National Academies Press: OpenBook

Population Dynamics of Senegal (1995)

Chapter: 5 Mortality

« Previous: 4 Fertility
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

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

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

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

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

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

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

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

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

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

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

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

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

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

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

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

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

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

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

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

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

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

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

FIGURE 5-4 Trends in child mortality (5q0 per 1,000), comparison between two large cities— Dakar-Pikine and Saint-Louis—and rural areas in Senegal. NOTES: DHS-I-rural: Direct estimates of the risk of death by period, based on data from the 1986 DHS-I [4] conducted solely among rural women residing in rural areas since childhood. DHS-I-Dakar: Direct estimates of the risk of death by period, based on data from the 1986 DHS-I conducted solely among women residing in the Dakar region at the time of the survey and those residing in urban areas since childhood. Pikine: 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 Pikine to women residing in Pikine at the time of the survey. Saint-Louis: Saint-Louis vital records. SOURCES: Antoine and Mbodji (1991); Diop (1990); Garenne et al. (1992:Table 5, p. 24).

dating the start of the child mortality decline for the entire Fissel-Thiénaba region covered by the Survey on Child Mortality in the Sahel of 1981-1984 [16]. However, this decline probably began still earlier than in Niakhar/Ngayokhème and apparently progressed very rapidly during the 1970s. During the period 1981-1984, child mortality in Thiénaba was 25 percent lower than that in Fissel (2q0 was 202 per 1,000 in Fissel, as compared with 149

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

FIGURE 5-5 Trends in child mortality (5q0 per 1,000), rural areas, Senegal. SOURCES: Antoine and Mbodji (1991); Diop (1990); Garenne et al. (1992:Table 5, p. 24).

per 1,000 in Thiénaba); this is attributable to an improved health infrastructure and more frequent use of health services in Thiénaba (Mbodji, 1988).

Mlomp presents a special case. This rural area, about 500 km from Dakar, is 50 km from Ziguinchor, the regional capital. In Mlomp, the population benefited from the establishment of a private dispensary and a maternity clinic in 1961, which soon after their establishment were providing high-quality health services to a large majority of the area's residents (Pison et al., 1993). An early and very rapid decline in child mortality occurred in this area in the mid-1960s; the risk of dying before age 5 was cut fourfold in 20 years.

With the exception of Mlomp, the onset of child mortality decline appears to be correlated with distance from Dakar, a factor that is, in turn, correlated with the availability of health services. The educational level of women and household income level were uniformly low in all of these areas during this time, so these factors cannot explain the observed differences in timing. The relatively rapid trend toward declining child mortality in rural

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

areas beginning in the late 1970s can, however, be linked in large part to infrastructure decentralization and to the new health policy implemented during that period (see the appendix to this chapter for a discussion of these issues).

Geographic Variations

Regional Differences

Table 5-4 shows that the western grand region, encompassing the Dakar and Thiès administrative regions, is clearly different from the other grand regions—the center (including the administrative regions of Louga, Diourbel, Kaolack, and Fatick), the northeast (Saint-Louis and Tambacounda), and the south (Ziguinchor and Kolda)—where child mortality is 1.5 times greater.1 Despite the overall decline in mortality, the ranking of these three grand regions remained approximately the same between 1968 and 1988—an appreciably lower mortality in the northeast and a higher mortality in the south, with the center remaining in the middle.

TABLE 5-4 Variations in Child Mortality (5q0 or 4q1) by Grand Regions, 1968-1992

 

1968-1972

1976-1985

1987-1988

1988-1992

Grand Region

5q0

Ratioa

5q0 or 4q1

Ratio

4q1

Ratio

5q0 or 4q1

Ratio

5q0

 

 

 

 

 

 

 

 

West

183

 

156

 

 

111

 

Northeast

253

1.4

200

1.3

 

183

1.6

Center

304

1.7

244

1.6

 

170

1.5

South

335

1.8

262

1.7

 

195

1.8

4q1

 

 

 

 

 

 

 

 

West

 

89

 

50

 

60

 

Northeast

 

109

1.2

83

1.7

84

1.4

Center

 

166

1.9

101

2.0

106

1.8

South

 

164

1.8

103

2.1

108

1.8

NOTE: Results are from univariate analysis of data from national surveys (1978 WFS, 1986 DHS-I, 1992-1993 DHS-II, and 1988 census).

a Ratio between regional 5q0 and 5q0 from the first category (west).

SOURCES: 1968-1972: Guèye and Sarr (1985); 1976-1985: Ndiaye et al. (1988); 1987-1988:1988 census (unpublished tables); 1988-1992; Ndiaye et al. (1994)

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Departmental Differences

Since the mortality data for children aged 1-4 from the 1988 census appear to be fairly reliable, we have used them to study variations in child mortality on a more discriminating scale—departmental.

Table 5-5 shows mortality rates for children aged 1-4 in 1987-1988 by department and rural or urban area, based on the 1988 census. Excluding the region of Dakar, mortality in this age group varies by a factor of two. At one extreme are the departments experiencing low rural mortality—Kébémer, Louga, and Tivaouane—in which the annual mortality rate for this age group is 19 per 1,000; at the other extreme are departments having high rural mortality—Fatick (37), Bambey (38), Mbacké (33), and Sédhiou (33). On average, rural mortality is lower in the north and higher in the south (see Figure 5-6). This distinction is particularly clear in the western half of Senegal, where the vast majority of the population resides. If, leaving Saint-Louis in the north, one travels southward keeping to the city of Saint-Louis' longitude, one passes in turn through the departments of Dagana (annual mortality rate for children aged 1-4 of 20 per 1,000), Louga (19), Kébémer (19), Tivaouane (19), Bambey (38), Fatick (37), and Foundiougne (29). Mortality abruptly doubles when one goes from Tivaouane to Bambey. The gradient still exists on a north-south line closer to Dakar, such as from Tivaouane through Thiès to Mbour, but the pattern is complicated by an additional low-mortality effect associated with proximity to Dakar.

Several departments constitute exceptions to this north-south division. Bakel has a mortality rate lower than that of Matam, the neighboring department farther to the north, and the departments of Kaolack and Oussouye, where mortality is lower than the average for the south.

This north-south contrast in rural mortality among children aged 1-4 in 1987-1988 could be explained by a more marked underreporting of deaths in the northern departments during the census. The lower mortality rate in the north is, however, confirmed by earlier surveys. We found this especially in the DHS-I survey, in which the northeastern grand region has a mortality rate one-third lower than that of the central and southern grand regions (see Table 5-4). The northeastern grand region in the WFS and the two DHS surveys corresponds approximately to the northern portion of the country we are describing here.

The line separating the north (lower mortality) and the south (higher mortality) lies parallel to the isohyet (equal average annual rainfall) lines. The lower rainfall and shorter rainy season in the north may explain the contrast: they may be the underlying factors associated with diarrhea—more frequent during the rainy season—and malaria. Malaria is a major cause of death among children aged 1-4, as discussed below. The role of this disease in child mortality differentials may have been particularly high

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

TABLE 5-5 Annual Mortality Rates for Children Aged 1-4 (per 1,000) by Department and Rural/Urban Residence: Deaths in Last 12 Months According to 1988 Census

Region and Department

Rural

Urban

Dakar

 

 

Dakar

7.6

Pikine

9.2

Rufisque

14

10

Diourbel

 

 

Bambey

38

22

Diourbel

30

19

Mbacké

33

17

Fatick

 

 

Fatick

37

20

Foundiougne

29

20

Gossas

27

20

Kaolack

 

 

Kaffrine

29

12

Kaolack

23

11

Nioro du Rip

30

28

Kolda

 

 

Kolda

31

14

Sédhiou

33

24

Vélingara

24

12

Louga

 

 

Kébémer

19

1.4

Linguère

20

18

Louga

19

7.3

Saint-Louis

 

 

Dagana

20

10

Matam

25

30

Podor

26

2.9

Tambacounda

 

 

Bakel

20

19

Kédougou

31

23

Tambacounda

27

8.7

Thiès

 

 

Mbour

24

13

Thiès

23

9.2

Tivaouane

19

7.5

Ziguinchor

 

 

Bignona

27

15

Oussouye

23

24

Ziguinchor

28

17

 

SOURCE: 1988 census (unpublished tables)

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

FIGURE 5-6 Departmental variations in annual mortality rate (per thousand) of children aged 1-4 years in rural and urban areas, 1987-1988. SOURCE: Unpublished tables, 1988 census.

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

in 1987 and 1988 because of the probable decrease in mortality from a number of infectious diseases due to increases in the proportion of children vaccinated. For example, measles must have diminished throughout Senegal during these years, following the accelerated EPI in 1986-1987.

The relatively low rural mortality rate in the department of Bakel, which stands out as an exception in the south, may be the result of the relative wealth of Soninke villages, which make up a major part of the department's population and which receive earnings from migrant workers abroad. A portion of such remittances is invested in dispensaries and medicines (Condé and Diagne, 1986).

The low rural mortality in the department of Oussouye, again an exception in the south, is corroborated by the Mlomp study [10], which showed in particular a very low level of malaria-caused mortality linked to the antimalaria program implemented by the private dispensary in this rural area (Pison et al., 1993). The abundance of dispensaries in this department probably contributes to its low child mortality as well. The department of Kaolack, although located in the south, also enjoys low rural mortality, which is associated with its proximity to a major town, Kaolack, and thus with better-than-average access to health care, despite its high incidence of mosquitos and its relatively poor sanitation.

Age and Sex Patterns in Child Mortality

Age Patterns

Senegal has been noted for having an unusual age pattern of child mortality, with very high mortality for the ages 1-5 and comparatively low mortality before age 1. This pattern was noted by Cantrelle (1969) on the basis of data from the Niakhar/Ngayokhème population study [11], and has been confirmed by national-level surveys since.2 In the 1960s and 1970s, the probability of dying between the ages of 1 and 5 (4q1) was 50 percent higher than the probability of dying in the first year of life, though in the 1980s, 4q1 declined more rapidly than 1q0 and 4q1 for Senegal as a whole and from a variety of subnational studies. Figure 5-7 compares the infant and child mortality patterns with the historical patterns embodied in the Coale-Demeny (1983) model life tables, plotting 4q1 against 1q0. The unusual nature of the Senegalese age pattern of child mortality is immediately evident, as is the recent convergence toward the historical norms.

The tendency of 4q1 to fall more rapidly than 1q0 is encountered everywhere. For example, in Dakar, which had experienced a pronounced decrease in mortality since the beginning of the century, 1q0 and 4q1 were already low during the 1960s: 1q0 was 58 per 1,000 in 1973, while 4q1 was

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

75 per 1,000. Thereafter, 1q0 stagnated, at least until 1980, while 4q1 continued its decline by falling almost 50 percent between 1973 and 1988 (see Table 5-6). The same type of change occurred in Saint-Louis (Table 5-6 and Figure 5-7). The period 1955-1988 saw a steady decline in both 1q0 and 4q1, but the pace of the decline was more rapid for 4q1 than for 1q0. As a result, from being higher than 1q0, 4q1 became lower. The decline in 4q1 accelerated in the middle of the 1980s.

Niakhar/Ngayokhème and Bandafassi (Table 5-6 and Figure 5-7) show the same type of change over time in rural areas. The decline of 4q1 in the mid-1980s was particularly rapid, especially in Bandafassi. The Mlomp rural area is an exception to the general pattern (see Table 5-6 and Figure 5-7). There, the reduction of 1q0 preceded and occurred more rapidly than that of 4q1.

Table 5-7 shows the changes in age-specific mortality rates over time for several rural areas with greater age detail. Until the mid-1980s, mortality rates were higher between ages 6 months and 1 year than between ages 1 and 5 months and remained high until age 24 months. This increase in risk, especially well-defined in Niakhar/Ngayokhème, but present virtually everywhere else, decreased or disappeared altogether beginning in the mid-1980s.

High mortality for ages 1-4 has been attributed to a combination of three factors: a relatively high incidence of infectious diseases, poor climate, and widespread malnutrition. Because of their seasonal or epidemic character in this country, malaria and measles are particularly problematic.

