The impact of oedema on anthropometric measurements in nutritional surveys: a case study from Zaire

Bulletin ofthe World Health Organization, 62 (1): 145-150(1984) © World Health Organization 1984 The impact of oedema on anthropometric measurements in nutritional surveys: a case study from Zaire ROBERT R. FRANKLIN,' LUSAMBA N. DIKASSA,2 & WILLIAM E. BERTRAND3 Six surveys for protein-energy malnutrition were carried out in sequence in Bas-Zaire beginning at the end of 1978 at the estimated height of the famine and continuing throughout the recovery period. Utilizing a stratified multi-stage sampling technique, over 1000 children aged 6 months to 6 years were measured anthropometrically and examined for the presence or absence of bilateral pedal oedema in each survey. The proportions of children who were less than 80% of the reference median weight-for-height and who had oedema decreased, respectively, from 12.8% and 14.4% initially to 2.1% and 1.8% in the final survey. The proportion of oedematous children who werefound to be not less than 80%o of the reference median weight-for-height was high, varying from 71.7 ± 7.0% to 94.4 ± 10.6%. The importance of these findings for the interpretation of anthropometric data from nutritional surveys in developing countries is discussed. The current concept of protein-energy malnutrition (PEM) is that kwashiorkor and nutritional marasmus are two different clinical pictures at opposite poles of a single continuum (1). Of the two, kwashiorkor usually leads to death within a period of many days to several weeks, whereas victims of untreated nu- tritional marasmus may survive for up to several months (2). In addition, the synergism between kwashiorkor and certain infectious diseases has been demonstrated to increase both mortality and mor- bidity (3). While the public health importance of kwashiorkor is clear, most current nutrition surveys emphasize indicators that can be insensitive to the presence of kwashiorkor in populations. The basic problem described in this paper is the impact of oedema on interpretation of body weight measure- ments. There is general agreement that oedema is the cardinal clinical sign of kwashiorkor, distinguishing it from nutritional marasmus, and without which the diagnosis of kwashiorkor should not be made (2, 4). While there are many other signs and symptoms of kwashiorkor, they appear to be both less sensitive and specific (4). It is also generally recognized in clinical settings that the presence of oedema in kwashiorkor i Assistant Professor, Department of Biostatistics and Epi- demiology, Tulane School of Public Health and Tropical Medicine, 1430 Tulane Avenue, New Orleans, LA 70012, USA. Formerly Co-Director, Zaire National Nutrition Planning Centre. Requests for reprints should be sent to this author. 2 Assistant Professor, Department of Public Health, University of Zaire, Kinshasa, Zaire. 3 Associate Professor and Chairman, Department of Biostatistics and Epidemiology, Tulane School of Public Health and Tropical Medicine, New Orleans, USA. can increase the body weight in children. Workers in nutritional rehabilitation programmes often observe a weight reduction (or lack of weight gain) during the first week or two of therapeutic feeding in patients with kwashiorkor. This reduction is usually found accompanying the loss of oedema (2). In recent years biochemical evaluations of children with severe protein-energy malnutrition during nu- tritional rehabilitation have more clearly defined the physiological processes involved. Patrick et al. in Jamaica, using tritiated water, found that children with severe malnutrition and oedema on admission had elevated proportions of total body water. Furthermore they found a mean decrease of 15% of the body weight after loss of oedema. Children with- out oedema did not show elevated total body weights if their diets had been limited to maintenance energy requirements (5). The authors explained that some of the findings in earlier studies showing high levels of total body water in non-oedematous children were probably invalid because the measurements were taken after rehabilitation had started and when extra- cellular fluid volumes were