A simple and reliable method for estimating haemoglobin.

A simple and reliable method for estimating haemoglobin G.J. Stott1 & S.M. Lewis2 A new colour scale has been devised for estimating haemoglobin levels by matching the blood sample with ten levels of haemoglobin (3, 4, 5, 6, 7, 8, 9, 10, 12 and 14 gldl) on the scale. Preliminary results show good correlations with spectrophotometric readings. The new device is being field tested and if the initial promise is confirmed, will provide a simple and reliable method for estimating haemoglobin where laboratory facilities are not available. Methods for assessing haemoglobin levels by match- ing a drop of blood on a piece of blotting paper against a colour scale have been widely used in health centres in developing countries for the detec- tion of anaemia. In theory, they are attractive because of their simplicity, portability and low cost. In practice, they are so grossly inaccurate, especially at lower haemoglobin levels, that they have little value. Nevertheless, it is realized that if a suitable ver- sion of a device based on the principle of direct com- parison of fresh blood with a reliable colour scale could be developed it would serve a very useful pur- pose, not only for field use but also in hospitals, health centres, clinics and general practice, especial- ly in situations where a reasonably reliable assess- ment, rather than great accuracy and precision, is what is needed to permit a clinical decision and action. The first problem was therefore to identify the factors responsible for the wide margin of error found in the available colour scales and to overcome them so as to develop a reliable method suitable for field use. Causes of inaccuracy and imprecision The main sources of error in the use of the haemo- globin colour scales currently available are: (1) The colours of the printed scales vary between manufacturers and may not even look like blood at all, particularly at lower haemoglobin levels, perhaps because they were not prepared 1 Programme on Health Technology, World Health Organization, 1211 Geneva 27, Switzerland. Requests for reprints should be sent to this author. 2 Senior Research Fellow in Haematology, Royal Postgraduate Medical School, London, England. Reprint No. 5609 by matching with fresh blood. (2) The absorbent papers supplied with these colour scales also show much variation and are unsuitable because: - the blood spreads unevenly and too widely, caus- ing dilution of the colour; - the blood takes too long to be absorbed, especial- ly at higher haemoglobin levels (from several seconds to many minutes), which is inconvenient for practical use; and - the paper is too thin and therefore very translu- cent when damp. (3) The design of the haemoglobin scale book- lets, with their absorbent papers in front and the colour scale at the back, makes it impossible to use them without light entering from behind the scale; the blood stains on the test paper, being damp and translucent, then become unmatchable. (4) Even if the colour scale and the test and background papers are held close together, light is reflected from the white background through the damp blood stain, giving too bright a colour which cannot be matched. (5) The circular apertures of 5-6 mm diameter in the colour scales, through which the blood stains are viewed, are too small to permit proper matching because the margin casts a shadow. Development of a new method By eliminating, as far as possible, the identified sources of error it proved possible to develop a de- vice and colour scale with absorbent papers that gave reproducible results and an acceptable degree of accuracy. Blood standards The study was based on the colours observed with a set of blood samples of known haemoglobin content, Bulletin of the World Health Organization, 1995, 73 (3): 369-373 © World Health Organization 1995 369 G.J. Stott & S.M. Lewis as measured by spectrophotometry in accordance with the ICSH reference method.a Their haemo- globin levels were adjusted by adding autologous plasma to obtain samples in the range 3-14 gIdl, in 1 g/dl steps. Absorbent test papers Desirable qualities in an absorbent paper for use with a haemoglobin colour scale are: - the paper should absorb the blood drop quickly and lose its sheen almost immediately; - the paper should be thick enough not to allow too much reflection of light through it when held against a pale background while still damp. (Note: even if the sheen disappears almost at once, the blood stain remains damp for 15-20 minutes depending on the ambient temperature and humidity); and - the paper when stained with blood at low hae- moglobin levels should have as uniform a colour as possible without too much interference from white fibres which prevents satisfactory matching. Absorbent papers have very different qualities depending on their texture, composition, pore size, flow characteristics, fluid spread and pH. A number of Whatman filter and chromatographic papers were examined, namely: 3 MM Chr, 4 Chr, 17 Chr, 31 ET Chr, BP 87, D 28, F 178-10, F 427-01, and F 427- 02; and also papers 557, 