Correspondence: Dr N. R. van den Broek, Wellcome Trust, PO Box 30096, Blantyre 3, Malawi.
Objective To determine prevalence of anaemia in pregnancy in southern Malawi, and to establish if an ‘at risk’ group can be identified for targeted intervention.
Design Prospective cross-sectional study.
Setting A semi-urban hospital and a rural health clinic in southern Malawi.
Population Pregnant women attending the antenatal clinic for their booking visit.
Methods Haemoglobin concentration (g/dL) measured from a capillary blood sample (fingerprick) by battery operated HemoCue machine.
Results In urban Blantyre 57% of women were anaemic by WHO standards (haemoglobin < 11.0 g/dL) and 3.6% were severely anaemic (haemoglobin < 7.0 g/dL). The prevalence was higher in the rural area; 72% and 4.0%, respectively. Primiparae were at slightly increased risk for overall anaemia and severe anaemia but the effect of targeting this group alone for interventions would mean at least 65% of anaemic women and over half of the women with severe anaemia would be excluded. When the effect of age was examined separate from gravidity, adolescents were not found to be at increased risk of anaemia. Anaemia was more prevalent in the wet season and in women who booked late for antenatal care.
Conclusion The prevalence of anaemia in pregnancy in our population is unacceptably high and deserves more attention. Prophylaxis of all women rather than an ‘at risk’ group based on age or gravidity is recommended.
Anaemia in pregnancy, particularly severe anaemia, is associated with an increased risk of maternal mortality1–3, which, in most developing countries, continues to be unacceptably high4–6. In 1993, the World Bank ranked anaemia as the eighth leading cause of disease in girls and women in developing countries7.
In Africa, the prevalence of anaemia in pregnancy is estimated to be 35% to 75%8. The actual prevalence rates for many individual countries is however not known. At the African Regional Consultation on the Control of Anaemia in Pregnancy (1989), it was recommended that simple studies of prevalence and aetiology should be undertaken for each region. In the last published tabulations from the WHO Maternal Health and Safe Motherhood anaemia database (1992), estimates for the region of South East Africa were based on a total sample of about 10,000 women with only one (unpublished) study from Malawi reporting the outcome of screening 319 women8. Up-to-date information from most countries is still scanty; the few published studies describe small samples and are usually from hospital-based populations. There is a lack of data from rural areas. It has been suggested that the prevalence of anaemia may depend on the season, increasing in relation to malaria in the wet season, or in relation to food shortage at the end of the dry season9. Published prevalence rates almost never reflect measurements performed all year round.
In a review of epidemiological criteria for the assessment of severity and magnitude of anaemia in pregnancy, the following categories were proposed: mild anaemia (haemoglobin 9.0–10.9 g/dL), moderate-to-severe anaemia (haemoglobin 7.0–8.9 g/dL) and severe anaemia (haemoglobin < 7.0 g/dL)10. This is in accordance with the WHO definitions of anaemia in pregnancy, which are: overall anaemia (haemoglobin < 11.0 g/dL) and severe anaemia (haemoglobin < 7.0 g/dL)1. In areas with high prevalence programmes for the control of anaemia can be expected to reduce the prevalence of severe anaemia and moderate-to-severe anaemia. Prevalence of overall anaemia and mild anaemia might initially remain unchanged.
Most sub-Saharan countries, including Malawi, currently have national policies to prevent and treat anaemia in pregnancy. These include the provision of haematinics and malarial prophylaxis for all pregnant women. Despite a lack of information about prevalence before and after the implementation of many such programmes, a general disappointment with the persistently common problem of anaemia in pregnancy has been expressed1,2,11,12. It has been suggested therefore that interventions should be more focused. Suggested target groups have been: women with severe anaemia2, adolescents13,14, primi-gravidae15,16 or conversely multigravidae13,17. However, there are insufficient data to indicate which group can rightly be identified as being particularly at risk of anaemia overall or severe anaemia. We need to estimate the baseline prevalence of anaemia to determine the possible effectiveness of target prophylaxis.
In this paper we examine the distribution of haemoglobin values obtained by screening women at their first antenatal clinic attendance. The percentages of women with anaemia (haemoglobin < 11.0 g/dL) and severe anaemia (haemoglobin < 7.0 g/dL) were determined and compared in a semi-urban and a rural population and at different times of the year. In an attempt to identify a ‘risk group’, univariate and multivariate logistic regression was performed to determine the relative risk of age, gravidity, trimester and season for anaemia overall and for severe anaemia.
