Rethinking anaemia surveillance

Abstract

Iron-deficiency anaemia affects about 1·3 billion people1Gillespie S Kevany J Mason J Controlling iron deficiency. United Nations Administrative Committee on Coordination, Subcommittee of Nutrition (ACC/SCN), Geneva1991: 1-93Google Scholar and is the most common form of malnutrition in the world. In the past few years, there have been renewed efforts to control iron-deficiency anaemia. At the 1990 World Summit for Children, leaders from more than 159 countries agreed to try to reduce the prevalence of iron-deficiency anaemia by a third by the turn of the century.2FAO, WHO.Preventing specific micronutrient deficiencies: major issues for nutrition strategies. Theme paper no 6. WHO, Geneva1992Google Scholar, 3UNICEF.The state of the world's children, 1995. Oxford University Press, Oxford1995: 1-89Google Scholar In 1993, the World Bank4World Bank.World development report: investing in health. Oxford University Press, New York1993: 1-329Google Scholar ranked anaemia as the eighth leading cause of disease in girls and women in developing countries, and concluded that iron-deficiency anaemia could be effectively controlled with existing cost-effective interventions.Surveillance systems for anaemia are fundamental to public-health decision-making and for monitoring the progress of national and worldwide goals to control anaemia.5Mason JB Habicht J-P Tabatabai H Valverde V Nutritional surveillance. WHO, Geneva1984: 1-194Google Scholar Effective anaemia surveillance is essential to identify which populations are most in need of intervention and to assess whether programmes are working. As the World Declaration and Plan of Action for Nutrition6FAO, WHO.World declaration and plan of action for nutrition. FAO, WHO, Geneva1992: 1-39Google Scholar states, the international community should “strengthen micronutrient surveillance capabilities and activities by devising indicators to monitor… strategies for achieving national goals”. There is a need for simple indicator cut-offs that enable progress to be tracked and compared across regions, and for criteria on prevalence that identify the most severely affected populations. Such surveillance activities differ from population assessments of the underlying causes or interventions and from the clinical assessments of individuals, for which other methods are recommended.7Expert Scientific Working GroupSummary of a report on assessment of the iron nutritional status of the United States population.Am J Clin Nutr. 1985; 42: 1318-1330PubMed Google Scholar, 8Yip 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 Iron nutrition in health and disease. John Libby and Co, London1996: 31-48Google Scholar, 9Tyler RD Cowell RL Classification and diagnosis of anaemia.Comp Haematol Int. 1996; 6: 1-16Crossref Scopus (10) Google Scholar Haemoglobin concentration is the key indicator of iron-deficiency anaemia for surveillance purposes, especially since the advent of the HemoCue system (HemoCue, AB, Angelholm, Sweden) which measures haemoglobin concentration within seconds from a drop of whole blood without need of electricity.10von Schenck H Falkensson M Lundberg B Evaluation of “HemoCue”, a new device for determining hemoglobin.Clin Chem. 1986; 32: 526-529PubMed Google Scholar The current criteria for the assessment of the severity and magnitude of iron-deficiency anaemia in populations are the prevalence of any anaemia and severe anaemia (table 1).Table 1Epidemiological criteria for assessment of severity and magnitude of iron-deficiency anaemiaCriterionPopulation prevalence category (%)HighModerateLowAny anaemia*Haemoglobin concentration <110 g/L in pregnant women or young children, <120 g/L in schoolchildren or non-pregnant women, <130 g/L in adult men.>4010–391–9Severe anaemia†Haemoglobin concentrations <70 g/L.>101–9<0·1–0·9* Haemoglobin concentration <110 g/L in pregnant women or young children, <120 g/L in schoolchildren or non-pregnant women, <130 g/L in adult men.† Haemoglobin concentrations <70 g/L. Open table in a new tab Little progress seems to have been made against iron-deficiency anaemia. Based on surveillance data from 1975 to 1990, the United Nations reported that the prevalence of anaemia worldwide had not declined.” Similarly, in sub-Saharan Africa, a region challenged by war and economic instability, and in Southeast Asia, a region characterised by economic emergence and declining rates of other forms of malnutrition, the prevalence of iron-deficiency anaemia has not fallen. A midterm report of the objectives of the World Summit for Children warns that the goal to reduce anaemia by 2000 is “unlikely to be met without a significant acceleration of effort over the next 6 years”.3UNICEF.The state of the world's children, 1995. Oxford University Press, Oxford1995: 1-89Google ScholarIs it true that