5
Mortality

INTRODUCTION

Throughout the world, wherever data are available, we know that mortality has declined over the last century. In regions where long time-series data are unavailable, our knowledge of mortality trends over the recent past is drawn from sources such as ad hoc surveys of women's birth histories. In sub-Saharan Africa, available information shows that mortality was very high in the middle of the twentieth century and that it has declined substantially since. In the 1950s, 30-40 percent of newborn children died before reaching age 5 (Hill, 1992, 1993). In the 1970s, the proportion was much lower—10-25 percent. Adult mortality has probably also decreased substantially over this period, though the data to document the decline are scanty (Timæus, 1993).

Both health programs and economic development have probably contributed to the mortality decline in African countries. In the past, the development of health services in Senegal mirrored the development in industrialized countries: most individual services were provided through hospitals, supplemented by disease-control public health measures. Specific programs were organized to control infectious diseases, such as those for the eradication of smallpox or the control of leprosy. Some of these programs reached into rural areas, but most health care services were located in urban settings. The reorientation of health policy towards primary health care is one of the great changes of the modern era. As was decided at the World Health



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 113
Population Dynamics of Senegal 5 Mortality INTRODUCTION Throughout the world, wherever data are available, we know that mortality has declined over the last century. In regions where long time-series data are unavailable, our knowledge of mortality trends over the recent past is drawn from sources such as ad hoc surveys of women's birth histories. In sub-Saharan Africa, available information shows that mortality was very high in the middle of the twentieth century and that it has declined substantially since. In the 1950s, 30-40 percent of newborn children died before reaching age 5 (Hill, 1992, 1993). In the 1970s, the proportion was much lower—10-25 percent. Adult mortality has probably also decreased substantially over this period, though the data to document the decline are scanty (Timæus, 1993). Both health programs and economic development have probably contributed to the mortality decline in African countries. In the past, the development of health services in Senegal mirrored the development in industrialized countries: most individual services were provided through hospitals, supplemented by disease-control public health measures. Specific programs were organized to control infectious diseases, such as those for the eradication of smallpox or the control of leprosy. Some of these programs reached into rural areas, but most health care services were located in urban settings. The reorientation of health policy towards primary health care is one of the great changes of the modern era. As was decided at the World Health

OCR for page 113
Population Dynamics of Senegal Assembly in 1978 at Alma-Ata, primary health care aims to provide limited but affordable services at the community level in an integrated manner using community health workers. Following the 1978 conference, Senegal reformulated its health care policy, placing greater emphasis on primary health care in rural areas. The number of hospitals, health posts, community health workers, and pharmacies increased throughout Senegal, particularly in areas outside of Dakar. In 1981, the Expanded Programme on Immunization (EPI) was established, with the purpose of increasing vaccination coverage in rural as well as in urban areas. The results of the EPI efforts have been notable. Vaccination coverage has increased substantially over the past couple of decades, particularly in rural areas, narrowing the inequalities in vaccination coverage between urban and rural areas (see the appendix to this chapter for more detail on health programs in Senegal). From the early 1980s, efforts toward the provision of better health services have taken place against a back-ground of economic crises and more recently of the HIV/AIDS epidemic. This chapter reviews the levels and trends in mortality in Senegal. The next two sections address data sources and national levels, trends, and geographical variations in child mortality and adult mortality, respectively. The final section presents conclusions. CHILD MORTALITY Sources and Quality of Data Five surveys and one census supply data that permit estimation of the national level of child mortality: the 1960-1961 Demographic Survey (DS) [1], the 1970-1971 National Demographic Survey (NDS) [2], the 1978 World Fertility Survey (WFS) [3], the 1986 Demographic and Health Survey (DHS-I) [4], the 1988 census[18], and the 1992-1993 Demographic and Health Survey (DHS-II) [7] (see Appendix A). The type and quality of data gathered vary among surveys, as do the methodologies used. Accordingly, it is preferable to focus initially on a simple, robust indicator of child mortality, such as the probability that a newborn will die before the age of five (5q0), also called the under-5 mortality rate. The advantage of this indicator is that it is less sensitive to age-reporting errors (for ages of surviving children or ages at time of death) than is the infant mortality rate or the probability of death prior to 1 year, 1q0, which is more often used. Moreover, 5q0 is less sensitive to biases linked to the estimation method (Cantrelle et al., 1986; Fargues and Khlat, 1989). The 1960-1961 DS [1] gathered information on the number of births and surviving children of each woman interviewed. These data allow the