The epidemiological patterns of measles are unique, especially in rural areas. First, the transmission of measles is seasonal; transmission is highest at the end of the dry season, when most of the migrations occur and when the climate favors transmission. Second, because the density of the population is low in rural areas, the population is grouped in villages distant from one another, so the disease is epidemic. For several years, there may be no measles cases, and then an epidemic occurs and affects many children, nearly all those born since the last epidemic. As a result, children tend to get measles when they are rather old. Third, the average family size is large, which favors secondary transmission of the disease (between siblings or children living in the same family). Such transmissions are associated with increased severity and higher case fatality ratios.

Child mortality (for ages 1-4) is also high due to malaria. Due to the climate, its transmission is also seasonal. A short rainy season, during which malaria is transmitted, alternates with a long dry season, during which the transmission is interrupted. Children tend to be infected for the first time at a higher age than if the rate of transmission was the same throughout the year.

The disappearance of this pattern of relatively high mortality for ages

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

1-4 in the late 1980s is related to health programs that focused on these diseases, including, in particular, measles immunization and antimalaria chemoprophylaxis. These programs were instituted earlier and were more effective than programs focused on improved pregnancy follow-up and delivery conditions, which affect primarily early mortality before 6 months of age. This fact explains the more rapid decline in 4q1 than in 1q0. The particular situation of Mlomp, where pregnancy monitoring and delivery in

TABLE 5-6 Age Pattern of child Mortality: National- and Regional-Level Variation and Change, 1955-1992

Level and Year or Period

1q0 per 1,000

4q1 per 1,000

5q0 per 1,000

National Level

 

 

 

1963-1967a

119

198

293

1968-1972a

123

183

283

1970b

102

199

281

1971-1975c

120

189

287

1973-1977a

112

170

262

1976-1980c

96

155

236

1978d

94

149

229

1978-1982e

90

119

199

1981-1985c

86

114

191

1983-1987e

84

109

185

1987-1988f

83

1988-1992e

68

68

131

Dakar

 

 

 

1973

58

75

128

1980

60

43

101

1987-1988

36

Saint-Louis

 

 

 

1955-1959

124

142

249

1960-1964

117

136

237

1965-1969

109

112

209

1970-1974

101

112

202

1975-1979

82

85

160

1980-1985

73

65

133

1986-1988

42

24

66

Ngayokhème

 

 

 

1963-1967

223

340

485

1968-1972

214

342

480

1973-1977

182

291

421

1978-1980

167

242

369

1984-1986

101

202

282

1987-1991

104

142

230

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

Level and Year or Period

1q0 per 1,000

4q1 per 1,000

5q0 per 1,000

Bandafassi

 

 

 

1970-1985

197

202

360

1986-1992

140

131

253

Mlomp

 

 

 

1930-1964

168

225

355

1965-1974

86

141

215

1975-1984

56

97

148

1985-1991

50

71

112

NOTE: See Appendix A for description of surveys.

a WFS

b 1970-1971 NDS

c DHS-I

d 1978-1979 survey

e DHS-II

f 1988 census

SOURCES: National level: 1970-1971 NDS (Cantrelle et al., 1986); 1978-1979 survey (Cantrelle et al., 1986); 1988 census (unpublished tables); WFS (Rutstein, 1983); DHS-I (Ndiaye et al., 1988); DHS-II (Ndiaye et al., 1994). Sub-National level: Dakar: Cantrelle et al. (1986); Saint-Louis: Diop (1990); Nagayokhème: 1963-1980: Cantrelle et al. (1986); 1984-1991: Project Niakhar (1992); Bandafassi: Pison and Desgrées du Loû (1993); Mlomp: Pison et al. (1993)

maternity units was expanded relatively early, before the other health programs were expanded, shows the reverse effect—a reduction in 1q0 that occurred earlier and more rapidly than that in 4q1. A portion of early deaths was thus eliminated well before the other health programs had the time to produce an effect on later deaths.

Sex-Related Differences in Mortality

The national surveys, the 1978 WFS, 1986 DHS-I, and 1992-1993 DHS-II, and the 1988 census reveal an excess mortality among boys as compared with girls, which appears to increase as mortality declines (see Table 5-8).3

This excess mortality of boys could be tied to an underreporting of the mortality of girls in the surveys. However, with the exception of the 1989-1991 population in Niakhar/Ngayokhème, excess male mortality is found in all three of the population studies where we do not expect such underreporting

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

FIGURE 5-7 Comparison of the pattern of infant and child mortality versus historical patterns embodied in the Coale-Demeny model life tables.

to be a problem (Niakhar/Ngayokhème [11], Bandafassi [8], and Mlomp [10]). Moreover, again with the exception of the 1989-1991 population in Niakhar/Ngayokhème, data from Niakhar/Ngayokhème and Bandafassi show a trend toward a widening of the gap between the sexes over time (see Table 5-8).

Several mechanisms probably play a role in the widening of the boy-girl gap, one of which is linked to the widespread practice of vaccination. When vaccination levels were low (before the EPI), girls probably died more frequently from measles (and perhaps pertussis) than did boys. The elimination or reduction of the incidence of these two diseases thus proved more advantageous to girls than to boys. It appears that, in addition to antimeasles protection, the measles vaccine used produces a beneficial effect on child survival independent of the disease itself, perhaps through a nonspecific stimulation of the immune system, and that this effect was more pronounced in girls than in boys (Aaby et al., 1993).

Nutritional Status and Childhood Growth

The DHS-I and DHS-II took measurements of the weight and height of a sample of children aged 6-36 months. These data make it possible to

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

TABLE 5-7 Change in Age Pattern of Child Mortality in Selected Rural Areas

 

Bandafassi

Niakhar/Ngayokhème

Mlomp

Age (in months)

Cohort 1970-1985

Cohort 1986-1992

Cohort 1963-1967

Period 1984-1990

Cohort 1930-1964

Cohort 1965-1974

Cohort 1975-1984

Period 1985-1991

Probability of dying in 1 month period, by age group

0

92

84

49

54

50

42

32

31

1-5

10.6

6.5

11.8

5.6

7.8

4.1

1.4

2.9

6-11

11.4

5.0

19.0

6.0

15.3

4.4

3.1

1.4

12-23

8.2

4.5

17.8

5.9

6.9

4.5

1.9

2.2

24-35

6.2

3.7

12.4

 

5.7

3.3

2.3

1.2

36-47

2.9

2.3

4.5

 

5.5

3.5

2.5

0.7

48-59

1.5

1.3

1.8

 

3.0

1.5

2.0

0.3

Probability of dying (1q0, 4q1, and 5q0)

0-11 (1q0)

197

140

210

112

168

86

56

54

12-59 (4q1)

202

131

354

159

225

141

97

51

0-59 (5q0)

360

253

490

253

355

215

148

102

NOTE: See Appendix A for description of surveys.

SOURCES: Bandafassi: Lagarde et al. (forthcoming); Niakhar/Ngayokhème: Cantrelle et al. (1986) and Project Niakar (1992); Mlomp: Lagarde et al. (forthcoming)

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

TABLE 5-8 Mortality Differences Between Male and Female Children, National Level and Selected Rural Areas

 

5q0 (per 1,000)

4q1 (per 1,000)

Population Covered and Period

Male

Female

Ratio m/f

Male

Female

Ratio m/f

National Level

 

 

 

 

 

 

1970-1978

278

265

1.05

175

176

0.99

1981-1985

198

183

1.08

117

111

1.05

1988-1992

95

79

1.20

1987-1988 (urban)

 

47

40

1.17

1987-1988 (rural)

 

122

106

1.15

1987-1988 (national)

 

94

82

1.15

Niakhar/Ngayokhème

 

 

 

 

 

 

1963-1981

431

421

1.02

305

291

1.05

1984-1988

298

263

1.13

192

171

1.12

1989-1991

186

184

1.01

108

111

0.97

Bandafassi

 

 

 

 

 

 

1971-1985

415

412

1.01

248

255

0.97

1986-1992

267

228

1.17

139

122

1.14

Mlomp

 

 

 

 

 

 

1985-1992

109

71

1.54

53

39

1.36

 

SOURCES: 1970-1978:1978 WFS (Rutstein, 1983); 1981-1985:1986 DHS-I (Ndiaye et al., 1988); 1987-1988:1988 census (unpublished tables); 1988-1992:1992-1993 DHS-II (Ndiaye et al., 1994); Niakhar/Ngayokhème, 1963-1981: Cantrelle et al. (1986); Niakhar/Ngayokhème, 1984-1991: Project Niakhar (1992); Bandafassi: Pison et al. (1993); Mlomp: Lagarde et al. (forthcoming)

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

assess the magnitude of malnutrition, and especially the frequency of wasting and stunting, in this age group. Wasting signifies low weight for height in relation to an international reference standard and may arise from recent or ongoing diseases, especially diarrhea and dehydration, as well as from acute food shortages. Low height for age, or stunting, indicates delayed growth, and may have multiple causes, in particular, a dietary deficiency over a long period, but also episodes of recurrent diseases. The generally adopted definitions of wasted and stunted are being more than two standard deviations below the median of a reference population (Waterlow et al., 1977), so by this definition, even within a well-nourished population some children may appear wasted or stunted.

The extent of wasting in the Senegalese population varies by season. Neither the DHS-I nor the DHS-II survey found particularly high levels of wasting among Senegalese children. The proportion of children aged 6-36 months in whom the weight/height ratio was more than two standard deviations below the reference median was only 6 percent in 1986 and 12 percent in 1992-1993 (Table 5-9). The prevalence of stunting was more frequent: nearly one in four of DHS-I and DHS-II children aged 6-36 months had a height for age more than two standard deviations below the reference median (Table 5-10).

The differences in the percentages of wasting among children aged 6-36 months in DHS-I and DHS-II may have several origins, and it is difficult to know the relative importance of any particular factor. Because the proportion of children who were stunted did not change appreciably, the increase in wasting is most likely due to short-term factors. Unfortunately, since both DHS-I and DHS-II were taken at different times of the year (DHS-I was in the field between April and July, 1986; DHS-II was in the field between November 1992 and March 1993), the short-term factors causing the differences in wasting are likely due to seasonal variation. Other possible causes for the differences may arise from (1) a difference in the measurement methods (of height or weight) and in the errors or biases between surveys; (2) annual variations in wastage due to epidemics (1992 was a measles epidemic year); or (3) the elimination of a supplementary feeding component to the national growth surveillance program in 1989. In the absence of more detailed information, it is not prudent to attribute the more frequent wasting of children in DHS-II to any single factor.

Stunting especially afflicts children aged 1 year or older. Before 1 year of age, children tend to rely on breastmilk and are infrequently affected. After age 1, when the child has been weaned, access to other sources of nutrition becomes important, and the incidence of stunting tends to be correlated with the household's socioeconomic status. Children in urban areas have a somewhat better dietary status than rural children, but the differences are not significant. More important differences relate to the literacy

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

of the mother. Among children whose mothers know how to read, 1 percent were wasted and 14 percent stunted in 1986, as compared with 7 percent and 24 percent, respectively, of children of illiterate mothers.

Causes of Death

Reliable statistics for causes of death among children are unavailable at the national or regional level. DHS-I asked questions to determine cause of death for children; however, a detailed assessment of the quality of these data by DHS staff concluded that they should be used with caution (Boerma et al., 1994). Indeed, in 66 percent of cases of neonatal death and 42 percent of the cases of death under 5 years of age, the mother was unable to give a cause of death. Despite the problems in the data, they show that neonatal tetanus, which was attributed to 14 percent of all deaths, was the primary cause of death among neonatals. Diarrhea, malaria, acute respiratory infections, and measles were the leading causes of death for children between 1 and 4, accounting for 73 percent of childhood deaths (Boerma et al., 1994).

Data on causes of child mortality, based largely on verbal autopsy techniques, are also available from rural population studies such as those at Mlomp [10], Niakhar/Ngayokhème [11], and Bandafassi [8]. Table 5-11 (neonatal) and Table 5-12 (1 month to 5 years) give the breakdown of causes of death in these areas in the second half of the 1980s.

