SEARCH

SEARCH BY CITATION

Keywords:

  • anaemia;
  • haemoglobin;
  • altitude;
  • cigarette smoking;
  • pregnancy
  • anémie;
  • hémoglobine;
  • altitude;
  • tabagisme;
  • grossesse
  • anemia;
  • hemoglobina;
  • altitud;
  • fumar;
  • embarazo

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Objective  To provide researchers with an unambiguous definition of anaemia using haemoglobin.

Methods  Review of recommendations by expert groups and review of the literature.

Results  This report provides an unambiguous approach to haemoglobin adjustments to define anaemia using international criteria. When determining anaemia using haemoglobin, it is important to account for pregnancy, altitude, cigarette smoking, and possibly ethnicity after removing unlikely values. These haemoglobin adjustments are presented.

Conclusion  Recommendations for defining extreme haemoglobin values and for reporting anaemia and haemoglobin results are provided, and software programs to determine anaemia are described.

Objectif:  Fournir aux chercheurs une définition sans ambiguïté de l’anémie basée sur l’hémoglobine.

Méthode:  Examen des recommandations par des groupes d’experts et revue de la littérature.

Résultats:  Ce rapport fournit une approche sans ambiguïté des ajustements de l’hémoglobine pour définir l’anémie sur base des critères internationaux. Lors de la détermination de l’anémie sur base de l’hémoglobine, il est important de tenir compte de la grossesse, de l’altitude, du tabagisme et de l’origine ethnique, après élimination des valeurs improbables. Ces ajustements de l’hémoglobine sont présentés.

Conclusion:  Des recommandations pour la définition de valeurs extrêmes d’hémoglobine et pour le report des résultats de l’anémie et de l’hémoglobine sont fournies. Des logiciels pour déterminer l’anémie sont aussi décrits.

Objetivo:  Proveer a los investigadores con una definición no ambigua de anemia, utilizando la hemoglobina.

Método:  Revisar las recomendaciones realizadas por grupos de expertos así como la literatura existente.

Resultados:  Este informe provee un acercamiento no ambiguo a los ajustes de hemoglobina para definir la anemia, utilizando criterios internacionales. Cuando se determina la anemia usando la hemoglobina, es importante tener en cuenta las variables de embarazo, altitud, si se es fumador o no, así como posiblemente la etnicidad, después de haber sacado los valores dudosos. Los ajustes de hemoglobina son presentados.

Conclusión:  Se dan recomendaciones para definir valores extremos de hemoglobina y para reportar resultados de anemia y hemoglobina. Se describen programas de software para determinar anemia.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Anaemia is an important public health problem that negatively affects the health of individuals and the economic potential of populations. In non-industrialised countries, the prevalence of anaemia is estimated to be 52% in pregnant women and 39% in children under 5 years; and in industrialised countries, 23% and 20%, respectively (UNICEF, UNU, WHO 2001). Assessments of anaemia prevalence in populations assist in determining its magnitude and distribution. At the population level, it is important to correctly classify anaemia status to estimate a valid prevalence. Haemoglobin is affected by several factors that need to be accounted for when determining whether an individual is anaemic: age, sex, pregnancy, altitude, cigarette smoking and ethnicity. Two publications provide information on haemoglobin cutoff values and adjustments (UNICEF, UNU, WHO 2001; Nestel & The INACG Steering Committee 2002). There are minor differences between these documents and in some instances exact boundaries between categories are unclear. Our aim was to provide an unambiguous definition for anaemia, and to discuss minimum and maximum acceptable haemoglobin values, use of haemoglobin standard deviation as a measure of data quality, adjusting cutoffs for severe anaemia in pregnancy, and recommendations on the presentation of anaemia and haemoglobin results from cross-sectional surveys.

Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Two reports and supporting documents were reviewed that address adjustment of haemoglobin levels (UNICEF, UNU, WHO 2001; Nestel and The INACG Steering Committee 2002). For differences between the UNICEF/UNU/WHO and INACG documents, we present the former because it was based on a consensus of a larger group of experts and endorsed by international agencies.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Recommended haemoglobin cutoffs and adjustments for determining anaemia

The cutoffs for haemoglobin by age, and for those 15 years and older, by sex, not pregnant, at an altitude <1000 m, non-smokers, and not of African extraction are depicted in Table 1. Individuals with a haemoglobin concentration below their cutoff are considered anaemic; those with a haemoglobin concentration greater than or equal to the cutoff are considered not anaemic. Before determining whether an individual is anaemic, there is a need to take into account pregnancy, altitude, cigarette smoking and ethnicity, as appropriate. Table 2 provides adjustments for these factors. The adjustments can be subtracted from or added to the haemoglobin cutoff values presented in Table 1 or, as discussed later, subtracted from or added to each individual’s observed haemoglobin value to derive an ‘adjusted’ haemoglobin value.

Table 1.   Haemoglobin cutoffs to define anaemia in non-smoking, non-pregnant, non-African extraction individuals living at an altitude <1000 m
Age and sexHaemoglobin cutoff (g/dl)
  1. UNICEF/UNU/WHO and INACG.

Children (males and females)
 ≥0.5, <5.0 years11.0
 ≥5.0, <12.0 years11.5
 ≥12.0, <15.0 years12.0
Non-pregnant females ≥ 15.0 years12.0
Men ≥ 15.0 years13.0
Table 2.   Adjustments to haemoglobin cutoffs and individual haemoglobin values for pregnancy, altitude, cigarette smoking and ethnicity
Trimester of pregnancyAdjustment to haemoglobin cutoff value (g/dl)Adjustment to individual haemoglobin value (g/dl)
  1. The INACG document has adjustments by trimester; the UNICEF/UNU/WHO and INACG documents provide adjustments for pregnant vs. not pregnant. For altitude, the haemoglobin adjustments are based on the UNICEF/UNU/WHO document except for the highest two altitude values, which are based on the INACG document. A more accurate adjustment for altitude can be made based on the application of a formula.

First−1.0+1.0
Second−1.5+1.5
Third−1.0+1.0
Trimester unknown−1.0+1.0
Altitude (m)
 <1000No adjustmentNo adjustment
 ≥1000, <1250+0.2−0.2
 ≥1250, <1750+0.5−0.5
 ≥1750, <2250+0.8−0.8
 ≥2250, <2750+1.3−1.3
 ≥2750, <3250+1.9−1.9
 ≥3250, <3750+2.7−2.7
 ≥3750, <4250+3.5−3.5
 >4250, <4750+4.5−4.5
 ≥4750, <5250+5.5−5.5
 ≥5250+6.7−6.7
Cigarettes smoked per day
 <10No adjustmentNo adjustment
 ≥10, <20+0.3−0.3
 ≥20, <40+0.5−0.5
 ≥40+0.7−0.7
 Smoker, amount unknown+0.3−0.3
Ethnicity
 African extraction−1.0+1.0

Adjustments for altitude are necessary to account for a reduction in oxygen saturation of blood (Hurtado et al. 1945; CDC 1989). The adjustment for altitude can be based on those presented in Table 2 or the following formula:

  • image

where the Hb adjustment is the amount added to the haemoglobin cutoff in Table 1 or subtracted from each individual’s observed haemoglobin level. As shown in Table 2, there are large differences in haemoglobin adjustments at higher altitudes, and the differences between the tabular vs. the formula approach for adjusting for haemoglobin can result in slightly different prevalence estimates of anaemia. Using data from a national survey among children under 5 years in Afghanistan, the tabular approach to adjusting haemoglobin levels resulted in a prevalence of 37.9% (Afghanistan Ministry of Health and UNICEF 2005), whereas the formula approach estimated a prevalence of 36.6% (data not shown).

Cigarette smoking increases carboxyhaemoglobin, resulting in an increase in haemoglobin (Nordenberg et al. 1990). Cigarette smoking was defined as ‘more than 100 cigarettes ever smoked and considered a cigarette smoker at the time of the interview’ and adjustments for cigarette smoking are presented in Table 2 (Nordenberg et al. 1990). Note the difference between the smoking categories in Table 2 compared with those in the UNICEF/UNU/WHO and INACG documents; Table 2 categorises data by average number of cigarettes per day; the other documents present the average number of packs of cigarettes smoked per day. The original data were based in the US where most cigarette packs have 20 cigarettes; however in some countries, cigarettes come in package sizes from 10 to 50 cigarettes per pack (Scollo et al. 2003). Neither the UNICEF/UNU/WHO nor INACG documents provide information about the age or sex groups to which the smoking adjustments should be applied. The original article that recommended haemoglobin adjustments for smoking was based on US white women aged 18–44 years (Nordenberg et al. 1990). It would seem reasonable to apply smoking adjustments to younger individuals, where in some areas there are high levels of initiation in children as young as 10 years, as well as to pregnant women, and men.

