7] poses the risk of compromised academic achievement due to deficiency-related cognition impairment.
In our recent study [
8], of 42 female college students with varying levels of body iron, women with lower
vs. higher body iron status took significantly longer to strategize their movements on a test of planning and working memory. This study was the first known to demonstrate that iron deficiency without anemia can impair cognitive function in adults. These findings are in agreement with those from observational and intervention studies performed on women of reproductive age [
9,
10,
11,
12,
13,
14,
15]. Important to demonstrating causality, iron supplementation trials show parallel improvements in hematologic and cognitive variables [
9,
11,
12,
14].
While iron supplementation is an inexpensive means to treat iron deficiency, it is associated with side effects that compromise compliance [
16,
17,
18]. The current project used a whole-food approach in correcting the cognitive impairment associated with iron deficiency. Observational studies and intervention trials support the benefit of increased meat intake in maintaining higher body iron status in premenopausal women [
7,
19,
20], but no known investigations have examined the effect of beef consumption on cognitive performance in iron-deficient women. The present study tested the hypothesis that moderate consumption (here defined as 3 oz, 3 times weekly) of beef improves iron status and cognitive function in young women. Moderate intake of beef, a popular source of bioavailable iron, was selected as a reasonable dietary intervention that could be readily adopted by women. The control treatment of various non-beef foods contained levels of kcals and protein similar to those provided by the beef lunches.
3. Results
3.1. Demographics and Baseline Measurements
Figure 1 depicts the flow of participants through the study. Most of the 43 women who completed the study were of Caucasian race (
n = 37), with the remaining being of mixed (
n = 2 in each lunch group) or Latino (
n = 1 in each lunch group) descent. At baseline, mean BMI, age, menstrual cycle duration and flow intensity, and measures of iron status and blood lipids were not significantly different between the beef and non-beef lunch groups (
Table 2 and
Table 3). Baseline body iron ranged from −2.73 to 11.64 mg/kg in the group of 43 women and was negative for one and three women in the beef and non-beef group, respectively. Four women in the beef group and 6 women in the non-beef group had abnormal serum ferritin (<14 ng/mL) and/or high TfR (>4.4 mg/L) at baseline. Baseline Hb was below the altitude-adjusted cutoff of 123 g/L [
38] for 2 and 3 women in the beef and non-beef groups, respectively.
Figure 1.
Volunteer flow diagram.
Figure 1.
Volunteer flow diagram.
Table 2.
Baseline characteristics of intervention participants.
Table 2.
Baseline characteristics of intervention participants.
Variable | All Women (n = 43) | Beef (n = 22) | Non-beef (n = 21) |
---|
Age (years) | 21.14 ± 0.38 | 21.70 ± 0.62 | 20.56 ± 0.43 |
Body weight (kg) | 64.80 ± 1.52 | 64.43 ± 2.36 | 65.18 ± 1.98 |
BMI (kg/m2) | 23.27 ± 0.51 | 23.76 ± 0.77 | 22.76 ± 0.67 |
Table 3.
Hematology and iron status measures (n = 43).
Table 3.
Hematology and iron status measures (n = 43).
