Table 1.
Description and results of cross-sectional studies examining iron status or anemia and dimensions of academic performance and cognitive function.
| Author (Year, Country) | Study Design | Study Population 1 | Exposure or Nutrition/Anemia Status | Learning/Cognition Outcome Assessed and Measure | Key Results 2 |
|---|---|---|---|---|---|
| Abalkhail et al. (2002, Saudi Arabia) [18] | Cross-sectional study |
n = 800 school children aged 9–21 y Age: 14 ± 2.6 y |
Prevalence of anemia was assessed using Hb estimated by Refletron (Boehringer Mannheim). | School grades were classified according to the national school grading classification. | Higher percentage of anemia among students with marks < 70% (fail/pass) than students with good, very good or excellent grades (no statistical analyses). |
| Anuar Zaini et al. (2005, Malaysia) [19] | Cross-sectional study |
n = 1405 standard four primary students, aged 9–10 y Age: 9.7 ± 0.5 y (68% were 10 y) |
Prevalence of anemia from finger-prick blood samples with a HemoCue®. | School grades in Malay language (comprehension and written), math, English, and science. RCPM for intelligence. | Severe anemics had higher scores in Malay language comprehension and writing, math, and English; lower science scores and RCPM. Moderate anemics had lower science scores and RCPM (no statistical analyses). |
| Aquilani et al. (2011, Italy) [20] | Cross-sectional study |
n = 48 high school girls aged 14–15 y Age: 14.6 ± 0.7 y |
Daily iron intake (mg) was assessed by a student-kept weighted 7 day food record and analysis was conducted using a computer system designed by the research group. | School achievement was assessed by mid-year curriculum performance in written math, oral math, and written Italian. | Students with satisfactory school performance had higher iron intakes than those with unsatisfactory scores **. Iron intake was significantly positively correlated with written math (r = 0.43 ***), oral math (r = 0.40 **) and written Italian (r = 0.37 *). |
| Cai and Yan (1990, China) [21] | Cross-sectional study |
n = 58/478 middle school students aged 13–15 y | Prevalence of IDA was assessed using a 5 mL venous blood sample for Hb, ferritin, and free erythrocyte porphyrin. | IQ was tested using the Bourden–Wisconsin test. School marks in Chinese, math, and English class were taken from school records. | No significant difference in scores for verbal IQ, performance IQ, total IQ, or school marks by subject for students with IDA compared with those without IDA. |
| Carruyo-Vizcaíno et al. (1995, Venezuela) [22] | Cross-sectional study |
n = 213 middle-class adolescents ages 12–17 y Age: 13.8 ± 1.3 y |
Prevalence of ID and anemia was determined using a CBC, SI, TIBC, TS, and ferritin. | IRA was the ratio between the number of subjects approved over the total number of subjects taken. The final GPA from grades of each subject from three periods of the school year | IRA scores positively correlated with ferritin levels < 20 ug/L (r = 0.411 **); negatively correlated with Hb, iron, and ferritin in the total population (not shown); anemic males scored worse than non-anemic males * and all non-anemic adolescents **; no differences in average scores for any other blood parameters. Hb in females was negatively correlated with GPA; no differences in final GPA were found for any blood parameters. |
| Dissanayake et al. (2009, Sri Lanka) [23] | Cross-sectional study | n = 188 Sinhalese students age 13–15 y | Prevalence of ID and IDA was assessed by Hb, determined by the indirect cyanmethemoglobin method, and ferritin. | RPM for intelligence. School marks in science, math, social science, and Sinhala language, and total marks. | No significant relationship was observed between IQ or school performance and iron status or severity of ID. |
| El Hioui et al. (2012, Morocco) [24] | Cross-sectional study |
n = 259 primary school children aged 6–16 y Age: 10.2 ± 2.48 y |