OCR for page 113
Population Dynamics of Senegal indirect estimation of the risk of death between 0 and 5 years of age. The method applied here is that of Brass, as adapted by Trussell and described in United Nations (1983a). The 1970-1971 NDS [2] gathered prospective data on births and deaths occurring during a 12 month period in a sample of households. Three visits were made to each household at 6 month intervals. This method of data collection reduces the risk of omitting childhood deaths, so the data are of fairly good quality. The probability of dying by age 5 was calculated directly by relating deaths to person-years of exposure to risk. The WFS [3], DHS-I [4], and DHS-II [7] collected birth histories for all sampled women, providing information for each live-born child on date of birth and, if the child had died, age at the time of death. These data permit direct calculation of the risks of death for different periods preceding the survey. A bias that often affects this kind of information is that some children who die young are omitted, thereby causing underestimation of mortality. In general, the risk of omission is greater for older than for younger women and for events further in the past. Another kind of error concerns the identification of the correct date of birth, which is often known only approximately, particularly in environments where illiteracy is prevalent. It appears that, in the WFS, births tended to be shifted toward the survey date, a phenomenon that leads to slight overestimation of the mortality rate at a given date if mortality is declining. This bias is not apparent in the other surveys. In fact, in the DHS-I, some births were probably moved from the most recent 5 year period to an earlier period, which could, conversely, lead to an underestimation of mortality. A third common type of error is in the reporting of age at death, both rounding to numbers of years (such as "1 year") and exaggerating age at death. This error may reduce the infant mortality rate relative to young child mortality, but has little effect on our chosen indicator, 5q0. The 1988 census [18] provided data about deaths occurring during the preceding 12 months. This type of data is normally of poor quality, because of respondents' difficulty in situating the beginning of the previous 12 months. Respondents tend either to over- or (more typically) underestimate mortality, depending on whether the beginning of the period is mistakenly placed before or after the true beginning of the reference period. Low reliability also results from the omission of certain deaths, especially those occurring shortly after birth, as in all retrospective surveys. With respect to the 1988 census, it is possible that the first bias, tied to the difficulty in accurately situating the beginning of the preceding 12 months, was less important than usual because the major Muslim holiday, the Korité, took place exactly 1 year prior to the census and was thus used as point of reference (see Chapter 4, footnote 6 for details). The census data on child mortality can be evaluated by comparison

OCR for page 113
Population Dynamics of Senegal with other sources, namely the ongoing small-scale surveys conducted in three rural areas, Niakhar/Ngayokhème [11] (Project Niakhar, 1992), Bandafassi [8] (Pison and Desgrées du Loû, 1993), and Mlomp [10] (Lagarde et al., forthcoming). Because the census data and data from the small-scale surveys do not exactly coincide in terms of area, comparisons can be made only between each study area and the rural part of the department in which that area is located. Comparisons were focused on the probability of dying before age 1 and the probability of dying between age 1 and 5 (1q0 and 4q1, respectively). The results appear in Table 5-1. In each case, the census estimate of 1q0 for the rural area of the department falls substantially below the estimate of the corresponding local study. The census estimates amount to 30 percent, 38 percent, and 41 percent of the local study estimates, respectively, for rural Fatick as compared with Niakhar/Ngayokhème, rural Oussouye as compared with Mlomp, and rural Kédougou as compared with Bandafassi. On the other hand, census estimates of 4q1 differ little from those of the local studies: 141 per 1,000 in rural Fatick as compared with 152 in Niakhar, 90 per 1,000 in rural Oussouye as compared with 71 in the Mlomp study area, and 117 per 1,000 in rural Kédougou as compared with 121 in Bandafassi. Local comparisons thus indicate underreporting by as much as two-thirds of the deaths of children below the age of 1 year recorded in the census. Conversely, the local comparisons indicate a relatively high level of recording of deaths among those aged 1-4. These conclusions are supported by national-level data. Table 5-2 gives the national estimates from the 1988 census and the national surveys, DHS-I and DHS-II. The infant mortality rate reportedly fell from 86 per 1,000 in 1981-1986 (DHS-I estimate) to 34 per 1,000 in 1987-1988 (census estimate), or a decline of more than one-half (60 percent) within 4 years, before rising thereafter to 68 per 1,000 (or 100 percent) in 1988-1992 (DHS-II estimate). These short-term fluctuations are highly implausible. Differentials among the various estimates most likely result from a pronounced underreporting in the census of deaths of children less than 1 year of age. On the other hand, mortality of children aged 1-4 reportedly fell from 114 per 1,000 in 1981-1986 to 83 per 1,000 in 1987-1988, and to 68 per 1,000 in 1988-1992, which is a plausible decline. Thus, the usual bias of retrospective studies—the omission of early deaths—is encountered in the census. On the other hand, the omission of deaths does not seem to be a problem above 1 year of age. Moreover, as noted above, the bias arising from the difficulty in temporally situating the beginning of the preceding 12 months was probably less of a factor than usual because of the Korité. Comparison of deaths registered in vital records in the city of Saint-Louis

OCR for page 113
Population Dynamics of Senegal TABLE 5-1 Comparison of Mortality Estimates from the 1988 Census and Selected Local Studies Region Source Date 1q0 (per 1,000) 4q1 (per 1,000) Department of Fatick (rural) 1988 census 1987-1988 36 141 Portion of the Niakhar and Tataguine arrondissements Niakhar study 1987-1988 122 152 Department of Oussouye (rural) 1988 census 1987-1988 19 90 Portion of the Loudia Wolof arrondissement Mlomp study 1985-1991 50 71 Department of Kédougou (rural) 1988 census 1987-1988 58 117 Portion of the Bandafassi arrondissement Bandafassi study 1986-1991 140 121 NOTES: Method of calculation for the 1988 census: annual mortality rates were calculated as the ratio of the number of deaths in the last 12 months to the average population in the last 12 months. The latter was estimated from the census adjusted downward by 1.5 percent, the annual growth rate of the population being around 3 percent. The mortality rates were then converted into the corresponding probabilities of dying. See Appendix A for description of surveys. SOURCES: Bandafassi study: Pison and Desgrées du Loû (1993); Mlomp study: Lagarde et al. (forthcoming); Niakhar study: Project Niakhar (1992); 1988 census: 1988 census (unpublished data) between May 1987 and May 1988 with those reported in the census yields the same conclusions: deaths occurring before 1 year of age are less than half as numerous in the census as in vital records, that is, 99 as compared with 210. For ages 1-4, however, the two sources are much closer: 109 deaths according to the census and 84 according to the vital records (Diop, 1990). This analysis of the quality of data from the 1988 census thus gives an unexpected result: aside from deaths of children under 1 year of age, the data gathered concerning deaths during the last 12 months seem relatively complete. At the national level, if one discards the census estimate for 1q0 and combines the census estimate for 4q1 with 1q0 from another source, estimates