Neonatal tetanus and prematurity are the two principal causes of neonatal mortality in Bandafassi and Niakhar/Ngayokhème. Neonatal tetanus accounts for 19 and 37 percent, respectively, and prematurity accounts for 10 and 26 percent, respectively, of all neonatal mortality in these two study areas (see Table 5-11). These two causes represent nearly one-half of neonatal mortality, which is 51 and 78 per 1,000, for Niakhar/Ngayokhème and Bandafassi, respectively.

The situation in Mlomp is quite different although our knowledge of cause of death is based on far fewer cases than the two other areas. Deaths resulting from neonatal tetanus and prematurity are almost 10 times less frequent than in Niakhar/Ngayokhème or Bandafassi, an indication of the success of the pregnancy monitoring/maternity unit delivery policy in practice in the Mlomp area for several decades. Thus, Mlomp provides an indication of the gains to be expected from improved pregnancy and delivery conditions throughout Senegal. Today, however, the average situation in the rural areas is closer to that in Niakhar/Ngayokhème and Bandafassi than to that in Mlomp.

Three causes of death predominate for postneonatal mortality among children aged 1 month to 5 years (Table 5-12). The most important of these is the diarrhea-malnutrition complex, which causes approximately one out

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

of three deaths in Mlomp and Niakhar/Ngayokhème, and almost one death out of five in Bandafassi. The two other principal causes of death are malaria and respiratory infections, each of which accounts for 6-20 percent of deaths in Niakhar/Ngayokhème and Bandafassi.

The insignificant role of measles and pertussis during the periods under consideration should be noted. These diseases were major causes of mortality in such rural areas in the past. Measles caused 28 percent of mortality among children aged 1 month to 5 years in Bandafassi during the period 1970-1985 (Pison and Desgrées du Loû, 1993). The pronounced decline of measles and pertussis as causes of death resulted from the EPI, particularly its acceleration campaign in 1987.

As in the case of neonatal mortality, postneonatal mortality in Mlomp is very different from that in the two other study areas. The mortality level resulting from all causes combined is four times lower in Mlomp, with lower risk of death from each of the causes except accident-related deaths, which are no less frequent in Mlomp than in the other two areas. The decline of malaria, malnutrition, and vaccine-preventable diseases (measles, pertussis) as causes of mortality is spectacular in Mlomp, and it again reflects the effectiveness of health programs, especially the antimalaria campaign and vaccinations.

Associations Between Socioeconomic Factors and Child Mortality

As with the fertility data reported in Chapter 4, the 1978 WFS, 1986 DHS-I, and 1992-1993 DHS-II allow the study of child mortality differentials by some important socioeconomic characteristics, such as place of residence, ethnicity, and level of instruction. The existence of the three comparable surveys allows us to analyze differences over time. In addition, the 1988 census allows us to analyze socioeconomic differences among departments of the country.

Tables 5-13 and 5-14 show variations in child mortality depending on the literacy of the mother and her ethnic group (Table 5-13), and according to urban or rural residence (Table 5-14).

Differences Resulting from Educational Level and Urban or Rural Setting

In the 1970s and early 1980s, children of illiterate mothers were between 2 and 2.5 times more likely to die before age 5 than those of literate mothers (Table 5-13). Children in rural areas died twice as often as those in urban areas (Table 5-14).

These differences in the probability of dying before age 5 are linked because the highest percentages of educated women live in urban areas.

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

TABLE 5-9 Nutritional Status of Children Aged 6-36 Months: Percentage Distribution in Percent of Weight for Height in Comparison with the International Distribution of Reference

 

Standard Deviation from the Reference Median

Sociodemographic Characteristic

At Least 2 Standard Deviations Below the Norm

Between 1 and 1.99 Standard Deviations Below the Mean

Between -0.99 and +0.99 Standard Deviations from the Mean

Between 1 and 1.99 Standard Deviations Above the Mean

More than 2 Standard Deviations Above the Mean

Sample Size

1986 DHS-I

 

 

 

 

 

 

Age (in months) —

 

 

 

 

 

 

6-11

2.0

15.0

65.4

13.1

4.6

153

12-23

8.1

29.5

55.0

6.3

1.1

271

24-36

5.7

25.6

64.9

2.4

1.4

211

Residence

 

 

 

 

 

 

Urban

3.5

21.2

65.8

6.5

3.0

231

Rural

7.2

26.7

57.9

6.7

1.5

404

Literacy of mother

 

 

 

 

 

 

Does not know how to read

6.7

26.7

58.7

6.0

1.9

535

Knows how to read

1.0

14.0

72.0

10.0

3.0

100

Total

5.8

24.7

60.8

6.6

2.0

635

1992-1993 DHS-II

 

 

 

 

 

 

Age (in months)

 

 

 

 

 

 

6-11

12.3

470

12-23

15.8

816

24-36

7.5

743

Total

11.9

 

 

 

 

2029

 

 

 

Normal Distributiona

 

 

 

 

2.3

13.6

68.0

13.6

2.3

 

NOTE: See Appendix A for description of surveys.

a Reference distribution: National Center for Health Statistics/Centers for Disease Control/World Health Organization.

SOURCES: Ndiaye et al. (1988, 1994)

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
This page in the original is blank.
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

TABLE 5-10 Nutritional Status of Children Aged 6-36 Months: Percentage Distribution in Percent of Height for Age in Comparison with the International Distribution of Reference

 

Standard Deviation from the Reference Median

Sociodemographic Characteristic

At Least 2 Standard Deviations Below the Norm

Between -1 and -1.99 Standard Deviations Below the Mean

Between -0.99 and +0.99 Standard Deviations from the Mean

Between 1 and 1.99 Standard Deviations Above the Mean

More than 2 Standard Deviations Above the Mean

Sample Size

1986 DHS-I

 

 

 

 

 

 

Age (in months)

 

 

 

 

 

 

6-11

8.5

29.4

55.6

2.6

3.9

153

12-23

26.9

34.7

36.2

2.2

0.0

271

24-36

27.5

31.3

37.9

2.8

0.5

211

Residence

 

 

 

 

 

 

Urban

17.7

26.4

52.8

2.6

0.4

231

Rural

25.5

35.6

34.9

2.5

1.5

404

Literacy of mother

 

 

 

 

 

 

Does not know how to read

24.3

34.6

37.6

2.2

1.3

535

Knows how to read

14.0

20.0

62.0

4.0

0.0

100

Total

22.7

32.3

60.8

6.6

2.0

635

1992-1993 DHS-II

 

 

 

 

 

 

Age (in months)

 

 

 

 

 

 

6-11

10.9

470

12-23

23.4

816

24-36

28.8

743

Total

22.5

 

 

 

 

2029

 

 

Normal Distributiona

 

 

 

 

 

2.3

13.6

68.0

13.6

2.3

 

NOTE: See Appendix A for description of surveys.

a Reference distribution: National Center for Health Statistics/Centers for Disease Control/World Health Organization.

SOURCES: Ndiaye et al. (1988, 1994)

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
This page in the original is blank.
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

TABLE 5-11 Neonatal Mortality (probability of a child born alive dying before age 28 days), by Cause of Death: Comparison Among Three Rural Areas: Mlomp, Niakhar/Ngayokhème, and Bandafassi

 

Mlomp 1985-1989

Niakhar/Ngayokhème 1984-1991

Bandafassi 1986-1990

Cause of Death

Number

of Deaths

Probability

of Dying (per 1,000)

Number

of Deaths

Probability

of Dying

(per 1,000)

Number

of Deaths

Probability

of Dying

(per 1,000)

Congenital malformations

9

12.0

11

1.2

1

0.7

Complications of labor

3

4.2

3

0.3

5

3.6

Prematurity

2

2.8

128

13.8

14

10.1

Tetanus

1

1.4

180

19.4

26

18.8

Other cause

11

15.2

55

5.9

24

17.3

Undetermined

1

1.3

95

10.3

39

28.2

No information

6

--

20

--

30

--

Total

33

36

492

51

139

78

 

SOURCES: Mlomp: Pison et al. (1993); Niakhar/Ngayokhème: Project Niakhar (1992); Bandafassi: Pison and Desgrées du Loû (1993)

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

TABLE 5-12 Probability of Dying Before Age 5 for a Child Alive at Age 1 Month, by Cause of Death: Comparison Among Three Rural Areas: Mlomp, Niakhar/Ngayokhème, and Bandafassi

 

Mlomp 1985-1989

Niakhar/Ngayokhème 1984-1991

Bandafassi 1986-1990

Cause of Death

Number of Deaths

Probability of Dying (per 1,000)

Number of Deaths

Probability of Dying (per 1,000)

Number of Deaths

Probability of Dying (per 1,000)

Gastroenteritis

14

15.7

470

56.5

24

19.8

Malaria, fever

2

2.2

199

23.9

37

30.4

Pneumonia

8

9.0

190

22.8

14

11.5

Malnutrition

0

0.0

176

21.2

20

16.5

Measles

0

0.0

74

8.9

1

0.8

Pertussis

1

1.1

74

8.9

2

1.6

Meningitis

1

1.1

40

4.8

13

10.7

Cholera

0

0.0

39

4.7

0

0.0

Accident

2

2.2

8

1.0

1

0.8

Other

7

7.9

59

7.1

11

9.1

Undetermined

6

6.7

380

45.7

95

78.2

No information

1

121

43

Total

42

46

1,830

205

261

179

 

SOURCES: Mlomp: Pison et al. (1993); Niakhar/Ngayokhème: Project Niakhar (1992); Bandafassi: Pison and Desgrées du Loû (1993)

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

For example, in 1986, among women aged 15-49, 54 percent of urban dwellers had attended school for at least 1 year, compared with 6 percent of those living in rural areas (Ndiaye et al., 1988). In addition, the Dakar urban area probably contributes significantly to heightening the contrast between urban and rural areas because Dakar represents a large proportion of Senegal's total urban population, and mortality in Dakar is lower than anywhere else in Senegal.

Multivariate analysis allows the researcher to assess the independent effect of a set of explanatory variables on a dependent variable, which in this case is the probability of dying before age 5. A discrete-time hazard model has been applied to the WFS and DHS-I data because it permits us to model accurately both deaths and exposure to the risk of dying (see Tables 5-15 through 5-17. (Further detail on the methodology is provided in Appendix C). For example, area of residence and mother's educational level produce separate independent effects. Having a mother who attended school reduces by nearly one-third the probability of dying before age 5, independently of urban or rural residence. Similarly, the probability of dying before age 5 is from one-third to one-half lower in urban than in rural areas, controlling for level of education.

TABLE 5-13 Change in child Mortality (5q0) According to Mother's Literacy and Ethnicity

 

1968-1972

1976-1985

Characteristic

5q0

Ratioa

5q0

Ratioa

Literacy

 

 

 

 

Mother knows how to read

112

 

106

 

Mother does not know how to read

277

2.5

225

2.1

Ethnicity

 

 

 

 

Wolof

254

 

197

 

Poular

231

0.9

212

1.1

Serer

289

1.1

226

1.1

Mandingo

392

1.5

261

1.3

Total

264

 

209

 

NOTE: Results are from univariate analysis of data from national surveys 1978 WFS, 1986 DHS-I. See Appendix A for description of surveys.

a Ratio between 5q0 of the category and 5q0 of the first category.

SOURCES: 1968-1972: data from Guèye and Sarr (1985); 1976-1985: Ndiaye et al. (1988)

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

TABLE 5-14 Variations in Child Mortality (5q0 or 4q1) by Urban/Rural Residence

 

1968-1972

1976-1985

1987-19881

1988-1992

Residence

5q0

Ratioa

5q0 or 4q1

Ratioa

4q1

Ratioa

5q0 or 4q1

Ratioa

5q0

 

 

 

 

 

 

 

 

Urban

153

 

135

 

 

102

 

Rural

314

2.1

250

1.9

 

184

1.8

4q1

 

 

 

 

 

 

 

 

Urban

 

71

 

41

 

50

 

Rural

 

164

2.3

104

2.5

107

2.1

NOTE: Results are from univariate analysis of data from national surveys 1978 WFS, 1986 DHS-I, 1992-1993 DHS-II, and the 1988 census. See Appendix A for description of surveys.

a Ratio between 5q0 of the category and 5q0 of the first category.

