1Since the classic study by Jean Meuvret published in Population in 1946, historical economists and demographers have regularly sought to further our understanding of past upheavals. By studying the city of Antananarivo over the period 1976-2000, Dominique WALTISPERGER and France MESLÉ show how demographic tools can also be used to accurately measure the effects of contemporary crises. They identify the most hard-hit population categories and the most prevalent diseases during the 1986 crisis which, at the height of the food shortage, reduced life expectancy at birth by 13 years for males and 8 years for females compared with 1976.

2This in-depth analysis sheds light on the ways in which food shortage leads to health crisis, and helps to identify risks for the future.

3In Africa, information on mortality by cause is very seldom available. As a rule, the data are highly fragmentary, so that there are no statistics on causes of death at the national or even sub-national level. Long-term series are available only for small populations monitored by demographic surveillance sites (INDEPTH, 2002). Some isolated studies have been conducted in cities such as Bamako, Dakar and Saint-Louis du Sénégal, but they cover short periods of time (Diop, 1990; Fargues and Nassour, 1988) or focus exclusively on infant mortality (Cantrelle et al., 1986).

4There is also a dearth of studies concerning the impact of economic crises on entire populations. They are very often confined to the effects on infant and child mortality (Brunet-Jailly, 1996; Barbieri and Vallin, 1996), and almost never examine the causes of death.

5Thanks to a systematic analysis of deaths registered at the Municipal Hygiene Office (Bureau Municipal d’Hygiène: BMH) of Antananarivo, the capital of Madagascar, a long-term series on cause-specific mortality is available for that city. A book published in 1998 offered a full discussion of the data collected and reported an initial analysis for the 1984-1995 period (Waltisperger et al., 1998). It has since been possible to fill in the series for earlier years back to 1976, and to expand the database regularly with the new years available. This very rich data source can now be used to analyse cause-specific mortality trends in Antananarivo over the 25-year period 1976-2000. In particular, it provides new information on the health impact of the economic crisis, which was exceptionally acute in Madagascar in the 1980s.

6The results reported here cannot, of course, be extrapolated to the entire island. The economic and health conditions in Tananarive—today renamed Antananarivo in the Hova language—are unquestionably more favourable than those of the rest of the country. The city is Madagascar’s administrative, political, economic, and cultural capital. It is situated on the island’s central highlands (altitude: 1,250-1,450 meters), in the territory of the largest Malagasy ethnic group: the Merina. Tananarive means “city of the thousand,” a reference to the thousand warriors who captured the locality in the seventeenth century under the orders of King Andrianjaka. Occupying a rectangle of almost 80 km², it is built on a range of hills forming the upper town, overlooking the valleys crossed by two rivers and the marshlands of the lower town. Today, the capital has a population of one million and a density of 12,000 inhabitants per km²—a figure in stark contrast with the national average of 30 inhabitants per km². However, its population growth is relatively moderate by comparison with the demographic explosions in some African capitals (Antoine et al., 2000). The economic crisis has not spared Antananarivo and has probably curbed its expansion. In this respect, the analysis of the city’s mortality data since the mid-1970s is of great interest.

7An initial study (Garenne et al., 2002), centred on the famine years, provided an estimate of excess mortality due to the 1985-1987 famine by comparing expected deaths with those actually recorded. In this article, we look at a longer period and discuss changes by age group and detailed cause. We try to distinguish the consequences of the economic crisis from the effects more clearly attributable to the ongoing health transition.

8After a brief chronology of the main events in the political and economic history of Madagascar since independence, we present the data used before analysing the main trends in age-specific mortality in the past 25 years. In the final section, we examine causes of death in order to identify the main factors in the health crisis induced by the economic crisis; this will enable us to describe Madagascar’s position in the health-transition process.

I – A highly unstable political and economic situation

9Since 1960, when Madagascar became independent from France, its social and political history has been characterized by extreme instability, with severe consequences on living conditions. The table in Appendix I lists the main political events in the country’s history during the past several decades.

