Birth weights in two rural hospitals in the United Republic of Tanzania.

Bulletin of the World Health Organization, 66 (5): 653-658 (1988) © World Health Organization 1988 Birth weights in two rural hospitals in the United Republic of Tanzania J. VAN ROOSMALEN1 Low birth weight has been designated by WHO as an indicatorfor monitoringprogress towards health for all by the year 2000. Data on birth weight distribution in rural areas ofAfrica, where the percentage of low-birth-weight infants is estimated to be high, are, however, relatively scarce. Such data are therefore presented here for births in two rural hospitals in the south-western highlands of the United Republic of Tanzania. Analysis of the data indicated that infants of birth weight > 2500 g have a relatively high risk ofperinatal mortality. This is interpreted as the contribution of cephalopelvic disproportion toperinatal mortality. It is concluded that the prevention ofprolonged labour will probably have more impact than provision of neonatal intensive care facilities on lowering perinatal mortality. Low birth weight is a strong determinant ofneonatal and infant mortality as well as reflecting maternal nutritional status (1). The results of studies of low birth weight are therefore ofpublic health importance. Data on the incidence of low birth weight on a worldwide scale are collected and regularly updated by WHO. Globally, incidences range from 7% in Europe to 14% in Africa and 31 % in middle south Asia (1-2). The percentage of low-birth-weight infants in a community has been accepted by WHO as one of the indicators for monitoring progress towards health for all by the year 2000, and a global target of less than 10% for this indicator has been set (3). Little information is available about the distri- bution of birth weights in Africa. Here, such data are therefore presented for two rural hospitals in the south-western highlands of the United Republic of Tanzania. Some differences in the methodological procedures used to calculate the percentage of low- birth-weight infants in the same population are also discussed. MATERIALS AND METHODS Birth weight data are presented on the following: -2319 consecutive births (including 34 pairs of twins) in Lugarawa hospital from 1976 to 1979, where 24% of all births in Ludewa district took place; and l Formerly, Medical Officer, Lugarawa hospital (1976-79) and Mbozi hospital (1979-83). Requests for reprints should be sent to Dr van Roosmalen at: Bloemendaalseweg 244, 2051 GN, Overveen, Netherlands. -4372 consecutive births (including 107 pairs of twins and two sets of triplets) in Mbozi hospital from 1980 to 1983, where 13% of all births in Mbozi district occurred. Lugarawa hospital is the only hospital in Ludewa district (area: about 7000 km2; 75 000 inhabitants). Most women who give birth in this hospital are subsistence farmers who cultivate maize and beans. Mbozi district hospital is the only hospital in the district (area: about 10 000 kmi2; 250 000 inhabi- tants). People in this area grow coffee in addition to maize and beans, and both districts are situated in the south-western highlands of the country. All newborns in these hospitals are weighed on a beam balanco by nurse-midwives within 0.5 h of birth. The balance used in Lugarawa is graduated in pounds and ounces and the readings were expressed in kilograms using a conversion chart. Low birth weight is defined by WHO as < 2500 g (4). In the present study, infants who weighed <1000 g were registered as abortions and were therefore not included. The percentage of low-birth-weight infants was calculated using the following four denominators: the total number of births, the number of live births, the number of singleton births, and the number of singleton live births. Since data on gestational age and the age of the women were unreliable, and very often unknown, these variables were not used in the study. The birth weight distribution of infants delivered at the Mbozi hospital was investigated in relation to parity and maternal height as well as to the sex of the infant. Missing observations were few: the birth