This is the old United Nations University website. Visit the new site at

Previous Page Table of Contents Next Page

Plausible evidence of effectiveness of an iron-supplementation programme for pregnant and post-partum women in rural Bangladesh


Rebecca J. Stoltzfus, Jyatsnamoy Chakraborty, Amy Rice, Bénédicte de la Brière, and Andres de Francisco
Rebecca Stoltzfus and Amy Rice are affiliated with the Center for Human Nutrition in the Department of International Health at the Johns Hopkins School of Public Health in Baltimore, Maryland, USA. Jyatsnamoy Chakraborty and Andres de Francisco are affiliated with the International Centre for Diarrhoeal Disease Research, Bangladesh (ICDDR,B) in Dhaka, Bangladesh. Bénédicte de la Brière is affiliated with the International Food Policy Research Institute in Washington, DC.


Since 1978 iron and folk add pills have been distributed, along with other family-planning and health services, to women in Matlab subdistrict, Bangladesh, through fortnightly visits of community health workers. In 1986 the iron dose was set at three pills daily (198 mg of iron) during pregnancy and two pills daily for 6 months post-partum. Haemoglobin concentrations were determined in a representative sample of 218 women who gave birth from June through August 1994 and who were followed prospectively from 0.5 to 9 months post-partum. At 0.5 months post-partum, 88% of women reported that they had taken iron pills on the previous day. The prevalence of anaemia declined from 36% at 0.5 months to 9% at 9 months post-partum, with an overall average prevalence of 23% in this period. To estimate the impact of the programme, these results were compared to anaemia results from representative samples of non-pregnant women in Matlab in 1975 and 1976 and in three other rural communities in Bangladesh in 1996. The estimated reduction in the prevalence of anaemia resulting from the programme ranged from 48% to 70%, and the estimated increase in haemoglobin concentration ranged from 0.9 to 2.1 g/dl. This evidence suggests that the Matlab programme has been highly effective in controlling anaemia.


Iron-deficiency anaemia is the most prevalent form of malnutrition, affecting around 50% of pregnant women worldwide [1]. Iron supplements are inexpensive, widely available, and efficacious, and iron-supplementation programmes have been carried out in many places throughout the world over the last two decades. Still, the prevalence of iron-deficiency anaemia does not appear to be declining [2,3]. Although most large-scale programmes have not been evaluated, several evaluations have found programmes to be ineffective [4], with no significant decrease in the prevalence of anaemia in the population. To our knowledge, there have been no published reports providing evidence of the effectiveness of a sustained, large-scale iron-supplementation programme.

Although iron-deficiency anaemia is prevalent in many parts of the world, the burden of anaemia in women is highest in South Asia, including Pakistan, India, Nepal, and Bangladesh. Worldwide there are around 420 million anaemic women, of whom 335 million (80%) live in Asia and 160 million (38%) live in South Asia [1]. Clearly, to make substantial progress against the global burden of anaemia in women, effective programmes are needed in South Asia.

We present evidence suggesting that a community-based health-care programme that delivers iron supplements to women during pregnancy and the post-partum period in rural Bangladesh has been highly effective in reducing anaemia in women. Haemoglobin and supplemental iron consumption data were collected as part of a study of vitamin A supplementation and breastfeeding. This provided an opportunity to estimate the impact of an iron-supplementation programme that has been operating for two decades without any evaluation. Because these data were not collected with the intent of evaluating the iron-supplementation programme, the need for additional evaluation data from this programme is highlighted. The elements of the programme that might be keys to its success are discussed.


Description of the Matlab iron-supplementation programme

This iron-supplementation programme is part of the community-based Maternal Child Health Family Planning Project (MCH-PP) of the International Centre for Diarrhoeal Disease Research, Bangladesh (ICDDR,B). Since 1966 ICDDR,B has operated a demographic surveillance system in Matlab thana (subdistrict). The surveillance system provides continuous recording of all births, deaths, migrations, and changes in marital status in a population of 210,000. In 1977 the MCH-FP programme was implemented in half of the surveillance area. In the other half, the health services were provided by the regular government programme. The focus of the MCH-FP programme was initially family planning, but in 1986 it was expanded to include interventions more clearly directed at maternal and child health.