Prevalence of IDA and anemia was assessed by CBC and ferritin. | RPM for intelligence. School achievement was assessed by the students’ scores in math GPA, cumulative GPA, and rank. | More anemic children had an intellectual deficit *; RPM performance related to Hb level ***. Ferritin was correlated with math (r = 0.5 *) and cumulative GPA (r = 0.37 *); math GPA was related to Hb level *; iron status related to school achievement ****. |
| Goudarzi et al. (2008, Iran) [25] | Cross-sectional study |
n = 540 students aged 11–17 y Age: 14.9 ± 1.2 y |
Prevalence of ID was assessed by SI, TIBC, and ferritin. | RPM for intelligence. | No significant difference in IQ scores or IQ classification among students with ID, IDA or normal iron status. |
| Halliday et al. (2012, Kenya) [17] | Cross-sectional RCT baseline analyses |
n = 2400 students aged 5–18 y Age: 10.3 y |
Prevalence of anemia was assessed using a portable hemoglobinometer. | Attention was assessed by pencil-tap test and the code transmission test. RPM for non-verbal reasoning. | Anemia status was not associated with attention, literacy, non-verbal reasoning, comprehension, or numeracy skills. |
| Halterman et al. (2001, USA) [26] | Cross-sectional study |
n = 5398 children aged 6–16 y (61.3% were 6–11 y, 38.7% were 12–16 y) | Prevalence of ID and IDA was determined by TS, ferritin, erythrocyte protoporphyrin, and Hb. | WISCR: verbal component (digit span) and performance examination (block design). WRAT: math and reading components. | For all categories, scores lowered with diminishing iron status. IDA and ID did not score differently than normal status for reading and digit span (ns). For reading, block design, and digit span the % scoring below average did not differ by iron status. ID was not at increased odds of scoring below average for reading, block design, or digit span (ns) but IDA scored lower than children with normal status *. IDA and ID had lower math scores * and had higher risk of scoring below average (OR 2.3; 95% CI: 1.1,4.4). |
| Hutchinson et al. (1997, Jamaica) [27] | Cross-sectional study |
n = 800 rural students in grade 5 aged 9–13 y Age: 10.8 ± 0.6 y |
Prevalence of anemia was assessed by portable hemoglobinometer. Samples were obtained from 769 children. | WRAT: reading, spelling and math subtests. | Hb was significantly positively correlated with reading and spelling scores but not correlated with math scores. |
| Ivanovic et al. (2004, Chile) [28] | Cross-sectional study |
n = 4509 students ages 5–22 y Age: 10.4 ± 3.5 y |
Daily iron intake (% of adequacy) from 24 h dietary recall data by individual interviews. | School achievement was evaluated through standard Spanish-language and math achievement tests designed for the study. | Iron intake (% daily value) was correlated with scholastic achievement for the whole sample (r = 0.065 ***). By grade, this positive correlation was only significant in grade 4 high school students (r = 0.142 *). NS for grade 1 high school, grade 6 or 8. |
| Ji et al. (2017, China) [29] | Cross-sectional study |
n = 428 elementary school students aged 11–14 y Age: 12.0 ± 0.4 y |
Prevalence of ID from Hb and SI. | CNB was used for performance accuracy and speed in four neurobehavioral domains. Chinese version of the WISCR was used to measure intelligence. | Only one difference in mean raw CNB scores was found * which was ns after adjustment. ID had longer reaction times on tests of mental flexibility and capacity for abstraction and the test of special processing ability *. High SI had slower speed on tests of spatial processing ability * and had decreased abstraction ability and mental flexibility *. Iron status was associated with the full-scale IQ score (ns). |
| Kharat and Waghmare (2015, India) [30] | Cross-sectional study | n = 74 adolescent girls aged 18–19 y | Prevalence of anemia was assessed by Hb concentration, tested by the cyanmethemoglobin method. | Cognitive performance was assessed with P300 using an odd ball paradigm with an RMSEMG EP II machine. | Anemic group had delayed P300 latencies as compared with the control group ****. The P300 amplitudes were larger in the girls in the control group than the anemic group *. |
| Masalha et al. (2008, Israel) [31] |