efforts to control iron-deficiency anaemia are failing, despite substantial health and economic improvements in some parts of the world? Table 2 shows haemoglobin concentrations for different populations of women. Although all the women are from less developed countries, they were not selected purposively. These data are from studies conducted since 1990 at the Center for Human Nutrition, Johns Hopkins University, Baltimore, USA; none of the studies should be interpreted as nationally representative. The Nepal data were collected in a continuing community-based study12Dreyfuss ML, Shrestha J, Khatry SK, et al. The prevalence of anemia among pregnant and lactating women, and among their infants in Sarlahi District. J Nepal Med Assoc (in press).Google Scholar in the Terai region, which is representative of the northern Gangetic flood plain of India and southern Nepal. Protein-energy malnutrition, vitamin A deficiency, and hookworm infection14Dreyfuss ML Shrestha JB Khatry SK et al.Relationship between iron status and helminth infection among pregnant women in Nepal.FASEB J. 1996; 10 (abstr).: A730Google Scholar are common in women from this region, but there is no programme to provide iron supplements to these women. The Zanzibar data were collected from a rural community-based survey on Pemba Island (Chwaya HM, unpublished observations), where acute malnutrition is common in children, and hookworms and Schistosoma haematobium and Plasmodium falciparum are endemic.15Stoltzfus RJ Chwaya HM Tielsch JM Schulze KJ Albonico M Savioli L Epidemiology of iron deficiency anemia in Zanzibari school children.Am J Clin Nutr. 1997; 65: 153-159Crossref PubMed Scopus (287) Google Scholar Iron supplements are provided to pregnant women by the Zanzibar Ministry of Health, but coverage is low because of limited tablet supply. The other three areas represent populations where malnutrition is declining and public-health infrastructures are reasonably good. In Peru and Indonesia, iron supplementation for pregnant women has been a long-standing health policy, and tablets are widely available. The Peru data were collected from women who attended antenatal clinics in low-altitude urban areas (Zavaleta N, Fukomoto M, unpublished observations), and the Indonesia data are from a rural community-based study in Central Java.16Stoltzfus RJ Hakimi J Miller KW et al.High dose vitamin A supplementation of breast-feeding Indonesian mothers: effects on the vitamin A status of mother and infant.J Nutr. 1993; 123: 666-675Crossref PubMed Scopus (157) Google Scholar The Shanghai population comprised women who received antenatal care at urban public hospitals.17Zhou L-M Yang W-W Hua J-Z Deng C-Q Tao X Stoltzfus RJ Hemoglobin concentration in early pregnancy and low birth weight.FASEB J. 1997; 11 (abstr).: A654Google Scholar Although this population is poor, acute malnutrition is rare and antenatal care is very good; iron supplements are available but are used by less than 10% of pregnant women.17Zhou L-M Yang W-W Hua J-Z Deng C-Q Tao X Stoltzfus RJ Hemoglobin concentration in early pregnancy and low birth weight.FASEB J. 1997; 11 (abstr).: A654Google ScholarTable 2Haemoglobin concentrations in different populations of womenPopulationNo% of population with haemoglobin concentrations (g/L)Below anaemia cut-off*Anaemia defined as haemoglobin concentrations <110 g/L in pregnant women or <120 g/L in non-pregnant women.<100<90<80<70Nepal, 3 months post partum61381·428·413·6<6·3<2·4Central Java, 3 months post partum14671·915·12·7<2·1<2·1Zanzibar, not pregnant58371·726·213·0<8·9<5·3Nepal, pregnant105269·840·520·8<9·5<4·9Shanghai, pregnant82966·225·35·5<0·8<0·2Peru, pregnant67044·314·54·3<0·1<0·1* Anaemia defined as haemoglobin concentrations <110 g/L in pregnant women or <120 g/L in non-pregnant women. Open table in a new tab From all we know about these populations—eg, their dietary adequacy, burdens of infectious diseases, and availability of health services—we would expect to find striking differences in the prevalence and severity of iron-deficiency anaemia. A good indicator should be able to separate these populations according to different degrees of risk of iron-deficiency anaemia. However, based on the current criteria for any anaemia, all the populations would be classified as areas with a high prevalence of iron-deficiency anaemia. By contrast, based on the prevalence of severe anaemia, none of the populations would have high rates of anaemia. But this opposition simply does not make sense. Either iron-supplementation programmes and general health indicators have no bearing on the prevalence of iron-deficiency anaemia, or the current criteria do not provide the information needed for the population burden of iron-deficiency anaemia to be assessed accurately. The answer lies in the lower end of the distribution of