OCR for page 113
Population Dynamics of Senegal TABLE 5-2 Comparison of National-Level Mortality Estimates from the 1988 Census (data on deaths in the last 12 months) and Estimates from DHS-I and DHS-II Source Period 1q0 (per 1,000) 4q1 (per 1,000) 1986 DHS-I 1981-1986 86 114 1988 census 1987-1988 34 83 1992-1993 DHS-II 1988-1992 68 68 NOTE: See Appendix A for description of surveys. SOURCES: 1986 DHS-I: Ndiaye et al. (1988); 1992-1993 DHS-II: Ndiaye et al. (1994); 1988 census (unpublished data) for 5q0 can be calculated. For example, combining the midpoint between the 1q0 indicated by the DHS-I for the period 1981-1986 and that of the DHS-II for 1988-1992, or 77 per 1,000, with the 4q1 estimate of 83 per 1,000 from the census gives an estimate of 5q0 for the period June 1987-May 1988 of 154 per 1,000, as seen in Table 5-3. Mortality Levels and Trends Among Children Under Age 5 National Level Table 5-3, illustrated in Figure 5-1, shows 5q0 estimates for all of Senegal obtained from the five national surveys and the census for different dates and periods. The table also summarizes the type of data gathered and the method of estimation employed. Although these measurements result from different surveys and estimation techniques, they are fairly consistent. Broadly speaking, in the 45 years following the end of World War II, child mortality (5q0) declined by two-thirds, falling from an estimated level of 373 in 1946 to 131 in the years 1988-1992. The decline appears to have occurred rather slowly until the early 1970s, with 5q0 falling only to 281 per 1,000 by 1970, a 25 percent decline in 25 years. The reduction seems to have accelerated thereafter, with 5q0 falling by more than 50 percent in the next 20 years, from 281 per 1,000 in 1970 to 131 in 1988-1992. The Case of Dakar Multiple sources of information make it possible to track the trend in 5q0 in Dakar over time (see Figure 5-2). The estimates from the various

OCR for page 113
Population Dynamics of Senegal TABLE 5-3 Proportion of Children Dying Before Age 5 in Senegal, 1945-1990 Source Period Reference Date 5q0 (per 1,000) 1960 DS — 1946.2 373a     1951.6 343a 1970-1971 NDS 1970-1971 1970.5 281b 1978 WFS 1964-1968 1966.1 293c   1969-1973 1971.1 283c   1974-1978 1976.1 262c 1986 DHS-I 1971-1975 1973.5 287c   1976-1980 1978.5 236c   1981-1986 1983.0 191c 1988 census 1987-1988 1987.9 154d 1992-1993 DHS-II 1978-1982 1980.6 199c   1983-1987 1985.6 185c   1988-1992 1990.6 131c a Estimate calculated using an indirect method with data on number of children born and number of children surviving (Hill, 1992). b Estimate calculated using a direct method with data on deaths collected by multiround survey. c Estimate calculated using a direct method with data on women's birth histories (status of each child—living or dead—and age at death, if deceased). d Estimate calculated using a direct method with data on deaths within the last 12 months (with a correction due to underreporting of deaths under 1 year of age). (See the discussion earlier in this section.) SOURCES: 1960: DS (Hill, 1992); 1970-1971: NDS (République de Sénégal, 1974) 1978: WFS (Rutstein, 1983); 1986: DHS-I (Ndiaye et al., 1988); 1992-1993: DHS-II (Ndiaye et al., 1994); 1988:1988 census (unpublished data) sources are broadly consistent, and differences among them are due primarily to variations in methodology and data quality, as well as to differences in the populations of children studied. Garenne et al. (1992) analyzed the birth histories of women from the WFS and DHS-I in the region of Dakar who had lived in the city since childhood. These series are shown on Figure 5-2 as WFS-Dakar and DHS-I-Dakar. The DHS-I-Dakar series seems more plausible than that of the WFS-Dakar. For the period 1975-1979, the former matches almost perfectly the estimate from vital registration (Dakar-2), while the WFS-Dakar

OCR for page 113
Population Dynamics of Senegal FIGURE 5-1 Trends in child mortality (5q0 per 1,000), Senegal, 1945-1992. SOURCES: Hill (1992); Ndiaye et al. (1988, 1994); Rutstein (1983); Antoine and Mbodji (1991); and unpublished data from the 1988 census. series shows a pronounced, and not very plausible, increase of mortality around 1960, probably reflecting poorer-quality data for earlier periods. Note that taken together the WFS-Dakar and DHS-I-Dakar sequences indicate a stagnation of 5q0 during the period 1965-1983 at a level of approximately 100-150 per 1,000. This stagnation is also found in the birth history data collected in Pikine in the 1986 Dakar-Pikine Survey [15] if one considers only children born in Pikine to mothers native to the Dakar urban area (Antoine and Diouf, 1992). The level of 5q0 for the same period from this survey was also about 100 per 1,000 (data not shown). Restricting the analysis to children whose mothers have lived in Dakar since childhood excludes children born in Dakar to migrant mothers from rural areas. The mortality of the latter group is higher (Antoine and Diouf, 1992; Garenne et al., 1992). However, the differential between the level of child mortality for all children from Dakar and the estimated level for the