SOURCES: 1968-1972: data from Guèye and Sarr (1985); 1976-1985: Ndiaye et al. (1988); 1987-1988:1988 census (unpublished tables) 1988-1992: Ndiaye et al. (1994)

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

These findings differ somewhat from those of Cantrelle et al. (1986), who concluded, based on an analysis of data from the WFS, that the most important factor determining child mortality was urban or rural residence, and that the educational level of the parents, in particular of the mother, exerted only a modest influence.

The effects of mother's educational level became slightly more pronounced between the WFS (1973-1977) and the DHS-I (1981-1985), while the urban-rural differences declined. However, the disparities among regions persisted during the same period.

This finding appears to confirm in part the analysis of the 1986 DHS-I survey performed by Barbieri (1989), who concluded that educational level of the mother was the main factor explaining the variations in child mortality. For example, for mothers living in the same setting, urban or rural, attendance through the end of primary school halved the risk of mortality of their children before the age of 5. However, Barbieri found that, surprisingly, given the same educational level of the mother, children living in urban rather than rural areas faced a higher risk of death under age 5. No similar conclusion was drawn in our analysis. In both the WFS and DHS-I, the probability of dying by age 5 in urban areas is significantly lower than in rural areas after allowing for mother's education.

The survey conducted in 1986 in Pikine [15], a sprawling expansion of Dakar, further clarifies the respective influences of residence and mother's educational level on child mortality in the urban context (Antoine and Diouf, 1992). This survey gathered women's life histories while recording the changes in their socioeconomic status over their entire life span so as to learn of their situation at the time of each event, i.e., a birth or death of a child. This innovation allows the researcher to analyze the effects of different migration behaviors and changes in socioeconomic circumstances on mortality.

The Pikine population includes a major percentage of migrants, thereby making it possible to compare the migrant children's mortality with that of children born in Pikine. The rate for children born in rural areas was 2.5 times higher than the rate for children born in Pikine. However, Antoine and Diouf (1992) found that immigrant mothers rapidly adapt to urban living conditions. Their analysis suggests that it is not the place of origin (urban or rural) of the mothers that ultimately determines child mortality differences, but other factors, such as educational level.

Pikine provides further support for the effect previously estimated for the entire country of a one-third reduction in mortality when the mother has attended school. However, the education factor interacts with access to clean water. Children born in households that obtain their water from a standpipe have a 1.3 times higher probability of dying under age 5 than children born in households having an inside faucet. When the parents are

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

illiterate, access to clean water does not affect the probability of child survival. Similarly, having educated parents does not lower mortality below that of the children of illiterate parents when water is obtained from a standpipe. An inside faucet produces lower child mortality only if at least one of the parents is literate, and the effect then becomes pronounced—mortality is halved.

The detailed multivariate analyses (Table 5-15 for the period 1973-1977 and Table 5-16 for the period 1981-1985) reveal that regional differences, in particular the lower mortality in the western grand region in relation to the three other grand regions, stem from the larger urban population and higher level of maternal education in the former. However, urbanization and maternal education do not fully explain the child mortality differences. Even taking these two factors into account, gaps among the grand regions remain. Table 5-15 thus reveals, for 1973-1977, a north-south gradient, according to which mortality is lowest in the northeast. In the center, mortality is average and at the same level as in the west. In the south, it is the highest. This north-south gradient may be related to the fact that the northern regions are better served with water and electricity than those in the south (see the section Economic Background: Regional Disparities in Socioeconomic Indicators in Chapter 2).

Table 5-16, for the years 1981-1985, shows that small changes occurred in these patterns in 8 years. First, the differentials between the western region and the three others widened. This was the result of a decline in rural mortality in the west (in the rural areas of the region of Thiès, since the region of Dakar incorporates virtually no rural areas) that was more rapid than that in the other grand regions. Second, the center, and no longer the south, had the highest urban mortality in Senegal.

The 1988 census throws additional light on a more discriminating geographic scale, that of the departments.4Table 5-5 gives mortality rates for children aged 1-4 in 1987-1988 by administrative departments and rural/urban residence, according to the 1988 census. In the following analysis, we omit the three departments within the Dakar region, since either they have no rural population with which the urban population can be compared (Dakar and Pikine), or the rural population is extremely small (Rufisque). For the remaining departments, rural and urban mortality rates have a correlation coefficient of 0.50 (see Figure 5-8). Within a department, rural mortality is almost always greater than urban mortality, on average by 77 percent.

The contrast between urban and rural areas outside the Dakar region is often greatest in departments whose capital is a major city. Rural excess mortality is thus high in the departments of Louga (2.7); Thiès (2.6); Kaffrine (2.4); Tivaouane (2.4); Kolda (2.2); Kaolack (2.1); and Dagana, which includes Saint-Louis (2.0).5 It is low or zero, and for some a slight urban

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

FIGURE 5-8 Relation between child mortality (4q1 per 1,000) in rural areas (x-axis) and in urban areas (y-axis) in the same department. NOTE: The three departments of the region of Dakar are not included because of the nonexistence, or small size, of the rural population. SOURCE: Unpublished tables, 1988 census.

excess mortality is even recorded, in the departments of Matam (0.8), Oussouye (1.0), Bakel (1.1), Nioro du Rip (1.1), and Linguère (1.1), whose capital towns are merely small cities.

Ethnic Differences

Compared with differences stemming from educational level or urban or rural residence, differences among ethnic groups are neither very pronounced nor very consistent at the national level. However, the excess mortality of Mandingo children in rural areas should be noted (Tables 5-15 and 5-16).

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

Measurements by ethnic group performed locally as part of small-scale population studies covering several ethnic groups reveal that the situation varies according to region and group. At Thiénaba and Fissel [16], there were, for example, no differences in mortality in 1981-1984 between the two ethnic groups represented, the Wolof and the Serer (Mbodji, 1988). In Bandafassi and Ndemene, on the other hand, mortality was lower among the Peul (belonging to the Poular group) than among the other major ethnic group in the region—the Malinké (belonging to the Mandingo group) in Bandafassi and the Wolof in Ndemene (Cantrelle et al., 1980; Pison et al., 1993). According to Cantrelle et al., the lower mortality rate among the Peul can be attributed to the fact that the Peul are predominantly cattle rearers and thus have more milk to give their children, which raises the children's nutritional status (Cantrelle et al., 1980). But these differences in mortality may also stem from differences in housing and household structure. The Peul tend to live in smaller households than other ethnic groups. In Bandafassi, for example, the average size of a Peul compound is 13 persons, as compared with 22 for the Malinké (Pison et al., 1993). In addition, the density in the Peul villages is lower than in the Malinké villages. Mortality due to certain infectious causes, such as measles, has been shown to increase with overcrowding and the frequency of intrafamily contagions (Aaby et al., 1993); these are more important in large households.

Quality of Housing and Household Facilities

Multivariate analysis of the data from the 1986 DHS-I shows a correlation between child mortality and the type of dwelling, toilet, and household water supply (see Table 5-17). Once these characteristics are considered, the type of place of residence, whether urban or rural, becomes insignificant, implying that the beneficial influence of living in urban as opposed to rural areas results largely from access to clean water and sanitation, as well as to better dwelling conditions.

The relative importance of these socioeconomic variables is not the same in rural as in urban areas. In rural areas, the type of dwelling is the only significant factor among the three. In urban areas, neither the type of dwelling nor the type of toilet is significant, but the type of household water supply is. Children living in households where water is drawn from an inside well or faucet or from a borehole or pump experience a nearly 40 percent lower probability of dying before age 5 as compared with those living in households obtaining their water from an outside well or faucet or a river.

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

TABLE 5-15 Variations in Mortality of Children Less Than 5 Years of Age (relative risk) According to Selected Demographic and Socioeconomic Characteristics, 1973-1977, Odds Ratios

 

 

Models with Socioeconomic Variables and Regional Variables

Models with Socioeconomic Variables and Ethnic Variables

Variable

Category

Senegal

Urban

Rural

Senegal

Urban

Rural

Sex

Male/female

1.09

1.05

1.10

1.08

1.05

1.09

Twins

Twin/singleton

1.38

2.28

1.21

1.22

2.50

0.98

Birth order

1

0.98

0.64

1.09

0.95

0.57

1.06

 

2 to 6 (omitted)

 

 

 

 

 

 

 

7 or more

1.10

0.94

1.12

1.07

0.98

1.09

Age of mother

Less than 20 years

1.18

1.49

1.11

1.16

1.67

1.06

 

20-34 years (omitted)

 

 

 

 

 

 

 

35 years

0.87

1.14

0.83

0.85

1.04

0.83

Survival of preceding child

Dead/surviving or no child

1.20a

0.85

1.26a

1.19a

0.76

1.24a

Mother's instruction

Instruction/no instruction

0.77a

0.83

0.65a

0.77a

0.86

0.62a

Residence

Urban/rural

0.46a

0.46a

Region

West (omitted)

 

 

 

 

 

 

 

Center

0.99

1.11

0.92

 

Northeast

0.81

0.71

0.77

 

South

1.26a

1.89a

1.15

Ethnicity

Wolof (omitted)

 

 

 

 

 

 

 

Poular

0.92

1.10

0.95

 

Serer

1.09

1.48

1.10

 

Mandingo

1.32a

1.02

1.47a

 

Diola

0.93

1.61

0.92

 

Other

0.78

0.33a

1.01

a p < 0.05, two-tailed test.

SOURCE: Multivariate analysis using WFS datafiles, period 1973-1977.

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

TABLE 5-16 Variations in Mortality of Children Less Than 5 Years of Age (relative risk) According to Selected Demographic, Geographic, and Ethnic Characteristics, 1981-1985, Odds Ratios

 

 

Model with Socioeconomic Variables and Regional Variables

Model with Socioeconomic Variables and Ethnic Variables

Variable

Category

Senegal

Urban

Rural

Senegal

Urban

Rural

Sex

Male/female

1.05

1.02

0.93

1.06

0.98

0.95

Twins

Twin/singleton

2.31a

2.62a

1.57

2.29a

2.58a

1.63

Birth order

1

1.05

0.72

1.17

1.05

0.79

1.15

 

2 to 6 (omitted)

 

 

 

 

 

 

 

7 or more

0.94

0.75

0.84

0.91

0.68

0.81

Age of mother

Less than 20 years

1.01

1.32

1.04

1.01

1.39

1.06

 

20-34 years (omitted)

 

 

 

 

 

 

 

35 years

0.94

1.04

1.08

0.93

0.87

1.09

Survival of preceding child

Dead/surviving or no child

0.99

0.62

0.96

1.04

0.69

0.97

Mother's instruction

Instruction/no instruction

0.65a

0.68

0.79

0.60a

0.66a

0.62

Residence

Urban/rural

0.78a

0.70a

Region

West (omitted)

 

 

 

 

 

 

 

Center

1.28a

1.62a

1.57a

 

Northeast

1.00

0.67

1.31

 

South

1.33a

1.11

1.48a

Ethnicity

Wolof (omitted)

 

 

 

 

 

 

 

Poular

0.88

0.48a

0.85

 

Serer

0.87

0.49a

0.93

 

Mandingo

1.08

0.53

1.30

 

Diola

1.10

0.93

0.73

 

Other

1.00

1.00

0.73

a p < 0.05, two-tailed test.

SOURCE: Multivariate analysis using DHS-I datafiles, period 1981-1985.

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

TABLE 5-17 Variations in Mortality of Children Less Than 5 Years of Age (relative risk) According to Selected Demographic and Socioeconomic Characteristics, 1981-1985, Odds Ratios

 

 

Socioeconomic Variable and Variable on the Quality of Housing Considered Together

Variable

Category

Senegal

Urban

Rural

Sex

Male/female

1.08

1.09

0.95

Twins

Twin/singleton

2.16a

2.64a

1.55

Birth order

1

1.05

0.74

1.15

 

2 to 6 (omitted)

 

 

 

 

7 or more

0.92

0.75

0.82

Age of mother

Less than 20 years

0.96

1.19

1.02

 

20-34 years (omitted)

 

 

 

 

35 years+

0.93

0.89

1.10

Survival of preceding child

Dead/surviving or no child

1.00

0.69

0.93

Mother's instruction

Instruction/no instruction

0.70a

0.78

0.89

Residence

Urban/rural

0.88

-

-

Quality of construction

Superior/inferior

0.71a

1.11

0.78a

Type of toilet

Superior/inferior

0.82a

1.06

0.84

Water supply

Superior/inferior

0.78a

0.63a

1.22

a p < 0.05, two-tailed test.