Matlab thana is located 45 km south-east of Dhaka, in a wet and fertile delta area that is interlaced with small rivers and irrigation canals. Rice, jute, and fish are the main agricultural products. The majority of the population are Muslim, but a significant minority (12%) are Hindu. Even though current governmental policies have improved female literacy, most women are functionally illiterate, and about half of adult women have never attended school.

The health services provided by ICDDR,B consist of a central clinic and four subcentre clinics that provide basic medical care. Through the MCH-FP programme, there has also been a system of community health workers who provide primary health-care and basic family-planning services at the household level. All households with a married woman of child-bearing age are visited fortnightly by one of 80 community health workers. These women provide family-planning services, immunizations, breastfeeding promotion, oral rehydration salts for diarrhoea, vitamin A capsules for children under five years of age, and prenatal care, including iron supplements and safe delivery kits, and they are trained to screen people for a variety of health conditions and refer them to health-care facilities. As part of the demographic surveillance system, these health workers also ascertain reproductive status and vital events at each visit.

The community health workers are educated young housewives recruited from the community where they are assigned to work. They are paid staff of the ICDDR,B. After recruitment they are given one month of training, followed by supervisory meetings every two weeks. These meetings focus on the problems the community health workers face daily. An experienced group of paramedical staff conduct training, supervise the work, and provide necessary backup to the community health workers.

The provision of iron supplements to pregnant women by community health workers began in 1978 and has been ongoing ever since with no formal evaluation [5]. The iron tablets are brown-coloured and contain 200 mg of ferrous fumarate (66 mg elemental iron) and 200 g of folic acid. The tablets are procured from local sources at a cost of US$5 per thousand tablets. Beginning in 1978, women were instructed to take one tablet daily from the time a pregnancy was detected by the health worker until six months post-partum. In 1986 the dose was increased to three tablets daily during pregnancy and two tablets daily post-partum. When a pregnancy is detected, the health worker gives the woman a one-month supply of tablets. At subsequent visits, the health worker reviews that the woman is taking the tablets correctly, but reported compliance or occurrence of side effects is not recorded. Initially, acceptance of the tablets was low, and women were suspicious of taking them and complained of side effects. Over time, the tablets gained popularity. The most commonly reported side effects are black stool, nausea, and constipation.

Data collection

Matlab 1994-95 data

Recent maternal haemoglobin data (hereafter called Matlab 1994 - 95) were collected as part of a randomized trial of vitamin A supplementation and b-carotene supplementation to women in the Matlab MCH-FP intervention area, the RETIBETA study. The details of the main study have been reported elsewhere [6], but the salient points are also described here. We attempted to enroll in the study all women who delivered infants in the MCH-FP area between June 14 and August 29,1994, and who could be assessed with their live infants at two weeks post-partum. The age distribution, education, and religion of the 218 women enrolled in the study were very similar to those of all women in the MCH-FP area who delivered infants around that time.

Beginning at two weeks post-partum, women in the study were randomly allocated to receive one of three interventions until nine months post-partum: one 200,000 IU vitamin A capsule, followed by daily placebo capsules; daily 7.5 mg b-carotene capsules; or daily placebo capsules. At 0.5,3,6, and 9 months post-partum, a random 50% subsample of the enrolled women was brought to the hospital for a health assessment; 93% of enrolled women completed the nine months of follow-up. Haemoglobin concentration was determined from a venous sample of blood using the HemoCue system (Angelholm, Sweden). A control cuvette was checked daily, and no problems were encountered with the instrument. The women’s vitamin A status was also assessed at these times. Vitamin A status improved with vitamin A supplementation, but haemoglobin concentrations and anaemia prevalence rates did not differ between treatment groups, and there was no evidence of any interactions among either form of vitamin A supplementation, iron supplementation, and haemoglobin concentration (data not shown). Therefore, the three treatment groups are combined in these analyses.