OCR for page 113
Population Dynamics of Senegal FIGURE 5-2 Child mortality trends in Dakar (5q0 per 1,000), multiple sources. NOTES: DHS-I-Dakar: Direct estimates of the risk of death by period, based on data from the 1986 DHS-I [4] conducted solely among women residing in Dakar since childhood. WFS-Dakar: Direct estimates of the risk of death by period, based on data from the 1978 WFS [3] conducted solely among women residing in Dakar since childhood. Dakar-1: Direct estimates of the risk of death by period, based on data from the 1986 Dakar-Pikine Survey [15]. Estimates include only children born in Dakar to women residing in Pikine at the time of the survey. Dakar-2: Mean 1975-1979 estimate from vital records. Pikine: Direct estimates of the risk of death by period, based on data from the 1986 Dakar-Pikine Survey [15]. Estimates included only children born in Pikine to women residing in Pikine at the time of the survey. SOURCES: Antoine and Mbodji (1991); Garenne et al. (1992:Table 5, p. 24, Table 7, p. 26); Cantrelle et al. (1986:Table 76, p. 112). subgroup of children whose mothers have lived in Dakar since childhood may have changed over time. Data from the Pikine survey allow us to analyze these changes (see Figure 5-2). The estimates from the Pikine survey encompass all of the children covered by this survey and born in either Dakar (Dakar-1 in Figure 5-2) or Pikine (Pikine in Figure 5-2), regardless of their mothers' origin. These estimates show a mortality level

OCR for page 113
Population Dynamics of Senegal that (1) was apparently higher during the 1960s than the mortality level of children whose mothers had lived in Dakar since childhood, and (2) declined during the two decades between the early 1960s and the early 1980s (5q0 fell from 200 per 1,000 at the beginning of the 1960s to less than 150 per 1,000 in the early 1980s). Overall mortality was probably not as high as that of children born to women residing in Pikine (a relatively low socioeconomic section of Dakar) regardless of their origin, and not as low as that of children born to women who had resided in Dakar since childhood. The most plausible scenario is that during the period 1960-1985, mortality declined at a relatively slow pace, especially at the end of the period. Rural Areas Figure 5-3 shows the trend of 5q0 in all rural areas in Senegal, according to the WFS, DHS-I, and DHS-II. These measurements were obtained using a method similar to the preceding one, that is, by selecting rural women who had lived in rural areas since childhood (Garenne et al., 1992, and unpublished tables from DHS-II). For the region of Dakar, the WFS series seems less plausible than either of the DHS series, especially before 1965. If only DHS estimates are considered, the trend shows stagnation at a level between 350 and 400 per 1,000 until the early 1970s, followed by a rapid decline beginning in the late 1970s: according to the DHS-I, 5q0 reportedly fell from 369 per 1,000 in 1970-1974 to 231 per 1,000 in 1980-1984, a reduction of 37 percent in 10 years. The DHS-II shows a similarly rapid decline, but with consistently lower mortality than in the DHS-I: according to the DHS-II, 5q0 fell from 303 per 1,000 in 1968-1972 to 133 per 1,000 in 1988-1992. Figure 5-4 compares the trends of 5q0 in rural areas, those discussed earlier in the region of Dakar, and the Saint-Louis series estimated from that city's vital records (Diop, 1990). The contrast between the mortality levels in the cities (Dakar, Pikine, and Saint-Louis) and those in rural areas is enormous during the period 1960-1975. The Saint-Louis series shows that the differentials are long-standing, resulting from the early, sizable decline in mortality in the cities beginning during the first half of the century. At the end of the 1970s, the situation changed. Because of the pronounced decline in mortality in rural areas and a slowing of the decline in the Dakar region, the gap between Dakar and rural areas narrowed to a factor of around two at the beginning of the 1980s. The rural areas did not evolve as a homogenous entity. Figure 5-5 shows some localized differences recorded by reliable surveys. In the various regions studied, we find, more or less, the pattern observed for the

OCR for page 113
Population Dynamics of Senegal FIGURE 5-3 Trends in child mortality (5q0 per 1,000) in rural areas, 1978 WFS, 1986 DHS-I, and 1993 DHS-II. NOTE: Only children whose mothers lived in rural areas since childhood are included. SOURCE: Garenne et al. (1992). whole of rural areas in Senegal: a plateau, followed by a rapid decline. Differences among regions appear mainly with respect to timing. In Bandafassi, the study area furthest from Dakar (700 km), the decline in child mortality began only in the mid-1980s, coincident with the accelerated 1986-1987 EPI (Pison and Desgrées du Loû, 1993) (see the appendix to this chapter for discussion of the EPI). In Niakhar/Ngayokhème, 150 km from Dakar, the decline began earlier, in the early 1970s. The decline during that period is attributed mainly to reduced rainfall, leading to reduced incidence of malaria (Cantrelle et al., 1986); however, the decline persisted at the same pace after the dry years ended, so that other factors undoubtedly already at work must have continued the trend. The rural areas of Fissel and Thiénaba are somewhat closer to Dakar (120 and 90 km, respectively). Thiénaba is 15 km from the major city of Thiès and only 10 km from Khombole, the only rural maternal-child health center in Senegal, which began operation in 1957. Earlier measurements are not available for