SOURCE: Multivariate analysis using DHS-I datafiles, period 1981-1985.

Associations Between Socioeconomic Level and Child Mortality from the 1988 Census

The 1988 census allows us to analyze associations between child mortality (measured by 4q1) and selected socioeconomic variables at the department level.6 But it is important to bear in mind that while associations may suggest, they do not prove, causal mechanisms. Not surprisingly, the correlation coefficients7 between child mortality and all the socioeconomic indicators are negative; a higher socioeconomic status is related to lower child mortality (see Table 5-18).

In rural areas, the strongest indicator of child mortality seems to be whether households have electricity. In urban areas, however, having a latrine or toilet is the strongest indicator.

Figure 5-9, which shows the association between child mortality and the proportion of households having a latrine or toilet, is illustrative of the

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

TABLE 5-18 Correlation Coefficients Between Socioeconomic Indicators and Child Mortality (4q1)

Socioeconomic Indicator

Rurala

(n = 28)

Urbana

(n = 30)

Nationala

(n = 30)

Nationalb

(n = 10)

Housing

-0.34

-0.44

-0.69

Electric

-0.45

-0.51

-0.77

Latrine

-0.10

-0.60

-0.71

Water

-0.30

-0.09

-0.68

Composite

-0.34

-0.44

-0.75

Ever school

-0.66

School enrollment rate (male)

-0.50

School enrollment rate (female)

-0.52

NOTE: See Chapter 2 for details on how the socioeconomic indicators were constructed.

a These coefficients are calculated using data at the departmental level. There are only 28 cases in the rural category because the departments of Dakar and Pikine do not have rural data.

b These coefficients are calculated using data at the regional level.

SOURCE: 1988 census (unpublished tabulations)

negative relationship between child mortality and socioeconomic status. Here, data for rural and urban residences for each department are shown.

Data on schooling are available only at the regional level, so there are only ten data points. However, a similar analysis reveals that child mortality is negatively related to various indicators of education. The strongest relationship is between child mortality and the proportion of the population aged 6 and older with some schooling.

Summary and Conclusions

Senegal has undergone a continuous decrease in child mortality since World War II, the only period for which data are available. In 45 years, from 1945-1990, the risk that a newborn would die before the age of 5 fell by a factor of three, declining from approximately 400 per 1,000 to 130 per 1,000. This decrease accelerated toward the end of the 1970s and the beginning of the 1980s. A new health policy, which emphasized primary health care and was implemented during this period, may have played a role. The proliferation of health infrastructures (which had previously been highly concentrated in Dakar) in the various regions and the implementation of the Expanded Programme on Immunization (EPI) probably contributed significantly to the accelerated decline.

Acceleration of the mortality decline accompanied implementation of

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

FIGURE 5-9 Relationship between child mortality (4q1) and level of sanitation, 1988. SOURCE: Unpublished tables, 1988 census.

the structural adjustment policies carried out in the early 1980s, an indication that the policies as implemented did not produce the adverse effect that might have been imagined.8 Public and private assistance programs set up by nongovernmental organizations undoubtedly made an important contribution in this regard.

Urban and rural areas experienced different rhythms of mortality change. Urban areas, with Dakar in the lead, had already experienced a significant decline in mortality in the first half of the century. Conversely, mortality remained high in rural areas until much later. The decline began in earnest in these areas only in the 1970s or 1980s. This trend, which started in the rural areas surrounding Dakar, gradually expanded to include the most remote regions. Once the decline had begun in rural areas, it spread rapidly, thereby narrowing the gap between rural and urban areas. Acceleration of the EPI in 1987 contributed heavily to this narrowing of the gap.

In the 1970s and 1980s, the principal differences in child mortality

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

occurred according to the educational level of the parents and urban or rural residence. In the late 1980s, mortality in the rural areas was appreciably higher in the southern half of Senegal than in the northern half, perhaps because of the stronger impact of malaria in the former region. Socioeconomic differences, while perceptible, were not very great on the local level, a sign of relative equality of opportunity and of the important role of infrastructure and health programs in the observed levels and trends.

Continued mortality decreases in the coming years will depend substantially on the continuation of health programs, in particular of the sustained vaccination initiative, improved pregnancy monitoring, and enhanced conditions of delivery. Much remains to be done in all three of these areas.

ADULT MORTALITY

Three broad types of data are used here to estimate adult mortality in Senegal: changes in cohort size from one census to another, information on deaths by age and sex in the 12 months before the 1988 census, and information on survival of parents from the 1988 census (see Table 5-19). Further details on each method are given below.

Data Sources

A number of data sources provide information about adult mortality in Senegal. First, intercensal survival methods can be used with the population estimates from the 1960-1961 DS [1] and the 1970-1971 NDS [2] and counts from the 1976 [17] and 1988 [18] censuses. Second, the 1988 census collected information on household deaths in the year before the census, by age and sex. In combination with the census populations, the information on deaths can be used to calculate age-specific mortality rates (ASMRs), and from them all other life table parameters. Third, the 1988 census also collected data on the survival of mothers and fathers, the basic information needed to apply the orphanhood method of adult mortality estimation (Timæus, 1991). Finally, the small-scale population studies [8-16] conducted in various parts of the country collected detailed prospective information on deaths by age and sex.9 Results obtained using these four sources are discussed below.

Results

Intercensal Survival Methods

A number of methods have been developed for deriving adult mortality estimates from successive census counts (United Nations, 1983a; Preston

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

TABLE 5-19 Methods of Calculating Adult Mortality

Method

Data Required

Source

1. Intercensal survival methods

 

 

(a) intercensal survival probabilities

Age distribution by sex from two consecutive censuses.

United Nations (1983a)

(b) intercensal growth rates

Intercensal growth rates for 5-year age groups.

Preston and Bennett (1983)

2. Distribution of deaths by age

 

 

(a) direct calculation

Deaths in a year, classified by 5-year age group and sex.

Shryock and Siegel (1976)

(b) growth balance methods

Midyear population classified by 5-year age groups and sex; see 2(a).

Brass (1975); United Nations

(1983a); Hill (1987)

3. Survival of Parents

 

 

(a) maternal orphanhood data: Brass method

Survival probabilities of mother alive (or dead), classified by 5-year age groups of women. Mean age of childbearing.

United Nations (1983a)

(b) paternal orphanhood data: Timæus method

Survival probabilities of father alive (or dead), classified by 5-year age groups of men. Mean age of childbearing.

Timæus (1991)

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

and Bennett, 1983). The Preston and Bennett method, which uses age-specific growth rates in its computation, is particularly appropriate for the 12-year intercensal period in Senegal. However, single-year survivorship ratios were calculated first to explore data quality and revealed systematic age-reporting bias.

Figure 5-10 shows the observed single-year survivorship ratios, for females and males. The ratios are extremely erratic. The points appear to divide into three sets: one set of very high ratios (corresponding to ages ending in 7 in 1976, and thus ending in 9 in 1988, such as 47-59); one set of very low ratios (corresponding to ages ending in zero in 1976, and thus ending in 2 in 1988, such as 50-62); and one set of values in between. This pattern is most unusual. In most African (and other developing country) census age distributions, there is a strong tendency for ages ending in zero (and, to a lesser extent, 5) to be inflated at the expense of ages ending in 9 and 1. In the 1988 census, this over-reporting of age affects ages ending in 9, accounting for the high survivorship ratios for ages ending in 9 in 1988, whereas in 1976 the typical pattern is observed, accounting for the low survivorship ratios for ages ending in zero in 1976.

Neither pattern is correct, of course. It is not the case that there were really more people with ages ending in 9 than ages ending in other digits in 1988, or that there were more people with ages ending in zero in 1976. The error is one of ascribing to some convenient age a number of people of surrounding ages who do not know their ages exactly. The change in reporting pattern is inconvenient for analytical purposes, however. For data tabulations by standard 5-year age groups, heaping on zero tends to exaggerate the true age of the population of a 5-year age group (because those of ages close to, but below, the figure ending in zero are categorized in the higher 5-year age group), whereas heaping on 9 categorizes people at ages below their true ones. Thus the change in age-reporting practice makes comparisons between 1976 and 1988 particularly problematic.

To make the 1988 data correspond to standard patterns, these data have been "repreferenced," with the excess numbers reported on ages ending in 9 being shifted to the next higher age ending in zero. The repreferencing was performed by subtracting from ages ending in 9 (and those ending in 4) the difference between the observed number and the average of the four single ages on either side and adding this difference to the adjacent age ending in zero (or 5).

The corrective measures taken into account for the 12-year intercensal period, as well as for age misreporting, allow comparison of the survival ratios with the Coale-Demeny West family of model life tables. Along with the 1976-1988 survivorship ratios from the censuses, Figure 5-10 shows the survivorship ratios from levels 5, 10, 15, and 20 of the Coale-Demeny West family of model life tables. It is immediately clear that the data suffer from

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

FIGURE 5-10 Observed single-year survivorship ratios and ratios from various levels of Coale-Demeny "West" life tables.

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

distortions more serious than just erratic noise. The survivorship ratios decline much less sharply with age than any of the model patterns, such that for young adults, the ratios suggest levels of adult mortality in the range of levels 10-15 in the Coale-Demeny tables (e[15]—life expectancy at age 15—in the range 40-50 years), but the ratios for older adults indicate adult mortality well below any of the models shown. Since we know from the various small-scale studies that age patterns of adult mortality resemble the models quite closely, the census data are probably distorted by substantial age exaggeration, particularly from middle age onwards.

The Preston-Bennett technique has been applied to the three intersurvey periods, 1960-1970, 1970-1976, and 1976-1988. Essentially, this method uses intercensal growth rates for each age group to convert an average of the age distributions into a stationary (or life-table) age distribution, from which expectations of life at each age can be obtained. This method, as well as being convenient for census intervals that are not multiples of five, is also relatively robust to local age misreporting patterns, such as digital preference, that are repeated from one census to the next (since age-specific growth rates are little affected by such repeated misreporting), though it is sensitive to substantial biases in age reporting and to changes in coverage completeness.

The expectations of life at each age from 15-55 estimated by the Preston-Bennett method are shown by sex and time period in Table 5-20; also shown are the corresponding Coale-Demeny West model life-table levels. The model life-table levels provide a convenient index for comparing estimates across ages and sexes.

The intercensal survival results are disappointing. The 1960-1970 survival data are not entirely implausible, but the male/female difference (median Coale-Demeny levels of 12.2 for females, 5.8 for males) and the extreme variability in estimated levels across age groups (suggesting massive age misreporting) do not inspire confidence. The 1970-1976 period is far worse: the mortality estimates are uniformly ridiculously low. The growth rates of both the male and female population over age 70 exceed 10 percent per year, and for males the rates of growth of the population over 45 all exceed 6 percent per year. The 1970 population estimate and the 1976 census count are clearly not consistent. Such an outcome is not uncommon with population totals coming from sample surveys analyzed together with those from censuses; censuses have often been found to do a substantially better job of coverage than surveys.