Variable * | Baseline | Midpoint (week 8) | Endpoint (week 16) | Absolute change (endpoint-baseline) | P value Main effect time | P value Main effect baseline measure | P value time × baseline measure |
---|
Beef | Non-beef | Beef | Non-beef | Beef | Non-beef | Beef | Non-beef |
---|
Hb (g/L) † | 143.4 ± 2.4 | 138.5 ± 0.3 | 145.6 ± 0.2 | 141.5 ± 0.2 | 146.2 ± 0.2 | 140.7 ± 0.2 | 2.8 ± 1.8 | 2.1 ± 1.9 | <0.0001 | <0.0001 | <0.0001 |
Hct (%) | 41.6 ± 0.7 | 40.2 ± 0.7 | 42.5 ± 0.6 | 40.8 ± 0.5 | 42.4 ± 0.5 | 40.9 ± 0.5 | 0.8 ± 0.5 | 0.8 ± 0.7 | <0.0001 | <0.0001 | <0.0001 |
RBC (millions/mm3) | 4.7 ± 0.1 | 4.7 ± 0.1 | 4.8 ± 0.1 | 4.7 ± 0.1 | 4.8 ± 0.1 | 4.7 ± 0.1 | 0.1 ± 0.1 | 0.0 ± 0.1 | <0.0001 | <0.0001 | <0.0001 |
MCV (µm3) | 88.2 ± 0.7 | 87.1 ± 1.1 | 88.2 ± 0.8 | 86.9 ± 1.0 | 88.1 ± 0.8 | 87.3 ± 1.0 | −0.1 ± 0.3 | 0.2 ± 0.6 | <0.05 | <0.0001 | <0.05 |
MCH (pg) | 30.4 ± 0.3 | 30.0 ± 0.5 | 30.3 ± 0.4 | 30.1 ± 0.5 | 30.4 ± 0.4 | 30.1 ± 0.5 | 0.0 ± 0.2 | 0.0 ± 0.3 | NS | <0.0001 | NS |
MCHC (%) | 34.5 ± 0.2 | 34.4 ± 0.2 | 34.3 ± 0.3 | 34.6 ± 0.2 | 34.5 ± 0.2 | 34.4 ± 0.3 | 0.0 ± 0.2 | 0.0 ± 0.3 | <0.05 | <0.0001 | <0.05 |
Body iron (mg/kg) †† | 6.5 ± 0.6 | 5.2 ± 0.8 | 6.5 ± 0.7 | 5.0 ± 3.8 | 7.0 ± 0.6 | 6.0 ± 0.5 | 0.5 ± 0.4 | 0.9 ± 0.5 | <0.001 | <0.0001 | <0.0001 |
Serum ferritin (ng/mL) | 33.6 ± 4.8 | 28.3 ± 4.2 | 34.0 ± 4.8 | 28.5 ± 4.6 | 41.6 ± 6.8 | 29.2 ± 3.0 | 8.0 ± 3.9 | 0.9 ± 2.7 | <0.0001 | <0.0001 | <0.0001 |
TfR (mg/L) † †† | 3.1 ± 0.2 | 3.8 ± 0.4 | 3.1 ± 0.2 | 3.8 ± 0.3 | 3.2 ± 0.2 | 3.4 ± 0.2 | 0.1 ± 0.1 | −0.4 ± 0.2 | <0.0001 | <0.0001 | <0.0001 |
Serum iron (µg/L) | 1027.8 ± 89.1 | 1034.7 ± 115.8 | 921.8 ± 57.6 | 915.2 ± 94.0 | 955.5 ± 66.2 | 1064.8 ± 73.6 | −72.0 ± 97.5 | 30.0 ± 112.9 | <0.0001 | <0.0001 | <0.0001 |
Tf (mg/dL) | 285.2 ± 9.0 | 290.0 ± 9.5 | 285.5 ± 9.6 | 297.4 ± 9.8 | 285.5 ± 10.5 | 288.3± 10.8 | 0.3 ± 6.2 | −15.5 ± 16.7 | NS | NS | NS |
Tf saturation (%) | 31.4 ± 3.3 | 31.5 ± 3.9 | 27.6 ± 2.4 | 26.7 ± 3.3 | 28.8 ± 2.5 | 31.1 ± 2.8 | −2.6 ± 2.9 | −0.4 ± 3.2 | <0.0001 | <0.0001 | <0.0001 |
3.2. Effects of Intervention and Baseline Iron Status on Change in Iron Status
There were significant effects of time, baseline measurement, and their interaction on all iron status and hematologic measures except Tf and MCH (
Table 3). Lunch group had a significant main effect on TfR and Hb and the interaction of lunch group and baseline measurement was significant for body iron and TfR.