The women’s reported consumption of iron tablets was ascertained during a dietary interview that included questions on intakes of all types of supplements. Neither the interviewers nor the mothers were aware of any special interest in the iron-supplementation programme. The interviewers were independent of the community health workers who distributed the tablets. Before starting the 24-hour recall in the dietary interview, the women were asked, “Other than the RETIBETA capsule, did you take any vitamin capsule, pill, or liquid yesterday?” Women who responded yes were then asked, “What kind of capsule, pill, or liquid did you take?” MCH-FP iron-folate tablets were recorded as one possible response to this question. Women were not prompted to recall any specific type of vitamin preparation.

All women gave their informed consent to participate in the study, which was reviewed and approved by the Committee on Human Research of Johns Hopkins University and the Ethical Review Committee of ICDDR,B.

Comparison data

To estimate the impact of the iron-supplementation programme on haemoglobin concentrations of the Matlab study sample, non-intervention data were needed for comparison. Two types of comparison data are presented. A survey of haematocrit values of non-pregnant women in Matlab was used as a historical comparison (hereafter called Matlab 1975-76 data). These data were collected in 1975-76 from a representative sample of 2,445 married women below the age of 50 years [7]. These data allow a pre-post-intervention estimate of programme impact, as the iron-supplementation programme did not begin until 1978.

The second type of comparison data consists of other contemporary data from Bangladesh. These data are from a haemoglobin survey conducted by the International Food Policy Research Institute in 1996 in three rural areas of Bangladesh apart from Matlab (hereafter called IFPRI 1996). In these surveys, haemoglobin concentrations of all women in survey households were determined in a representative sample of households, using the HemoCue method and capillary blood samples. The Manikganj site (Saturia thana) is 90 minutes by car from Dhaka, off the main road running west from Dhaka to Jessore. It is a vegetable-growing region, with intensive agriculture for the markets of Manikganj and Dhaka. The Mymensingh study area (Pakundia, Kishoreganj, and Gaffargaon thanas) is a traditional rice-growing area two hours north of Dhaka. The Jessore site is eight hours south-west of Dhaka, near the Indian border. The economy consists of subsistence agriculture and an active labour market based on fruit trees, date juice, rickshaw pulling, and embroidery. In all three study sites, there is no regular programme to provide iron supplements to pregnant women. However, iron tablets have replaced the “placebo” pills in monthly cycles of oral contraceptives used by 20% of women in these survey areas. Out of 28 pills per packet, 7 pills contain 75 mg of iron as ferrous fumarate. Therefore, this comparison population does not represent a non-intervened population, but one in which iron supplementation reaches many fewer women overall and none during pregnancy, is intermittent (seven days per month), and is of lower dose than the Matlab programme. The efficacy of this one week per month supplementation strategy has not been measured; however, iron supplementation one day per week improved iron status in non-pregnant women in Indonesia [8] and the United States [9].

Data analysis

To compare the Matlab 1994-95 data with those from other studies, we needed a summary figure of haemoglobin concentration in the study sample over the post-partum study period. Because the majority of women contributed two haemoglobin measurements, we took the average of all available haemoglobin measurements for each woman and used this value to obtain overall mean haemoglobin concentration and anaemia prevalence. The two haemoglobin measurements for each woman were taken at random intervals post-partum. Thus this overall average value represents a theoretical midpoint of the period from 0.5 to 9 months post-partum.