10This turbulent history has resulted in the country’s steady impoverishment. Apart from a few years’ respite, GDP per capita has consistently declined since 1970, shrinking more than 30% in 25 years. Most dramatically, it plunged 45% between 1971 and 1996, a fall unmatched in any other country in peacetime (Figure 1). Since 1995, the situation has improved somewhat, but output per capita remains very low.

Figure 1

Real gross domestic product (GDP) in Madagascar since 1960

Real gross domestic product (GDP) in Madagascar since 1960

11This impoverishment is partly due to the growth gap between the population and food production: while the population nearly doubled (+92%) from 1960 to 2000, food production rose only 38%. It is largely the result of political choices that proved to be economically disastrous.

12At independence, the new regime introduced an agricultural policy aimed at increasing the area under rice cultivation and its yield. Rice production rose 48% between 1961 and 1968. After 1972, first under the transitional military regime, then after Didier Ratsiraka seized power in 1975, rice processing and trading were gradually nationalized. Between 1975 and 1982, the price paid to producers fell 25%, while most investment was channelled into manufacturing. This policy progressively discouraged producers and aggravated the rice shortage. By 1982, the situation forced the government to appeal to lenders and implement an initial structural adjustment programme, paving the way for the liberalization of agriculture. At the time, two systems functioned side by side: the State-controlled system, which set the selling price of rice (220 Malagasy francs per kilogram in early 1985), and a parallel market where prices were far higher (over 400 Malagasy francs in August 1985). Also in 1985, controls on producer prices were maintained whereas those on consumer prices were scrapped. The rice stocks collected by the State were depleted by September, and the market was left to the private sector, which had driven prices above 500 Malagasy francs per kg by year-end (Figure 2).

Figure 2

Rice prices in Madagascar (Malagasy francs per kg)

Rice prices in Madagascar (Malagasy francs per kg)

13The acute food shortage and the concomitant sale of spoiled products peaked in 1986. That year marked the start of a nationalization program and a reorganization of the production system. A buffer stock was set up to dampen the impact of cyclical shortages. The State’s total withdrawal from commercial and financial activities from 1996 on, and a series of incentives to producers (such as arrangements for seed production and a credit system), led to a modest revival of domestic production. The exemption from import taxes on foodstuffs in the event of shortage improved food security and curtailed speculation. The final outcome, however, was very negative. Between 1961 and 1995, food consumption per capita fell 34% in volume terms (Ravelosoa and Roubaud, 1996). In 1993, the SEECALINE study based on the household survey (INSTAT, 1995) showed that two-thirds of the population failed to meet its daily calorie requirement. In 1999, a new round of the same survey (INSTAT, 2000) revealed that households continued to spend more than 70% of their income on food. Rice production per capita has been steadily declining for 30 years, but there was no unusually steep fall in 1986. The shortage that year was due less to a production shortfall than to the disorganized transportation and distribution systems and to speculation (Régnard, 2003).

14The capital did not escape this food crisis, as demonstrated by two phenomena: first, the 20% decline in per capita rice consumption between 1982-1983 and 1986-1987; second, the increased reliance on the parallel market for rice supplies when the shortage was at its peak (Table 1).

Table 1

Rice consumption in Antananarivo by supply source in 1982-1983 and 1986-1987 (kg per capita)

Source Period 1982-1983 1986-1987 Official market 132 64 Parallel market 10 51 Own-account production 16 11 Total 157 126 Source: World Bank, 1989.

Rice consumption in Antananarivo by supply source in 1982-1983 and 1986-1987 (kg per capita)

15In 1995, the country suffered another crisis because of a sharp fall in rice production after Hurricane Géralda destroyed part of the crop. The consequences, which are better known (Ravelosoa and Roubaud, 1996), bear no comparison with those of the 1980s crisis. The 1995 fall in rice production due to bad weather caused a 5% decline in consumption compared with earlier years. In 1986-1987, consumption in Antananarivo had fallen 20% from its 1982-1983 level (World Bank, 1989). Moreover, the very limited crisis of 1995 was followed by a significant improvement in living conditions for the capital’s households, whose average income gained 53% between 1995 and 2001 (Razafindrakoto and Roubaud, 2002).