weight of two and the sex of 35 infants; and the parity of three 4924 -653- 654 J. VAN ROOSMALEN and the height of 54 women. In contrast, in Luga- rawa, the birth weight distribution could only be calculated for the last 1219 of the total of 2319 births studied, since for the first 1100 births it was recorded only whether or not the birth weight was less than 2500 g. The distribution found for the restricted number of births did not differ from that for Mbozi and will not be referred to further here. Birth-weight-specific perinatal mortality in Mbozi was also compared with data on 24 897 births in a community of mixed ethnic origin in Amsterdam, the Netherlands.a RESULTS Birth weight distribution The birth weight distributions of singleton and multiple births in Mbozi hospital, both of which are approximately Gaussian, are shown in Fig. 1. Out of 4372 births, 678 (16%) infants had low birth weight. The corresponding proportion for Lugarawa was 402 (17%) out of 2319 live births. The mean weight of singleton births in Mbozi was 2946 g (standard deviation (SD): 476 g; range: 1000-4700 g), while the mean weight of multiple births was much lower at 2249 g (SD: 433 g; range: 1100-3550 g). For purposes of comparison, the mean birth weights of infants born in a selection of hospitals or clinics in east and southern Africa are also shown in Table 1. a DOORNBOS, J. F. R. & NORDBECK, H. J. Perinatal mor- tality - obstetric riskfactors in a commnunity ofmixed ethnic origin in Amsterdam. (Ph.D. Thesis, University of Amsterdam, 1985). 9 Singleton births In= 42011 Multiple births In= 1711 30 .. _ 053 .- .-30~~~~~~~3 20- 15 6'w 10~~~~~~~~~~~~~~~2311~~~~~~~~~~1 10 i 12.6 0~~~~~~~~~~ ~ 41 Birth weight range (g) Fig. 1. Birth weight distribution for singleton and multiple births in two rural communities in the United Republic of Tanzania. Figures in italics indicate the number of infants in each weight interval. Birth weight, parity, and maternal height in Mbozi Primiparae in Mbozi delivered singleton infants with a mean birth weight of 2844 g, more than 100 g below the mean birth weight for multiparae (Table 2). In Mbozi 13% ofwomen who were taller than 150 cm delivered low-birth-weight infants, while 26% of women who were less than 146 cm delivered low- birth-weight infants. Finally, the proportion of low- birth-weight infants born to women whose height was 146-150 cm was 18%. Birth weight and sex of infants in Mbozi In Mbozi male infants had a greater mean birth Table 1. Mean birth weights of infants in some communities in east and southern Africa All Live Singleton Live singleton Area (period) births (g) births (g) births (g) births (g) United Republic of Tanzania Dar es Salaam (1975-76)° 2991 - - Lugarawa (1 976-79) 2830 - - Mbozi (1980-83) 2919 2934 2946 2962 lkwiriri (1972-80)b - - 3009 Kenya Machakos (1975-78)c - 3133 Lesotho Quthing (1979-80)d - 3052 ' See reference (5). See reference (6). C See reference (7). d LAMBERS, M. D. A. Early growth of the Mosotho child. (Thesis, University of Amsterdam, 1984). BIRTH WEIGHTS IN RURAL TANZANIA 655 weight than that of their female counterparts (Table 2). However, the birth weight differences between the sexes diminished with increasing parity: for example, there was a 167-g difference between male and female infants for primiparae; a 115-g differ- ence for parities of 1-4, and a 70-g difference for parities > 5. Birth weight and perinatal mortality The relation between birth weight and perinatal mortality can be seen from Fig. 2, which shows the birth-weight-specific perinatal mortality rates for singleton births in Mbozi. For purposes of compari- son, Table 3 shows the birth-weight-specific perinatal mortality rate for all births in Mbozi together with analogous data from a recent study of a large com- munity of Dutch, Negroid, Asian, and Mediterranean origin living in Amsterdam, the Netherlands. For all weight groups, the risk of perinatal death was greater in Mbozi than in Amsterdam, but the relative risks depended on the particular weight groups -the highest relative risk (5.1) being for infants of birth weight )2500 g and the