haemoglobin concentrations. The cumulative distributions in table 2 show the expected differences between the populations. There is only a 1·8-fold difference in the prevalence of anaemia between the populations; however, if haemoglobin concentrations below 90 g/L (moderate-to-severe anaemia) are used as the indicator, there is a 7·7-fold difference between populations. Some comparisons are illuminating. The prevalence of anaemia is similar among non-pregnant women in Central Java and Zanzibar. However, the prevalence of moderate-to-severe anaemia is five times higher in Zanzibar than in Java. The rate of anaemia in pregnant women is similar in rural Nepal and Shanghai, but the prevalence of moderate-to-severe anaemia is 2·5 times higher among Nepalese women than among women from Shanghai.Severe anaemia (haemoglobin concentration <70 g/L) is closely linked to risk of mortality, but is difficult to monitor because the prevalence rarely exceeds 5–7%, even in populations with a high prevalence of iron-deficiency anaemia. Furthermore, this indicator does not always distinguish between populations with very different haemoglobin distributions—eg, Nepalese and Central Javanese women in the post-partum period (table 2). Changing the cut-off for haemoglobin concentration in anaemia surveillance is such a simple idea that it seems unimportant, but it is not. Let us return to two key objectives of anaemia surveillance: to identify population groups in greatest need of intervention and to monitor progress in anaemia control. The first objective enables donors and governments to use scarce health resources wisely and the second enables them to set goals and see if they are met. How would public-health decision-making be influenced by surveillance of moderate-to-severe anaemia—ie, a haemoglobin cut-off of 90 g/L?First, public-health resources for anaemia control would be allocated differently. The prevalence of all anaemia or moderate-to-severe anaemia give different indications of the severity of anaemia in the population groups (table 3). The current perception is that anaemia is widespread in all countries. Fund donors, who need to address health issues that are well-defined and amenable to solution, find it difficult to focus their programme efforts on anaemia. Governments of developing countries are overwhelmed with health problems, and may not feel compelled to tackle anaemia when their better-off neighbours report similar rates of anaemia. But, table 3 shows that there are differences in the prevalence of iron-deficiency anaemia between countries and regions, and highlights the need for action in Nepal and Zanzibar.Table 3Iron-deficiency anaemia in different populations by prevalence of all anaemia and prevalence of moderate-to-severe anaemiaPopulationNoPrevalence of iron-deficiency anaemia (%)Defined by prevalence of all anaemia*Haemoglobin concentrations <110 g/L in pregnant women, <120 g/L in women who are not pregnant.Defined by prevalence of moderate-to-severe anaemia (haemoglobin <90 g/L)Nepal, 3 months post partum6138114Central Java, 3 months post partum146723Zanzibar, not pregnant5837213Nepal, pregnant10527021Shanghai, pregnant829666Peru, pregnant670444* Haemoglobin concentrations <110 g/L in pregnant women, <120 g/L in women who are not pregnant. Open table in a new tab Second, in developing countries, the prevalence of moderate-to-severe anaemia is more likely to indicate progress in anaemia control if other health improvements are being made. The prevalence of all anaemia in Indonesian women has remained above 50% during the past 30 years, even though progress has been made in most other health indicators, especially nutrition. It would be interesting to know whether the prevalence of moderate-to-severe anaemia has declined or whether it has always been low. I suspect it has declined greatly. Indeed, the lower end of the distribution of a homoeostatically controlled indicator, such as haemoglobin concentration, would be likely to change more rapidly than the upper end. If a successful programme for anaemia control were implemented in Nepal or Zanzibar, the distribution of haemoglobin concentrations might first shift to become like Indonesia—ie, with high prevalence of all anaemia but low prevalence of moderate-to-severe anaemia. This distribution would represent substantial progress, but would not be revealed with surveillance that relied on the prevalence of all anaemia.If we begin to use the prevalence of moderate-to-severe anaemia for surveillance purposes, it is important not to lose sight of the prevalence of mild anaemia. A survey of haemoglobin concentrations provides data on both mild and moderate-to-severe anaemia. For countries where the prevalence of mild or all anaemia has already fallen below 50%, such as Latin America, the Middle East, and North Africa, the prevalence of all anaemia will measure the lower end of the haemoglobin distribution and is the best indicator for monitoring progress in the control of iron-deficiency anaemia. A suggestion to monitor the prevalence of moderate-to-severe anaemia in regions where the prevalence of iron-deficiency anaemia is high does not imply that there are no health risks associated with mild anaemia. Mild iron-deficiency anaemia affects cognitive capacity in children,18Cook JD Lynch SR The liabilities of iron deficiency.Blood. 