The Matlab 1975-76 data consisted of haematocrit rather than haemoglobin values. So that these data could be directly compared with the contemporary haemoglobin data, mean and SD haematocrit values were converted to estimated haemoglobin values using a factor of 0.33 [1]. The criterion for anaemia in non-pregnant women is 12 g/dl haemoglobin or 36% haematocrit [10]. The prevalence of anaemia according to this cutoff was not reported by Huffman et al. [7]. We derived an estimated prevalence of haematocrit values less than 36%, based on the reported mean and SD, assuming a normal distribution of values.

The IFPRI 1996 survey included all women in survey households. Two samples were constructed, designed to be comparable to the Matlab 1975 - 76 and Matlab 1994-95 samples, respectively. The first sample included all married women below 50 years of age, the sample definition in the Matlab 1975-76 study. For this sample definition, haemoglobin and anaemia values are reported separately for each of the three survey communities, to reveal the variability in anaemia between communities. The second sample included all women who re- ported having given birth within the past nine months. This sample definition approximates the Matlab 1994-95 sample. For this sample, haemoglobin and anaemia values are reported for all three communities combined, as the total number of women meeting this definition was only 62.

To make comparisons between study samples, we calculated 95% confidence intervals around the mean haemoglobin values and the prevalences of anaemia [11]. The programme impact was estimated as the percent reduction in anaemia prevalence in the Matlab 1994-95 sample as compared with either the historical (Matlab 1975-76) or contemporary (IFPRI 1996) samples.


Matlab 1994-95 sample

At all times during the post-partum period, the average haemoglobin concentration was above 12 g/dl (table 1). The haemoglobin concentration improved dramatically from the first six months to the ninth month post-partum, reaching a high of greater than 14 g/dl. The prevalence of moderate anaemia, defined as haemoglobin concentration less than 10 g/dl, was 4.7% in the first month post-partum but declined to nil by nine months post-partum. The percentage of women who reported that they consumed iron tablets was 88% immediately post-partum but declined markedly after six months post-partum, the time at which the community health workers were scheduled to stop distributing the tablets.

Comparability of population samples

A potential flaw in the use of historical control data that span a 20-year interval is the effect of secular trends. In particular, the family-planning programme and other health interventions that have been carried out in Matlab might have led to an improvement in women’s nutritional status that is independent of the iron-supplementation programme. However, the differences in heights and weights of women in the Matlab 1994-95 and 1975-76 samples were small (table 2). Although the women in the 1994-95 sample were on the average 1.6 cm taller, the weights of the women in the two samples were nearly identical (0.5 kg difference).

A noteworthy difference between the samples is the age of the women, with the 1975-76 sample including a much greater proportion of women over the age of 30. This difference is explained by the fact that the 1975-76 sample was a representative sample of all non-pregnant women of reproductive age, whereas the 1994-95 sample was a representative sample of all women who had recently delivered an infant. The probability of delivering an infant is higher in women 20 to 30 years old than in older women.

The heights and weights of the women in the IFPRI 1996 sample of women of reproductive age were slightly greater than those of the women in the Matlab 1994-95 sample. The IFPRI 1996 post-partum subsample was similar to the Matlab 1994-95 sample in weight (0.7 kg heavier), height (1.5 cm taller), and percentage of women above 30 years of age (4% difference).

Estimates of programme impact

The average haemoglobin values for women in the Matlab 1994-95 data yielded a population prevalence of 23% for anaemia, whereas the prevalence of anaemia in the Matlab 1975-76 sample was 60% (table 3). In the IFPRI 1996 samples, the prevalence of anaemia was 44% to 59% in non-pregnant married women, and 76% in women in the first nine months post-partum. The estimated impact of the Matlab programme in 1994-95 was a 62% reduction in the prevalence of anaemia, as compared with the historical Matlab data. The estimated impact was a 48% to 61% reduction in the prevalence of anaemia as compared with contemporary non-pregnant women of reproductive age, and a 70% reduction as compared with contemporary women in the first nine months post-partum. The programme impact on average haemoglobin concentration was an increase of 0.9 to 2.1 g/dl.