Cross-sectional study | n = 67 fourth, fifth, and sixth graders ages 9–11 y. | Prevalence of anemia was assessed using venous blood was used. | Academic Achievement Index was calculated as the ratio of all marks achieved of all approved courses over the total. Low achievement was classified as scores < 80%. | Of the 14 children with anemia, 6 had low academic achievement scores (42.9%). (no statistical analyses reported.) |
| More et al. (2013, India) [32] | Cross-sectional study | n = 87 girl aged 12−15 y studying in sixth to ninth standard | Screening for anemia and ID was performed by CBC and ferritin. | School achievement was assessed by math score from the final term exam on report cards. Multicomponent Test for verbal learning, memory, and attention; PGI test; and Bhatia battery performance test. | Scholastic performance, IQ, and scores of mental balance, attention and concentration, verbal memory, and recognition were decreased in iron-deficient girls, both anemic and non-anemic, as compared with the non-iron-deficient girls *. |
| Nagalakshmi et al. (2015, India) [33] | Cross-sectional study |
n = 60 rural school children aged 9–12 y Age: 10.4 ± 1.1 y |
Hb level was assessed using Sahli’s acid hematin method. | Visual reaction time; whole-body reaction time, and MMSE. | Whole-body reaction time was negatively correlated with Hb ***. Visual reaction time and MMSE were negatively correlated with Hb (ns). |
| Nemati et al. (2005, Iran) [34] | Cross-sectional study |
n = 170 adolescent girls Age: 12 y |
Prevalence of IDA and anemia from venous blood samples. Measured Hb, hematocrit, MCV, TIBC, and ferritin. | “Educational progression including average test score of base class primary school for schoolgirls”. Test scores (/20) were classified as low (10–15) and high (15.1–20). | Anemics had lower test scores than those without anemia *. IDA had significantly lower test scores than those without IDA *. ID did not have significantly lower test scores than those without ID. Hb was correlated with average test score (r = 0.171 *). Ferritin was not significantly correlated. |
| Olson et al. (2009, Philippines) [35] | Cross-sectional study |
n = 322 rural school students aged 7–18 y Age: 12.1 (95% CI: 11.7,12.4) y |
Prevalence of IDA and anemia from a CBC by hematology analyzer on venous blood samples. Ferritin, sTfR were also measured for iron status. | WRAML, verbal fluency, and PNIT. | Students with IDA and NIDA had lower non-verbal intelligence scores than students with no anemia **. After adjustment, anemia status showed no effect on WRAML learning index, but children with NIDA scored worse than children without anemia on the verbal memory component *. Anemia status, regardless of type, had no significant effect after adjustment on verbal fluency. |
| Ortega et al. (1993, Spain) [36] | Cross-sectional study |
n = 64 middle-class adolescents aged 15–18 y Age: 15.9 ± 0.8 y |
Iron intake was quantified using the 5 day “food consumption registration” technique. Fasting venous blood samples were used for a CBC measured using a Coulter S analyzer Plus. SI was also measured. |
Spanish TEA for verbal, reasoning, and calculus. IQ percentile (IQ < or > 100) is calculated from total scores. The attention test consisted of clearly crossing out all the letters that were accompanied by two apostrophes and the hits, errors, omissions, and speed were recorded. School grades for Latin, Spanish language, foreign language, geography, religion-ethics, math, physics-chemistry, physical education, and technical-professional activities were obtained. | In girls, ID was associated with lower scores for verbal, calculus, school aptitude, and IQ *; higher IQ had higher Hb *; iron status was not associated with school grades. In boys, ID was associated with lower factor scores for verbal, reasoning, school aptitude, attention speed, grades in physics and chemistry *; ferritin was positively associated with IQ percentile *. Overall, Hb was associated with calculus score (r = 0.2905 *), but not attention, verbal scores, reasoning scores, or overall TEA global scores. Ferritin was associated with attention hits (r = 0.3434 *) and speed (r = 0.3989 *). Iron intake was negatively associated with attention hits (r = −0.2874 *) but not IQ percentile. SI was associated with none of the above scores. |