OCR for page 113
Population Dynamics of Senegal APPENDIX HEALTH INFRASTRUCTURE AND PROGRAMS Health Infrastructures Health statistics make it possible to monitor the progress of the health infrastructure in Senegal during the last 30 years (see Table 5-29). The number of hospitals increased threefold between 1960 and 1988, reflecting the policy to equip each region with a hospital and to divide some hospitals into two in the large cities. The number of hospital beds has not grown proportionally; it has not even kept pace with population growth. Thus, despite the proliferation of hospitals, the supply of beds per inhabitant has declined. The number of health centers has not changed much, continuing to reflect the policy of having one center for every departmental capital. Such centers are normally run by a physician and are equipped with hospital beds. The number of dispensaries, on the other hand, has increased sharply, having more than tripled between 1960 and 1988. These dispensaries, which are run by nurses, are found throughout the country. They are normally located in the district ("arrondissement") capitals or rural communities. Maternity clinics, of which there are not very many, were concentrated in the towns until 1977. Beginning in 1978, the primary health care policy led to the construction of a large number of such clinics in rural areas. In 1988, there were nearly as many rural maternity clinics as dispensaries, according to health statistics. In addition, the number of family planning clinics increased substantially in the late 1980s (Osmanski et al., 1991). (See the section Contraception in Chapter 4 for more discussion of family planning.) Though Senegal's health infrastructure has not always kept pace with population growth (as in the case of hospital beds), it was decentralized during the period 1960-1988 by virtue of hospital, dispensary, and maternity clinic construction virtually everywhere in the country. In 1960, the Dakar region, which had 14 percent of the population, had 3 out of 5 hospitals (60 percent) and the vast majority of hospital beds. In 1988, Dakar had 22 percent of the population, but only 6 out of 16 hospitals (38 percent) and one-half of the hospital beds (Table 5-30). The distribution of facilities between Dakar and the rest of the country, while remaining unequal, appears to have improved appreciably. Nevertheless, health personnel remain very concentrated in Dakar, where two-thirds of the country's physicians, pharmacists, and dentists, and approximately one-half of its nurses and mid-wives, are found (Table 5-30). Thus while the distribution of health resources in Senegal has improved over time, the Dakar region has remained relatively overserved, with more than 50 percent of resources serving less than one-quarter of the population as of 1988.

OCR for page 113
Population Dynamics of Senegal TABLE 5-29 Change in Health Infrastructure, 1960, 1980, 1988 Infrastructure 1960 1980 1988 Hospitals 5 12 16 Hospital beds 2,400 3,523 5,179 Inhabitants per hospital bed 1,300 1,580 1,650 Health centera 34 35 47 Dispensaryb 201 376 659 Rural maternity clinic — 189 502 Population (in millions) 3.1 5.6 6.9 a Headed by a physician. b Headed by a nurse. SOURCES: République du Sénégal (1988); Osmanski et al. (1991) Health Programs Before 1978 As noted above, Senegal introduced primary health care in 1978. Before that time, public health programs to improve hygienic conditions and control disease had been developed primarily in the towns, building on existing health infrastructures. Rural areas, poorly served by these infrastructures, benefited only from periodic visits of mobile teams from the Major Endemic Diseases Department.12 However, the activities of these mobile teams began to deteriorate following independence in 1960. Numerous programs were implemented before 1978. However, each carried out a specific activity: for example, smallpox eradication and control of leprosy were carried out by mobile teams; maternal and child health (MCH) services were provided by MCH centers in the towns beginning in the 1950s, and growth monitoring and distribution of supplementary foodstuffs were done by the Nutrition and Health Protection Program 13 beginning in the 1960s. Two of these specific programs, the antimalaria campaign and vaccinations, are discussed in greater detail below. Antimalaria Campaign Malaria, which is endemic in Senegal and one of the major causes of child mortality, was the focus of specific eradication programs beginning in 1953 (Cantrelle et al., 1986). Between 1953 and 1961, an eradication trial was conducted in the region of Thiès and the western part of the region of Fatick, in which homes were sprayed with

OCR for page 113
Population Dynamics of Senegal TABLE 5-30 Public and Private Health Resources Located in the Region of Dakar, 1988 Resource Total for Senegal Total for the Region of Dakar Percentage in the Region of Dakar Health Infrastructure Hospitals 16 6 38 Hospital beds 5,179 2,565 50 Health centers 47 7 15 Dispensaries 659 78 12 Medical Personnel Doctors 407 280 69 Pharmacists 200 133 67 Dentists 58 42 72 Nurses 934 375 40 Midwives 474 239 50 Population (in millions) 6.9 1.5 22   SOURCES: République du Sénégal (1988) DDT combined, after 1957, with chemoprophylaxis. This program was a failure. In 1963, another antimalaria program, using chloroquine-based chemoprophylaxis (and named ''chloroquinization"), was launched throughout Senegal. It appears to have been marked by both failures and successes and to have affected the various regions unevenly, though there was little follow-up or evaluation. On the whole, its effects on morbidity and mortality due to malaria appear to have been limited (Garenne et al., 1985). This program, as well as the eradication trial, nevertheless helped popularize chemoprophylaxis in the affected regions. In 1979, this program ended. Thereafter, malaria prevention was incorporated into primary health care. Vaccinations Until the Expanded Programme on Immunization (EPI) was instituted in 1981, vaccinations had been administered in two ways. The first was through mass campaigns conducted in successive operations by mobile teams combing the rural and urban areas. For example, smallpox vaccination campaigns were implemented as part of the worldwide eradication effort against this disease. Vaccinations against yellow fever and meningitis were also organized during epidemic upsurges of these diseases. In addition, measles vaccinations were administered between 1967 and 1969 throughout the country.