TABLE 1. Haemoglobin concentration, prevalence of anaemia, and self-reported consumption of iron tablets in Matlab, 1994-95, according to time post-partum

Time post-partum (mo)


% Prevalence of iron tablet consumption

Hbb concentration (g/dl)

% Prevalence of anaemia

Hb <12.0 g/dl

Hb <10.0 g/dl




12.5 ± 1.8c






12.4 ± 1.4












14.1 ± 1.4


. 0

a. Haemoglobin concentration was measured in 218 women; for most women measurements were made at two different times post-partum.
b. Hb, Haemoglobin.
c. Plus-minus values are means ± SD.
TABLE 2. Anthropometric and age characteristics of population samples



Mean weight - kg (95% CI)

Mean height - cm (95% CI)

Age > 30 yr -%

Matlab 1994-95


40.9 (40.3 to 41.5)

149.5 (148.8 to 150.2)


Matlab 1975-76a


40.4 (40.2 to 40.6)

147.9 (147.7 to 148.1)


IFPRI 1996 (non-pregnant married women <50 yr)



41.6 (40.8 to 42.3)

149.9 (149.2 to 150.5)




41.5 (40.6 to 42.4)

149.7 (149.0 to 150.3)




42.8 (42.0 to 43.5)

150.3 (149.6 to 150.9)


IFPRI 1996 (women <9 mo post-partum)

All communities


41.6 (40.4 to 42.8)

151.0 (149.7 to 152.2)


Abbreviations: CI, confidence interval; IFPRI, International Food Policy Research Institute.
a. From Huffman et al. [7].


We have documented a low prevalence of anaemia in women in Matlab during the post-partum period, accompanied by a high self-reported consumption of iron supplements. Both figures are remarkable, given the consistent reports in the literature of prevalent and severe anaemia in South Asian women [1, 3] and the poor coverage of iron supplementation documented in other programmes [4]. The obvious question is, “Can the low rates of anaemia in this population be attributed to the iron-supplementation programme?” Because an evaluation was not designed into the programme, we constructed comparison data from existing anaemia surveys and tried to assess the extent and direction of bias that is inherent in non-randomly allocated comparison groups. We draw the following conclusions from these analyses:

» The current prevalence of anaemia is substantially lower in Matlab today than it was before iron supplementation began, and it is substantially lower than in similar samples of women in three other rural Bangladeshi communities today.

» Women in the comparison samples consumed no supplemental iron, or much less supplemental iron than contemporary women in Matlab were reported to consume.

» The difference in the prevalence of anaemia between the samples is not attributable to differences in age or in general nutritional status of the women, as reflected by their weights and heights.

» The most plausible explanation at hand is that the low rates of anaemia in Matlab are related to the high consumption of supplemental iron provided by the MCH-FP programme. However, this inference is ad hoc and merits a more rigorous investigation.

Our best estimates of programme impact from these data are a 62% reduction in anaemia based on historical Matlab data and a 70% reduction based on a contemporary sample of women at a similar stage post-partum. To our knowledge, this is the first evidence in the published literature of substantial impact from a long-standing health-care programme that delivers iron supplements to pregnant and post-partum women.

TABLE 3. Haemoglobin concentration and prevalence of anaemia in non-pregnant women in Matlab 1994-95, Matlab 1975-76, and IFPRI 1996 surveys



Hb - g/dl (95% CI)

% Prevalence of anaemia (95% CI)

% Impact of Matlab programme on anaemiab

Matlab 1994-95c


12.9 (12.7 to 13.1)

23 (21 to 25)

Matlab 1975-76d


11.7 (11.7 to 11.8)

60 (58 to 62)


IFPRI 1996 (non-pregnant married women <50 yr)e



12.0 (11.8 to 12.1)

47 (41 to 53)




11.3 (11.1 to 11.5)

59 (52 to 65)




11.9 (11.8 to 12.1)

44 (38 to 50)