Post-hoc comparisons using the Tukey-Kramer adjustment showed no significant differences between lunch groups within time points for any iron status measure. For the group of 43 women, body iron and ferritin measures were significantly higher at endpoint vs. baseline and midpoint [body iron: p = 0.050 (endpoint vs. baseline) and p = 0.020 (endpoint vs. midpoint); ferritin: p = 0.042 (endpoint vs. baseline) and p = 0.021 (endpoint vs. midpoint)]. Differences between baseline and endpoint for Hb and Hct did not reach significance (p = 0.087, p = 0.088, respectively).
Baseline iron status had a significant effect on absolute change (endpoint-baseline) in iron status measures, such that women with lower vs. higher iron status displayed a greater magnitude of improvement in iron levels. Changes in body iron and ferritin were significantly affected by baseline levels (body iron: p < 0.0001); ferritin: p < 0.0001. There was a significant effect of lunch group (p = 0.0021), baseline TfR (p < 0.0001), and their interaction (p = 0.0005) on change in TfR. This interaction reflected a stronger relationship between baseline TfR and change in TfR in the non-beef vs. beef group: In correlational analysis, baseline TfR was significantly inversely correlated with change in TfR for the non-beef group (Spearman ρ = −0.66, p = 0.001) but not the beef group (ρ = −0.18, p = 0.42).
Seventeen women were classified as ferritin responders to either lunch intervention (
Table 4). Responders
vs. non-responders displayed a significantly greater percent increase in ferritin (
p < 0.0001), percent decrease in TfR (
p = 0.006), percent increase in Hb (
p = 0.009), and absolute increase in body iron (
p < 0.0001). The number of responders in the beef and non-beef groups was 10 and 7, respectively. There were no significant interactions between responder class and lunch group on iron status measures.
Table 4.
Iron status measures in intervention women classified as ferritin responders or non-responders.
Table 4.
Iron status measures in intervention women classified as ferritin responders or non-responders.
Variable * | Ferritin Responder (n = 17) | Ferritin Non-responder (n = 26) | p value |
---|
Ferritin, percent change | 100.75 ± 17.51 | −12.74 ± 4.08 | p < 0.0001 |
TfR, percent change | −8.23 ± 3.69 | 5.90 ± 3.16 | p = 0.006 |
Body iron, absolute change (mg/kg) | 2.67 ± 0.36 | −0.65 ± 0.24 | p < 0.0001 |
Hb, percent change | 5.23 ± 1.76 | 0.11 ± 0.98 | p = 0.009 |
3.3. Effect of Intervention and Iron Status on Cognitive Test Performance
Verbal Recognition Memory (VRM): Lunch group and session had significant main effects on free recall of correct targets, with more words recalled by women in the beef vs. non-beef group (p = 0.007), and during the second vs. first session (p = 0.008). The higher word recall by women in the beef group was not statistically significant when compared within sessions (Tukey’s post-hoc, p > 0.05). Body iron and ferritin had significant negative effects on free recall women, with fewer words recalled by women with higher vs. lower body iron (p = 0.002) and ferritin (p = 0.001). Recognition of targets and distractors were not affected by lunch group, body iron, or session.
Changes in VRM scores between ferritin responders vs. non-responders were not significantly different.
One Touch Stockings of Cambridge (OTS): Mixed models ANOVA with repeated measures showed significant effects of body iron (p = 0.032) and ferritin (p = 0.015) on latency to correct choice for the higher-difficulty tasks (moves 4–6): Women with higher iron status required less time to make a correct choice. The effect of body iron (p = 0.058) but not ferritin (p = 0.024) was attenuated when all moves (1–6) were included in analyses. Mean number of choices to correct choice and number of problems solved on first choice were not significantly affected by body iron or ferritin. Lunch group did not have a significant effect on test measures.
Women classified as ferritin responders
vs. non-responders showed a significantly greater improvement in latency to first choice between baseline and endpoint for moves 1–6 (
p = 0.007;
Figure 2). Mean number of choices to correct choice and number of problems solved on first choice were not significantly affected by responder class.
Figure 2.