The mortality estimates obtained from an intercensal survival analysis of the 1976 and 1988 censuses are surprisingly high. For both males and females, the levels for ages 40 and below are modestly consistent (in the range of 7.9 to 9.1 for males, 6.1 to 10.5 for females), both between ages

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

TABLE 5-20 Intercensal Survival Estimates of Adult Mortality

 

1960-1970

1970-1976

976-1988

 

Male

Female

Male

Female

Male

Female

Age

e(x)

Level

e(x)

Level

e(x)

Level

e(x)

Level

e(x)

Level

e(x)

Level

15

62.6

>24

53.0

17.1

83.5

>>24

73.4

>>24

39.1

7.9

44.7

10.5

20

48.4

19.3

38.8

8.8

75.5

>>24

64.9

>>24

36.2

8.5

38.4

8.4

25

35.2

10.7

30.9

4.9

73.4

>>24

56.2

>>24

33.3

8.9

33.4

6.9

30

27.6

6.6

28.8

5.5

75.8

>>24

49.4

>24

30.3

9.1

29.5

6.1

35

23.1

5.0

29.6

9.4

81.1

>>24

44.2

>24

26.9

9.1

27.4

7.0

40

19.4

3.8

28.4

12.2

83.9

>>24

37.6

23.2

23.6

9.0

24.9

7.6

45

16.7

3.5

26.2

14.6

84.0

>>24

32.5

22.9

22.0

11.3

24.0

10.9

50

14.7

4.3

22.8

14.4

74.5

>>24

28.8

23.6

20.0

13.8

22.5

14.5

55

13.1

5.8

20.2

17.8

60.3

>>24

26.5

>24

17.5

15.8

19.9

16.9

60

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

14.5

16.5

16.9

18.7

65

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

12.0

18.2

14.0

20.2

70

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

n.a.

10.4

22.3

12.6

23.9

NOTE: See Appendix A for description of surveys and censuses; n.a., not available.

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

and between sexes. The estimated mortality is very high, implying a probability of surviving from age 15-60 of less than 50 percent. Above age 40, the model levels rise (mortality level falls) steeply with age, exceeding level 20 by age 70. Clearly, the estimates based on the population above age 40 are affected by systematic age exaggeration, whereby people report themselves to be older than they really are (over and above any local age misreporting).

The very high estimates of adult mortality (based on the population below age 40), higher than those from any of the small-scale area studies, suggest yet another problem with the data from the two censuses. It is likely that the 1976 census had relatively higher coverage than the 1988 census, giving rise to the impression of high intercensal mortality. The coverage change could have come about as a result of either overenumeration in 1976 or underenumeration in 1988; the comparisons available do not address the question of which census is relatively better, but merely demonstrate that they are different.

In summary, the available age distribution data suffer from three patterns of error that affect adult mortality estimation: the inconvenient but not disastrous preference for ages ending in certain digits, with the unusual wrinkle of a preference for ages ending in 9 in 1988; a disastrous (from the point of view of adult mortality estimation) tendency to exaggerate age; and a disastrous (from the same point of view) change in census coverage. However one interprets the relative coverage of the various sources, it remains the case that essentially no conclusions about levels or trends of adult mortality can be obtained from the available sequence of population estimates or counts.

Age-Specific Mortality Rates Based on Deaths Reported in the Year Before the 1988 Census

The 1988 census included a series of questions on deaths in the household in the previous 12 months. (The French version of the census questionnaire asks for deaths in the previous 12 months, but in practice interviewers are believed to have asked for events since "la Korité," an important Muslim holiday, the date of which fell almost to the day 1 year before the reference day of the census; see Chapter 4, footnote 6 for more details.) Information collected included the name, sex, and age at death of each deceased.

The data resulting from these questions permit the direct calculation of ASMRs, and thus of all standard life-table functions (see, for example, Shryock and Siegel, 1976). Although it is clear that the operative age preference affecting age at death is for ages ending in zero,10 the reported 1988 population age distribution, despite its anomalous digital preference

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

for age ending with 9, has been used in the denominators. Table 5-21 shows the ASMRs, the life-table survivorship column, the expectation of life at each age from 15-70, and the corresponding Coale-Demeny model life-table level corresponding to each expectation of life.

The consistency of the model life-table levels implied by the expectations of life at adult ages is remarkable, especially if one omits from consideration the 70+ age group, which is affected by assumptions about the age distribution of the elderly. Reasonable consistency is not very surprising, of course, since computing a survivorship function from a set of age-specific death rates, and then computing expectations of life at successive ages, imposes a substantial degree of smoothing on the raw data.

An attempt to assess the completeness of death reporting through growth balance methods was overwhelmed by biases in age reporting (see Appendix D). However, the evidence discussed above that reporting of deaths of children aged 1-5 was close to complete, combined with the consistency of the age-specific estimates between the ages of 15-70, suggest that the 1988 death data are of usable quality for ages after infancy. Thus the 1987-1988 life table is accepted as a reasonable estimate of adult mortality for the period for Senegal. Almost 30 percent of males and about 27 percent of females die between the ages of 15 and 60. Although male risks are higher, males do slightly better than females in terms of Coale-Demeny model levels, implying that the male disadvantage is slightly smaller in Senegal than in the historical experience embodied in the models.

Estimates Based on Survival of Parents

The 1988 census included questions for all respondents on survival of mother and father. The proportion of respondents of a given age with mother still alive reflects the female adult survivorship from roughly the average age of childbearing to that age plus the age of the respondents. Proportions with father still alive provide similar measures for adult males. Standard methods exist for converting proportions with surviving mother or father into conventional life-table measures. The methods used here are those described by United Nations (1983a) for mothers and by Timæus (1991) for fathers.

Table 5-22 shows proportions with mother surviving, implied survivorship from age 25, implied Coale-Demeny West mortality level, and time reference period for respondents aged 15-19 to 45-49. The estimates cover the period 1971-1980. Early in the period, mortality was high (West level around 12.5, probability of dying from age 15-60 around 37 percent), but appears to have improved dramatically by 1980 (West level around 16, probability of dying between ages 15 and 60 around 28 percent).

Table 5-23 shows proportions with father surviving, implied survivorship

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

from age 35, implied Coale-Demeny West mortality level, and time reference period for respondents aged 5-9 to 35-39. The estimates cover the period 1974-1983. Early in the period, adult male mortality was quite high (West level around 15, probability of dying between ages 15 and 60 around 37 percent), but appears to have fallen sharply over the period for which estimates are available, to quite low levels in the early 1980s (West level about 19, probability of dying between ages 15 and 60 around 26 percent).

These rates of change in adult mortality in quite a short period are very high, and though they are theoretically possible, the potential role of typical data errors must be taken into account. Mortality estimates based on reports of parental survival by young respondents are often distorted downward; this is often due to an adoption effect, whereby young children, one of whose parents has died, are adopted by a relative who is treated, at least for all data collection purposes, as the true parent. The effect of this error for older respondents is limited, however, because the parent, whether true or adoptive, dies within the respondent's own memory span. Thus the most accurate adult mortality estimates from data on survival of parents are those based on older respondents. On this basis, the orphanhood data indicate that in the early 1970s, the probability of dying between the ages of 15 and 60 was about 37 percent, for both males and females.

The orphanhood estimates combined with the 1988 life table indicate substantial declines in adult mortality. For males, the probability of dying between 15 and 60 fell from 37 to 30 percent from the early 1970s to the late 1980s. For females, the decline was even more pronounced, from 37 to 27 percent over the same period. The corresponding increases in expectation of life at age 15 are from about 48.3 to 50.0 for males and from about 47.0 to 51.5 for females.

Evidence from Small-Scale Studies

Senegal is unusual by African standards of demographic data availability with respect to the number of small-scale, intensive population studies that have been conducted [8-16]. Data from three of these studies are discussed here. The earliest such population study was established in the arrondissement of Niakhar [11] in 196211; Bandafassi [8], in eastern Senegal, has been observed since 1970, and Mlomp [10], in the south, has been under observation since 1984. All three areas provide mortality information collected longitudinally, typically by annual rounds of data collection and are thus less prone than retrospective data to omission, misplacement of events in time, and event-related age misreporting. On the other hand, the populations under observation are quite small, so it is not possible to follow trends closely, at least in adult mortality. Nor can these local-area mortality measures be regarded as representative of Senegal as a whole. However,

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

TABLE 5-21 Adult Mortality Indicators, 1988 Census: Deaths in Last 12 Months

Age

Deaths 1987-1988

Population 1988

Population 1976

Mid-Year Population 1987-1988a

ASMR

1(x)b

5Lxc

e(x)d

"West" Level

Males

15-19

1,214

333,468

246,856

329,344

.00369

.8335

4.1309

50.01

16.9

20-24

1,130

260,066

202,963

257,416

.00439

.8183

4.0486

45.89

17.1

25-29

1,092

239,588

170,075

236,212

.00462

.8005

3.9585

41.85

17.2

30-34

1,169

181,011

128,383

178,454

.00655

.7822

3.8505

37.77

17.3

35-39

900

158,738

121,084

156,971

.00573

.7570

3.7335

33.94

17.7

40-44

963

105,298

107,486

105,398

.00914

.7356

3.5990

29.85

17.8

45-49

837

106,186

94,850

105,697

.00792

.7027

3.4479

26.13

18.4

50-54

1,160

83,553

77,424

83,296

.01393

.6754

3.2677

22.08

18.1

55-59

1,042

79,531

65,454

78,895

.01321

.6299

3.0525

18.48

18.3

60-64

1,683

58,463

55,569

58,345

.02885

.5895

2.7568

14.57

16.7

65-69

1,233

51,690

35,135

50,870

.02424

.5100

2.4098

11.44

16.0

70+

9,277

70,302

70,882

70,331

.13190

.4516

3.4237e

7.58

9.3

unknown

 

5,209

 

 

 

 

 

 

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

Females

15-19

1,269

376,720

264,355

371,215

.00342

.8619

4.2744

51.60

16.0

20-24

1,194

299,951

224,342

296,354

.00403

.8473

4.1958

47.44

16.1

25-29

1,197

292,445

202,120

287,989

.00416

.8304

4.1108

43.36

16.1

30-34

1,226

198,166

154,403

196,123

.00625

.8133

4.0063

39.21

16.0

35-39

995

180,976

136,742

178,882

.00556

.7882

3.8893

35.38

16.3

40-44

932

113,623

117,486

113,786

.00819

.7666

3.7592

31.30

16.3

45-49

787

113,075

89,241

111,969

.00703

.7358

3.6181

27.50

16.6

50-54

1,086

80,670

74,150

80,390

.01351

.7104

3.4404

23.39

16.3

55-59

846

80,400

52,797

79,006

.01071

.6639

3.2363

19.85

16.9

60-64

1,348

49,944

49,591

49,931

.02700

.6292

2.9548

15.80

15.7

65-69

838

46,281

27,586

45,296

.01850

.5495

2.6305

12.72

16.2

70+

8,035

70,069

66,794

69,906

.11494

.5008

4.3571

8.70

11.1

unknown

 

 

2,302

 

 

 

 

 

 

NOTE: Probability of dying, 15-60, 45q15 (= 1 - 1(60)/1(15)) = .293 for males and .270 for females.

a Computed by calculating age-specific growth rate, reducing 1988 population by half year's growth.

b Separation factor of 2.6 used throughout; 5qx = 5 * 5ASMRx / (1 + (5 - 2.6) * 5ASMRx).

c5Lx = 2.6 * 1(x) + 2.4 * 1(x + 5).

d e(x) = ∼wx 5Lx / 1x.

e L(70+) = l(70+) / wASMR70+.

SOURCE: 1988 census (unpublished tables); Coale and Demeny (1983)

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

TABLE 5-22 Adult Female Mortality Estimated from Survival of Mother, 1971-1980

Age

N

Proportion with Mother Alive

1(25+N)/ 1(25)

Mortality Level

Reference Date

15-19

20

0.9034

0.8940

16.1

1980.6

20-24

25

0.8577

0.8491

15.6

1978.7

25-29

30

0.7825

0.7751

14.4

1977.0

30-34

35

0.7076

0.7011

14.1

1975.6

35-39

40

0.5793

0.5688

12.6

1974.0

40-44

45

0.4665

0.4474

12.3

1972.8

45-49

50

0.3466

0.3131

12.3

1971.7

 

SOURCE: Coale and Demeny (1983); 1988 census (unpublished tabulations)

age patterns of mortality and broad patterns of change over time may well reflect a wider reality than the areas themselves.