Women attending the antenatal clinic at two sites were studied. The Queen Elizabeth Central Hospital in Blantyre is a teaching hospital which, in practice, serves both as a secondary referral centre and as a primary facility for an urban and a semi-urban population of about one million people. Antenatal clinics for those attending for their first visit (‘booking clinic’) are held once a week. We conducted screening for anaemia every other week for one year (July 1997 to June 1998 inclusive). Namitambo Health Centre is situated about an hour by car out of Blantyre in the middle of a rural area with no other health facilities. The health centre serves a population of 60,000 people. Booking clinics are held once a week. Screening for anaemia at Namitambo Health Centre was carried out every week over two years (April 1997 to March 1999 inclusive).
After signed (Namitambo) or verbal (Blantyre) consent had been given, a capillary blood sample (fingerprick) was obtained and haemoglobin concentration was determined using a battery operated HemoCue system18,19. All the women were offered screening and no-one refused to undergo a fingerprick. Age and parity were recorded and trimester estimated by examination of the fundal height or from the woman's last recorded menstrual period. As is current practice in Malawi, all women with a haemoglobin concentration of less than 6.0 g/dL were admitted to hospital and offered a blood transfusion. In accordance with the policy of the Ministry of Health in Malawi all women were given a supply of iron tablets (ferrous sulphate 200 mg with folic acid 0.25 mg) and three tablets of sulfadoxine-pyrimethamine to prevent malaria.
Data were entered and preliminary analysis was done using SPSS6.0 for Windows. Regression analyses were done using GENSTAT 5 release 3.2 for Windows. Ethical Permission for this study was obtained from the Malawi Health Sciences Research Committee.
In Blantyre 4708 women and in Namitambo 2728 women had haemoglobin estimations performed at their first antenatal visit. Complete information on date of screening, age, trimester at booking, parity and haemoglobin concentration were available for 4646 women from Blantyre and 2293 from Namitambo.
Prevalence of anaemia and severe anaemia
The distribution of haemoglobin values obtained in each site is presented in Fig. 1 and Fig. 2. In Blantyre the range of values was 4.2–18.6 g/dL with a mean of 10.5 g/dL and a median (25th, 75th centiles) of 10.6 g/dL (9.4–11.8). The prevalence of all anaemia (haemoglobin < 11.0 g/dL) was 57.1% and the prevalence of severe anaemia (haemoglobin < 7.0 g/dL) was 3.6%.
In Namitambo the range of values obtained was: 1.4–17.6 g/dL with a mean of 10.1 g/dL and median (25th, 75th centiles) of 10.0 g/dL (8.9–11.1). The prevalence of all anaemia (haemoglobin < 11.0 g/dL) was 71.7% and the prevalence of severe anaemia (haemoglobin < 7.0 g/dL) was 4.0%. If severe anaemia is defined as a haemoglobin of < 8.0 g/dL, the prevalence in Blantyre is 8.5% and in Namitambo 10.5%.
The difference in prevalence rates for anaemia overall is statistically significant for the rural area (Namitambo) compared with the semi-urban/urban area of Blantyre. (odds ratio (OR) 2.01; 95% confidence interval (CI) 1.82–2.26, adjusted OR 2.34; 95% CI 2.08–2.63) The difference in prevalence rates of severe anaemia is however not significant on univariate analysis, and just significant on multivariate analysis (OR 1.19; 95% CI 0.92–1.54, adjusted OR 1.38; 95% CI 1.05–1.81).
Multiple linear regression was carried out to determine the effect of age, gravidity, trimester and season on haemoglobin levels. Results are presented in Table 1. Age was not found to be a significant determinant of haemoglobin levels in either population, but gravidity was significant. Trimester at booking and season were found to be significant only in the population studied in Blantyre. Univariate and multivariate analysis was then performed to identify risk factors for overall anaemia and severe anaemia at each site. The results are presented in Table 2 and Table 3.
Table 1. Multiple linear regression for haemoglobin concentration at booking visit by age, gravidity and trimester.
Queen Elizabeth Central
Hospital (n= 4646)
Centre (n= 2293)
Table 2. Univariate and multivariate logistic regression of risk factors for anaemia ([Hb] < 11.0 g/dL) and severe anaemia ([Hb] < 7.0 g/dL) in an urban centre, Queen Elizabeth Central Hospital, Blantyre (n= 4646).
OR (95% CI)
Adjusted OR (95% CI)
Table 3. Univariate and multivariate logistic regression of risk factors for anaemia ([Hb] < 11.0 g/dL) and severe anaemia ([Hb] < 7.0 g/dL) in a rural area, Namitambo Health Centre (n= 2293).