In the US, healthy individuals of African extraction have lower haemoglobin levels than do other ethnic groups independent of iron deficiency and therefore it was recommended that their cutoff value should be lower (Table 2; Dallman et al. 1978; Perry et al. 1992; Johnson-Spear & Yip 1994; Beutler & West 2005). However, in terms of identifying iron deficiency, some groups do not agree with the recommendation of having separate haemoglobin cutoff values for those of African extraction (CDC 1999).

Adjustment of cutoff values vs. adjustment of individual haemoglobin values

There are two ways to apply the haemoglobin adjustments for pregnancy, altitude, smoking and ethnicity. One way is to adjust the anaemia cutoff value; the other is to adjust each individual’s observed haemoglobin value (Table 2). The first method is presented in the UNICEF/UNU/WHO and INACG documents. For example, to determine whether a 2-year-old child living at an altitude of 2000 m is anaemic, one could take the haemoglobin cutoff value at sea level (11.0 g/dl) and adjust the cutoff for altitude (+0.8 g/dl), yielding an adjusted cutoff value of 11.8 g/dl. This approach could be used at health clinics located at different altitudes or by survey teams to determine anaemia status of patients or survey participants.

The second approach, presented in the INACG document, is to adjust each individual’s haemoglobin value and compare the adjusted haemoglobin to the appropriate cutoff values presented in Table 1. In essence, each individual’s haemoglobin is adjusted to estimate what his or her haemoglobin level would be at sea level (<1000 m), not pregnant nor a cigarette smoker. This approach would most likely be used when analysing survey or clinic-based electronic data files. For example, a 2-year-old male living at an altitude of 2000 m would have his observed haemoglobin adjusted through the subtraction of 0.8 g/dl, and the resulting value compared with the haemoglobin cutoff value of 11.0 g/dl from Table 1. Both approaches lead to the same conclusion as to whether an individual is anaemic; however, the second approach has an advantage in the analysis of haemoglobin data from populations. Namely, the SD of the haemoglobin concentrations can be used to assess the quality of the haemoglobin data in a population, such as from a cross-sectional survey or a clinic-based reporting system. In general, a smaller SD around the mean haemoglobin is considered to represent better data quality than a larger SD. Factors that affect the width of the haemoglobin distribution include biological variability, distribution of the various causes of anaemia within the population, use of venous vs. capillary blood, and the quality of procedures and methods for the collection and testing of haemoglobin. SDs may differ by age, sex and other factors. In the authors’ empirical experience, haemoglobin SDs from cross-sectional surveys or surveillance systems with apparently acceptable quality technique using the HemoCue® system tend to be in the 1.1–1.5 range. The use of the haemoglobin SD can be useful within a survey to compare survey teams or in clinic-based surveillance systems to compare data quality among clinics.

The SD is generally smaller when based on adjusted individual haemoglobin values (adjusted for altitude and other factors) than when based on observed haemoglobin values. For example, we present data on children from Afghanistan aged 6–59.9 months (Table 3; Afghanistan Ministry of Health and UNICEF 2005). In order to focus attention on the effect of altitude on haemoglobin and anaemia, these data are presented without taking into account the complex sample design. At higher altitudes, the mean observed haemoglobin is higher. The haemoglobin cutoff value for defining anaemia differs in each of the altitude categories; in this age group, the cutoff would range from 11.0 to 12.9 g/dl going from low to high altitudes. The mean and SD of the observed (i.e., not adjusted for altitude) and adjusted (adjusted for altitude) haemoglobin are presented. Note that the SD is larger for the overall average observed mean haemoglobin than the overall adjusted mean haemoglobin because the SD for the former is affected by both the within-group and between-group variability. The between-group variability is reduced when based on adjusted haemoglobin values, and therefore the overall SD is smaller. Also shown in Table 3 is a comparison of the prevalence of anaemia based on the observed haemoglobin values (i.e., not adjusted for altitude) and haemoglobin adjustment for altitude, with the former resulting in an underestimate of the prevalence.