IFPRI 1996 (women <9 mo post-partum)

All communities


10.8 (10.4 to 11.2)

76 (63 to 85)


Abbreviations: CI, confidence interval; Hb, haemoglobin; IFPRI, International Food Policy Research Institute.
a. Defined as Hb < 12.0 g/dl.
b. Defined as the ratio of the prevalence of anaemia in Matlab 1994-95 to the prevalence in the comparison sample, subtracted from 100%.
c. Calculated from the average of the haemoglobin values available for each woman, usually two values.
d. Estimated from haematocrit values of Huffman et al. [7]. No iron supplements were provided in the study area.
e. Twenty percent of women reported taking oral contraceptive pills in which 7 of the 28 pills contained 75 mg of iron per pill.
What are the major threats to the validity of these conclusions? To begin with, we have identified two possible sources of bias in the comparisons between these data sets. First are the assumptions we made in estimating the prevalence of anaemia from the Matlab 1975-76 data. We used a conversion factor of 0.33 between haematocrit and haemoglobin. This factor depends on the mean cell haemoglobin concentration [12], which will vary among individuals in any study sample. On a population basis, anaemia prevalence rates based on haematocrit tend to be lower than those based on haemoglobin when “equivalent” cut-offs for the two indicators are based on this factor [13]. This bias would cause us to underestimate programme impact. We also assumed a normal distribution of haematocrit values. Examination of numerous haemoglobin distributions from a variety of iron-deficient populations shows that in general the assumption of normality is a reasonable one [14]. When distributions are not normal, they are skewed to low values. If the Matlab 1975-76 data were skewed to low values, this also would bias downward our estimate of the prevalence of anaemia and cause us to underestimate the programme impact.

Second, the Matlab 1994-95 sample consisted of women in the post-partum period, whereas the Matlab 1975-76 sample consisted of all non-pregnant married women below the age of 50 years. Because of these different sampling strategies, the women in the Matlab 1994-95 sample were younger and earlier in the post-partum period than those in the historical comparison sample. The effect of these biases on comparison of the two Matlab samples is illustrated in the comparison of the two IFPRI 1996 samples (table 2).Women in the post-partum period had significantly lower haemoglobin values than all women of reproductive age, even though they were younger. This is almost certainly because women shortly post-partum are still recovering from the substantial iron costs of pregnancy. In summary, these biases would tend to lead us to underestimate programme impact. No bias that we have identified could spuriously create the greater than 50% reduction in the prevalence of anaemia that we observed.

The preceding discussion relates to biases we can identify in the variables we measured. However, the most important threat to validity lies in differences in the samples that we did not measure and therefore cannot account for. Unmeasured differences between the historical and contemporary Matlab samples might include general socio-economic improvements, such as greater mobility of women, changes in infectious disease patterns, and changes in reproductive patterns, with contemporary women having fewer babies and starting at a later age. Despite economic development, however, there is no evidence that dietary intake has improved over the last three decades in rural Bangladesh. Food-intake data collected by the government of Bangladesh show a decline in food availability per capita, which they attribute to the fact that the rate of population growth is outpacing the rate of increase in food production [15]. This may explain the lack of improvement in women’s weights between the Matlab 1975-76 and Matlab 1994-95 samples. Unmeasured differences between the contemporary IFPRI and Matlab samples might include differences in reproductive patterns and safe motherhood practices that could have reduced blood loss at delivery in Matlab. We cannot rule out the influences of these variables, and a more rigorous evaluation should account for them.

Evidence of programme impact would be stronger if it could be demonstrated that the haemoglobin concentrations of women who consumed many pills were higher than those of women who consumed few or no pills. Iron-supplementation trials of pregnant women have shown clearly that the impact on haemoglobin is directly related to the duration of supplementation [16]. According to our 24-hour-recall information, women who reported consuming a pill on the previous day at three, six, and nine months post-partum had anaemia rates 25% to 50% lower than women who did not report consumption, but these differences were not statistically significant. The strength of this evidence is limited by our small sample size. Furthermore, we assessed consumption on only four days in the post-partum period (i.e., the day before each clinic visit), although women received supplements throughout most of pregnancy and up to six months post-partum. Prospective information on pill consumption in larger samples of women should be ascertained in future evaluations of this programme.