maintained in the face of increased destruction of body solids. Therefore, if weight measurements are used as indi- cators of nutritional status based on assumptions that they reflect changes in soft tissues, one would expect oedematous children to have spuriously high body weights and, correspondingly, spuriously high calcu- lated values for nutritional status. Thus, it might be anticipated that the use of weight as an indicator of nutritional status in population surveys would present problems in areas where kwashiorkor is highly preva- lent. The oedema present in kwashiorkor cases would 43u -145- R. R. FRANKLIN ET AL. raise the body weights which would be falsely inter- preted as an increase in body soft tissue mass. Classification systems for PEM vary greatly in their treatment of oedema. Garrow, McLaren, and Burgess have each developed composite systems in which oedema and/or the presence of other signs of kwashiorkor are used in combination with certain anthropometric measurements to determine the pres- ence of kwashiorkor (6-8). In the Wellcome classifi- cation system, oedematous individuals who measure 60-80%7o of the reference median weight-for-age are classified as having kwashiorkor, while oedematous individuals below 60%0 are classified as having maras- mic kwashiorkor (9). The most popular classification systems, as cur- rently used, exclude oedema entirely from their evalu- ation of individuals. The Waterlow system differ- entiates between weight-for-height as a measure of wasting (indicating acute PEM) and height-for-age as a measure of stunting (indicating chronic PEM) (10). In the Jelliffe system, children are classified by weight-for-age according to whether they fall in the range 80-90% of the reference median, 70-80%, or less than 600/ (4). In the Gomez classification, chil- dren are categorized as having first- , second- , or third-degree malnutrition if their weight-for-age is in the range 76-90%, 61-75%7o, or less than 61%, respectively, of the reference median (11). It is interesting to note, however, despite current use of these systems, that in at least the cases of Jelliffe and Waterlow the authors clearly stated their belief that obvious clinical cases of kwashiorkor (especially those demonstrating oedema) should be classified as severely malnourished, above and beyond all con- siderations of anthropometric data (4, 10). Bengoa even suggested a modification of the Gomez scheme to include all oedematous cases as third-degree malnutrition without consideration of weight (12). However, in the absence of a standardized protocol on how to deal with oedematous cases, all three classification systems of PEM are now usually em- ployed and cases of kwashiorkor are rarely treated separately. Probably the most widely used approach to popu- lation-based surveys of PEM is the one exemplified by the national nutrition surveys undertaken by governments in some developing countries, in cooperation with the United States Agency for International Development (USAID) and usually the Centers for Disease Control of the United States (CDC). These studies stress the classification ofPEM by the Waterlow and Gomez systems. Oedema rates are invariably reported for each country as a whole and sometimes for smaller areas within the country. However, oedema rates are rarely considered in con- junction with anthropometric measurements when determining the overall rates of acute protein-energy malnutrition (13-21). The data to be presented here demonstrate the problems in this approach for studies in areas where kwashiorkor is highly prevalent. METHODOLOGY The study was carried out in Zaire from 1978 to 1980. There were two separate components: a base- line study which essentially defined the problem, and several follow-up survey rounds which monitored the resolution of the problem. The baseline study was performed at the approximate height of the famine, in November and December 1978, in response to an emergency request for evaluation of reported famine conditions. A CDC Epidemiological Intelligence Officer came to Zaire to assist in study design and implementation. The Bas-Fleuve portion of that study was only part of a larger study which