Change in response time by ferritin response group; Change (endpoint-baseline) in One-Touch Stockings of Cambridge latency to first choice for move categories 1–6 for ferritin responders (n = 17) and non-responders (n = 26). Main effect of responder group, p = 0.007.
Figure 2.
Change in response time by ferritin response group; Change (endpoint-baseline) in One-Touch Stockings of Cambridge latency to first choice for move categories 1–6 for ferritin responders (n = 17) and non-responders (n = 26). Main effect of responder group, p = 0.007.
Spatial Working Memory (SWM): Latency to first response was significantly affected by body iron (p = 0.012), lunch group (p = 0.0003), box number (p < 0.0001), and session (p = 0.031): Greater speed was seen in women with higher vs. lower body iron, in the non-beef vs. beef group, in the less difficult (lower) box number trials, and in session 2 vs. 1. Token search time was significantly affected by body iron (p = 0.018), group (p = 0.003), box number (p = 0.008), and session (p = 0.001): Greater speed was demonstrated in women with higher body iron, in the non-beef group, in the lower box number tasks, and in session 2. Post-hoc analysis showed faster latency to first response and token search time in the non-beef group only in session 1 (p < 0.05). Women with higher vs. lower ferritin showed faster speed in latency to first response (p = 0.038) and token search time (p = 0.017).
SWM strategy showed a significant effects of group (p = 0.018), box number (p < 0.0001) and session (p = 0.048), with better strategy demonstrated in women in the non-beef vs. beef group, in trials with fewer box numbers, and during the session 2 vs. 1. Post-hoc analysis showed no significant differences between groups within sessions, (p > 0.05). No main effects of body iron or ferritin on strategy were seen. Error scores showed no effect of body iron, ferritin, or group.
Ferritin responders demonstrated significantly greater improvement in strategy between baseline and endpoint than ferritin non-responders (
p = 0.007,
Figure 3).
Figure 3.
Change in strategy by ferritin response group; change (endpoint-baseline) in Spatial Working Memory strategy score for six- and eight-box problems in ferritin responders (n = 17) and non-responders (n = 26). A lower score and larger negative change indicate better strategy. Main effect of responder group, p = 0.007; * p < 0.05 between-group comparison for eight-box problem.
Figure 3.
Change in strategy by ferritin response group; change (endpoint-baseline) in Spatial Working Memory strategy score for six- and eight-box problems in ferritin responders (n = 17) and non-responders (n = 26). A lower score and larger negative change indicate better strategy. Main effect of responder group, p = 0.007; * p < 0.05 between-group comparison for eight-box problem.
Rapid Visual Processing (RVP): Latency to respond showed significant effects of ferritin (p = 0.023) and block (p < 0.0001): Faster speed was seen in women with higher vs. lower ferritin and in the earlier vs. later blocks. The effect of body iron on latency to respond approached but did not reach significance (p = 0.09). Neither lunch group nor session had significant effects on latency to respond.
Lunch group, session, and block had significant main effects on total hits: More hits were achieved in the beef vs. non-beef group (p = 0.0038) and during session 2 vs. 1 (p < 0.0001) and fewer hits were achieved as blocks progressed from 5 to 7 (p = 0.043). Consistently, total misses were significantly lower in the beef vs. non-beef group (p = 0.006) and in session 2 vs. 1 (p < 0.0001). Correct rejections were significantly higher in the beef vs. non-beef group (p = 0.009) and in session 2 vs. session 1 (p < 0.0001). The higher number of hits and correct rejections and lower number of misses by women in the beef vs. non-beef group was not statistically significant when compared within sessions (p > 0.05). Neither body iron nor ferritin had a significant main effect on RVP hits, misses, or correct rejections.
Ferritin responders vs. non-responders tended to show greater improvement in correct rejections (p = 0.056). No other effects on change in RVP scores were seen for responder class.