The low prevalence of anaemia in Matlab at nine months post-partum (9.2% with haemoglobin levels below 12.0 g/dl, and none with levels below 10.0 g/dl) is particularly noteworthy in a rural South Asian population. The substantial improvement in haemoglobin concentrations of Matlab women during the post-partum period probably reflects the effect of continued iron supplementation during a period that is physiologically favourable to iron status. This is because iron contained in the additional red cell mass needed during pregnancy is made available, and iron requirements are relatively low because menses are suppressed during full lactation and the iron content of breastmilk is low [17]. Furthermore, women have low iron stores following pregnancy, and this facilitates the absorption of iron. Thus women tend to be in positive iron balance (i.e., a state of improving iron status) during the post-partum period, especially when consuming supplements.

Some characteristics of the MCH-FP iron-supplementation programme that might account for its apparent success are generalizable lessons. A review of six large-scale iron-supplementation programmes carried out in 1991 identified several major constraints to programme success. These included “low accessibility and utilization of antenatal care, inefficient supply and distribution of supplement... inadequate training and motivation of first line health workers, insufficient and inappropriate counselling of mothers, and failure of effective screening and referral procedures” [4]. It was revealing that the major constraints highlighted in the review had much more to do with the health-care system and health-care providers than with non-compliant behaviour by the recipients. The system of community health workers in Matlab overcomes these constraints. In the context of this well-functioning distribution system, the compliance with and effectiveness of daily iron supplementation with 198 mg of iron as ferrous fumarate appear to be very high. Nevertheless, even if compliance can be achieved at higher doses of iron, we should not give women three iron pills daily if less iron would be equally effective, since reducing the dose would reduce side effects and increase efficiency by reducing costs and logistical difficulties.

Some aspects of the MCH-FP programme are unique. The health-care system in Matlab coordinates with governmental services, but the system of community health workers is sustained with a high level of donor support and with the supervision of an international research centre. The Matlab programme represents one example of an effective iron-supplementation programme. It remains to be demonstrated under what conditions governmental or non-governmental programmes can achieve similar success.

Several questions are unanswered by this evaluation and should be addressed in a further investigation. How many pills do women in Matlab typically consume during pregnancy and the post-partum period? Do women who take fewer pills have lower haemoglobin concentrations? What is the prevalence of anaemia in a representative sample of non-pregnant women and in pregnant women? What benefits do women and community health workers perceive from the iron-supplementation programme?

In 1986, when the Matlab regimen was revised to three pills daily, the iron dose given in the Matlab programme exceeded the international recommendation, which was 120 mg of iron daily during pregnancy, when anaemia is prevalent [18]. The dose given in Matlab is more than triple the current recommendation of 60 mg [10, 19]. This provides a unique opportunity to evaluate the new recommendation in a programmatic context. Could the observed low rates of mild anaemia be sustained if the dose of iron were reduced to around 60 mg (one pill) daily?

As compared with other forms of micronutrient malnutrition, global progress against iron-deficiency anaemia has been slow, if not invisible. This has led to numerous discussions of what we are doing wrong and what innovations can spur progress [4,20-22]. To substantially reduce anaemia of pregnancy on a global scale, it is likely that prenatal iron supplementation must be combined with interventions that improve women’s iron stores before pregnancy [9]. Nonetheless, the high coverage and apparent effectiveness of the Matlab pre- and post-natal iron-supplementation programme sound a timely note of optimism and a reminder to strive to implement better what we know can work, while also striving to discover new and complementary approaches.