included most of the entire Bas-Zaire region. Owing to the severity of the conditions found in Bas-Fleuve, most of the food relief was subsequently directed to that subregion. In July 1979 a surveillance system was established there to monitor improvements in nu- tritional status. Five monitoring rounds were per- formed from July 1979 to February 1980. The data presented here will be restricted to information con- cerning only the Bas-Fleuve subregion. Descriptions of the findings for the entire region have been presented elsewhere (22, 23). A stratified, multistage sampling technique recom- mended by CDC and the World Health Organization was used. "Collectivities" (the local administrative units) in the two survey zones were listed by sizes of population from the 1977 census information. The total population was subdivided by the number of sampling units, 19 villages, which the teams would be able to survey in the time allocated. A population sampling interval was thus obtained. This sampling interval was used for the systematic selection of the collectivities. Two underrepresented groups, refugees and residents of larger urban areas, were sampled in- dependently and constituted an enrichment sample of nine sites making a total sample of 28 sites. Within collectivities, the villages were selected randomly from a map. The greatest map distance across a collectivity (its major axis) was measured and also the greatest distance perpendicularly. Each was subdivided into 10 equal parts and the second digit of the random number was used to identify the village nearest to this point. In each village, at least 25 children between 6 months and 6 years of age were examined. Any village with less than 25 children was supplemented with children from a neighbouring 146 NUTRITIONAL OEDEMA AND ANTHROPOMETRIC MEASUREMENTS village to achieve an adequate sample size. All the children were examined in those villages that had 25-60 children. In urban areas, house-to-house ran- dom sampling techniques were used. Three teams, each comprising two auxiliary nurses and a supervisor, performed the anthropometric measurements. Each child was weighed to the nearest 100 g with a portable Salter scale and measured for height or length to the nearest millimetre using a portable measuring device (a "microtoise"). Pedal oedema was assessed by pressing for 3 seconds on the dorsum of both feet and reporting if pitting was present. Data were recorded on coding forms for punching and verification at IBM Zaire. Analysis of the data was performed at Tulane University utilizing a DEC 20/60 computer. Calculation of anthropometric indices was performed using the CDC FORTRAN pro- gram based on the National Center for Health Statis- tics (NCHS) reference population sample. Other cal- culations were performed using the Statistical Package for the Social Sciences (SPSS) package of programs. For the monitoring rounds, the original sample sites from the baseline study were used with the addition of enrichment sites using identical sampling procedures. Two teams consisting of 3 auxiliary nurses and 1 supervisor performed the surveys. These teams continually surveyed the area from July 1979 to February 1980. Each village was visited once every 1+-2 months. All other aspects of the study method- ology were similar to that in the baseline study. RESULTS The proportions of children in the baseline survey who were found to have bilateral pedal oedema or who were less than 80%o of the reference median weight-for-height are shown in Table 1. The pro- portions were nearly equal, 14.4%o with oedema and 12.8% acutely wasted. However, only 4.1 of the children were found to be both acutely wasted and oedematous. Therefore if calculation of the pro- portion of children with acute PEM had been based on anthropometric criteria alone, the total would have been grossly underestimated; 71.7%o (114/159) of the children with oedema would not have been con- sidered to be acutely malnourished based on weight- for-height measurements alone. The corresponding data for each of the monitoring rounds are presented in Tables 2 and 4. Despite the decreasing prevalence of oedema and wasting in successive rounds, in