| Sen and Kanani (2006, India) [37] | Cross-sectional study | n = 350 low-income adolescent girls aged 9–14 y | Prevalence of anemia was measured with Hb by the cyanmethemoglobin method. | Gujarati version of WISC: digit span test for short-term memory, maze test for visual–motor coordination, Clerical task for concentration and ability to discriminate, and visual memory test for short-term memory. | Girls with anemia performed worse on the digit span test and visual memory tests in both 9–11 and 12–14 age ranges *. No difference in performance on the maze test or clerical task by anemia statuses. |
| SoonMyung et al. (2004, Korea) [38] | Cross-sectional study |
n = 193 adolescent girls aged 11–14 y Age: 12.6 ± 0.6 y |
Prevalence of anemia through Hb was measured using an Automatic Blood Cell Counter. SI, TIBC, and ferritin were also measured. | Questionnaire regarding clinical symptoms of anemia was administered. Decreased ability to concentrate and poor memory were measured using Likert-type scales. | Hb and ferritin were not significantly correlated with decreased ability to concentrate and poor memory. |
| Teni et al. (2017, Ethiopia) [39] | Cross-sectional study |
n = 442 adolescent girls aged 10–19 y Age: 14.2 ±1.7 y |
Prevalence of anemia measured by the HemoCue (Hb 301) system. | Average scores in the school were obtained from the school records. | Anemic girls were more likely to show low academic performance, compared with non-anemic girls (AOR = 1.7; 95% CI: 1.2, 2.7 *). More anemic girls had academic performance below the mean compared with non-anemic girls (71.1 vs. 64.5%) (no statistics analyses). |
| Thalanjeri et al. (2016, India) [40] | Cross-sectional study | n = 30 school going children both males and females between the ages of 9 and 13 y | Prevalence of anemia was assessed through venous blood was collected for a CBC using a semi-auto hematology analyzer. | Visual memory test and RPM. | RPM scores were lower in anemic children as compared with non-anemic children ***. No significant correlation between Hb and the visual memory test. |
| Walker et al. (1998, Jamaica) [41] | Cross-sectional study | n = 452 adolescent girls aged 13–14 y in grade 8 | Prevalence of anemia using Hb measured by an automated method on a Cell Dyn 700 cell counter. | School achievement using the WRAT for spelling, reading, and arithmetic. Scores on the test were converted to grade levels. | Anemia was associated with lower achievement levels in reading and spelling **. |
| Webb and Oski (1973, USA) [42] | Cross-sectional study | n = 193 students ages 12–14 y in a junior high school low SES black community | Prevalence of anemia assessed by CBC using the Coulter Counter, Model S. | School achievement using the composite score of the Iowa Tests of Basic Skills. | Anemic subjects differed from non-anemic subjects in composite scores achieved *. Anemic girls aged 12 y scored better than non-anemic girls. All other anemic subjects scored worse than non-anemic subjects. |
CBC, complete blood count; CNB, Penn Computerized Neurocognitive Battery; GPA, grade point average; Hb, hemoglobin; ID, iron deficiency; IDA, iron-deficiency anemia; IQ, intelligence quotient; IRA, Academic Performance Index; MCV, mean corpuscular volume; MMSE, mini-mental state examination; OR, odds ratio; PNIT, Philippines non-verbal intelligence test; RCPM, Raven’s Colored Progressive Matrices; RPM, Raven’s Progressive Matrices; SI, serum iron; sTfR, soluble transferrin receptor; TEA, Test of Educational Ability; TIBC, total iron-binding capacity; TS, transferrin saturation; WISCR, Wechsler Intelligence Scale for Children-Revised; WRAML, Wide- Range Assessment of Memory and Learning; WRAT, Wide-Range Achievement Test-Revised. 1 Age is presented as the mean age ± SD unless otherwise stated. 2 Effect estimates, when available, are presented with associated significance value (ns p > 0.05, * p < 0.05, ** p < 0.01, *** p < 0.001, and **** p < 0.0001).