An evaluation of methods of screening for anaemia*

Bulletin ofthe World Health Organization, 62(1): 115-120(1984) © World Health Organization 1984 An evaluation of methods of screening for anaemia* J. E. STONE,' W. K. SIMMONS,2 P. J. JUTSUM,3 & J. M. GURNEY4 Screening methodsfor anaemia were selectedfor testing on the grounds ofcheapness, simplicity, sturdiness, accuracy and independence of mains electricity or batteries. The methods evaluated were the copper sulfate method, the Dare haemoglobinometer, the Lovibond comparator, the A. 0. Spencer haemoglobinometer, and the Tallqvist method. A new device, the Carib haemoglobin comparator, was developed. The Dare and Lovibond instruments were found to be inaccurate in the laboratory. The other instruments were tested by primary health care workers in clinics in Jamaica. The Carib haemoglobin comparator and the copper sulfate method were found to be accurate, easy to use, and cheap. Both methods are considered to be useful for screening for anaemia at primary health care level. Anaemia is found throughout the world, and can be particularly serious when it occurs in pregnant women (1, 2). In general, diagnosis of anaemia depends on measurement of the concentration of haemoglobin in the blood. This can be done by assess- ing: (1) the colour of the blood (since haemoglobin is the main pigment); (2) the erythrocyte volume fraction; or (3) the specific gravity of the blood. There is a need for a simple screening method for anaemia (2) for use by public health workers in the field. The present study was carried out to deter- mine the most appropriate method of screening for anaemia in clinics or during home visits by primary health care workers in Jamaica. The available screen- ing methods were reviewed and tested for reliability and ease of use, both in the laboratory and in the field. In addition, we developed and tested a new screening method, based on colorimetry. REVIEW OF AVAILABLE METHODS Any method of screening or monitoring individuals for anaemia at primary care level should be cheap, simple to operate, sturdy enough for field use, depen- dent neither on mains electricity nor batteries, and reasonably accurate. It should also use a minimum of materials that require regular replacement and should give immediate results. * From the Caribbean Food and Nutrition Institute, PO Box 140, Kingston 7, Jamaica. Requests for reprints should be addressed to Dr Simmons. ' Formerly Research Assistant. 2 Public Health Nutritionist. 3 Systems Analyst. 4 Formerly Director. Present address: Nutrition, World Health Organization, 1211 Geneva 27, Switzerland. After a careful review of the available technology, we concluded that several methods were worthy of further study. These were: the copper sulfate method (3, 4), the Dare haemoglobinometer (5, 6), the Lovi- bond comparator (7), the A. 0. Spencer haemo- globinometer (8, 9), and the Tallqvist method (10, 11). These methods are described below. Copper sulfate method The copper sulfate method is based upon the obser- vation that the specific gravity of blood is greatly influenced by its erythrocyte volume fraction. Serum protein does not have a large effect on specific gravity, and can thus be neglected. A blood droplet is allowed to fall into copper sulfate solution, with a specific gravity equivalent to that of blood with a haemoglobin content of 100 g/l, from about 1 cm above the surface. The momentum of the droplet takes it to 1-2 cm below the surface within about 5 seconds. The movement of the drop over the next 5-10 seconds is observed. If it continues to fall, the haemoglobin level is judged to be more than 100 g/l. If it rises, the haemoglobin is taken as less than 100 g/l, and the test is repeated in copper sulfate solution with a specific gravity equivalent to a haemoglobin level of 80 g/l. Consequently, the haemoglobin level of every sample is categorized as below 80 g/l, between 80 and 100 g/l, or over 100 g/l. Detailed instructions for preparation of the solutions and use of the method in clinics are given elsewhere (12). Dare haemoglobinometer A small glass reusable chamber is filled with whole blood by capillary action. (Alternatively, a disposable chamber of blotting-paper may be used.) The 4385 -115- J. E. STONE ET AL. chamber is illuminated with a battery-lit bulb and viewed through a red filter; the intensity of emergent light is compared with a graded standard. Lovibond comparator A measured quantity of whole blood is diluted with a measured quantity of dilute ammonia solution made up with distilled water and placed in a compar- ator tube. Haemolysis occurs resulting in a clear red solution. Ammonia solution is placed in an identical comparator tube. Coloured glass filters are rotated in front of the clear tube and a comparison of colour intensity made by viewing transmitted daylight or electric light through both tubes. A. 0. Spencer haemoglobinometer A drop of whole blood is