3.4. Relationship of Iron Status and Cognitive Function in All Women (n = 54) with Baseline Measures
Baseline demographic and iron status data from the larger group of 54 women who began the study were similar to those from the group of 43 women who completed the intervention (
Table 5). None of the 11 women who did not complete the study had below-normal iron status. Like the group of women who completed the study, the group inclusive of all women with baseline data displayed an effect of iron status on word recall, response speed, and spatial working memory.
Table 5.
Baseline characteristics of all women enrolled.
Table 5.
Baseline characteristics of all women enrolled.
Variable | Women (n = 54) |
---|
Age (years) | 21.70 ± 0.41 |
Body weight (kg) | 64.99 ± 1.30 |
BMI (kg/m2) | 23.47 ± 0.44 |
Body iron (mg/kg) | 6.20 ± 0.46 |
Ferritin (ng/mL) | 32.93 ± 2.88 |
TfR (mg/L) | 3.30 ± 0.17 |
Hb (g/L) | 140.94 ± 1.49 |
VRM: There were significant main effects of body iron (p = 0.034) and ferritin (p = 0.026) on free recall of correct target words, with more words recalled by women with lower vs. higher iron status. Iron status did not have a significant effect on recognition of targets or distractors.
OTS: There was a significant main effect of body iron (p = 0.010) and ferritin (p = 0.020) on latency to first choice: Women with higher iron status demonstrated greater speed. Neither the number of problems solved on first choice nor the number of choices to correct showed significant effects of body iron or ferritin.
SWM: Faster response time was displayed in women with higher iron status: Effect of body iron and ferritin on latency to first response, p = 0.047 and p = 0.060, respectively. There was a tendency toward shorter token search time in women with higher iron status: Effect of body iron and ferritin, p = 0.051 and p = 0.066, respectively. Strategy scores were significantly better in women with higher iron status: Effect of body iron and ferritin, p = 0.003 and p = 0.005, respectively.
RVP: The effect of ferritin on latency to respond approached significance (p = 0.07). Otherwise, there were no significant effects of body iron or ferritin on RVP scores.
3.5. Effect of Intervention on Blood Lipids (n = 43)
There was no significant effect of lunch group or time on blood lipids; nor did lunch intervention have a significant effect on change in blood lipids between baseline and endpoint.
3.6. Dietary Assessment
Women underreported dietary intake according to a cut-off point of 1.2 for the plausible ratio of energy intake to calculated basal metabolic rate [
39,
40]: ratio = 0.94 ± 0.06 (FFQ 1) and 0.76 ± 0.02 (FFQs 2–5). Mean reported kcal intake was higher for FFQ 1 [1612 ± 137 kcal (6.745 ± 0.573 MJ)] than for FFQs 2–5 [1310 ± 40 kcal (5.481 ± 0.167 MJ)] (main effect of FFQ number,
p = 0.0006). Mean absolute and adjusted macro- and micronutrient intakes did not differ significantly between lunch groups at baseline. Baseline adjusted iron intake was significantly correlated with body iron (ρ
= 0.33,
p = 0.015;
n = 43). No correlations were seen between body iron and adjusted intake of meat iron, heme iron, zinc, vitamin C, or protein at baseline.
Women reported lower adjusted protein intake during the intervention compared to baseline (
p = 0.004), and women in the non-beef reported lower intakes of adjusted heme iron and adjusted meat iron during the intervention compared to those in the beef group (
p < 0.05;
Table 6). Burger and steak intake frequency was significantly different between baseline and the intervention (
p < 0.0001), with higher intakes reported by women in the beef group (adjusted
p < 0.01) and lower intakes reported by women in the non-beef group (adjusted
p < 0.05).
Table 6.
Reported dietary intake for lunch groups.
Table 6.
Reported dietary intake for lunch groups.