1986; 68: 803-809PubMed Google Scholar increases the risk of preterm delivery in pregnant women,17Zhou L-M Yang W-W Hua J-Z Deng C-Q Tao X Stoltzfus RJ Hemoglobin concentration in early pregnancy and low birth weight.FASEB J. 1997; 11 (abstr).: A654Google Scholar, 19Scholl TO Hediger ML Fischer RL Shearer JW Anemia vs iron deficiency: increased risk of preterm delivery in a prospective study.Am J Clin Nutr. 1992; 55: 985-988Crossref PubMed Scopus (432) Google Scholar and reduces work output in all individuals.20Haas JD Fairchild MW Summary and conclusions of the International Conference on Iron Deficiency and Behavioural Development, Oct 10–12, 1988.Am J Clin Nutr. 1989; 50: 703-705Google Scholar Public-health interventions to control iron-deficiency anaemia are needed in all the populations listed in table 2. But the addition of a lower haemoglobin cut-off for surveillance purposes is important because the health risks associated with moderate-to-severe anaemia are substantially greater; it is within this range that anaemia contributes directly to mortality among women and children.A helpful analogy is the use of the prevalence of moderate-to-severe stunting or wasting (more than 2 SD below the reference mean) to monitor protein-energy malnutrition worldwide, even though many children who are 1 SD below the reference are malnourished and experience adverse health risks.21WHO Expert Committee.Physical status: the use and interpretation of anthropometry. WHO Technical Report Series no 854. WHO, Geneva1995: 1-452Google Scholar Similarly, use of low birthweight defined as below 2500 g to monitor rates of low birthweight does not mean that the risk of mortality for a baby who weighs 2600 g is the same as that for a 3300 g baby.22Shapiro S McCormick MC Starfield BH Krisher JP Bross D Relevance of correlates of infant deaths for significant morbidity at 1 year of age.Am J Obstet Gynecol. 1980; 136: 363-373PubMed Scopus (137) Google Scholar Surveillance indicators are chosen because they serve the purposes of surveillance, not because they indicate the point of minimum risk. Despite effective interventions, control of iron-deficiency anaemia has not proved to be easy. We need to intensify our surveillance of anaemia so that the progress we make can be assessed, especially in those countries with a high prevelence of anaemia. A simple change in the haemoglobin cut-offs used for anaemia surveillance would sharpen our focus.I thank Keith West, Michele Dreyfuss, Li-Ming Zhou, Nelly Zavaleta, and Hababu Mohammad Chwaya for making their data on anaemia prevalence available for my analysis, and Jim Tielsch for discussions that spawned these ideas. This study was supported by cooperative agreements DAN 0045 and DAV 5116 between the Center for Human Nutrition, Johns Hopkins University and the Office of Health and Nutrition, US Agency for International Development. Iron-deficiency anaemia affects about 1·3 billion people1Gillespie S Kevany J Mason J Controlling iron deficiency. United Nations Administrative Committee on Coordination, Subcommittee of Nutrition (ACC/SCN), Geneva1991: 1-93Google Scholar and is the most common form of malnutrition in the world. In the past few years, there have been renewed efforts to control iron-deficiency anaemia. At the 1990 World Summit for Children, leaders from more than 159 countries agreed to try to reduce the prevalence of iron-deficiency anaemia by a third by the turn of the century.2FAO, WHO.Preventing specific micronutrient deficiencies: major issues for nutrition strategies. Theme paper no 6. WHO, Geneva1992Google Scholar, 3UNICEF.The state of the world's children, 1995. Oxford University Press, Oxford1995: 1-89Google Scholar In 1993, the World Bank4World Bank.World development report: investing in health. Oxford University Press, New York1993: 1-329Google Scholar ranked anaemia as the eighth leading cause of disease in girls and women in developing countries, and concluded that iron-deficiency anaemia could be effectively controlled with existing cost-effective interventions. Surveillance systems for anaemia are fundamental to public-health decision-making and for monitoring the progress of national and worldwide goals to control anaemia.5Mason JB Habicht J-P Tabatabai H Valverde V Nutritional surveillance. WHO, Geneva1984: 1-194Google