OCR for page 113
Population Dynamics of Senegal The second method of administering vaccinations was through MCH centers. These centers dispensed standard vaccines against tuberculosis (BCG), diphtheria, pertussis, and tetanus (DPT), and polio, as well as measles, beginning in 1968. Unlike the vaccinations given through the mass campaign system, these vaccinations were administered on a regular basis: mothers brought their children to the MCH centers on the days scheduled for vaccination sessions. Since the MCH centers existed only in the towns, this system benefited primarily the urban population. The only exception was the rural area of Khombole in the region of Thiès, which had the only rural MCH center in the country, built in 1958 by the Dakar Medical School. After 1978 In 1978, following the recommendations of the Alma-Ata Conference on primary health care, Senegal reformulated its health care policy with an emphasis on primary health care. Paralleling the effort towards decentralization of the major health facilities (hospitals and dispensaries), this policy led to the training of community health workers and the establishment of village pharmacies and maternity clinics. The EPI, a component of primary health care, was organized as a separate program. It was given special attention in Senegal, since this country was a testing ground for assessing the ability to achieve EPI objectives in an African country. Data gathered to assess the program's effectiveness are examined in detail in the following section. Expanded Programme on Immunization (EPI) Organization The EPI was initiated in Senegal in 1981. The program was designed to extend vaccination coverage to rural areas, which were at that time not well served, and to improve coverage in urban areas. Its objective was to protect children against seven diseases: tuberculosis, diphtheria, tetanus, pertussis, polio, measles, and yellow fever. It relied on fixed vaccination centers and mobile teams to achieve this objective. In urban areas, the MCH centers operated as fixed centers, as they had done previously. Their activities were supplemented in the rural areas by dispensaries, which began to administer vaccinations systematically at fixed sites. In addition, the rural dispensaries provided coverage for people within a 15-km radius by means of traveling vaccination teams. Mobile teams were established in rural areas to administer vaccinations beyond the 15-km radius served by the fixed dispensary centers. Vaccinations were administered

OCR for page 113
Population Dynamics of Senegal to young children and pregnant women, the latter being given tetanus vaccinations to protect their newborns against neonatal tetanus. Since its beginning, the EPI has undergone two major acceleration efforts, one in early 1987 and the other in early 1990. These initiatives led to training and mobilization of administrative and health personnel, media information campaigns (especially radio), and the outfitting of dispensaries with new equipment, especially in 1987. Changes in Vaccination Coverage No reliable measurements of vaccination coverage before 1984 are available for Senegal at the national level.14 In 1984, 1987, and 1990, however, three national vaccination coverage surveys [5] were conducted (Claquin et al., 1987; OCCGE-Muraz, 1990). These surveys, along with the 1986 DHS-I [4] and the 1992-1993 DHS-II [7], supply measurements of vaccination coverage (Ndiaye et al., 1988; Ndiaye et al., 1994). The vaccination coverage surveys show that the percentage of children aged 12-23 months who were completely vaccinated15 increased from 18 percent in July 1984 to 35 percent in July 1987, and to 55 percent in June 1990 (see Table 5-31). The 1987 and 1990 percentages probably represent maxima for the period 1987-1990. Indeed, in both years the surveys were carried out just after an acceleration phase, and the average for the period is likely somewhat lower. Detailed analysis of vaccination dates confirms that the increased coverage rate coincided with the two acceleration campaigns that took place in the first trimester of 1987 and the first trimester of 1990. Data from the DSH-I and DHS-II are not strictly comparable, either with each other or with the vaccination coverage surveys, because of their differing treatment of children who lacked health or immunization cards. However, both surveys confirm the substantial increase in vaccination coverage between 1986—just before the first EPI acceleration campaign—and 1992-1993. Table 5-31 shows that, in 1987 and 1990, the vast majority of children in Senegal were vaccinated against at least one disease; 92 percent had received the BCG vaccine. The much lower proportion of completely vaccinated children is attributable mainly to the fact that many of these children did not receive the required second or third dose of DPT and polio vaccines, and that some of them were not vaccinated against measles (or received it at an improper age). The same situation was noted in 1986. These discrepancies, of major importance in 1986 and 1987, were partly overcome in 1990, and this contributed to the appreciable rise in the proportion of completely vaccinated children—from 35 percent in 1987 to 55 percent in 1990. Coverage for each particular vaccine (with the exception of DPT3 and polio 3) increased less than the percentage receiving all vaccinations.