Lunch Group | Baseline | Intervention |
---|
| Adj protein (g) | Adj iron (mg) | Adj heme iron (mg) | Adj meat iron (mg) | Adj protein (g) | Adj iron (mg) | Adj heme iron (mg) | Adj meat iron (mg) |
Beef | 44 ± 2.1 | 7.3 ± 0.3 | 0.5 ± 0.1 | 1.0 ± 0.2 | 41 ± 1 * | 7.3 ± 0.2 | 0.6 ± 0.0 † | 1.2 ± 0.1 † |
Non-Beef | 45 ± 1.8 | 7.5 ± 0.4 | 0.5 ± 0.1 | 1.1 ± 0.1 | 40 ± 1 * | 7.7 ± 0.2 | 0.3 ± 0.0 | 0.4 ± 0.0 |
4. Discussion
This study tested the efficacy of moderate beef consumption in ameliorating cognitive impairment associated with low iron status. The current findings support a positive relationship between iron status and cognitive function, but they do not show that moderate intake of beef improves iron status or cognitive performance in women with decreased iron status to a greater degree than non-beef protein foods. Young women with higher vs. lower body iron and ferritin performed better on tests of planning speed and spatial working memory and women classified as ferritin responders vs. non-responders displayed more pronounced improvements in tests of planning speed, spatial working memory strategy, and attention following the intervention.
These results replicate and extend findings from our prior observational investigation showing a significant relationship between planning speed and body iron in young college women [
8]. Both studies tested women of similar age, BMI, and education level and found slowed performance on a task of planning ability (Stocking of Cambridge in the present study and Tower of London in the previous study) in women with lower body iron. The current study further reports a significant effect of iron status on response latencies in tests of spatial working memory and rapid visual processing, and search time and strategy in a test of spatial working memory. The results are consistent with those reported by Murray-Kolb and Beard [
14], who found iron deficiency-related impairments in multiple domains of cognitive performance in college-educated women 18–35 years of age. These investigators showed that severity of iron depletion correlated with degree of cognitive impairment in three groups of women: iron-deficient anemic, iron-deficient non-anemic, and iron-sufficient. Following supplementation with 60 mg elemental iron/day for 16 weeks, ferritin responders
vs. non-responders showed significant improvements in attention, learning, and memory task accuracy, but not task speed. Increased speed was seen only in women classified as Hb responders. The different findings between the present study and that of Murray-Kolb and Beard regarding ferritin response and improved task speed might be related to differences in participant population and treatment. The current study included fewer women with a more narrow range of iron status levels and a less potent iron dose. Regarding the finding of an inverse effect of iron status on verbal free recall in the present study, this is inconsistent with existing evidence of improved memory with higher iron status [
41,
42] and cannot be readily explained in the context of the overall results presented here. The absence of this effect in the present study’s ferritin responder and non-responder groups suggests the finding might be an artifact.
Other intervention studies of females of reproductive age demonstrate corrections in cognitive impairment and iron status in response to iron supplementation [
9,
11,
12,
13,
43]. The doses of elemental iron used in these studies ranged from 18 to 195 mg/day and whether naturally iron-rich foods provide sufficient iron to induce similar responses is not known to have been reported prior to the present study. Dietary interventions using iron-fortified foods have shown efficacy in improving iron status in adults. In a 12-week intervention in young adult non-anemic women, Hoppe
et al. [
44] showed comparable increases in body iron and ferritin resulting from either consumption of blood-based crisp bread (35 mg iron, 27 mg being heme) or supplementation with iron (35 mg and 60 mg). Karl
et al. [
45] demonstrated that consumption of iron-fortified bars (2 per day, 27.9 mg iron each) protected iron-deficient anemic female soldiers against training-associated declines in iron status better than a control food. Using a lower level of fortification, Haas
et al. [
46] substituted high-iron rice for local rice for 9 months in a group of Filipino young women and showed a significant increase in body iron in non-anemic participants. This study demonstrated that a relatively small increase in dietary iron intake (1.41 mg/day increase above control) over an extended period can significantly improve iron status.