Scholar Effective anaemia surveillance is essential to identify which populations are most in need of intervention and to assess whether programmes are working. As the World Declaration and Plan of Action for Nutrition6FAO, WHO.World declaration and plan of action for nutrition. FAO, WHO, Geneva1992: 1-39Google Scholar states, the international community should “strengthen micronutrient surveillance capabilities and activities by devising indicators to monitor… strategies for achieving national goals”. There is a need for simple indicator cut-offs that enable progress to be tracked and compared across regions, and for criteria on prevalence that identify the most severely affected populations. Such surveillance activities differ from population assessments of the underlying causes or interventions and from the clinical assessments of individuals, for which other methods are recommended.7Expert Scientific Working GroupSummary of a report on assessment of the iron nutritional status of the United States population.Am J Clin Nutr. 1985; 42: 1318-1330PubMed Google Scholar, 8Yip 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 Iron nutrition in health and disease. John Libby and Co, London1996: 31-48Google Scholar, 9Tyler RD Cowell RL Classification and diagnosis of anaemia.Comp Haematol Int. 1996; 6: 1-16Crossref Scopus (10) Google Scholar Haemoglobin concentration is the key indicator of iron-deficiency anaemia for surveillance purposes, especially since the advent of the HemoCue system (HemoCue, AB, Angelholm, Sweden) which measures haemoglobin concentration within seconds from a drop of whole blood without need of electricity.10von Schenck H Falkensson M Lundberg B Evaluation of “HemoCue”, a new device for determining hemoglobin.Clin Chem. 1986; 32: 526-529PubMed Google Scholar The current criteria for the assessment of the severity and magnitude of iron-deficiency anaemia in populations are the prevalence of any anaemia and severe anaemia (table 1). Little progress seems to have been made against iron-deficiency anaemia. Based on surveillance data from 1975 to 1990, the United Nations reported that the prevalence of anaemia worldwide had not declined.” Similarly, in sub-Saharan Africa, a region challenged by war and economic instability, and in Southeast Asia, a region characterised by economic emergence and declining rates of other forms of malnutrition, the prevalence of iron-deficiency anaemia has not fallen. A midterm report of the objectives of the World Summit for Children warns that the goal to reduce anaemia by 2000 is “unlikely to be met without a significant acceleration of effort over the next 6 years”.3UNICEF.The state of the world's children, 1995. Oxford University Press, Oxford1995: 1-89Google Scholar Is it true that efforts to control iron-deficiency anaemia are failing, despite substantial health and economic improvements in some parts of the world? Table 2 shows haemoglobin concentrations for different populations of women. Although all the women are from less developed countries, they were not selected purposively. These data are from studies conducted since 1990 at the Center for Human Nutrition, Johns Hopkins University, Baltimore, USA; none of the studies should be interpreted as nationally representative. The Nepal data were collected in a continuing community-based study12Dreyfuss ML, Shrestha J, Khatry SK, et al. The prevalence of anemia among pregnant and lactating women, and among their infants in Sarlahi District. J Nepal Med Assoc (in press).Google Scholar in the Terai region, which is representative of the northern Gangetic flood plain of India and southern Nepal. Protein-energy malnutrition, vitamin A deficiency, and hookworm infection14Dreyfuss ML Shrestha JB Khatry SK et al.Relationship between iron status and helminth infection among pregnant women in Nepal.FASEB J. 1996; 10 (abstr).: A730Google Scholar are common in women from this region, but there is no programme to provide iron supplements to these women. The Zanzibar data were collected from a rural community-based survey on Pemba Island (Chwaya HM, unpublished observations), where acute malnutrition is common in children, and hookworms and Schistosoma haematobium and Plasmodium falciparum are endemic.15Stoltzfus RJ Chwaya HM Tielsch JM Schulze KJ Albonico M Savioli L Epidemiology of iron deficiency anemia in Zanzibari school children.Am J Clin Nutr. 1997; 65: 153-159Crossref PubMed Scopus (287) Google Scholar Iron supplements are provided to pregnant women by the Zanzibar Ministry of Health, but coverage is low because of limited tablet supply. The other three areas represent populations where malnutrition is declining and public-health infrastructures are reasonably good. In Peru and Indonesia, iron supplementation for pregnant women has been a long-standing health policy, and