lowest (1.2) for those of birth weight 1500-1999 g. A simple logistic regression analysis of the data using a weighted least-squares approach indicated that for every 100-g decrease in birth weight the perinatal mortality increased by 18% in Mbozi and by 24% in Amsterdam. DISCUSSION The mean birth weights in Lugarawa and Mbozi are among the lowest that have been reported for African countries. The average birth weights of infants who died in the perinatal period and of multiple births are much lower than those of live-born singletons (Table 1). It has been estimated that the proportion of low- birth-weight infants in east Africa "dropped" from 14% in 1979 to 13% in 1982 (2). The data reported Perinatal mortality rate (per 10001 0 100 200 300 400 500 60 700 800 900 1000 perinastal n monalitV 1000 21 20 12S0 34 20 1500 42 21 1750 48 6 2000 142 19 2250 266 28 2000 602 29 2750 931 29 3000 1053 31 3250 600 10 3500 311 7 3750 116 5 40001 30 0 4250 4 0 0 4500 1 0 All 4201 220 Fig. 2. Birth-weight-specific perinatal mortality rate (per 1000) for singleton births in Mbozi, United Republic of Tanzania. for the United Republic of Tanzania, however, indicate that there was an "increase" from 13% in 1979 to 14.4% in 1982 (2). It has been suggested that such changes indicate a deterioration in the situation (2), but a deduction based on such small differences is problematic for the reasons indicated below. (1) Almost all low-birth-weight data are derived from hospitalpopulations. Clearly, it will be difficult for some years to come to obtain population- based birth weight data in developing countries: for example, even the well-known population-based Machakos study in Kenya refers only to hospital- based data (7). Moreover, the few population-based data that have been reported involved small sample sizes. b (2) Different definitions oflow birth weight. Exten- sion of the definition of low birth weight to include also infants who weighed exactly 2500 g at birth increased the proportion of low-birth-weight infants in Lugarawa and Mbozi by 2.7% and 2. 1 %, respec- b SRI KARDJATI, M. Maternal nutrition profile and birth- weight in rural villages in Sampang, Madura (Indonesia). (Thesis, Airlangga University, Surabaya, 1985). Table 2. Mean birth weight, parity, and sex of singleton births in Mbozi district, United Republic of Tanzania, 1980-83 Male Female Parity infants (g) n infants (g) n Both sexes n 0 2926 (425)a 505 2759 (455) 484 2844(447) 989 1-4 3043 (468) 1136 2928 (449) 1105 2986 (462) 2241 .,5 2994(549) 453 2924 (491) 492 2958 (521) 945 All 3004 (479) 2094 2888 (466) 2081 2946 (476) 4175 a Figures in parentheses are standard deviations. 656 J. VAN ROOSMALEN Table 3. Birth-weight-specific perinatal mortality in Mbozi, United Republic of Tanzania, and Amsterdam, Netherlands Perinatal mortality Relative risk Birth weight range (g) Mbozi (n = 4372) Amsterdam (n = 24 897)a (Mbozi/Amsterdam) 1000-1499 69.2%; n= 45 49.0%; n= 79 1.4 011-1.8)b 1500-1999 23.5%; n= 28 20.0%; n= 73 1.2 (0.8-1.7) 2000-2499 10.1%; n= 50 5.4%; n = 65 1.9 (1.3-2.7) >2500 3.1%;n=113 0.6%; n= 139 5.1 (4.0-6.5) All 5.4%; n = 236 1.4%; n = 356 3.8 (3.2-4.4) Data from Doombos & Nordbeck. (see footnote a, p. 654). Figures in parentheses are the 95%-confidence intervals. tively (Table 4). There is no reason for not complying with the WHO-recommended limit of birth weights <2500 g; however, many studies still appear in which low birth weight is defined as .2500 g (8-12). (3) Inclusion of birth weights less than 1000 g. Although it is recommended that national statistics should include data on all deliveries of birth weight > 500 g, different registration practices for highly premature births in various countries (13) prompted WHO to recommend the use of standard perinatal statistics, whereby calculation of rates is restricted to birth weights > 1000 g.c c FIGO STANDING COMMITTEE ON PERINATAL MORTALITY AND MORBIDrry: Report of the commitee following a workshop on monitoring and reporting perinatal mortality and morbidity, 19-21 March 1982, Heidelberg, Federal Republic of Germany. Unpublished document, International