placed in a glass chamber and haemolysed by agitating gently with a stick impregnated with saponin. The chamber is covered with a glass plate and placed in front of a battery-lit bulb emitting green light; the intensity of colour of the sample is compared with a graded standard on a split screen. Tallqvist method A drop of whole blood is placed on a strip of stan- dard blotting-paper and allowed to dry. The intensity of colour is compared with a range of standard colours printed on paper. CARIB HAEMOGLOBIN COMPARATOR From the review of available methods for esti- mating haemoglobin level, it was concluded that none of the techniques entirely satisfied the criteria of ease of use, cheapness, and freedom from reliance on reg- ular supplies. A new instrument, the Carib haemo- globin comparator, was therefore designed and built in the Physics Department of the University of the West Indies. Specifications In order to keep down the construction costs, it was decided not to attempt to measure the haemoglobin level itself but to categorize each patient as severely anaemic, moderately anaemic, or not anaemic. This corresponds to the three categories given by the copper sulfate method, which had proved satisfactory when used in antenatal clinics in Jamaica (13, 14). In addition, it was decided that the apparatus should incorporate a filter to exclude all but the principal wavelength of light absorbed by haemoglobin; in this way, errors caused by confusion between colour dif- ferences and differences in transmission or reflection of light would be minimized. The instrument was thus designed to provide easily interpretable results, be independent of the electricity supply, and to have a low cost. A detailed report on the instrument has been published elsewhere (15). We should have liked to use untreated whole blood in the instrument, but it was found necessary to haem- olyse the samples. Commercially-produced saponin sticks for this purpose are expensive, but it was possible to produce saponin sticks very cheaply using toothpicks dipped in a saponin solution and dried. Method of use A drop of blood is placed in a transparent chamber, haemolysed with a saponin-coated stick, and covered with a glass plate. This is placed in the instrument alongside two filters (corresponding to haemoglobin levels of 80 g/l and 100 g/l) of a neutral density type. Commercial photographic filters of the appropriate densities were used. Daylight or interior lighting is viewed through the sample and comparison filters and through a green filter which removes colour differences between them. The intensity of emergent light is compared and the haemoglobin level is thus classified as > 100 g/l, between 80 and 100 g/l, or <80 g/l. LABORATORY TESTING All the techniques were tested and compared in the laboratory by two experienced technicians, who had spent at least an hour familiarizing themselves with each instrument. Non-anaemic blood was diluted with isotonic saline to give 6 samples with haemoglobin concentrations of 59 g/l, 82 g/l, 94 g/l, 97 g/l, 116 g/l, and 126 g/l, respectively. Each sample was tested 4 times on each instrument by each technician, in a random fashion. The technicians were unaware of the true haemo- globin levels of the samples. The cyanmethaemoglobin method (16) was used as the control against which the other methods were assessed. The values were read on a Beckman 26 spectrophotometer. All the samples used in the laboratory study were also assessed on a Coulter counter. Analysis Two methods were used for the statistical analysis of the results. The Dare and Spencer haemoglobin- ometers, which measure actual haemoglobin levels, were subjected to an analysis of variance. The Carib haemoglobin comparator and the copper sulfate method, which group samples as being above or 116 SCREENING FOR ANAEMIA below a given cut-off point, were assessed for sensi- tivity and specificity. The Tallqvist method was assessed by both techniques, and the Lovibond com- parator was considered as a screening instrument and treated accordingly. Results The standard deviations from the actual values of the results given by the Dare, Spencer, and Tallqvist instruments are given in Table 1. The Spencer instru- ment gave the most accurate results; the standard deviation for the Dare instrument was unacceptably high, with both disposable and reusable chambers. Table 2 sets out the sensitivity and specificity at 80 g/l and 100 g/l of the screening methods tested. Table 1. Standard deviations of the haemoglobin levels given by the different techniques Instrument Standard deviation (g/1) Dare with disposable chamber 19.1 Dare with reusable chamber 20.3 Spencer haemoglobinometer 3.8 Tallqvist method 9.6 Table 2. Sensitivities and specificities of the instru- ments tested in the laboratory Method Level Sensitivity' Specificity bMethod (g/lW) (% (% Carib haemoglobin 80 100 100 comparator 100 100 100 Copper sulfate 180 100 