Table 3.   Summary statistics of observed and adjusted haemoglobin values (g/dl) and prevalence of anaemia by altitude, children aged 6–59 months, Afghanistan 2004
Altitude (m)n Observed haemoglobin (g/dl) Adjusted haemoglobin† (g/dl)Prevalence (%) of anaemia based on observed and adjusted haemoglobin
MeanSDMeanSDObservedAdjusted†
  1. The calculation of means, SD and prevalence estimates in this table does not account for the complex sample design of the survey in order to demonstrate the effect of altitude adjustment. Adjustments for altitude were based on Table 2.

  2. †Adjusted for altitude.

<100037311.31.3711.31.3734.934.9
≥1000, <12505511.21.3411.01.3445.549.1
≥1250, <17509711.21.8710.71.8734.048.5
≥1750, <225022712.21.3811.41.3816.332.6
≥2250, <27507912.71.1411.41.145.138.0
≥2750, <32503913.31.3211.41.322.638.5
Total/avg.87011.71.5311.21.4326.437.1

Recommended haemoglobin cutoffs for determining severe anaemia in pregnancy

Severe anaemia in pregnancy is defined as a haemoglobin <7.0 g/dl (UNICEF/UNU/WHO 2001). This value is appropriate at sea level, but there should be adjustments for altitude and cigarette smoking. A table for these cutoffs is available from the authors.

Recommended minimum and maximum acceptable haemoglobin values

Unlikely haemoglobin values should be scrutinised and possibly discarded after taking into account pregnancy, altitude and smoking. Based on empirical information, the authors generally recommend minimum and maximum adjusted haemoglobin values of 4 and 18 g/dl, respectively, for non-pregnant women and children; for men, 4 and 20 g/dl, respectively. From a statistical viewpoint, the probability of observing values below the minimum or above the maximum value are <0.0002 using a range of likely mean and SD estimates (data not shown). These minimum and maximum values may not be appropriate in all settings. For example, the UNICEF/UNU/WHO document defines very severe anaemia in pregnancy as a haemoglobin <4 g/dl. The recommended approach to using these minimum and maximum acceptable values would be first to adjust observed haemoglobin values to account for pregnancy, altitude and number of cigarettes smoked per day and then remove values outside the acceptable range. If the adjustment to the haemoglobin cutoff approach is used, there will need to be different minimum and maximum cutoff values for the various altitude levels and smoking categories (information available from the authors).

Presentation of anaemia and haemoglobin results

To present anaemia and haemoglobin results, we recommend providing the prevalence of anaemia with 95% confidence intervals (CIs), and if the data were collected using a complex sample design, the design effect (see example in Table 4). For haemoglobin, we recommend providing the mean, 95% CI and SD. The results should be presented by age group and, when possible, by other demographic variables such as sex, socioeconomic status, urban/rural location and geographic area. Results for adult women should be provided separately for pregnant (by trimester if possible) and non-pregnant women.

Table 4.   Prevalence of anaemia and mean haemoglobin levels among pre-school children aged 6–59 months, Afghanistan 2004
Characteristics of pre-school childrennAnaemic†Haemoglobin (g/dl)†
Percentage (95% CI)Mean (95% CI) [SD]‡
  1. Point estimates and confidence intervals (CIs) account for cluster design of survey with sample weighting; sex was not noted on the survey for six children. Overall design effect for anaemia was 2.3. The overall severity of anaemia in this population is ‘moderate’ based on the following classification scheme: normal, prevalence of anaemia <5.0%; mild, 5.0–19.9%; moderate, 20.0–39.9% and severe, >40% (UNICEF, UNU, WHO 2001).

  2. †Anaemia and haemoglobin results adjusted for altitude.

  3. ‡The SD is presented as a measure of the quality of the haemoglobin values and does not account for the complex survey design.