Federation of Gynaecology and Obstetrics, London, 1982. (4) Different methods of calculating low-birth- weight rates. The use of different denominators to calculate low-birth-weight rates resulted in differ- ences in the percentage of low-birth-weight infants of up to 1.6% in Lugarawa and 3.4% in Mbozi (Table 4). Combining these differences with those discussed above in the section on the use of different definitions of low birth weight leads to a range in the proportion of low-birth-weight infants of 15.7% to 20% in Lugarawa and of 12.1% to 17.6% in Mbozi (Table 4). These considerable differences call for use of a uniform definition of low birth weight to facilitate comparison with other studies. Usually the proportion of low-birth-weight infants is expressed as a percentage of the total number of live births, as recommended by WHO (3). Calculated in this way, the proportion was higher (17.1%) in Lugarawa than in Mbozi (14.5%). In both these Table 4. Incidence of low birth weight (LBW) in east and southern Africa for different definitions and calculated in various ways Definition Singleton Live singleton Area (period) of LBW All births Live births births births United Republic of Tanzania Lugarawa (1976-79) 1000-2500 g 20.0% 19.4% 18.5% 18.0% 1000-2499 g 17.3% 17.1% 16.0% 15.7% Mbozi (1980-83) 1000-2500g 17.6% 16.5% 15.1% 14.0% 1000-2499 g 15.5% 14.5% 13.2% 12.1% Dar es Salaam (1975-76) <2501 g 15.2% - - lkwiriri (1972-80) <2500g - - - 10.6% Kenya Machakos (1975-78) 1000-2499 g - - 6.9% Lesotho Quthing (1979-80) <2500g - - 7.7% BIRTH WEIGHTS IN RURAL TANZANIA 657 districts there was almost no difference in the distri- butions of maternal height (population below 150 cm: 19% in Lugarawa and 18% in Mbozi). The socioeco- nomic circumstances in Mbozi district, with its coffee growing, were better than those in Ludewa district, where there is subsistence farming only. The recent low-birth-weight data from other rural areas in east and southern Africa shown in Table 4 indicate that in these areas the percentage of low- birth-weight infants is lower than that found in the present study. In Tanzania, the only data on low- birth-weight rates in urban areas (calculated using birth weights < 2500 g and all live births) are for Dar es Salaam, where the rate was 15.2% in 1975-76 (5). The comparable rates were 19.4% for Lugarawa and 16.5% for Mbozi. It is unlikely that these differences between urban and rural areas can be accounted for by differences in social class, since 97% of the study population in Dar es Salaam were of low socioeco- nomic status. There was, however, a social class gradient in the Dar es Salaam sample since 6.5% of low-birth-weight infants were from the highest and 15.6% the lowest extremes of the low social class group (5). A considerably smaller proportion of low- birth-weight infants (10.6%) was found in Ikwiriri, a village in rural Tanzania, based on 70% of all births that took place in the local health centre, and calcu- lated using live singleton births as the denominator. This is all the more surprising since the prevalence of malaria is higher in this area than in Lugarawa and Mbozi, and placental infection with malaria is con- nected with low birth weight. The high level of antenatal malaria and anaemia prophylaxis coverage in Ikwiriri village may, however, account for the rather low percentage of low-birth-weight infants born there (6). Birth weight and perinatal mortality As has also been observed previously, in the present study perinatal mortality increased with decreasing birth weight; however, surprisingly, the perinatal mortality rates of infants who weighed 1500-1999 g in Mbozi was almost the same as that found in Amsterdam. Facilities for neonatal intensive care were nonexistent in Mbozi. In part, the similar perinatal mortality rates in Mbozi and Amsterdam might arise because of ma- turity differences between infants of the same birth weight in the two locations (14, 15). Gestational age was not taken into consideration in the study, although it is an important determinant of perinatal mortality (16, 24), and this underlines that the