OR (95% CI)
Adjusted OR (95% CI)
* ≥ 20
Effect of age
Age was recorded for 4680 women in Blantyre and 2548 in Namitambo; 24% of women in Blantyre and 29% in Namitambo were under 20 years of age. Univariate analysis showed an increased risk for adolescents, but when age was corrected for gravidity and trimester at booking there was no significant increased risk for anaemia overall or severe anaemia in either the rural or urban setting. Very few girls were aged 15 or less: 38 (0.8%) in Blantyre and 34 (1.3%) in Namitambo. Data were analysed separately for girls aged less than 17 and those aged 17–19 years. Neither of these groups was found to have a significantly increased risk of anaemia at their first antenatal clinic visit compared with women 20 years or older. Ninety-five percent of girls under 17 and 80% of girls under 20 were primiparae. For mild anaemia the OR (95% CI) for under 17 (n= 118 Blantyre, n= 99 Namitambo) vs above 20 years was 0.96 (0.67–1.39) at Blantyre and 1.34 (0.83, 2.68) at Namitambo. For severe anaemia the corresponding OR (95% CI) were 0.98 (0.35–2.68) and 1.59 (0.69–3.76).
Effect of gestational age at booking
In Blantyre 80.8% of women booked in the first or second trimester, similar to the group in Namitambo (87.6%). In Namitambo trimester at booking was not a risk factor for anaemia; but in Blantyre the risk of anaemia overall and severe anaemia was increased in women who booked late in pregnancy.
Effect of gravidity
The proportion of primigravidae at booking was 33.4% in Blantyre and 29.7% in Namitambo (OR 1.19, 95% CI 1.07–1.33). Adjusted for age and trimester primiparae were at increased risk for anaemia overall and severe anaemia in both sites, when compared with grand multiparae. The increase in risk for primiparae when compared with women in the second, third or fourth pregnancy is relatively small and not statistically significant except for anaemia overall in Blantyre. If in these populations primiparae were selected as the target group for interventions to prevent and treat anaemia, 65% of all anaemic women and 56% of women with severe anaemia would be excluded. Similarly, in Namitambo 67% of all anaemic women and 55% of severely anaemic women would not receive treatment.
Effect of season
Prevalence rates for the dry season (May to October) and the wet season (November to April) were calculated. In Blantyre, the prevalence of anaemia overall was 61-6% (1704/2767) in the wet season and 50.8% (955/1879) in the dry season. The prevalence of severe anaemia was 4.1% in the wet season and 2.8% in the dry season. In Namitambo, the prevalence of anaemia overall was 74.9% (887/1184) in the wet season and 71.1% in the dry season (789/1109). The prevalence of severe anaemia was 4.9% in the wet season and 3.5% in the dry season. The reduction in prevalence of anaemia overall during the dry season is only just significant for Namitambo (OR 0.83; 95% CI 0.69–0.99) and significant for Blantyre (OR 0.64; 95% CI 0.57-0.73). The reduction in prevalence of severe anaemia in the dry season does not reach statistical significance in Namitambo but does in Blantyre (OR 0.71; 95% CI 0.51–0.99).
Prevalence of anaemia and severe anaemia
The definitions of anaemia overall in pregnancy of haemoglobin < 11.0 g/dL and of severe anaemia of haemoglobin < 7.0 g/dL are chosen arbitrarily20,21. It is not currently known what degree of anaemia is definitely associated with adverse maternal or fetal outcomes22. It has recently been suggested that in areas of high prevalence, cut off points for the diagnosis of anaemia should be lowered21,23. We have used the WHO criteria in this paper and have defined anaemia overall as haemoglobin < 11.0 g/dL and severe anaemia as haemoglobin < 7.0 g/dL1.
This study was limited to those women who attend the antenatal clinic. In Malawi we estimated that 93% of women attend on at least one occasion despite their being in our rural population, a mean distance of five kilometres from home to the nearest clinic24. As there is no reason to believe that women who are anaemic selectively visit the antenatal clinic, it may be assumed that the estimates of prevalence obtained in this study are a reasonable estimate for the population.
The prevalence of anaemia in pregnancy was found to be high in both urban and rural areas. Studies from other countries in the region report similar rates. In Mozambique the prevalence of anaemia overall was 58%, of severe anaemia 1%25. A study from rural Zaire reported a prevalence of 76% for anaemia overall and 3.7% for severe anaemia15. In coastal Kenya the prevalence was found to be 75.6% for anaemia overall and 9.8% for severe anaemia26. Two studies from Tanzania reported prevalence rates of 74.5% and 86.0% for anaemia overall and 7.0% for severe anaemia in an urban population23,27. In contrast, the prevalence rates for anaemia overall in pregnancy in industrialised countries are usually below 10%1.