each case over 70%o of the children with bilateral pedal oedema were found to be greater or equal to 80Vo of the reference median weight-for-height. If weight-for-age criteria were utilized (see Tables 3 and 4), at least 50% of the children with oedema in each round would not be classified as acutely malnourished based on less than 60%7o of the reference median weight-for-age (third- degree malnutrition in the Gomez classification). DISCUSSION Clearly a large proportion of the children in these surveys with bilateral pedal oedema did not satisfy the anthropometric criteria for acute PEM. In light of the material presented earlier, the most prob- able explanation is that their oedema levels had aug- mented their weights sufficiently to invalidate the weight-dependent anthropometric indices. If this in fact had taken place, two major questions should be addressed. First, should all those children with oedema be considered to have kwashiorkor and there- fore be categorized as acutely malnourished? Second, how do these findings influence the interpretation of current major nutritional surveys? While there are many causes of oedema of the lower extremities in human populations, the prevalence of these conditions is reduced dramatically if only the paediatric population is examined and only bilateral oedema considered. Probably the most common non- PEM causes of oedema would include: beriberi (thia- mine deficiency), nephrotic syndrome (and other renal pathological states), and to a much lesser extent hepatic and other cardiac causes. However, it is very difficult to make assumptions on the epidemic pro- portions of any of these diseases except perhaps for beriberi, which appears predominantly in popu- lations that have milled rice as a staple food (22). The Table 1. Number of children with acute PEM cross- classified by weight-for-height and presence or absence of pedal oedema, Bas-Fleuve, November-December 1978 Percentage Oedema of median weight- for-height Present Absent Total < 80%1' 45 (4.1)b 97 (8.7) 142 (12.8) > 80% 114 (10.3) 851 (76.9) 965 (87.2) Total 159 (14.4) 948 (85.6) 1107 (100) e Less than 80% of median weight-for-height, NCHS stan- dards; under CDC's acute undernutrition threshold; acutely wasted. b Figures in parentheses are percentages. 147 R. R. FRANKLIN ET AL. Table 2. Proportion of children with and without oedema classified by weight-for-height measurements during the Bas-Fleuve famine and recovery period With no oedema With oedema <80% 80% <80% 80% Total wt/ht wt/ht wt/ht a wt/ht Survey Dates no. of children No. % No. % No. % No. % Baseline Nov-Dec. '78 1107 97 8.7 851 76.9 45 4.1 114 10.3 Monitoring rounds: One Jul-Aug. '79 1422 35 2.5 1301 91.5 20 1.4 66 4.6 Two Aug-Sept. '79 1313 25 1.9 1229 93.6 13 1.0 46 3.5 Three Sept.-Oct. '79 1203 22 1.8 1137 94.5 12 1.0 32 2.7 Four Oct-Dec. '79 1159 18 1.6 1113 96.0 6 0.5 22 1.9 Five Jan-Feb. '80 1007 20 2.0 969 96.2 1 0.1 17 1.7 e Less than 80% of the reference median weight-for-height, NCHS standards; under CDC's acute undernutrition threshold. Table 3. Proportion of children with and without oedema classified by weight-for-age measurements during the Bas- Fleuve famine and recovery period With no oedema With oedema < 60% 60% <60% >60% Total wt/agea wt/age wt/agea wt/age Survey Dates no. of children No. % No. % No. % No. % Baseline Nov-Dec. '78 1107 85 7.7 863 78.0 47 4.2 112 10.1 Monitoring rounds: One Jul-Aug. '79 1422 51 3.6 1285 90.3 31 2.2 55 3.9 Two Aug-Sept. '79 1313 35 2.7 1219 92.8 27 2.1 32 2.4 Three Sept.-Oct. '79 1203 33 2.7 1126 93.6 21 1.8 23 1.9 Four Oct-Dec. '79 1159 37 3.2 1094 94.4 14 1.2 14 1.2 Five Jan-Feb. '80 1007 19 1.9 970 96.3 5 0.5 13 1.3 e Less than 60% of the reference median weight-for-age, NCHS standards; approximately equal to third-degree Gomez classifi- cation. staple foods in the Bas-Fleuve area are manioc and banana-plantain and no clinical evidence of any wide- spread vitamin B deficiencies were found in a con- current study of paediatric wards there during the famine. The second question that should be addressed is "If oedema interferes with the validity of anthropometric measurements based on weight, how does this in- fluence the results