Age group (mos.)
 ≥6, <129433.1 (21.7, 44.6)11.3 (11.0, 11.6) [1.2]
 ≥12, <2415659.6 (50.9, 68.4)10.4 (10.0, 10.9) [1.6]
 ≥24, <3617649.0 (40.4, 57.6)10.9 (10.6, 11.2) [1.3]
 ≥36, <4821636.5 (27.5, 45.5)11.3 (11.1, 11.6) [1.4]
 ≥48, <6022817.5 (11.4, 23.6)11.8 (11.6, 12.0) [1.1]
Sex
 Male46937.2 (32.1, 42.3)11.2 (11.0, 11.4) [1.4]
 Female39538.6 (31.8, 45.4)11.2 (11.0, 11.4) [1.4]
National total/avg87037.9 (32.9, 43.0)11.2 (11.0, 11.4) [1.4]

Software for determining anaemia

Computer programs are available to determine whether or not an individual is anaemic. A web-based program allows users to type in information for an individual (by age, sex, pregnancy, altitude and smoking). Other computer programs are for use with Epi Info version 6.0 (DOS), SPSS and SAS. These programs and additional tables for cutoff values are available at http://www.sph.emory.edu/~cdckms.

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

This report provides unambiguous definitions for the anaemia based on the recommendations of UNICEF, UNU, WHO and INACG. Some countries may have their own criteria for defining anaemia. The identification of individuals who are anaemic may lead to attempts to determine the cause of the anaemia. Identifying the cause of anaemia in individuals and the major causes in populations is important but beyond the scope of this article.

To assess the prevalence of anaemia, unlikely haemoglobin values should be removed and the effects of pregnancy, altitude and smoking need to be taken into account. This report makes recommendations regarding methods to account for pregnancy, altitude and smoking, and on the definition of extreme haemoglobin values. It also offers recommendations on assessing the quality of haemoglobin measurements and for presenting results. Careful attention to collecting, cleaning, analysing, and presenting anaemia and haemoglobin results is important to assure accurate estimates and to allow for comparisons between countries or within countries over time.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

We thank Drs. Bradley Woodruff and Dale Nordenberg, both formerly with the Centers for Disease Control and Prevention, and Dr Thomas Adamkiewicz of the Morehouse School of Medicine for their comments and the Afghanistan Ministry of Health and UNICEF/Afghanistan for use of data. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Afghanistan Ministry of Health and UNICEF (2005) Report of the National Nutrition Survey: Afghanistan, 2004. UNICEF, New York.
  • Beutler E & West C (2005) Hematologic differences between African-Americans and whites: the roles of iron deficiency and alpha-thalassemia on hemoglobin levels and mean corpuscular volume. Blood 106, 740745.
  • CDC (1989) CDC criteria for anemia in children and childbearing-aged women. Morbidity and Mortality Weekly Report 38, 400404.
  • CDC (1999) Recommendations to prevent and control iron deficiency in the United States. Morbidity and Mortality Weekly Report 47, 125.
  • Dallman PR, Barr GD, Allen CM & Shinefield HR (1978) Hemoglobin concentration in white, black, and oriental children: is there a need for separate criteria in screening for anemia? American Journal of Clinical Nutrition 31, 377380.
  • Hurtado A, Merino C & Delgado E. (1945) Influence of anoxemia on the hemopoietic activity. Archives of Internal Medicine 75, 284323.
  • Johnson-Spear MA & Yip R (1994) Hemoglobin difference between black and white women with comparable iron status: justification for race-specific anemia criteria. American Journal of Clinical Nutrition 60, 117121.
  • Nestel P & The INACG Steering Committee (2002) Adjusting Hemoglobin Values in Program Surveys. http://www.inacg.ilsi.org/file/HemoglobinValues2004.pdf.
  • Nordenberg D, Yip R & Binkin NJ (1990) The effect of cigarette smoking on hemoglobin levels and anemia screening. JAMA 264, 15561559.
  • Perry GS, Byers T, Yip R & Margen S. (1992) Iron nutrition does not account for the hemoglobin differences between blacks and whites. Journal of Nutrition 122, 14171424.
  • Scollo M, Younie S, Wakefile M, Freeman J & Icasiano F (2003) Impact of tobacco tax reforms on tobacco prices and tobacco use in Australia. Tobacco Control 12 (Suppl. II), II59II66.
  • UNICEF, UNU, WHO (2001) Iron deficiency anaemia assessment, prevention, and control – a guide for programme managers. (WHO/NHD/01.2).http://www.who.int/reproductive-health/docs/anaemia.pdf.