tablets are widely available. The Peru data were collected from women who attended antenatal clinics in low-altitude urban areas (Zavaleta N, Fukomoto M, unpublished observations), and the Indonesia data are from a rural community-based study in Central Java.16Stoltzfus RJ Hakimi J Miller KW et al.High dose vitamin A supplementation of breast-feeding Indonesian mothers: effects on the vitamin A status of mother and infant.J Nutr. 1993; 123: 666-675Crossref PubMed Scopus (157) Google Scholar The Shanghai population comprised women who received antenatal care at urban public hospitals.17Zhou L-M Yang W-W Hua J-Z Deng C-Q Tao X Stoltzfus RJ Hemoglobin concentration in early pregnancy and low birth weight.FASEB J. 1997; 11 (abstr).: A654Google Scholar Although this population is poor, acute malnutrition is rare and antenatal care is very good; iron supplements are available but are used by less than 10% of pregnant women.17Zhou L-M Yang W-W Hua J-Z Deng C-Q Tao X Stoltzfus RJ Hemoglobin concentration in early pregnancy and low birth weight.FASEB J. 1997; 11 (abstr).: A654Google Scholar From all we know about these populations—eg, their dietary adequacy, burdens of infectious diseases, and availability of health services—we would expect to find striking differences in the prevalence and severity of iron-deficiency anaemia. A good indicator should be able to separate these populations according to different degrees of risk of iron-deficiency anaemia. However, based on the current criteria for any anaemia, all the populations would be classified as areas with a high prevalence of iron-deficiency anaemia. By contrast, based on the prevalence of severe anaemia, none of the populations would have high rates of anaemia. But this opposition simply does not make sense. Either iron-supplementation programmes and general health indicators have no bearing on the prevalence of iron-deficiency anaemia, or the current criteria do not provide the information needed for the population burden of iron-deficiency anaemia to be assessed accurately. The answer lies in the lower end of the distribution of haemoglobin concentrations. The cumulative distributions in table 2 show the expected differences between the populations. There is only a 1·8-fold difference in the prevalence of anaemia between the populations; however, if haemoglobin concentrations below 90 g/L (moderate-to-severe anaemia) are used as the indicator, there is a 7·7-fold difference between populations. Some comparisons are illuminating. The prevalence of anaemia is similar among non-pregnant women in Central Java and Zanzibar. However, the prevalence of moderate-to-severe anaemia is five times higher in Zanzibar than in Java. The rate of anaemia in pregnant women is similar in rural Nepal and Shanghai, but the prevalence of moderate-to-severe anaemia is 2·5 times higher among Nepalese women than among women from Shanghai. Severe anaemia (haemoglobin concentration <70 g/L) is closely linked to risk of mortality, but is difficult to monitor because the prevalence rarely exceeds 5–7%, even in populations with a high prevalence of iron-deficiency anaemia. Furthermore, this indicator does not always distinguish between populations with very different haemoglobin distributions—eg, Nepalese and Central Javanese women in the post-partum period (table 2). Changing the cut-off for haemoglobin concentration in anaemia surveillance is such a simple idea that it seems unimportant, but it is not. Let us return to two key objectives of anaemia surveillance: to identify population groups in greatest need of intervention and to monitor progress in anaemia control. The first objective enables donors and governments to use scarce health resources wisely and the second enables them to set goals and see if they are met. How would public-health decision-making be influenced by surveillance of moderate-to-severe anaemia—ie, a haemoglobin cut-off of 90 g/L? First, public-health resources for anaemia control would be allocated differently. The prevalence of all anaemia or moderate-to-severe anaemia give different indications of the severity of anaemia in the population groups (table 3). The current perception is that anaemia is widespread in all countries. Fund donors, who need to address health issues that are well-defined and amenable to solution, find it difficult to focus their programme efforts on anaemia. Governments of developing countries are overwhelmed with health problems, and may not feel compelled to tackle anaemia when their better-off neighbours report similar rates of anaemia. But, table 3 shows that there are differences in the prevalence of iron-deficiency anaemia between countries and regions, and highlights the need for action in Nepal and Zanzibar. Second, in developing countries, the prevalence of moderate-to-severe anaemia is more likely