A national policy to ensure that all pregnant women receive iron tablets (ferrous sulphate 200 mg with 0.25 mg folic acid) together with antimalarial prophylaxis (sulphadoxine and pyrimethamine) has been in place for about six years in Malawi. In practice, tablets are not always available in antenatal clinics. Noncom-pliance is also often cited as a problem. Because of the lack of information on prevalence until now the effect of this national policy of prophylaxis of all pregnant women, whether anaemic or not, is difficult to predict. It is clear however, that anaemia in pregnancy is still a very common problem.
A policy of treating only women with severe anaemia in our population would seem to be inappropriate as it would exclude the majority of women from receiving treatment. Even if a haemoglobin concentration of less than 8.0 g/dL was taken as indicative of severe anaemia 90% of women with mild to moderate anaemia would be excluded.
In order to establish which group has severe enough anaemia to be considered for intervention, a simple accurate method for estimating haemoglobin concentration is necessary. The new WHO Colour scale might prove useful, especially at primary health care level28. In practice however most clinics in developing countries rely only on conjunctival inspection. In order to be able to determine the effect of anaemia on maternal and fetal mortality and morbidity of mild, moderate or severe anaemia, an accurate method of measuring haemoglobin levels is required.
Identification of a ‘risk group’
In a review on maternal health during adolescent pregnancy it was suggested that pregnant adolescents might be particularly at risk for anaemia29. Ogbeidi et al. from Nigeria report a similar observation13. However, the numbers of women were very small in these studies. In addition the influence of age has often not been separated from the effect of parity. Most other studies from Africa have confirmed the findings of our study and reported that age alone is not an important determinant of haemoglobin values15,23,25,30.
Although there was no effect of parity on haemoglobin levels in studies from Tanzania23 an increased risk in primiparae has been documented by a number of other workers15,26. The most frequently given explanation for this has been that primiparae are known to have an increased susceptibility to malaria17,26. However, Matteelli et al. reported that although malaria was associated with anaemia in primiparae, the highest prevalence of anaemia was in multiparous patients17. In Mali seasonal variation in the prevalence of anaemia for both multiparous and primiparous women was suppressed when women received malaria prophylaxis31. It is possible that in Blantyre the seasonal difference in the prevalence of anaemia is due to the effect of malaria in primiparae, although this requires further study. Rogerson et al. have recently demonstrated that if only primiparous women are targeted for malaria prophylaxis in Blantyre, 40% of women with peripheral parasitaemia at time of booking would be missed. Attending the antenatal clinic in the malaria season and young age rather than low gravidity were found to be risk factors for malaria (unpublished information).
Women in Blantyre who booked very late in pregnancy had a significantly increased risk for anaemia overall and for severe anaemia, compared with women who first attend in the first trimester of pregnancy. Although our data are cross-sectional this finding is in accordance with the expected drop in haemoglobin levels as a result of plasma volume expansion in the course of normal pregnancy; in addition it may be expected that as a result of fetal demand and underlying maternal disease untreated anaemia in early pregnancy is likely to get worse. It is also possible that the rate of malaria is higher in Blantyre than in Namitambo and that lack of early malaria prophylaxis contributed to the increased risk of anaemia in the urban women who booked late.
In conclusion, the prevalence of anaemia in pregnancy is high in southern Malawi and can be assumed to be similarly high in surrounding countries in the region. Recent research has shown that the provision of iron with folic acid alone is insufficient to combat nutritional anaemia32. Other nutritional deficiencies including vitamin A and vitamin B12 are associated with anaemia in this population and multiple-micronutrient supplements are probably required. The role of infection in the aetiology of anaemia, including HIV infection, also requires further study21,33,34. The effect of malaria in causing anaemia is likely to vary from region to region. Malaria prophylaxis as currently prescribed may be insufficient35.
In our population, primiparae have an increased risk of anaemia and severe anaemia, which is independent of age; however, giving prophylaxis to primiparae only will exclude more than half of all women with anaemia overall and with severe anaemia. We recommend that all pregnant women in southern Malawi should receive suitable prophylaxis rather than a targetted group only.
The authors would like to thank members of the department of Obstetrics and Gynaecology at the Queen Elizabeth Central Hospital in Blantyre and nurses at both antenatal clinics for their help in facilitating this project.