80100 ~ 100 100 80 100 58Lovibond 100 98 35 Tallqvist 100 88 75 No. of samples correctly indicated as being below the cut-off point a Sensitivity = x 100 True no. of samples with value below the cut-off point b Specificity = No. of samples correctly indicated as being above the cut-off point True no. of samples with value above the cut-off point x 100 The Lovibond instrument gave an unacceptable specificity at both levels. The Carib haemoglobin comparator, the copper sulfate method, and the Tall- qvist method gave satisfactory results. The time taken in the laboratory for each test and the assessments of the laboratory assistants regarding the ease of operation of each instrument are set out in Table 3. They found the copper sulfate method was easy to use; it was also the fastest. However, they pre- ferred the Spencer haemoglobinometer. They found both Dare instruments were hard to use and had difficulties with colour matching in the Dare, Lovi- bond, and Tallqvist methods. Stability of copper sulfate solution The specific gravity of copper sulfate solution kept in closed glass or plastic containers was checked regularly over a period of a year; no change was detected. Solution kept in open vessels in various environments showed a negligible change in specific gravity over an 8-hour period. We conclude that no errors would result from deterioration of the solution, either during storage or during a clinic session. Conclusions The Dare haemoglobinometers and Lovibond haemoglobin field kit performed unsatisfactorily. The Tallqvist method gave good results, but the laboratory technicians did not like it. Table 3. Time taken for each estimation and ease of operation of the methods tested Mean time Method for oneestimation Ease of Observations (s) operation' of technicians Carib haemo- Sample holders globin 84 4 need comparator improvement Copper sulfate 60 1 Easy to use Dare (disposable 152 6 Matching colours chamber) very difficult Dare (reusable 210 7 Matching colours chamber) very difficult Lovibond 73 5 ~~~Matching coloursLovibond difficult Spencer 92 2 Preferredinstrument Tallqvist 120 3 ~~Matching coloursTallqvist difficult a Ranked by the technicians from 1 to 7, in order of increasing difficulty. 117 J. E. STONE ET AL. The Carib haemoglobin comparator, copper sul- fate method, and Spencer haemoglobinometer functioned well and the copper sulfate solutions showed no deterioration with storage over a year or after exposure to the atmosphere for 8 hours. FIELD TESTING Four methods were chosen for field assessment in clinics. These were the Carib haemoglobin compar- ator, the copper sulfate method, the Spencer haemo- globinometer, and the Tallqvist method. The purpose was to assess the ease of operation and accuracy of each instrument as a screening device. Each method was tested in a separate urban govern- ment public health clinic in Jamaica. The technique was first demonstrated to the clinic nurses and community health aides in the respective clinics. Then the health aides were trained until they could use the method correctly and were confident and at ease with it. Two aides in each clinic were then chosen to test the blood of at least 50 patients using the method in which they had been trained. All those chosen had had at least 5 years of secondary education, and all but one had worked for at least 4 years as a health aide. The blood for the test was obtained either from a finger prick or intravenously with the free and informed consent of the patients. At the same time, 20 Al of venous blood were taken from each patient, added to 5 Al of Drabkin solution, and stored in a refrigerator in the laboratory. The haemoglobin level of this sample was estimated by the cyanmethaemoglobin method within, at most, 72 hours (usually within 24 hours) of being taken. Ease of operation The Carib haemoglobin comparator was well liked by the health aides who found no difficulty in its use. The necessity for washing and drying the sample holder between each test did not make the instrument unpopular. If the area where the tests were being carried out was dark, the examiner had to move to a sunlit area to read the result. The copper sulfate method was found to be very easy to use. However, some months after the com- pletion of the study, when the copper sulfate method was in use in many clinics, some lack of confidence in its use was expressed by nurses. There are thought to be three main reasons for this: (1) the test was felt to be less "scientific" and thus less reliable than other tests; (2) there was a lack of appreciation that the be- haviour of the drop over the first 10-15 seconds is the important factor; and (3) there was incomplete understanding of the concept of screening (so that, for example, "over 100 g/l" may seem not so useful a statement as, say, "112 g/l"). The first two reasons can be related to the need for careful training and demonstration rather than to a fault in the