Table 5-24 shows the Coale-Demeny West model life-table levels implied by the expectation of life at each age from 15-75 for the life tables from Ngayokhème (1963-1981), Niakhar-II (1984-1991), Bandafassi (1971-1991), and Mlomp (1985-1990). Figure 5-11 plots age- and sex-specific mortality rates against the rates from the four regional families of the Coale-Demeny (1983) model life-table systems, the model life-table levels being selected on the basis of the expectation of life at birth of the observed tables. Although the observed mortality rates fluctuate somewhat, no doubt because of the small numbers of observations, in most cases they track the West family mortality pattern fairly closely. The exception is the 1963-1981 Niakhar/Ngayokhème life table, which has clearly lower mortality in old age, given the level in early and middle age, than any of the model families. The life table also has lower rates in old age than the later 1984-1991 life table for Niakhar/Ngayokhème, despite substantially higher rates at earlier ages. Rather than assume that mortality at advanced ages increased while mortality at younger ages fell, and that the initial mortality pattern was very different from model patterns, it is concluded that the Ngayokhème pattern is affected by considerable exaggeration of age at initial data collection. Such exaggeration means mortality rates that appear to be for, say, those aged 60 are actually for, say, those aged 50, thus giving the appearance of low older-age mortality.

Although the mortality levels implied by the expectations of life rise at old ages in most of the tables, consistent with continued effects of age exaggeration, the levels at younger ages are reasonably consistent, with only two exceptions. Thus between ages 15 and 50, the male e(x) values fall in the range of Coale-Demeny West levels 16.7 to 18.1 for Niakhar-II

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

TABLE 5-23 Adult Male Mortality Estimated from Survival of Father, 1974-1983

Age

N

Proportion with Father Alive

1(35+N)/ 1(35)

Mortality Level

Reference Date

5- 9

10

0.956

0.9678

21.3

1983.7

10-14

15

0.915

0.9471

21.9

1981.7

15-19

20

0.821

0.8425

18.1

197.9

20-24

25

0.718

0.7269

16.2

1978.2

25-29

30

0.582

0.6021

15.6

1976.7

30-34

35

0.456

0.4505

14.8

1975.1

35-39

40

0.314

0.2918

13.9

n.a.

40-44

n.a.

0.205

n.a.

n.a.

n.a.

NOTE: Calculations assume mean age of childbearing = 37.3

SOURCE: Coale and Demeny (1983); 1988 census (unpublished tabulations)

1984-1991, and 16.9 to 18.0 for Mlomp 1985-1990; at the national level, the 1987-1988 census life table implies a West level of about 17.5. For females between the same ages, the range is 14.7 to 19.8 for Ngayokhème 1963-1981, 16.5 to 17.1 for Niakhar-II 1984-1991; 10.3 to 11.9 for Bandafassi 1971-1991, and 17.2 to 19.0 for Mlomp 1985-1990. At the national level, orphanhood estimates indicate a level of about 12.6 in the early 1970s, and the 1987-1988 census life table implies a West level of about 16.4. The exceptions to consistency are the values for Ngayokhème 1963-1981 males and Bandafassi males, both of which show a strong upward trend in model levels, from 14.2 at age 15 to 15.7 at age 50 in Ngayokhème, and from 13.4 at age 15, to 16.1 at age 50, to 18.6 at age 65 in Bandafassi.

Although the numbers of observations are not large, two of the small-scale studies provide some evidence of trends in adult mortality. For Ngayokhème, expectation of life at age 15 increases by about 3 years from its average for the period 1963-1981 to its average for the period 1984-1991, an increase of about one-fifth of a year per year if the decline was reasonably steady over time (see Table 5-25). For Bandafassi, Malinké villages have been under observation for 20 years, with expectation of life at age 20 increasing from 36 years in the period 1971-1976 to 43 years in the period 1986-1991 (Pison and Desgrées du Loû, 1993), an increase of almost half a year per year in a heavier mortality environment. Thus the regional-level studies do suggest substantial declines in adult mortality.

The small-scale studies also provide the only information available on the cause-of-death structure in Senegal. Verbal autopsies have been used to try to identify the probable cause of death in Niakhar-II from 1984-1990. Unfortunately, 50 percent of all deaths of persons over age 5 were from

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

TABLE 5-24 Coale-Demeny West Levels Implied by Expectations of Life at Each Age, Four Regional Studies:

 

Male

Female

Age

Niakhar/Ngayokhème

Bandafassi

Mlomp

Niakhar/Ngayokhème

Bandafassi

Mlomp

 

1963-1981

1984-1991

1971-1991

1985-1990

1963-1981

1984-1991

1971-1991

1985-1990

15

14.2

16.9

13.4

18.0

14.7

16.7

11.6

18.9

20

14.4

17.0

13.5

17.8

14.8

16.5

11.4

19.0

25

14.0

16.7

14.3

17.5

15.8

16.7

11.9

18.6

30

14.2

17.0

14.2

16.9

16.5

16.9

11.5

18.5

35

15.0

17.6

14.0

17.5

16.8

16.7

11.0

18.7

40

15.8

17.5

13.7

17.7

18.3

16.7

10.3

18.7

45

15.6

18.0

14.8

17.7

19.1

16.6

11.0

18.8

50

15.7

18.1

16.1

17.3

19.8

17.1

11.9

17.2

55

16.3

18.6

17.4

18.9

20.7

17.3

10.7

17.8

60

16.2

19.3

18.3

17.9

22.0

17.9

11.2

17.6

65

18.2

18.6

18.6

16.0

22.9

17.6

10.7

17.2

70

22.2

18.4

17.3

12.8

23.1

18.0

13.1

20.8

75

>24

20.8

23.0

14.3

>24

18.2

20.3

21.1

 

SOURCES: Niakhar/Ngayokhème, 1963-1981: Cantrelle et al. (1986); Niakhar/Ngayokhème, 1984-1991: Project Niakhar (1992); Bandafassi: Pison and Desgrées du Loû (1993); Mlomp: Lagarde et al. (forthcoming)

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

FIGURE 5-11a Ngayokhème life table 1963-1981 versus Coale-Demeny level 8 models.

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

FIGURE 5-11b Niakhar life table 1984-1991 versus Coale-Demeny level 13 models.

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

FIGURE 5-11c Bandafassi life table 1970-1990 versus Coale-Demeny level 9 (males) and level 8 (females) models.

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

FIGURE 5-11d Mlomp life table 1985-1990 versus Coale-Demeny level 17 (males) and level 18 (females) models.

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

TABLE 5-25 Estimates of Adult Mortality, National- and Regional-Level Sources

 

 

Male

Female

Study

Reference Date

45q15

e(15)a

45q15

e(15)

National

 

 

 

 

 

 

1970

n.a.

47.5

n.a.

48.5

 

1978

n.a.

48.9

n.a.

51.5

Orphanhood,

1988 census

1976

0.354

48.3

n.a.

n.a.

Orphanhood,

1988 census

1972

n.a.

n.a.

0.375

47.1

Deaths in year before 1988 census

1987-1988

0.295

49.9

0.271

51.5

Regional

 

 

 

 

 

Ngayokhème

1963-1981

0.386

46.9

0.361

49.8

Niakhar

1984-1990

0.331

50.0

0.279

52.4

Bandafassi

1971-1991

0.434

46.0

0.411

46.0

Mlomp

1985-1990

0.291

51.3

0.200

55.3

a The table in Cantrelle et al. (1986) refers to this figure as e(10), whereas the text refers to it as e(15); the magnitudes of the differences between these values and the (0)s cited indicate that they are probably e(15)s.

SOURCES: Local data for Ngayokhème, Niakhar, Bandafassi and Mlomp same as Table 5-24); 1988 census (unpublished tables); Cantrelle et al. (1986).

unidentified causes, and for another 10 percent no information was provided. The largest single probable cause of death recorded was cholera, which was a major killer in 1985 and 1987. Maternal mortality was the probable cause of 3 percent of deaths, diarrhea and malaria of 2 percent each, diseases of the lung nearly 5 percent, other infectious diseases nearly 4 percent, and accidents 2 percent.

Civil registration in the city of Saint-Louis has a long tradition of high quality, and deaths by cause have been tabulated for the period 1970-1988. The three largest groups, each contributing over 15 percent of deaths over age 20, are infectious and parasitic diseases, diseases of the circulatory system, and diseases of the digestive system. Accidents account for some 5 percent of deaths among males aged 20 and over, but only about 2.5 percent of female deaths.

Three tentative conclusions can be drawn from the small-scale studies. First, the age pattern of adult mortality in Senegal is fairly well approximated by the Coale-Demeny West family of model life tables. Second,

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

female adult mortality seems to be somewhat higher (relative to the historical differences embodied in the Coale-Demeny model life tables) than male adult mortality, except in Mlomp, where females seem to do slightly better than males. Third, for the areas under study for a sufficient period, adult mortality has fallen substantially through the 1960s, 1970s, and 1980s.

Variations in Adult Mortality by Region

Two types of information allowing the calculation of regional variations in adult mortality are available from the 1988 census: orphanhood data and deaths within the last 12 months. One problem with using orphanhood data for subnational populations is migration: the residence of the respondent (child) is not necessarily the same as the residence of the person exposed to risk (parent). To minimize this problem, special tabulations of survival of parents by region were produced from the census-data, including only those respondents born in the same region as that in which they were resident at the time of the enumeration. These data were then analyzed using the procedures described above.

It has been noted that the time trends in adult mortality indicated by data on survival of parents are unreasonably fast, particularly for males. The same problem affects most of the regions also, being especially marked in areas with initially high mortality. Table 5-26 summarizes the regional estimates of adult mortality derived from information on survival of parents. For adult male mortality, indicators of reference data, Coale-Demeny level, and implied 45q15 have been obtained by averaging the results for two broad age groups—a ''recent" estimate based on respondents aged 10-14 to 25-29, and an "earlier" estimate based on age groups 25-29 to 40-44. The presentation is similar for female adult mortality, except that the two age ranges used are 20-24 to 30-34 for the "recent" estimate, and 35-39 to 45-49 for the "earlier" estimate. Note that although the age ranges differ substantially, the reference dates of the "recent" and "earlier" periods differ by only about 5 years.

Discussion here focuses on the "earlier" estimates because of concerns that the "recent" estimates have been substantially distorted, but not equally for all regions. The female estimates are for mortality in the early 1970s, whereas the male estimates are for the mid-1970s. There are sharp differences in adult mortality among regions: in the two high-mortality regions (Tambacounda and Kolda), the probability of dying between the ages of 15 and 60 is around 50 percent; in the five middle-mortality regions (Diourbel, Fatick, Kaolack, Louga, and Ziguinchor), it is around 35 percent; and in the three low-mortality regions (Dakar, Saint-Louis, and Thiès), it is around 20 percent.

There is a strong correlation between the mortality conditions for males

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

TABLE 5-26 Estimates of Adult Mortality by Region Based on Survival of Parents

 

Recent Period

Earlier Perioda

Parent and Region

Date

Level

45q15

Date

Level

45q15

Fathers (male mortality)

Dakar

80.2

21.8

0.181

76.8

21.3

0.197

Ziguinchor

80.2

20.3

0.225

76.0

17.5

0.301

Diourbel

80.1

19.7

0.241

75.5

15.7

0.349

Saint-Louis

80.4

21.4

0.193

77.0

19.9

0.235

Tambacounda

79.6

15.8

0.346

73.1

9.4

0.525

Kaolack

80.0

19.2

0.254

75.3

15.0

0.367

Thiès

80.3

22.1

0.171

76.7

19.4

0.249

Louga

80.2

21.5

0.190

76.1

18.0

0.287

Fatick

80.0

19.0

0.260

75.3

15.1

0.364

Kolda

79.7

16.9

0.317

73.8

11.0

0.477

Mothers (female mortality)

Dakar

77.9

21.3

0.145

76.0

21.3

0.145

Ziguinchor

77.3

16.3

0.275

73.0

12.7

0.367

Diourbel

77.2

15.7

0.291

72.8

12.6

0.370

Saint-Louis

77.5

18.0

0.231

74.1

16.3

0.275

Tambacounda

76.2

9.6

0.454

70.2

6.6

0.544

Kaolack

77.1

15.3

0.301

72.6

12.3

0.378

Thiès

77.8

20.1

0.176

74.4

19.8

0.184

Louga

77.3

16.7

0.265

73.7

15.0

0.309

Fatick

77.3

16.5

0.270

73.3

14.1

0.332

Kolda

76.3

9.7

0.452

70.1

6.5

0.547

a For male mortality, "recent period" is based on age groups 10-29, "earlier period" on age groups 25-44; for female mortality, "recent period" is based on age groups 20-34, "earlier period" on age groups 35-49.