Meat-based interventions appear less frequently in the literature than supplementation and fortification studies, but their results indicate benefit in supporting iron levels in women. Lyle
et al. [
47] showed that a high food − iron + meat supplement diet protected iron status more effectively than iron supplementation in university women participating in an exercise program. Notably, total iron intake in the high food − iron + meat group was 11.8 ± 2.8 mg, significantly less than that in the 50 mg iron-supplemented group (57.8 ± 2.4 mg) The present study and that of Lyle
et al. differ in several respects, yet both studies show improved iron status in women consuming ~1.5 servings of meat per day. Further, beef was shown to be superior to poultry and fish in improving serum ferritin in intervention studies of adolescents [
48] and iron-deficient women [
49]. Lastly, systematic reviews support the efficacy of dietary iron interventions for improving iron status. A meta-analysis by Casgrain
et al. [
50] reported that iron supplementation significantly improved iron status and that supplementation form (pills, meat, and fortified food) was not a significant modifier of ferritin or TfR response. Altogether, findings from the present and previous studies indicate that food-based approaches, including regular, moderate meat consumption, are beneficial in protecting and increasing iron nutriture.
Parallel changes in blood iron levels and cognitive task performance are suggestive of a causal relationship between iron and brain function. Brain activity, measured by electroencephalography (EEG), is altered as a function of iron status. Wenger
et al. [
51,
52] recently reported significant relationships between iron status, attention and memory abilities, and EEG patterns in adolescents and women. Correcting iron-deficiency anemia with iron supplementation can reverse EEG abnormalities [
53] and the associated impairments in cognitive task performance [
12]. These studies build on early findings by Tucker
et al. [
54,
55] of associations between EEG patterns, cognitive ability, and iron status. The mechanisms underlying altered brain activity in relation to iron status are not fully understood but may involve disruptions in neuroendocrine function. Iron is a cofactor for enzymes synthesizing catecholamines, serotonin, and the thyroid hormones and alterations in their levels and activity are seen in iron-deficient humans and animals [
56,
57].
The unexpected finding of improved iron status in both lunch groups seems to suggest that regular (3 times weekly), nutrient-dense lunches improved the women’s diet quality enough to affect blood iron indicators. Women’s mean body weight increased 0.8 kg, which was not statistically significant (
p > 0.9) but could indicate greater kcal and nutrient intake during the intervention. This possibility is difficult to substantiate since women underreported dietary intake and physical activity was not measured. Poor diet quality in college students is common [
58,
59,
60] and the study lunches might have displaced less nutritious food normally consumed by the women in this intervention. The present study was designed to test the effect of the total nutrient package of beef, rather than the effect of a particular nutrient, on iron status and cognitive function. While adjusted intakes in the beef
vs. non-beef protein groups were greater for heme iron, meat iron, and absolute servings of steak and hamburger, no differences between the groups were seen in adjusted protein and adjusted total iron intakes. The lack of observable differences in iron status between the lunch groups could also be related to the moderate level of beef intervention used in this study, which was chosen to simulate a reasonable dietary strategy. A specific benefit of consumption of beef over other foods on iron status might require larger intakes over a longer duration to observe. A two-fold higher incidence of negative iron balance has been observed in British women reporting no
vs. high consumption of red meat [
61].
Strengths of this study include a participant population of women similar in education level, age, physical activity level, and BMI who were tested for iron status and cognitive function during the same menstrual phase and under the same fasted (blood draw) and fed (controlled pre-cognitive testing snack) conditions. Also, the study lunches were prepared for the participants and two of every three lunches were consumed in the presence of research staff. Lastly, one research staff member blinded to participant iron status performed all cognitive tests.
Limitations include the homogeneous participant population, which restricts generalizability of results to other groups. Further, this study did not restrict enrollment to women with low iron status, which would have allowed for improvements in iron status to be observed more clearly. While the women were instructed on allowable beef intake outside the study and dietary intake was monitored by monthly FFQs, this factor was not strictly controlled in these free-living participants. Also, the women were not blinded to treatment allocation. The women’s experience with computerized games could have affected performance on the cognitive tests [
62], but this was not measured. Lastly, a training effect of repeated cognitive testing is known to result in improved performance irrespective of intervening treatment. However, this effect was common across women and session was taken into account during data analysis.