OCR for page 113
Population Dynamics of Senegal TABLE 5-31 Vaccination Coverage of Children Aged 12-23 Months (in percent) According to the Date and the Vaccinea   Demographic and Health Surveyb Vaccination Coverage Survey (standard World Health Organization method) b Vaccine May-July 1986 December 1992-March 1993c July 1984 July 1987 June 1990 BCG 27 84 — 92 94 DPT1 + Polio 1d 27 77 — 81 91 DPT2 + Polio 2d 18 69 — 69 83 DPT3 + Polio 3d 10 59 — 47 63 Measles 20 57 — 63 76 Yellow fever 19 55 — 72 75 Completely vaccinated childrene 7 49 18 35 55 a Measure based only on the information contained on health cards or vaccination cards. When these documents were lost, the child was not counted as being vaccinated. These estimations are, therefore, the minimum. The only exception is the 1992-1993 DHS-II, where when the child did not have the documentation, the mother's statement was taken into consideration. Also, for this survey, the percentages are calculated for all of the children, whether they had documentation or not. b Because of differences in survey methodologies, the absolute rates are not comparable; their relative variations from one vaccination to another are, however. c Among all children, whether they have a health card or not, the proportion of them who have received each vaccination according to the health card or mother's statement. d In 1986 and 1992-1993, only DPT. e In 1987 and 1990, children vaccinated against seven diseases; in 1986 and 1992-1993, children vaccinated against six diseases (excluding the yellow fever vaccination). The definition of completely vaccinated children is not known for 1984. SOURCES: May - July 1986: Ndiaye et al. (1988); December 1992-March 1993: Ndiaye et al. (1994); July 1984: Claquin et al. (1987); July 1987: Claquin et al. (1987); June 1990; OCCGE-Muraz (1990) The coverage rate for the BCG vaccine, already remarkably high, did not change much, rising from 92 to 94 percent. The coverage rate for measles vaccine increased only slightly—from 63 to 76 percent. Vaccination coverage appears to have declined between June 1990 and December 1992-March 1993. The actual decline may be greater than that indicated in Table 5-31. The 1990 measurement is a minimum estimate. In that year, a child who did not have a vaccination card was considered not to be vaccinated. On the contrary, in 1992-1993, if a child did not have a vaccination card, the statements of the mother concerning vaccination were

OCR for page 113
Population Dynamics of Senegal taken on faith. This and other methodological differences between DHS-I and DHS-II make it difficult to draw firm conclusions. By region in the order of children completely vaccinated in 1990, Table 5-32 gives the proportions of children aged 12-23 months who were vaccinated against measles in 1987 and 1990, along with the proportion of children completely vaccinated in the same years. The regions receiving the poorest overall coverage in 1990 were Tambacounda, which also had the lowest coverage in 1987, Saint-Louis, Diourbel, Kaolack, and Louga. Ziguinchor was the region with the highest level of vaccination coverage in 1990, ahead, notably, of the region of Dakar. The regional differentials in vaccination coverage were more pronounced in 1990 than in 1987. In 1990, they ranged from 26 percent completely vaccinated in the region of Tambacounda to 70 percent in Ziguinchor, and from 54 percent vaccinated against measles in Tambacounda to 87 percent in Kolda. In 1987, they ranged from 25 percent completely vaccinated in Tambacounda to 57 percent in Ziguinchor, and from 54 percent vaccinated against measles in Kolda to 74 percent in Ziguinchor. Table 5-33 traces the progress of vaccination coverage in urban and rural areas between July 1984 and July 1987. In the region of Dakar, vaccination coverage increased only from 32 to 39 percent. In other urban areas, it rose from 25 to 37 percent. The strong vaccination campaign that took place in early 1987 thus had only a moderate impact in urban areas. Conversely, in rural areas, where coverage was relatively low in 1984, the accelerated campaign had a very strong impact, increasing coverage almost threefold from 12 to 34 percent. Speeding up the program in 1987 ultimately had as a consequence, in addition to a doubling of overall vaccination coverage, virtual elimination of the inequalities between urban and rural areas. This was, in fact, one of the objectives of the accelerated campaign—to reach all of the children in Senegal. We do not have separate measurements for urban and rural areas for the 1990 survey, so we do not know whether the additional 50 percent increase in vaccination coverage at the national level between 1987 and 1990 occurred equally in urban and rural areas; however, some speculation can be offered. The acceleration campaign of 1990 was inspired primarily by the realization that the progress made in areas of high population density, in urban areas in general and in Dakar in particular, had been so weak. Consequently, the initiative centered on these areas. It is therefore a virtual certainty that the 50 percent increase at the national level seen in Table 5-32 resulted from the combination of a strong increase in urban areas and a lesser increase (or even decrease) in rural areas. Thus the gaps between the urban and rural areas that were virtually eliminated by the 1987 acceleration have probably reemerged. The vaccination coverage survey conducted in 1992 in the Bandafassi

OCR for page 113
Population Dynamics of Senegal TABLE 5-32 Variations in Vaccination Coverage of Children Aged 12-23 Months (in percent), by Region and Year, 1987 and 1990   Vaccinated Against Measles Completely Vaccinated Regiona July 1987 June 1990 July 1987 June 1990 Ziguinchor 74 83 57 70 Kolda 54 87 27 63 Dakar 58 73 39 62 Thiès 64 85 31 62 Fatick 66 76 35 61 Louga 59 76 31 57 Kaolack 67 74 29 56 Diourbel 65 74 34 47 Saint-Louis 68 74 39 35 Tambacounda 63 54 25 26 Senegal 63 76 35 55 NOTE: A child is considered to be completely vaccinated if he/she has received vaccinations for the seven following diseases: tuberculosis, diptheria, tetanus, pertussis, polio, measles, and yellow fever. a In decreasing order according to the proportion of children completely vaccinated in June 1990. SOURCES: July 1987: Claquin et al. (1987: adapted from Table 2, p. 24); June 1990: OCCGE-Muraz (1990) TABLE 5-33 Change in Vaccination Coverage of Children Aged 12-23 Months (in percent) Between 1984 and 1987, According to Residence   Percentage of Children Completely Vaccinated Region July 1984 July 1987 Region of Dakar 32 39 Urban (outside of Dakar) 25 37 Rural 12 34 Senegal 18 35 NOTE: The 1990 vaccination coverage survey report does not give the results by the characteristics used in this table. SOURCES: Claquin et al. (1987:6 and Table 2, p. 24).