correct interpretation of the results of low-birth-weight studies is difficult (17). The results of the study demonstrate that, with in- creasing birth weight, the relative risk of perinatal mortality is greater in Mbozi than in Amsterdam. Also, the relative risk of perinatal death of "mature" as compared with that of low-birth-weight infants was much greater in Mbozi than in Amsterdam. This presumably arose because of the contribution of cephalopelvic disproportion to perinatal mortality. Obstructed labour was the most frequent cause of perinatal death in Mbozi (78 out of 298 perinatal deaths (26%) during the study period), and the majority of these deaths (66 of the 78) involved infants of birth weight > 2500 g.d This indicates that the prevention of prolonged and obstructed labour may be of great importance in reducing perinatal mortality, and this objective should therefore have a higher priority in rural areas than the provision of intensive neonatal care facilities. The proportion of low-birth-weight infants in a community can be reduced by providing dietary sup- plements to pregnant women who are nutritionally at risk (18, 19).e In this respect, high energy foods are more important than high protein supplements (20). A side-effect of such programmes to increase birth weights can, however, be an increase in the preva- lence of cephalopelvic disproportion of neonates (21). Selection of women at risk of delivering low-birth- weight infants is a far from easy task (22) -pre- pregnancy weight and height, maternal weight-for- height, and maternal weight gain during pregnancy are only some of the indicators (23, 24). Ultimately, however, the eradication of poverty and of illiteracy and the stimulation of rural socioeconomic develop- ment will eliminate the need for dietary supplements as indicated by nutritional studies in developed countries (20, 23). d VAN ROOSMALEN, J. Maternal health care in the south- western highlands of Tanzania. (Thesis, University of Leiden, 1988). ' See footnote b, p. 655. RESUME POIDS A LA NAISSANCE DANS DEUX HOPITAUX RURAUX DE RtPUBLIQUE-UNIE DE TANZANIE L'OMS a choisi l'insuffisance pond6rale a la naissance comme l'un des indicateurs des progres realises vers la sant6 pour tous d'ici I'an 2000. En Afrique, on dispose de peu de donnees sur la repartition des poids a la naissance. On a 658 J. VAN ROOSMALEN donc pr6sente ici les resultats obtenus pour 2319 naissances cons6cutives a 1'hopital de Lugarawa (1976-1979) et pour 4372 naissances a 1'hopital de Mbozi (1980-1983), tous deux situ6s dans les montagnes du sud-ouest de la Re- publique-Unie de Tanzanie. A Mbozi, le poids moyen a la naissance pour les accouchements simples etait de 2946 g (ecart type (a) =476 g), alors qu'il etait bien plus faible pour les naissances multiples avec 2249 g (a=433 g). En moyenne, les primipares ont donne naissance a des enfants plus petits que les multipares et les garcons etaient plus gros que les filles. La mortalite perinatale due a l'insuffisance ponderale a la naissance a l'hopital de Mbozi a ete comparee avec la mortalite observee sur un large echantillon de population a Amsterdam (Pays-Bas). Le risque de deces perinatal etait plus eleve a Mbozi qu'a Amsterdam pour tous les groupes de poids, mais avec un risque relatif variable en fonction du poids: le risque relatif le plus eleve (5,1) se retrouve pour les poids a la naissance > 2500 g et le risque relatif le plus faible (1,2) pour les poids compris entre 1500 et 1999 g. L'emploi de definitions de l'insuffisance ponderale a la naissance et de methodes de calcul differentes s'est traduit par des ecarts d'environ 5% dans le pourcentage de nouveau- nes presentant une insuffisance ponderale dans une meme population. Pour effectuer des comparaisons a l'6chelon international, il est par consequent recommande d'adopter des criteres uniformes dans ce domaine. Le risque de mortalite perinatale relativement eleve observe pour les enfants ayant un poids a la naissance superieur ou egal a 2500 g est probablement du' au fait que la disproportion foeto-pelvienne est responsable de davan- tage de complications chez les nourrissons normaux que chez les nourrissons ayant un faible poids a la naissance. La prevention des complications du travail aura probablement plus d'impact sur la diminution de la mortalite perinatale que la mise en place de services de soins intensifs pour les nouveau-nes. REFERENCES 1. WORLD HEALTH ORGANIZATION. The incidence of low birth weight-a critical review of available infor- mation. World health statistics quarterly, 33: 197-224 (1980). 