II – A rich source of data on more than 150,000 deaths

16In the absence of any continuous statistical tracking of the population, the effects of these severe crises on health and mortality are still not well known. Despite a tradition of vital-statistics registration going back to the early twentieth century, the statistical processing of the information gathered remains inadequate for producing reliable annual demographic indicators with nationwide coverage. Only the censuses and some surveys offer benchmarks for monitoring the main demographic trends (Razafimanjato et al., 2001). Despite routine registration of deaths and their causes at the Municipal Hygiene Offices (Bureau Municipal d’Hygiène, BMH), there is unfortunately no centralization or statistical processing of the data collected. Thanks to a research project launched in 1993 with support from UNICEF and assistance from INED, the registers of the Antananarivo BMH [1] since 1976 have been tabulated. Data series on deaths by sex, age, and cause are now available for the period 1976-2000, and can be used for refined analysis of the impact of the economic crisis on mortality in the capital.

17BMHs began registering deaths in 1921, the year of the last major plague epidemic. In Antananarivo, the BMH, which to the Municipal Directorate of Social Affairs, is responsible for preparing death certificates from reports filled by physicians (of the BMH itself or of the hospital, depending on the place of death). On presentation of this document, the civil registry office delivers the burial permit. The capital’s cemeteries are guarded and clandestine burials are practically impossible, ensuring that nearly all deaths are recorded.

18Since 1973, all registers have used the same format, defined by Dr. Randrianarivo, then Director of the Antananarivo-ville BMH. For each death, 14 information items are recorded:

• death registration number;
• birth date of the deceased;
• sex of the deceased;
• whether the deceased resided or not in Antananarivo;
• fokontany (neighbourhood) of residence of the deceased if (s)he resided in Antananarivo;
• fivondronana (sector) of residence of the deceased if (s)he did not reside in Antananarivo;
• date of death;
• hour of death;
• date on which death was reported to the BMH;
• place of death;
• main cause of death;
• associated cause of death (or external cause, if an accident);
• occupation of the deceased or his/her parents;
• kinship tie of reporting person to the deceased.

1 – Quality of data on cause-specific deaths

19The comparison of infant and child mortality computed from deaths registered at the BMH with those obtained from data from two Demographic and Health Surveys (DHS) conducted in 1992 and 1997 shows a very good coverage of deaths by the BMH (Garenne et al., 2002).

20In Madagascar, deaths must be reported within twelve days. This time limit is respected in more than 99% of cases, and over 80% of declarations are filled on the day of death or the day after. Compliance with the time limit is a well-established habit of Antananarivo residents and was not weakened by the political-economic crisis. The percentage of deaths reported within 12 days is very close to 100% for the entire period studied here (Figure 3).

Figure 3

Variations in death-reporting times, Antananarivo, 1976-2000

Variations in death-reporting times, Antananarivo, 1976-2000

21By contrast, the accuracy of declarations of age at death has distinctly improved in 25 years (Figure 4): the age was known to within a day in slightly over half the cases at the start of the period, compared with more than 80% at the end of the period. However, accuracy did not improve continuously throughout the period. Between 1980 and 1984, economic and political problems actually caused the quality to deteriorate: fewer than 40% of ages at death were known to within a day in 1981. But after, even when the food shortage was at its worst, the accuracy of declarations of age at death continued to improve.

Figure 4

Accuracy of age-at-death declarations, Antananarivo, 1976-2000

Accuracy of age-at-death declarations, Antananarivo, 1976-2000

22Accuracy varies substantially by age (Figure 5). The older the age, the less accurately it is known. The imprecision is mainly due to more inaccurate declarations of exact dates of birth of elderly persons. By contrast, the more recent the birth cohort, the more accurately the date of birth is declared, which explains most of the progress observed in the accuracy of ages declared at death. However, the decline in accuracy in the early 1980s concerns the youngest and oldest deceased alike, indicating a less effective registration system. The start of that decade saw a severe economic downturn and a record debt, which led to the implementation of the first structural adjustment plan. The plan, in turn, caused drastic cuts in public spending, presumably with adverse effects on the death registration system.