598 L, 2316, 2292 and 3469, produced by Schleicher and Schuell. Some were too thin, others too thick; and some absorbed the blood drops easily, but others poorly. It was concluded that the paper Whatman 31 ET Chr gave the best results, with the Schleicher and Schuell paper 2992 as second best. A drop of blood is absorbed and loses its sheen on these papers in a few seconds at high haemoglobin levels, and in one or two seconds at lower levels, and gives a regular round stain with limited spread. Colour standards To develop a colour scale corresponding to the appearance of the blood sample at the different hae- moglobin levels (3-14 g/dl), drops of blood were placed on test strips of Whatman 3 ET Chr paper and, as soon as the blood stains had lost their sheen, the spectral characteristics of their colours were measured by a computerized analytic spectrophotom- eter. a International Council on Standardization in Haematology. Journal of clinical pathology, 1978, 31: 139-143. To meet these specifications, printing inks (resistant to fading due to ultraviolet rays) were pre- pared from pigments of the three primary colours and a neutral diluent. The different shades were then printed on strips of acid-free paper at a defined ink thickness, and dried and varnished. Lighting Colour matching of a blood stain on absorbent paper with the shades on a colour scale is influenced by the type of light (daylight, tungsten, or fluorescent tube), the angle at which the light is reflected, and the visu- al interference from a white or coloured background. It was found that the best results were obtained when the scale is held at an angle of 450 in daylight (but not bright sunlight), with the light coming from behind or over the shoulder of the observer, and with a neutral pale-grey matt background to the colour scale. Effect of time on the colour of blood stains Blood stains on absorbent paper change in colour with time, as the haemoglobin converts to the reduced form and to methaemoglobin. This change begins after a couple of minutes and comparisons with the colour scale must therefore be made as soon as the sheen disappears. Preparation of a reliable colour scale With the experience from many studies of the various sources of error and how to overcome them, an improved device with a new colour scale was pre- pared with the following main features: - there are ten colour standards corresponding to the colour of blood stains on Whatman 31 ET Chr paper at haemoglobin levels of 3, 4, 5, 6, 7, 8, 9, 10, 12 and 14 g/dl; - the colour standards (20 x 40 to 60 mm) are printed in a continuous row without any separa- tion so as to allow matching of test strips also against one side of the colour scale; - circular apertures of 8-9 mm diameter are placed in the centre of each colour standard (to facilitate comparisons when test strips are placed behind the colour scale); - the colour scale is mounted on a rigid white polyvinyl chloride or polypropylene sheet or thick card with a neutral pale-grey matt back- ground of laminated board for easy cleaning; and - the test papers (12-15 x 60 mm) are supplied in small packets separately from the colour scale. 370 WHO Bulletin OMS. Vol 73 1995 Reliable method for estimating haemoglobin Practical features It is essential that a device of this type should: be inexpensive and not require batteries, cuvettes, chemicals or maintenance; be reliable, durable and replaceable at low cost; be suitable for use also by relatively junior health staff (provided they are instructed in the meth- od); allow the detection of mild, moderate or severe degrees of anaemia (whatever the age, sex, state of pregnancy, or altitude); permit health staff to detect improvement or deterioration in individual patients following therapy; provide the basis for a set of guidelines for the recognition and management of anaemia (espe- cially where laboratory facilities are not readily available) in children, pregnant women, adults and those with malaria or other parasitic condi- tions (e.g., ancylostomiasis); and allow the identification of potential blood donors (i.e., those with haemoglobin levels at or above 12 or 14 g/dl, depending on national norms). The ten colour standards in the scale make it possible to achieve a fair degree of precision in esti- mating haemoglobin levels (0.5 g/dl) but the number could be reduced if desired, although this would increase the margin of error at some haemoglobin levels. Matching of test strips may be done from behind or at the side of the colour scale, depending on local lighting conditions. The design of the card ensures that it remains closed when not in use so as to reduce the risk of colours fading. Besides the instructions for use, the device will include brief guidelines for the management of anae- mia under field conditions, which may be modified to suit the local health situation. Field studies Preliminary tests Prototype models of the device with the new colour scale were tested by independent observers. These tests showed that, with brief instruction and strict adherence to the simple technique, individual obser- vers-with