This work was funded by the United States Agency for International Development through cooperative agreement DAN-5116-1-00-8051-00 with Johns Hopkins University and grant HRN-5110A-00-3006-00 to the International Food Policy Research Institute.


1. World Health Organization. The prevalence of anaemia in women: a tabulation of available information. WHO/MCH/MSM/92.2. 2nd ed. Geneva: WHO, 1992.

2. UNICEF. The state of the world’s children 1995. Oxford: Oxford University Press, 1995.

3. United Nations Administrative Committee on Coordination/Subcommittee on Nutrition. Second report on the world nutrition situation. Vol. 1. Global and regional results. Geneva: ACC/SCN, 1992.

4. Gillespie S, Kevany J, Mason J. Controlling iron deficiency. United Nations Administrative Committee on Coordination/Subcommittee on Nutrition: Geneva, 1991.

5. Fauveau V, Chakraborty J. Women’s health and maternity care in Matlab. In: Fauveau V, ed. Matlab: women, children, and health. ICDDR,B Special Publication No. 35. Dhaka: International Centre for Diarrhoeal Disease Research, Bangladesh, 1994:109-38.

6. Rice AL. Impact of vitamin A and beta-carotene supplementation in lactating women in Bangladesh on biochemical indicators of maternal and infant vitamin A status. Doctoral dissertation, Johns Hopkins University, Baltimore, Md, USA, 1998.

7. Huffman SL, Wolff M, Lowell S. Nutrition and fertility in Bangladesh: nutritional status of nonpregnant women. Am J Clin Nutr 1985;42:725-38.

8. Gross R, Schultink W, Juliawati. Treatment of anaemia with weekly iron supplementation. Lancet 1994:344:821.

9. Viteri FE. Iron supplementation for the control of iron deficiency in populations at risk. Nutr Rev 1997;55:195-209.

10. World Health Organization/UNICEF/United Nations University. Indicators and strategies for iron deficiency and anaemia programmes. Report of a consultation, Geneva, 6-10 December 1993. Geneva: WHO, 1997.

11. Snedecor GW, Cochran WG. Statistical methods. 7th ed. Ames, la, USA: Iowa State University Press, 1980.

12. Gibson RS. Principles of nutritional assessment. New York: Oxford University Press, 1990.

13. Graitcer PL, Goldsby JB, Nichaman MZ. Hemoglobins and hematocrits: Are they equally sensitive in detecting anemias? Am J Clin Nutr 1981,34:61-4.

14. Yip R, Stoltzfus RJ, Simmons WK. Assessment of the prevalence and the nature of iron deficiency for populations: the utility of comparing haemoglobin distributions. In: Hallberg L, ed. Iron nutrition in health and disease. London: John Libby, 1996:31-48.

15. Government of Bangladesh. National plan of action for nutrition (NPAN). Dhaka: Government of Bangladesh, 1995.

16. Sloan NL, Jordan EA, Winikoff B. Does iron supplementation make a difference? Working Paper No. 15. Washington, DC: MotherCare/Population Council, 1992.

17. National Research Council. Recommended Dietary Allowances. 10th ed. Washington, DC: National Academy Press, 1989.

18. DeMaeyer EM. Preventing and controlling iron deficiency anaemia through primary health care. A guide for health administrators and programme managers. Geneva: World Health Organization, 1989.

19. Stoltzfus RJ, Dreyfuss ML. Guidelines for the use of iron supplements to prevent and treat iron deficiency anemia. Washington, DC: International Life Sciences Institute, 1998.

20. Galloway R, McGuire J. Special report. Daily versus weekly: How many iron pills do pregnant women need? Nutr Rev 1996;54:318-23.

21. Solomons NW. Daily versus weekly iron: We still might not be asking the right questions. Nutr Rev 1997;55:141-2.

22. Stoltzfus RJ. Rethinking anaemia surveillance. Lancet 1997;349:1764-6.

Previous Page Top of Page Next Page