Iran can be divided into 11 regions with 28 provinces (Supplementary file), based on ethnographic, demographic, epidemiologic, and socioeconomic factors (18). This study was cross-sectional and was conducted during the year 2001 in Iran. It was a part of the micronutrient survey, which covered the whole country and was conducted by the Ministry of Health (MOH) of Iran.

The study population was households and the sampling method was unequal cluster sampling with unequal household sizes randomly selected throughout the country. The clusters were sampled relative to the size of the urban and rural population of each province in all 11 regions. The center of the cluster was selected by systematic random sampling. There were 880 clusters in all 11 regions, 504 urban and 376 rural clusters.

Before the study was conducted, a team of health workers in each health care center in urban and rural areas were selected. Each team was assigned to pick up at least five people for the study from each subregion. Pregnant women were informed about the study process. Assigned pregnant women gave informed consent, and accepted to give blood samples and complete questionnaires. These pregnant women were targeted in their own provinces by written invitation at the nearest health care center that belonged to their region. If a woman was not able to go to the health care center within the due date, the survey staff would go to the woman's house and invite her to participate in the study. As an inclusion criterion, it was important that the pregnant women were otherwise healthy and did not have any kind of chronic diseases such as diabetes or heart disease. The blood samples were collected carefully by trained health workers. The total number of pregnant women who participated in the study was 4,368 from all regions of Iran, but data on vitamin A levels were only available in 3,270 women.

A questionnaire was designed by the MOH in Iran to collect information about mothers on general characteristics, health status, supplements if used, number of abortions, pregnancies, and deliveries.

From each participant, a minimum of 6.5 mL venous blood was collected and kept in dry ice and sent to an accredited district laboratory for centrifugation. The samples were coded and kept at −20°C until analysis. All samples were transported in a refrigerated vehicle to an accredited laboratory in Tehran. Vitamin A was analyzed as retinol by using high-performance liquid chromatography (HPLC) at the WHO collaborating laboratory in Tehran after the samples were transported there in a cold chain vehicle.

The mean age of the participating women was 26 years, ranging from 14 to 47 years. The mean parity was 2, ranging between 1 and 12 deliveries. Gestational age ranged from 22 to 42 weeks at the time of blood sampling. The number of included pregnant women in the 11 regions varied between 229 and 330 (Table 1). Mean (SD) values of serum vitamin A ranged from 0.77 (0.46) to 1.12 (0.43) µmol/L. In Table 1, the regional as well as national mean (SD) levels of vitamin A as well as prevalence rates within the categories deficient, insufficient, suboptimal, and optimal is shown. The combined prevalence rates within the categories deficient and insufficient (<0.7 µmol/L) are also shown. The highest mean plasma retinol level (1.12 µmol/L [0.43]) was found in central Iran (region 6). Deficiency was mainly seen in regions 2, 5, 7, 8, and 11. Only 6.6% showed deficiency (<0.36 µmol/L), while 18.3% had insufficiency (0.36–<0.7 µmol/L) indicating that the population prevalence below 0.7 µmol/L for the whole nation is well above 20% (24.9%) and should therefore be characterized as severe according to the WHO categories. In some parts of the country the prevalence below 0.7 µmol/L is over 30% (3, 11) with one region reaching the level of 45.9% (7).

In Fig. 1, the percentage distribution of vitamin A status is shown for the 11 regions. Deficiency was mainly seen in regions 2, 5, 7, 8, and 11. The highest optimal percentage of vitamin A was in regions 2, 4, 6, 9, and 11. More than half of pregnant women had serum vitamin A with suboptimal level, while vitamin A deficiency was only seen in less than 7% of all pregnant women in the study, and insufficiency in 18%.

No statistical differences were observed between rural and urban areas. The mean (SD) vitamin A level in rural and urban areas were 0.98 (0.4) and 1.00 (0.4) µmol/L, respectively.

Vitamin A in each age group was slightly lower in rural than that in urban areas. A two-way ANOVA for group analysis of variance was performed. The difference showed that the difference of vitamin A in both areas was not statistically significant. However, there was a statistically significant main effect of age on vitamin A level (p=0.005). Using the same test, this effect size was rather small (eta squared=0.003) (Fig. 2).

Vitamin A levels in pregnant women decreased by increasing maternal age (ANOVA, p=0.005) (Table 2). Women who were less than 23 years old had higher levels of serum retinol than the older ones. There was a weak negative but significant correlation between age and vitamin A status (Pearson correlation, r=−0.047, p=0.007).

Table 3 shows levels of vitamin A in pregnant women during different gestational ages. The mean vitamin A decreased as the gestational age increased (p=0.009, one-way ANOVA). The mean serum retinol decreased as the gestational age increased (p=0.009, one-way ANOVA).

There was no significant relation between parity and vitamin A status (F=0.79, p=0.45) as shown in Table 4.