no knowledge of the actual haemoglobin levels of the blood samples or of the colour scale equivalents-could vary in their readings but there was no obvious bias (Table 1, Fig. 1). The results showed good correlations. There were some random differences in readings, especially at higher levels (mean difference, 0.85 g/dl). The intercept and slope suggest that there is a small bias (i.e., constant error). This may be overcome, if necessary, by an appro- priate adjustment of the values assigned to the colour scale. Table 1: Comparison of colour scale device and spectrophotometric methods for haemo- globin estimation Colour scale Spectrophotometry 7.27 27 2.92 Conclusions 0.33 0.925 Differences in means (g/dl) SE of differences in means F-ratio at v = 26 t-test on difference in means at v = 26 Mean of differences (g/dl) SE of mean of differences t-test on differences in paired results at v = 26 Correlation coefficient (r) Intercept (a) Slope (b) No significant difference in means 1.011 99% probability of no significant difference in means 0.417 65-70% probability of no significant difference 0.85 0.13 6.405 Significant differences present between some individual results 0.9386 Perfect correlation would give r = 1.000; a = 0, b = 1 0.718 0.944 WHO Bulletin OMS. Vol 73 1995 Mean Hb (g/dl) No. of samples SD 6.94 27 2.90 371 G.J. Stott & S.M. Lewis Fig. 1. Correlation of colour scale readings at actual haemoglobin levels. 12- 0 0)<,10- -o 0) - a) 8- a) a) 6- a) a) Ir A.- bin levels; Mr K.J. Wynn, Printing Service, WHO, Geneva, for his sustained interest, technical advice and much prac- tical help; Mr P. Ender, SICPA, Lausanne, for developing the colour standards; the staff of the Metal and Woodwork Section, WHO, Geneva, for help in making prototypes; Professor H. Gilles, Liverpool School of Tropical Medicine, Liverpool, and Dr R.J. Guidotti, Maternal and Child Health, WHO, Geneva, for their views on the potential value of the device and colour scale; Dr Ph. Beris, H6pital Canto- nal, Geneva, for providing standardized blood samples; Dr Sriwanna Poolsuppasit, Provincial Hospital, Uttaradit, Thailand, and Dr Torpong Sanguensermsri, Faculty of Medicine, Chiang Mai, Thailand, and their colleagues for testing the colour scale, and Professor B. Mendelow, South African Institute for Medical Research, Johannes- burg, for participating in the multicentre study. Thanks are also due to Dr F.S. Antezana, Assistant Director-General, WHO, Geneva, for support received through the WHO Programme on Health Technology, Mrs Christine Klekr for secretarial assistance, and to many WHO colleagues for their participation in practical trials with the colour scale. New method (g/dl) Proposed multicentre study With these encouraging results, a larger double-blind trial involving other centres and observers, each using 10-20 blood samples at different haemoglobin levels, some of them being duplicates, is now under way. Preliminary observations indicate that a period of training is necessary to familiarize observers with the method and to enhance their ability to distinguish different shades of red. Conclusion We believe that this device, with its new and impro- ved colour scale (Fig. 2), meets the required criteria for haemoglobin estimation. If the initial promise is confirmed, the device and the test strips (of suitable absorbent paper) will be made available for use in mass surveys and screening, as well as in public health and clinical services. Acknowledgements We are grateful to the following persons for their kind col- laboration at different stages of the work described: the laboratory and nursing staff of the Joint Medical Service, WHO, Geneva, for taking blood and measuring haemoglo- Resume / Une methode simple et fiable de dosage de l'hemoglobine Les centres de sante des pays en developpement utilisent largement, pour la d6tection de l'an6mie, des m6thodes de dosage de l'h6moglobine qui consistent a comparer une goutte de sang d6po- s6e sur un morceau de papier buvard a une 6chelle color6e, mais les systemes employ6s sont si impr6cis, surtout pour de faibles concentrations d'h6moglobine, que les r6sultats ne sont guere valables. Une 6tude a donc ete entreprise pour re- chercher les facteurs responsables de cette im- portante marge d'erreur et mettre au point, en mo- difiant la conception, les mat6riaux et le mode d'emploi, une nouvelle m6thode et une nouvelle 6chelle de coloration qui donnent des resultats fiables et d'une pr6cision acceptable. Les teintes choisies pour la nouvelle echelle correspondent a dix concentrations d'h6moglobine dans le sang (3, 4, 5, 6, 7, 8, 9, 10, 12 et 14 g/dl). Les premiers r6sultats pr6sentent une bonne corr6lation avec les mesures spectrophotom6- triques. La nouvelle m6thode fait actuellement l'objet d'essais sur le terrain, et si les premieres esp6rances se confirment, elle constituera un moyen pratique et rapide de d6tection de l'an6mie. WHO Bulletin OMS. Vol 73 1995372 Reliable method for estimating haemoglobin Fig. 2. An example of a colour scale using eight colours (with apertures between the colours). WHO Bulletin OMS. Vol 73 1995 373