obtained from current major nu- tritional surveys?" The rates of oedema found in the USAID sponsored national nutritional surveys have been remarkably low. In Egypt, Liberia, Nepal, Sierra Leone, Sri Lanka, Togo, United Republic of Cameroon, and Yemen Arab Republic, the rates were all below 1 7o, and in Haiti the rate was 2.77o (13-21). Therefore, with the exception of Haiti, the rates were low enough to have had probably little effect on the overall conclusions drawn. However, it is interesting to note that, based on the data presented on the number of children with oedema, we calculated that approximately 8107o of the children with oedema in Haiti and 41(o in Nepal would not have been con- sidered acutely malnourished based on indices that are less than 80qo of the reference median weight-for- height (18, 20). Similarly the rates of kwashiorkor were very low in 148 NUTRITIONAL OEDEMA AND ANTHROPOMETRIC MEASUREMENTS Table 4. Proportion of children with oedema not classi- fied as acutely malnourished based on weight-for-height or weight-for-age Children with Children with Total oedema but oedema but no. > 80% wt/ht . 60% wt/age with Survey oedema No. %a No. %a Baseline 159 114 71.7 + 7.0 112 70.4 ± 7.1 Monitoring rounds: One 86 66 76.7 8.9 55 64.0 10.1 Two 59 46 78.0 ± 10.6 32 54.2 + 12.7 Three 44 32 72.7 ± 13.2 23 52.3 ± 14.8 Four 28 22 78.6 15.2 14 50.0 18.5 Five 18 17 94.4 ± 10.6 13 72.2 ± 20.7 a Percentage within 95% confidence limits. the CDC 1974 Sahel nutrition survey. In four of the countries, none of the children examined had oedema and in the fifth, Upper Volta, the rate was less than 1% (25). Therefore in most current major nutritional surveys the prevalence of oedema has been too low to affect the overall anthropometric findings. Based upon the case study from Bas-Zaire, specific directives should be included in the rapid nutritional status survey methodologies currently utilized con- cerning the potential importance of oedema and its measurement, particularly in famine situations. While the implications of oedema are perhaps ap- parent to persons with a good knowledge of nutrition, they are not necessarily understood by field workers who often perform and analyse the surveys. The seriousness of underestimating the population acutely at risk is such that every effort should be made to avoid it. ACKNOWLEDGEMENTS The authors are grateful to the following for advice and assistance in the preparation of this manuscript: Dr Milton Nichaman, Former Director, Nutrition Division, Centers for Disease Control, Atlanta, GA, and Dr John McKigney, Nutrition Advisor, Office of Nutrition, United States Agency for International Development, Washington, DC, USA. This study was conducted as part of Nutrition Planning Center USAID Project No. 660-0055. RESUME EFFET DE L'CEDtME SUR L'INTERPRtTATION DES MESURES ANTHROPOMtTRIQUES EFFECTUtES AU COURS DES ENQUSTES NUTRITIONNELLES: tTUDE DE CAS AU ZAYRE La plupart des enquetes actuelles sur la nutrition dans les pays en developpement mettent l'accent sur des indicateurs qui peuvent etre insensibles a la presence du kwashiorkor dans la population. Le probleme essentiel qu'expose le present article est celui de l'effet de l'cedeme sur l'inter- pretation des mesures du poids corporel. Les observations des medecins travaillant dans les centres de recuperation nutritionnelle aussi bien que les evaluations biochimiques faites recemment a l'aide de marqueurs radioactifs montrent que les enfants atteints de kwashiorkor ont un poids corporel eleve du fait de l'cdeme. Des lors, si l'on utilise le poids corporel comme indicateur de l'etat nutritionnel en se basant sur l'hypothese qu'il reflete l'evolution des tissus mous, les enfants cedemateux vont etre classes erronement comme bien nourris. Au cours de l'etude decrite ici, six enquetes successives sur la malnutrition proteino-energetique ont ete executees dans le Bas-Zaire; commencees a la fin de 1978, date que l'on estime correspondre au plus fort de la famine, elles se sont poursuivies pendant toute la periode de recuperation. En utilisant une technique de sondage a plusieurs degres par stratification, on a mesure les caracteristiques anthropome- triques de plus