This study showed that pregnant women in 2001 in this nationally representative study from Iran had a low vitamin A status since 25% of pregnant women had insufficient or deficient vitamin A levels. Using the WHO cut-offs for national prevalence of vitamin A deficiency, Iran should by this standard be considered a country with severe vitamin A insufficiency prevalence (<0.7 µmol/L) in 2001. In contrast, reports from the north of Iran, Marand district, in 2002 showed that 2.1% of child-bearing women had a deficiency level of vitamin A (<0.7 µmol/L) but 18% had values between insufficient or suboptimal level of vitamin A (19). Prevalence of deficiency levels (<0.36 µmol/L) was 6.6% nation-wide and similar to the average global prevalence of maternal vitamin A deficiency in developing countries (7%) (20). The highest prevalence of deficiency to our knowledge was reported from Nepal (31%) (21), about five times higher than in Iran. China had the lowest prevalence of deficiency, only 2% in 2001 (21, 22). In contrast, data from WHO in 2005 showed that on average 11.6% of pregnant women in Europe had vitamin A deficiency, ranging from 2.6 to 20.6% over the past 15 years (10). Maternal vitamin A deficiency was also correlated to a reduced intra-uterine growth rate and birth weight, and higher infant mortality 23–(25) .

The prevalence of vitamin A deficiency we found indicates a slight improvement in vitamin A status from severe to moderate deficiency during 1989 to 2001 (10). A general severe deficiency of vitamin A in cord blood in male infants in 1989 developed into moderate deficiency in 2001 in Iran (17). Despite the improvement in prenatal health care facilities, which in the meanwhile have been organized in each province supplying multivitamin (vitamins A+D) supplements free of charge through the Primary Health Care system. Supplementation of vitamin A to infants from 2 weeks of age until 2 years by MOH in the past decades may be responsible for a previous finding that vitamin A deficiency and insufficiency were rare in infancy (18). The higher prevalence of insufficient vitamin A levels in the south of Iran might be associated with lower contribution of animal and milk products to the total energy intake, being 6 and 1%, respectively, in comparison with central and north of Iran, where animal and milk products contributed to about 8 and 4%, respectively, of the total energy intake (26). A low access in the south of Iran to main sources of vitamin A (dairy products, vegetables, and fruits) may be related to the hot and dry climate in this region (27). One study in northwest of Iran showed that the important sources of vitamin A in child-bearing women were nuts and green leaves, products that were not typical of other populations (19). Vegetables and fruits provide carotenoids equivalent to more than 50% of vitamin A (4). MOH reported that the ‘intake to need’ ratio of vitamin A shows a scattered provincial distribution, from 60% in Sistanbalouchestanto to 100% in East Azarbayejan (26). This is in accordance with the consumption of the main sources of vitamin A in the south of Iran, that is, Sistanbalouchestan is low ranking in both urban and rural areas. This might also be related to socioeconomic status, drought, and low production of fruits and vegetables.

Referring to the UN Food and Agriculture Organization (FAO) regarding the nutritional intake in Iran (2002), the average daily intake of vitamin A was lower in villages than in towns, but insufficient in both, based on lower intake of animal products, fruits, and vegetables (26). In this study, the level of vitamin A deficiency was almost the same in both areas.

The association between the status of vitamin A and age has been reported earlier (10, 21) and has been related to factors such as dietary patterns and life-style that influence the nutritional status (28). The current study revealed that there was a significant correlation between gestational age and serum retinol. Similarly, among Nepalese pregnant women, even though the prevalence of vitamin A deficiency was higher only in late pregnancy, it lead to night blindness (9). This might be explained by an increased vitamin A-transfer to the fetus during late pregnancy. In contrast to our finding regarding the negative association of vitamin A with age, it has been reported elsewhere that night-blind women were more likely to be teenagers; however, the risk again increased with age among women aged above 30 (9). The difference could be explained by the fact that the Nepalese study reported data based on whether the women had night blindness during pregnancy or not, while our study reported serum retinol levels. In addition, we had no records of the socioeconomic status which could be an influencing factor here, as it has been shown in the previously mentioned study (9). Relation between serum retinol to parity. Similar to our findings, one study in Bangladesh showed no correlation between parity and serum retinol (29).

A limitation to our study was that investigations at the regional level in the 11 included regions did not include data on sociodemographic differences.

This study illustrates that the status of vitamin A was good in 75% of pregnant women in Iran. As the WHO goal is to eliminate vitamin A deficiency globally, this study shows that Iran has achieved a relatively good result historically in the drive to eliminate vitamin A deficiency only in some provinces but not at the national level. Therefore, the problem needs to be closely monitored. Since this data does not represent the current status of vitamin A in the country, further information on vitamin A status in pregnant women are required for a strict surveillance of this important problem and to ensure that the vitamin A status continues to improve in Iran.