method itself. The third emphasizes the need to teach the oper- ational consequences of these investigations. The Spencer haemoglobinometer was very well liked. The nurses and health aides found it easy to use and had no problems in colour matching. No prob- lems developed in the instrument during the two months of testing. The health aides had great difficulty in matching the colours with the Tallqvist method, which was not liked. Accuracy Few of the patients at the clinics had haemoglobin levels below 80 g/l, so that it was not possible to make meaningful comparisons at this level. The statistical indices of sensitivity and specificity (17) at 100 g/l were calculated for each instrument, as had been done in the laboratory testing. Positive and negative predictive values were also calculated (18). The positive predictive value indicates the proportion of positive results that are correct and the negative predictive value indicates the proportion of negative results that are correct. In comparing the results, which are shown in Table 4, it must be remembered that each method was tested Table 4. Sensitivity, specificity, and positive and neg- ative predictive values of four screening methods, at a haemoglobin level of 100 g/l Positive Negative Method Sensitivity Specificity predictive predictive value' value b Carib haemoglobin 72.6 96.5 84.2 94.3 comparator Copper sulfate 87.5 98.9 87.5 98.9method Spencer haemo- 77.5 95.9 93.9 88.6 globinometer method 60.5 59.1 46.0 72.2method~~~~~~~~~~~ a Positive predictive value = h Negative predictive value = No. of anaemic people so recognized Total no. diagnosed as anaemic x 100 No. of non-anaemic people so recognized x 100 Total no. diagnosed as non-anaemic 118 SCREENING FOR ANAEMIA in a different clinic. Thus the operators and the patients were different. The clinic test showed the Tallqvist method to be not sufficiently reliable, while the Carib haemoglobin comparator, the copper sulfate method, and the Spencer haemoglobinometer all performed with satis- factory accuracy under clinic conditions in Jamaica. CONCLUSIONS Our purpose in this study was to find an anaemia screening device for use at the primary health care level, that would be accurate, acceptable, cheap, simple to operate, sturdy, and easy to use, and that would give immediate results. We conclude that the Spencer instrument is ac- curate, acceptable, and sturdy in clinic use. It is, however, too expensive for use at the primary health care level in most countries. The copper sulfate method is an accurate screening technique, is very cheap and easy to use in primary health care, and is an excellent method for use in clinics. The copper sulfate solutions need to be made up at a central point and distributed every few months. The Carib haemoglobin comparator is accurate, simple to operate and portable and would be fairly cheap if manufactured on a large scale. We consider that it has great possibilities. None of the remaining methods- the Dare haemo- globinometer, the Lovibond comparator, and the Tallqvist method -performed satisfactorily. ACKNOWLEDGEMENTS This project, including the design and testing of the Carib haemoglobin comparator, was funded by the World Health Organization. We thank the Tropical Metabolism Research Unit and the Department of Haematology of the University of the West Indies for providing laboratory space and equipment, and Philbert Jahoor and Jacqueline Hibbert for performing the laboratory investigations. We acknowledge the contribution of Dr Richard Bone of the Department of Physics, Univer- sity of the West Indies, in developing, producing, and testing the prototype Carib haemoglobin comparator. We thank the Health Department of the Kingston and St Andrew Corporation, Jamaica, and the St Catherine Health Department for granting permission for the field testing, and the public health nurses, midwives, and community health aides of the Glen Vincent Memorial Clinic, Marverley Health Centre, Spanish Town Health Centre, and the Waltham Park Road Clinic. RtSUME EVALUATION DES MtTHODES DE DtPISTAGE DE L'ANtMIE Les instruments utilisables au niveau des soins de sante primaires pour depister les sujets exposes a l'anemie doivent etre bon marche, simples a utiliser, robustes, fonctionnant sans source d'electricite et raisonnablement precis entre les mains d'auxiliaires de sante; enfin, ils ne doivent n6cessiter au plus qu'un petit nombre d'6l1ments a remplacer regu- li6rement et ils doivent fournir des resultats immediats. AprFs avoir pass6 en revue les methodes disponibles, on a retenu, en vue d'essais de laboratoire, la methode au sulfate de cuivre et celles de Dare, Lovibond, A. 0. Spencer et Tall- qvist. De plus, on a mis au point et experimente un nouvel instrument, l'hemoglobinometre Carib. Avec ce dernier appareil, un comparateur colorimetrique, on hemolyse par la saponine une goutte de sang placee dans une chambre transparente. On observe la lumiere