de 1000 enfants ages de six mois a six ans et cherche a determiner si ces enfants presentaient ou non un oed6me bilateral du pied. Les proportions de ceux qui avaient moins de 80% du poids pour la taille mediane de reference et qui etaient porteurs d'cedeme se sont abaissees, respectivement, de 12,8% et 14,4% au debut des travaux a 2,1% et 1,8% lors de l'enquete finale. Quant a la proportion des enfants cedemateux qui n'avaient pas moins de 80% du poids pour la taille mediane de reference, elle etait elevee et oscillait entre 71,7 ± 7,0% et 94,4 ±10,6%. Les auteurs s'interrogent sur l'importance de ces observations pour interpreter les donnees anthropometriques recueillies au cours d'enquetes nutritionnelles dans les pays en developpe- ment. 149 150 R. R. FRANKLIN ET AL. REFERENCES 1. FERRO-LUZZI, A. & SPADONI, M. A. Protein-energy malnutrition. 3. Kwashiorkor and marasmus. Progress in food and nutrition science, 2: 528 (1978). 2. WHO Technical Report Series, No. 477, 1971 (Joint FAO/WHO Expert Committee on Nutrition: eighth report). Part II. Protein-calorie malnutrition. 3. CHANDRA, R. K. & NEWBERNE, P. M. Nutrition, immu- nity and infection: mechanisms of interaction. New York, Plenum Press, 1977. 4. JELLIFFE, D. B. The assessment ofthe nutritional status of the community. Geneva, World Health Organiz- ation, 1966. 5. PATRICK, J. ET AL. Total body water in malnutrition: the possible role of energy intake. British journal of nutrition, 16: 417-424 (1973). 6. GARROW, J.S. Kwashiorkor and marasmus in Jamaican infants. Archivos Latinoamericanos de Nutrici6n, 16: 146-154 (1966). 7. McLAREN, D. S. ET AL. A sample scoring system for classifying the severe forms of protein-calorie mal- nutrition of early childhood. Lancet, 1: 533-535 (1967). 8. BURGESS, H. J. L. ET AL. The health of children in Kisarawe district, Tanzania. Tropical and geographical medicine, 21: 39-46 (1969). 9. WATERLOW, J. C. Annex 5: Classification and definition of protein-energy malnutrition. In: Beaton, G. H. & Bengoa, J. M., ed., Nutrition in preventive medicine. Geneva, World Health Organization (Mono- graph Series No. 62), pp. 532-533. 10. WATERLOW, J. C. Classification and definition of protein-calorie malnutrition. British medicaljournal, 3: 566-569 (1972). 11. GOMEZ, F. ET AL. Mortality in second and third degree malnutrition. Journal of tropical pediatrics, 2: 77 (1956). 12. BENGOA, J. M. Recent trends in the public health aspects of protein-calorie malnutrition. WHO Chronicle, 24: 552-561 (1970). 13. MINISTRY OF RURAL DEVELOPMENT, GOVERNMENT OF ToGo. Togo nutrition status survey: October 1976- January 1977. Washington, DC, USAID, 1977. 14. MINISTRY OF HEALTH, GOVERNMENT OF SRI LANKA. Sri Lanka nutrition status survey: September 1975-March 1976. Washington, DC, USAID, 1976. 15. THE NUTRITION INSTITUTE, MINISTRY OF HEALTH, ARAB REPUBLIC OF EGYPT. Arab Republic of Egypt nutrition status survey. Washington, DC, USAID, 1978. 16. MINISTRY OF HEALTH AND SOCIAL WELFARE, REPUBLIC OF LIBERIA. Liberia national nutritional survey: December 1975-March 1976. Washington, DC, USAID, 1976. 17. GOVERNMENT OF SIERRA LEONE. Sierra Leone national nutrition survey. Washington, DC, USAID, 1978. 18. BUREAU OF NUTRITION, DEPARTMENT OF PUBLIC HEALTH AND POPULATION, REPUBLIC OF HAITI. Haiti nutrition status survey. Washington, DC, USAID, 1979. 19. GOVERNMENT OF UNITED REPUBLIC OF CAMEROON. United Republic of Cameroon national nutrition survey. Washington, DC, USAID, 1978. 20. HIS MAJESTY'S GOVERNMENT OF NEPAL. Nepal nutrition status survey: January-May 1975. Washington, DC, USAID, 1975. 21. MINISTRY OF HEALTH AND MINISTRY OF SUPPLY OF THE YEMEN ARAB REPUBLIC. The Yemen Arab Republic national nutrition survey. Washington, DC, USDHH/ USAID, 1979. 22. NKAMANY, K. ET AL. The consequences of the drought in Bas-Zaire. Disasters, 4: 55-64 (1980). 23. NKAMANY, K. ET AL. Investigation of nutritional consequences of the drought in Bas-Zaire, 1978. Kinshasa, Zaire National Nutrition Planning Center, 1979 (Report No. VIII). 24. AYKROYD, W. R. Conquest of deficiency diseases. Geneva, World Health Organization, 1970, p. 17. 25. KLOTH, T. I. Sahel nutrition survey. Atlanta, Center for Disease Control, 1974.