OCR for page 113
Population Dynamics of Senegal study area within the region of Tambacounda [8] furnishes an example of progress in a rural district during this period (Desgrées du Loû and Pison, 1994). Complete vaccination coverage, which was close to zero before 1987, rose to 45 percent for children aged 12-23 months (and 85 percent for children receiving at least one vaccination) after the first acceleration campaign in early 1987. Coverage declined thereafter, falling to 22 percent of children completely vaccinated (and 44 percent of children receiving at least one vaccination) in 1992, though there was a temporary increase in 1990. In this study area, the main factor determining the variation in the probability that a given child would be vaccinated was the distance from a dispensary. The progress in vaccination coverage was, therefore, quite different from village to village, depending on the distance from a dispensary. Villages located less than 10 km from a dispensary saw improved vaccination coverage between 1987 and 1992. In villages located a moderate distance away, 10-15 km, coverage remained constant into 1991, then deteriorated in 1992. In villages more than 15 km from a dispensary, the coverage rate declined rapidly over the 5-year period, with virtually no children receiving all vaccinations in 1992. This deterioration resulted mainly from the fact that the mobile teams, who were theoretically responsible for vaccinations in the villages and whose rounds in 1987 allowed almost the entire population to be served, did not make new rounds after 1987. In Senegal, as in many countries, improvements due to the EPI thus depend on the strengthening of activities in rural areas and in areas with the most difficult access through revitalizing and increasing the effectiveness of mobile teams. This high-priority task has been neglected since 1987. As noted above, the EPI includes the vaccination of pregnant women to protect newborns against the risk of neonatal tetanus. This work was evaluated by the 1990 vaccination survey, as well as the DHS-I and DHS-II. The 1990 survey revealed that, in all of Senegal, 37 percent of women who delivered during the preceding year had received two injections of the antitetanus vaccine (OCCGE-Muraz, 1990). Since it is believed that two injections give 80 percent protection, and since some unvaccinated women had some protection as a result of vaccinations during previous pregnancies, at least 32 percent of newborns ultimately received protection. The differentials among regions were highly pronounced: only 11 percent of newborns in the regions of Tambacounda and Kaolack were protected, as compared with 62 percent of those in the region of Dakar. The data supplied by the DHS are not easily compared with the data from the vaccination-specific surveys. The DHS-I indicates, however, that among children born during the 5-year period preceding the survey, between 1981 and 1986, mothers received at least one tetanus shot in 31 percent of cases. The DHS-II reveals that, for births during the period

OCR for page 113
Population Dynamics of Senegal 1987-1992, the proportion was 71 percent, thus indicating a strong increase in tetanus vaccinations among pregnant women over a 7 year period. NOTES 1.   The term "grand region" is used throughout this report in reference to the four WFS and DHS regions, as opposed to the ten administrative regions. 2.   This atypical mortality pattern has been observed in other sub-Saharan African countries as well, but it was discovered first in the Niakhar/Ngayokhème region in Senegal in the 1960s by Cantrelle. 3.   Male/female mortality differentials vary by age. Analysis using traditional age groups indicate that during the neonatal period, males usually experience a higher mortality than females, but that this differential often is reversed later in childhood. The reader is referred to Baye (1994), Desgrées du Loû (forthcoming), and Desgrées du Loû et al. (forthcoming) for a discussion of mortality differentials using nonclassical age groups. 4.   Recall that when using the census data, we restrict the study of mortality to the 1-4 age group, excluding those less than age 1, for whom the mortality data gathered from the census are unreliable. 5.   The departments of Kébémer and Podor have surprisingly low urban mor­tality rates, for unknown reasons. They have been omitted from these calculations. 6.   See Chapter 2 for a detailed description of the socioeconomic variables discussed in this section. 7.   A correlation coefficient is a measure of the association between two non-random variables. 8.   The structural adjustment policies implemented by Senegal are discussed in Chapter 2. See Rouis (1994) for an excellent description of these policies. See Working Group on Demographic Effects of Economic and Social Reversals (1993) for a discussion of the demographic responses. 9.   The quality of the data on adult mortality is uneven and needs to be evalu­ated before conclusions can be drawn. Much of the evaluation was conducted in terms of the age pattern of adult mortality, since such age patterns are sensitive to typical data errors. The small-scale studies in Bandafassi, Mlomp, and Niakhar/ Ngayokhème, though not nationally representative, provide a reasonable basis for determining the true age pattern of adult mortality. These studies suggest that the age pattern of adult mortality in Senegal is well approximated by the Coale-Demeny "West" family. Available national data will thus be compared with the West family for checks of data quality. 10.   Deaths in age groups starting with zero, such as 50-54, are typically as numerous as those in the previous age group starting with 5, 45-49 in the example. 11.   The boundaries of the study area changed somewhat over the years. How­ever, during 1963-1981 data was collected consistently in Ngayokhème, a part of the total study area. After 1984, the Niakhar study area changed again, extending farther north, but not as far south, as the old region. We refer to the latter study area as Niakhar-II to emphasize the minor change in the baseline population. 12.   Service des Grandes Endèmies.

OCR for page 113
Population Dynamics of Senegal 13.   Programme de Protection Nutritionnelle et Sanitaire. 14.   Routine health service statistics are available, but they are not reliable and show a strong tendency to overestimate actual vaccination coverage (OCCGE-Muraz, 1990). 15.   A child was considered completely vaccinated if he/she had been vacci­nated against seven diseases: tuberculosis, diphtheria, tetanus, pertussis, polio, measles, and yellow fever.