2. WORLD HEALTH ORGANIZATION. The incidence of low birth weight: an update. Weekly epidemiological record, 59: 205-211 (1984). 3. Development of indicators for monitoring progress towards health for all by the year 2000. Geneva, World Health Organization, 1981 (Health for All Series, No. 4). 4. WORLD HEALTH ORGANIZATION. International Classifi- cation ofDiseases. 1975 Revision. Volume 1. Geneva, WHO, 1977. 5. MBISE, P. R. & BOERSMA, E. R. Factors associated with low birthweight in the population of Dar es Salaam, Tanzania. Tropical and geographical medicine, 31: 21-32 (1979). 6. BANTJE, H. Birthweight distribution and antenatal care in Ikwiriri village, Tanzania. Tropical and geographi- cal medicine, 34: 213-223 (1982). 7. VOORHOEVE, A. M. ET AL. The outcome of pregnancy. In: Ginneken, J. K. & Muller, A. S., ed. Maternal and child health in rural Kenya, an epidemiological study. Beckenham, Croom Helm, 1984, pp. 223-240. 8. TAMBYRAA, R. L. & RATNAM, S. S. The small fetus: growth-retarded and preterm. Clinics in obstet- rics and gynaecology, 9: 517-537 (1982). 9. STRAHAN, M. Birthweights in a rural Solomon Island population. Journal oftropicalpediatrics, 30: 293-296 (1984). 10. EDITORIAL: Infant mortality and the low birthweight infant. Journal of the American Medical Association, 253: 826 (1985). 11. SCHELP, F. P. & PONGPAEW, P. Analysis of low birthweight rates and associated factors in a rural and an urban hospital in Thailand. Journal of tropical pediatrics, 31: 4-8 (1985). 12. DAVIS Tsu, V. & NEwTON, N. Appropriate technolo- gies for perinatal care. Advances in internal medicine and child health, 6: 166-194 (1986). 13. KEIRSE, M. J. N. C. Perinatal mortality rates do not contain what they purport to contain. Lancet, 1: 1166-1169 (1984). 14. ROOTH, G. Low birthweight revised. Lancet, 1: 639-641 (1980). 15. RANSOME-KUTI, D. Intra-uterine growth, birthweights and maturity of the African newborn. Acta paediatrica Scandinavica supplement, 319: 95-102 (1985). 16. VERLOOVE-VANHORICK, S. P. ET AL. Neonatal mor- tality in very low birthweight and very preterm infants. American journal of obs4tetrics and gynecology, 153: 929-930 (1985). 17. KEIRSE, M. J. N. C. Epidemiology and aetiology of the growth retarded baby. Clinics in obstetrics and gynaecology, 11: 415-436 (1984). 18. LECHTIG, A. ET AL. Effects of maternal nutrition on infant health: Implications for action. Journal of tropi- cal pediatrics, 28: 273-286 (1982). 19. WHARTON, B. A. Sorrento studies of birthweight. Case for international reference data. Acta paediatrica Scandinavica supplement, 319: 170-179 (1985). 20. RUSH, D. Effects of changes in protein and calorie intake during pregnancy on the growth of the human fetus. In: Enkin, M. & Chalmers, I., ed. Effectiveness and satisfaction in antenatal care. London, Heine- mann, 1982, pp. 92-113. 21. MTIMAVALYE, L. A. R. ET AL. The relationship between increasing birthweight and cephalopelvic dis- proportion in Dar es Salaam, Tanzania. Journal of obstetrics and gynaecology of the British Common- wealth, 81: 380-382 (1974). 22. HABICHT, J. P. & YARBROUGH, C. Efficiency in sel- ecting pregnant women for food supplementation during pregnancy. In: Aebi, H. & Whitehead, R., ed. Maternal nutrition during pregnancy and lactation. Bern, Hans Huber, 1980, pp. 314-336. 23. LECHTIG, A. ET AL. Effect of maternal nutrition on the mother-child dyad. Proceedings ofthe XIV Symposium of the Swedish Nutrition Foundation. Stockholm, Almquist & Wiksell International, 1979, pp. 74-93. 24. KRAMER, M. S. Determinants of low birth weight: methodological assessment and meta-analysis. Bulletin ofthe World Health Organization, 65: 663-737 (1987).