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

and females. Using the probability of dying between the ages of 15 and 60 as the indicator, the relationship is shown in the top panel of Figure 5-12. The strong correspondence by sex is powerful evidence that the measures are not hopelessly biased. It may be noted, however, from Table 5-26, that the relationship between male and female mortality levels appears to depend on the overall level of mortality. In the two regions with very heavy mortality, females do relatively much worse than males, with Coale-Demeny levels around 6 for females as against around 10 for males. For the five middle-level mortality regions, females do a bit worse than males (female levels around 13, male levels around 16). For the three low-mortality regions the male and female levels are quite similar.

Deaths by age reported in the year before the 1988 census provide another, and much more recent, source of information about adult mortality differences among regions. Table 5-27 shows the regional indicators of adult mortality calculated from the deaths reported in the census. The lower panel of Figure 5-12 plots the resulting female probabilities of dying between the ages of 15 and 60 against those for males by region. There is reasonable agreement between the sex-specific regional estimates, though the ranges of the estimates, for both males and females, are smaller than is the case for the orphanhood-based estimates.

Figure 5-13 compares the orphanhood-based mortality estimates with those based on deaths in 1987-1988. For males, there is no clear relationship, but for females a higher mortality estimate from survival of mother is associated with a higher mortality estimate from deaths in 1987-1988, though the relationship is not very strong. Orphanhood estimates have a broader range from best to worst and may be better measures of regional differentials than deaths in 1987-1988, for which omission might be lowest in low-mortality regions and highest in high-mortality regions, thus narrowing the differentials.

Figure 5-14 compares regional estimates of adult mortality with regional estimates of child mortality, in both cases using deaths in 1987-1988. Both for male and female adult mortality, there is a fairly clear positive relationship between child and adult mortality, though child mortality varies more in relative terms among regions than does adult mortality. These comparisons suggest that the regional adult mortality measures are of reasonably good quality.

Impact of Acquired Immunodeficiency Syndrome in Senegal

The HIV seroprevalence rate is unknown for Senegal as a whole. For four cities—Dakar, Saint-Louis, Kaolack, and Ziguinchor—a surveillance program was started in late 1989. It provides an estimate of the seroprevalence rate among pregnant women, which is a proxy for the rate in the general

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

FIGURE 5-12 Comparison of regional estimates of male and female adult mortality (45q15) using data on survival of parents and data on deaths the year before 1988 census. SOURCE: Unpublished tables, 1988 census.

population for these cities. For the period 1989-1992, the city of Saint-Louis, in the north, had the lowest prevalence rate (0.1 percent). The three other cities had similar rates, all between 1 and 2 percent: Dakar (1.0 percent), Kaolack (1.6 percent), and Ziguinchor (1.7 percent) (Comité National de Prévention du SIDA, 1991, 1993). Both strains of HIV (HIV-1 and HIV-2) are present in Senegal, but HIV-2 infection is more frequent, accounting for 70 percent of infections among pregnant women in Dakar, 81 percent in Kaolack, and 91 percent in Ziguinchor. The prevalence rate among pregnant women seems to have been stable over the period 1989-1992. In rural areas, local studies during the period 1986-1990 show very

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

TABLE 5-27 Regional Estimates of Adult Mortality Calculated from Deaths in the 12 Months Before the 1988 Census

 

Male

Female

Region

e(15)

45q15

e(15)

45q15

Dakar

50.7

0.220

52.8

0.173

Diourbel

45.7

0.346

46.5

0.325

Fatick

44.6

0.380

46.4

0.328

Kaolack

48.3

0.268

48.7

0.273

Kolda

46.2

0.336

44.7

0.374

Louga

49.6

0.244

50.3

0.240

Saint-Louis

46.1

0.348

48.2

0.275

Tambacounda

47.5

0.290

46.6

0.322

Thiès

47.9

0.304

49.5

0.250

Ziguinchor

45.8

0.359

46.2

0.336

low levels of HIV infection among the adult population, between 0 and 1 percent, with a gradient from north to south—the rates being higher in the south (Le Guenno et al., 1992).

Senegal is in the low range of African countries in terms of the AIDS epidemic in the general population in 1989-1992. Data on causes of death are not available, but it is unlikely that the impact of the epidemic on mortality was very high over this period. Even if HIV becomes much more prevalent, the impact on the age structure of the population and the population growth rate are not likely to be very large. (See Stoto [1993] for a discussion of the likely effects of HIV on the population of Africa.)

Socioeconomic Factors Linked to Adult Mortality

Adult mortality estimates have been developed only at the regional level. Consequently, there are only 10—not 30—observations of associations between adult mortality and all the socioeconomic indicators. This shortage of observations makes it more difficult to identify associations between adult mortality outcomes and socioeconomic indicators.

Table 5-28 shows the correlation coefficients between adult mortality and the same socioeconomic indicators discussed above by region. As is the case for child mortality, all of the relationships between adult mortality and socioeconomic indicators are negative. Interestingly, the negative relationships are stronger for female than for male mortality for all of the socioeconomic indicators used. In addition, the education indicators, particularly the current school enrollment rates, show a weaker relationship with adult mortality than do the housing indicators, of which water supply and electricity show the strongest relationships.

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

FIGURE 5-13 Comparison of orphanhood estimates of adult mortality (45q15) and estimates of adult mortality based on deaths in year before 1988 census, males and females, regional data. SOURCE: Unpublished tables, 1988 census.

Table 5-28 also shows the correlation coefficients between the same socioeconomic indicators and adult mortality for all the regions except Dakar. It is clear that the relatively high socioeconomic status of Dakar, coupled with the relatively low adult mortality, accounts for the relatively strong correlations seen in Table 5-28. When Dakar is excluded, all of the correlation coefficients that remain negative weaken. Moreover, several of the associations become positive.

Both mortality and the socioeconomic indicators reviewed above refer to the late 1980s, and can thus be regarded as coincident. Figure 5-15, however, relates the regional change in adult mortality—simply the difference

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

FIGURE 5-14 Comparison of regional estimates of male and female adult mortality with regional estimates of child mortality using data on deaths in year before 1988 census. SOURCE: Unpublished tables, 1988 census.

between the probability of dying between the ages of 15 and 60 in the early 1970s and that in the late 1980s—to the change in the proportion of females aged 15-49 with some education. To the extent that there is a relationship, it appears to be in a counterintuitive direction: regions with large changes in educational stock had small or negative declines in adult mortality, while regions with large changes in adult mortality had small increases in educational stock. Regions with very high mortality in the early 1970s apparently enjoyed large survivorship gains, while regions with low adult mortality experienced only small gains.

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

TABLE 5-28 Correlation Coefficients Between Socioeconomic Indicators and Adult Mortality (45q15)a; All Regions and All Regions Excluding Dakar

 

All Regions

All Regions Excluding Dakar

Indicator

Female

Male

Female

Male

Housing

-0.77

-0.41

-0.49

0.17

Electric

-0.80

-0.61

-0.69

-0.20

Latrine

-0.70

-0.61

-0.15

-0.17

Water

-0.84

-0.57

-0.66

-0.21

Composite

-0.82

-0.57

-0.63

-0.09

Urban

-0.73

-0.52

-0.00

0.32

Ever school

-0.53

-0.28

0.10

0.38

School enrollment rate (male)

-0.35

-0.10

0.08

0.40

School enrollment rate (female)

-0.37

-0.10

-0.31

0.10

NOTE: See Chapter 2 for details on how the socioeconomic indicators were constructed.

a These coefficients are calculated using data from the regional level, n = 10.

SOURCE: 1988 census (unpublished tabulations)

FIGURE 5-15 Relationship between change in male and female survivorship and change in the proportion of females with some education, ca. 1970-1988. SOURCE: Unpublished tables, 1988 census.

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

Summary and Conclusions

Table 5-25 summarizes the national adult mortality estimates regarded as being of acceptable quality, together with estimates from local studies. A variety of methods applied to a variety of data give rise to a puzzling variety of estimates of adult mortality in Senegal. Intercensal survival methods applied to the 1960, 1970, 1976, and 1988 surveys and censuses lead to very variable estimates of mortality. The estimates appear to have been affected by changes in population coverage from 1970-1976, and again from 1976-1988, the latter change being estimated on the order of 20 percent. Deaths in the year before the 1988 census give rise to mortality patterns that are highly consistent with model patterns, indicating an expectation of life at age 15 of about 50 years for males and 51.5 years for females, corresponding to probabilities of dying between ages 15 and 60 of about 30 percent and about 27 percent, respectively. Estimates of adult mortality based on survival of parents show sharply, indeed implausibly sharply, falling mortality over time, particularly for males. However, the parental survival estimates of mortality for the early 1970s seem quite reasonable. Female expectation of life at age 15 was probably about 47 years (probability of dying between ages 15 and 60 about 37 percent), whereas male expectation of life at age 15 around 1976 was probably about 48.3 years (probability of dying between ages 15 and 60 about 36 percent).

National data sources indicate a decline in adult mortality from the early 1970s to the late 1980s, with expectation of life at age 15 increasing by about 2 years for males between the mid-1970s and the late 1980s, and by about 4 years for females over the slightly longer period from the early 1970s to the late 1980s. The Niakhar/Ngayokhème study suggests a rather similar rate of increase in e(15) of around 2.5 years from the early 1970s to the late 1980s. Bandafassi data indicate a more rapid reduction of mortality, with e(20)—not very different from e(15)—rising by 7 years from the early 1970s to the late 1980s, though from substantially higher initial mortality levels.

Mortality declines have not, however, resulted in homogenous levels of adult mortality throughout Senegal. In fact, orphanhood measures, which may give the best indication of regional differences, show that these differences are sharp. Adult mortality is highest in the southeastern part of the country, where the probability of dying between ages 15 and 60 is around 50 percent, and lowest in the western part of the country, where it is about 20 percent.

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

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.

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

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

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

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.

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

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

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

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.

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

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

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

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

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

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).

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

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

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

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.

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×

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.

Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 113
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 114
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 115
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 116
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 117
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 118
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 119
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 120
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 121
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 122
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 123
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 124
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 125
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 126
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 127
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 128
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 129
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 130
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 131
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 132
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 133
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 134
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 135
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 136
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 137
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 138
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 139
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 140
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 141
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 142
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 143
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 144
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 145
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 146
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 147
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 148
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 149
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 150
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 151
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 152
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 153
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 154
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 155
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 156
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 157
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 158
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 159
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 160
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 161
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 162
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 163
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 164
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 165
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 166
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 167
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 168
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 169
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 170
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 171
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 172
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 173
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 174
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 175
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 176
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 177
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 178
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 179
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 180
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 181
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 182
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 183
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 184
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 185
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 186
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 187
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 188
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 189
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 190
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 191
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 192
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 193
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 194
Suggested Citation:"5 Mortality." National Research Council. 1995. Population Dynamics of Senegal. Washington, DC: The National Academies Press. doi: 10.17226/4900.
×
Page 195
Next: 6 Conclusions »
Population Dynamics of Senegal Get This Book
×
Buy Paperback | $55.00 Buy Ebook | $43.99
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

This volume, the last in the series Population Dynamics of Sub-Saharan Africa, examines key demographic changes in Senegal over the past several decades. It analyzes the changes in fertility and their causes, with comparisons to other sub-Saharan countries. It also analyzes the causes and patterns of declines in mortality, focusing particularly on rural and urban differences.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    Switch between the Original Pages, where you can read the report as it appeared in print, and Text Pages for the web version, where you can highlight and search the text.

    « Back Next »
  6. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  7. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  8. ×

    View our suggested citation for this chapter.

    « Back Next »
  9. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!