du jour ou la lumiere d'une source a incandescence en interposant sur le trajet lumineux d'une part l'echantillon d'autre part deux filtres neutres permettant des comparaisons d'intensite lumineuse (et correspondant, dans notre essai, a des taux d'hemoglobine de 80 et de 100 g/l) ainsi qu'un filtre vert qui 6limine les differences de couleur. L'observateur note si le sang lui apparait plus fonce que le filtre fonce, plus clair que le filtre clair ou d'intensite intermediaire et il en deduit l'ordre de grandeur du taux d'hemoglobine. Chaque instrument a ete essaye en laboratoire par deux techniciens qualifies et l'on a compare les resultats obtenus aux valeurs <exactes> fournies par la methode a la cyan- methemoglobine. Avec l'instrument de Dare, 1'ecart-type a ete trop eleve pour etre admissible, tandis que l'instrument de Lovibond ne presentait pas une specificite acceptable aux points limites (80 et 100 g/l). En revanche, l'hemoglobinometre Carib, la methode au sulfate de cuivre et la methode de Tall- qvist ont donne des resultats satisfaisants. Sur la base des resultats de laboratoire, I'hemoglobino- metre Carib, la methode au sulfate de cuivre, I'hemoglo- binometre, de Spencer et la methode de Tallqvist ont ete essayes dans des dispensaires de la Jamaique en vue d'etablir s'ils etaient suffisamment commodes a employer et precis pour servir de moyen de depistage. Chaque m6thode a ete experimentee par deux agents de la sante communautaire auxiliaires ayant fait des etudes secon- daires pendant au moins 5 ans. Ces auxiliaires ont eu beau- coup de mal A etablir la bonne correspondance de couleurs 119 120 J. E. STONE ET AL. avec la methode de Tallqvist, tandis qu'ils ont appreie les trois autres methodes. On a calcule la sensibilite, la specificite et la valeur pro- nostique negative ou positive (faux et vrais positifs) de toutes ces methodes et trouve des resultats satisfaisants dans tous les cas sauf pour la methode de Tallqvist. La conclusion est que I'hemoglobinometre de Dare, le comparateur de Lovibond et la methode Tallqvist ne donnent pas satisfaction. L'instrument de Spencer est fiable mais son cout en exclut l'emploi au niveau des soins de sante primaires; en revanche, la methode au sulfate de cuivre et l'hemoglobinometre Carib sont tous deux excellents en vue d'une utilisation a l'endroit meme ou sont mis en aeuvre les soins de sante primaires. REFERENCES 1. WHO Technical Report Series, No. 405, 1968 (Nutritional anaemias: report of a WHO Scientific Group). 2. WHO Technical Report Series, No. 580, 1975 (Control of nutritional anaemia with special reference to iron deficiency: report of an IAEA/USAID/WHO Joint Meeting). 3. PHILLIPS, P. R. ET AL. Measurement of specific gravities of whole blood and plasma by standard copper sulphate solutions. Journal ofbiological chemistry, 183: 305-330 (1950). 4. VAN SLYKE, D. D. ET AL. Calculation of hemoglobin from blood specific gravities. Journal of biological chemistry, 183: 349-360 (1950). 5. DARE, A. A new hemoglobinometer for the examin- ation of undiluted blood. Philadelphia medicaljournal, 6: 557 (1900). 6. Foy, H. & KONDI, A. Haemoglobin measurement in developing countries. Lancet, 2: 401 (1977). 7. WORLD HEALTH ORGANIZATION. Manual of basic tech- niques for a health laboratory. Geneva, World Health Organization, 1980. 8. SEIVERD, C. E. Hematology for medical technologists. 3rd ed., Philadelphia, Lea and Febiger, 1970. 9. ELWOOD, P. C. & JACOBS, A. Haemoglobin estimation: a comparison of different techniques. British medical journal, 1: 20-24 (1966). 10. TALLQVIST, T. W. Methode pratique d'evaluation directe de la quantite d'hemoglobine du sang. Archives of general medicine, 3: 421-425 (1900). 11. GAMMON, A. & BAKER, S. J. Studies in methods of haemoglobin estimation suitable for use in public health programmes. Indian journal of medical research, 65: 150-156 (1977). 12. CARIBBEAN FOOD AND NUTRITION INSTITUTE. The copper sulphate method for screening for anaemia: a manual for its use. Kingston, Caribbean Food and Nutrition Institute, 1982. 13. ANDRIANASOLO, R. ET AL. An evaluation of a simplified method for screening haemoglobin in the field. American journal of clinical nutrition, 32: 728-730 (1979). 14. ANDRIANASOLO, R. An evaluation of programs to control anemias of pregnancy in Jamaica. Ithaca, Faculty of the Graduate School, Cornell University 1980 (Ph.D. thesis). 15. BONE, R. A simple haemoglobin screening device. Appropriate technology, 9: 27-29 (1983). 16. HAINLINE, A., JR. Hemoglobin. In: Seligson, D., ed., Standard methods of clinical chemistry. New York, Academic Press, 1958, vol. 2, pp. 49-51. 17. TALLQVIST, T. W. Ein einfaches Verfahren zur directen Schatzung der Farbestarke des Blutes. Zeitschrift fur klinische Medizin, 40: 137-141 (1900). 18. HABICHT, J. P. Some characteristics of indicators of nutritional status for use in screening and surveillance. American journal of clinical nutrition, 33: 531-535 (1980).