to indicate progress in anaemia control if other health improvements are being made. The prevalence of all anaemia in Indonesian women has remained above 50% during the past 30 years, even though progress has been made in most other health indicators, especially nutrition. It would be interesting to know whether the prevalence of moderate-to-severe anaemia has declined or whether it has always been low. I suspect it has declined greatly. Indeed, the lower end of the distribution of a homoeostatically controlled indicator, such as haemoglobin concentration, would be likely to change more rapidly than the upper end. If a successful programme for anaemia control were implemented in Nepal or Zanzibar, the distribution of haemoglobin concentrations might first shift to become like Indonesia—ie, with high prevalence of all anaemia but low prevalence of moderate-to-severe anaemia. This distribution would represent substantial progress, but would not be revealed with surveillance that relied on the prevalence of all anaemia. If we begin to use the prevalence of moderate-to-severe anaemia for surveillance purposes, it is important not to lose sight of the prevalence of mild anaemia. A survey of haemoglobin concentrations provides data on both mild and moderate-to-severe anaemia. For countries where the prevalence of mild or all anaemia has already fallen below 50%, such as Latin America, the Middle East, and North Africa, the prevalence of all anaemia will measure the lower end of the haemoglobin distribution and is the best indicator for monitoring progress in the control of iron-deficiency anaemia. A suggestion to monitor the prevalence of moderate-to-severe anaemia in regions where the prevalence of iron-deficiency anaemia is high does not imply that there are no health risks associated with mild anaemia. Mild iron-deficiency anaemia affects cognitive capacity in children,18Cook JD Lynch SR The liabilities of iron deficiency.Blood. 1986; 68: 803-809PubMed Google Scholar increases the risk of preterm delivery in pregnant women,17Zhou L-M Yang W-W Hua J-Z Deng C-Q Tao X Stoltzfus RJ Hemoglobin concentration in early pregnancy and low birth weight.FASEB J. 1997; 11 (abstr).: A654Google Scholar, 19Scholl TO Hediger ML Fischer RL Shearer JW Anemia vs iron deficiency: increased risk of preterm delivery in a prospective study.Am J Clin Nutr. 1992; 55: 985-988Crossref PubMed Scopus (432) Google Scholar and reduces work output in all individuals.20Haas JD Fairchild MW Summary and conclusions of the International Conference on Iron Deficiency and Behavioural Development, Oct 10–12, 1988.Am J Clin Nutr. 1989; 50: 703-705Google Scholar Public-health interventions to control iron-deficiency anaemia are needed in all the populations listed in table 2. But the addition of a lower haemoglobin cut-off for surveillance purposes is important because the health risks associated with moderate-to-severe anaemia are substantially greater; it is within this range that anaemia contributes directly to mortality among women and children. A helpful analogy is the use of the prevalence of moderate-to-severe stunting or wasting (more than 2 SD below the reference mean) to monitor protein-energy malnutrition worldwide, even though many children who are 1 SD below the reference are malnourished and experience adverse health risks.21WHO Expert Committee.Physical status: the use and interpretation of anthropometry. WHO Technical Report Series no 854. WHO, Geneva1995: 1-452Google Scholar Similarly, use of low birthweight defined as below 2500 g to monitor rates of low birthweight does not mean that the risk of mortality for a baby who weighs 2600 g is the same as that for a 3300 g baby.22Shapiro S McCormick MC Starfield BH Krisher JP Bross D Relevance of correlates of infant deaths for significant morbidity at 1 year of age.Am J Obstet Gynecol. 1980; 136: 363-373PubMed Scopus (137) Google Scholar Surveillance indicators are chosen because they serve the purposes of surveillance, not because they indicate the point of minimum risk. Despite effective interventions, control of iron-deficiency anaemia has not proved to be easy. We need to intensify our surveillance of anaemia so that the progress we make can be assessed, especially in those countries with a high prevelence of anaemia. A simple change in the haemoglobin cut-offs used for anaemia surveillance would sharpen our focus. I thank Keith West, Michele Dreyfuss, Li-Ming Zhou, Nelly Zavaleta, and Hababu Mohammad Chwaya for making their data on anaemia prevalence available for my analysis, and Jim Tielsch for discussions that spawned these ideas. This study was supported by cooperative agreements DAN 0045 and DAV 5116 between the Center for Human Nutrition, Johns Hopkins University and the Office of Health and Nutrition, US Agency for International Development.