Figure 5

Proportion by age group of deaths for which the age is known to within a day, Antananarivo, 1976-2000

Proportion by age group of deaths for which the age is known to within a day, Antananarivo, 1976-2000

23The quality of cause-of-death registration displayed a more uneven profile. Unknown and ill-defined causes (codes 780 to 799 in the 9th revision of the International Classification of Diseases, ICD9) accounted for 13%-15% of causes of death until the early 1980s, then fell by half for a few years [2] (Figure 6). The excess-mortality period of the mid-1980s was also characterized by a lesser diagnostic accuracy. In those years, the proportion of deaths at home rose significantly to 65% of the total as against 58% in 1983-1985 (the same percentage as in the years 1989-1991). The crisis probably disrupted the functioning of medical services.

Figure 6

Proportion of deaths due to ill-defined or unknown causes, Antananarivo, 1976-2000

Proportion of deaths due to ill-defined or unknown causes, Antananarivo, 1976-2000

24In the early 1990s, the share of deaths due to ill-defined causes fell back to its 1985 level. It has since risen steadily. This growth is partly due to the greater proportion of deaths of persons aged 50 and over in total deaths. It is at the oldest ages that the share of deaths due to ill-defined causes (including senility) is the largest.

2 – Population estimates

25Estimating reference populations is more problematic. Two censuses were performed during the period of study: the first in 1975, the second in 1993 (INSRE, 1980; INSTAT, 1997). We have no population estimate for the interval. In the absence of data on births and migrations, we computed annual populations by interpolating between 1975 and 1993 the enumerations by sex and age group and extrapolating the numbers to 1994-2000 on the assumption that the growth rate remains unchanged. This approach prevents us from taking into account specific migration flows of which many have been particularly large during the most difficult years. We have no information for determining their direction or estimating their volume. However, in the mid-1980s, the degree of disorganization made travel extremely difficult, and massive flows of persons into or out of the capital are unlikely to have occurred. Moreover, if they had taken place, they would have been concentrated in specific age groups—probably younger persons of working age—and under-estimation of flows would have produced significant bias in the age-specific mortality curves. But this is not the case: the curves of annual changes in age-specific probabilities of dying do reveal some random irregularities but in no way indicate a consistent over- or under-estimation of a particular age group. Also, as we shall see later, the excess mortality observed at certain ages during the crisis concerns only a limited number of causes of death, whereas a consistent bias in the estimates of population by age would have produced a peak of comparable magnitude for all causes.

26Lastly, we observe a very good consistency between our estimates and the levels of infant and child mortality obtained from the 1992 and 1997 Demographic and Health Surveys [3] (Figure 7).

Figure. 7

Comparison of infant and child mortality (5q0) in Antananarivo from different sources

Comparison of infant and child mortality (5q0) in Antananarivo from different sources

27In sum, the fact that a specific timing of migration flows could not be taken into account might explain the slight irregularities that appear in the mortality rates, particularly at adult ages, but does not call into question the broader trends observed here.

28Before determining life expectancies, however, we smoothed the observed irregularities as follows. We compared each raw life table with the Princeton model tables (Coale and Demeny, 1983) and the United Nations model tables (1982). The age distribution of mortality over age 10 observed in Antananarivo proved to be very close to the UN general model. We smoothed each raw table with the aid of the model table giving the same life expectancy at 10 years (we performed no adjustment for mortality before age 10 [4]). Our smoothing method is based on a linear regression applied to survivors (logit(lx)) aged 15 years or more. The correlation (R²) between the raw data and the smoothed data consistently exceeds 0.99, which underscores both the good regularity of the raw data and the closeness of the mortality distribution in the capital to that of the chosen model.

29Next, we calculated cause-specific mortality rates by applying the distribution of different causes of death to the adjusted all-causes mortality rates for each age group.

III – Crisis, shortage, and mortality

30The change in life expectancy computed from these annual tables is shown in Figure 8. From 1976 to 1986, the mean length of life decreased substantially—more so among men, who lost more than 13 years of life expectancy (from 58.1 to 44.7 years), while women lost nearly 8 (from 60.9 to 53.0 years). Men experienced the steepest loss between 1984 and 1986, at the peak of the economic crisis, with a fall of almost 6 years in two calendar years. After these particularly damaging years, life expectancy started rising again, rather briskly in the late 1980s, then at a slower pace in the 1990s. As a result, by 2000, men had merely returned to their 1976 level while women had posted a 2.4-year gain over the 25-year period. For the post-1993 period, we estimated the age-specific population that serves as denominator in computing mortality rates on the assumption that the population would continue to grow at the same rate as in the 1975-1993 intercensal period. If the actual growth rate since 1993 exceeds the one we chose, our computation underestimates life expectancy in the most recent years. Accordingly, a doubling of the growth rates adopted since 1993 would lengthen life expectancy at birth by about 2.5 years by 2000. However, this potential underestimation does not invalidate the observed trend; if, on the contrary, we have overestimated growth, this means that the reality is even worse.

Figure 8

Life expectancy at birth in Antananarivo, 1976-2000

Life expectancy at birth in Antananarivo, 1976-2000

31The increase in mortality between the mid-1970s and the mid-1980s affected all age groups but to varying degrees (Figures 9 and 10). For both males and females, paradoxically, the age groups least affected were those at the two extremes of life. Infant mortality rose by “only” 10% for boys and 15% for girls, while mortality at age 70 rose 26% among men and 8% among women. Children aged 1-15 paid the heaviest price. Mortality between ages 5 and 10 was multiplied by 2.6 for boys and 2.9 for girls between 1976-1978 and 1984-1988. In 1986, the worst year of the crisis, male mortality at these ages was even 4.5 times as high as in 1976, and female mortality 3.5 as high. After this major peak, male excess mortality in 1984-1988 stayed at about 2 compared with 1976-1978 until age 35-40 before declining gradually with age. Among females, adult excess mortality was less pronounced, at close to 1.5 until age 30-35.

Figure 9

Comparison of age-group-specific probabilities of dying in Antananarivo for three periods: 1976-1978, 1984-1988, and 1997-2000

Comparison of age-group-specific probabilities of dying in Antananarivo for three periods: 1976-1978, 1984-1988, and 1997-2000

Figure 10

Ratio of age-group-specific probabilities of dying for 1984-1988 and 1997-2000 to the probabilities for 1976-1978, Antananarivo

Ratio of age-group-specific probabilities of dying for 1984-1988 and 1997-2000 to the probabilities for 1976-1978, Antananarivo

32From 1984-1988 to 1997-2000, the mortality rate started to improve again for all age groups but, again, with uneven effects. Sizable gains among the under-15s took child and youth mortality to well below its 1976-1978 levels. By contrast, over age 15, female mortality barely returned to the levels prevailing 20 years earlier, whereas male mortality in recent years has been running even higher than in the late 1970s.

33These highly contrasting patterns obviously reflect the serious political and economic upheavals experienced by Madagascar, and in particular its capital, in the past two decades. They were also influenced by the health programmes implemented during the same period. Very young children, still at least partly breast-fed by their mothers, were the prime target of the action programmes: they suffered less than children a few years older. The specific sensitivity of young men to the crisis by comparison with that of young women is fairly surprising. Analysis of changes in the causes of death should provide some explanations.

IV – Poverty-related causes of death

34After analysing the BMH registers, we coded the main causes of death according to the detailed list of the 9th revision of the International Classification of Diseases (WHO, 1977) and consolidated them for this study into 40 groups of causes (Appendix II). We distributed unknown or ill-defined causes of death proportionally by sex and age group among the well-defined causes in order to ensure series comparability over time [5].

1 – The predominant role of infectious diseases and nutritional deficiencies

35We began by consolidating the 40 groups of causes into 7 categories to analyse the contribution of these larger categories to the change in life expectancy [6] during the two periods already examined in the previous section: (1) from 1976-1978 to 1984-1988, when life expectancy declined (by 9 years for men and 6 years for women); (2) from 1984-1988 to 1997-2000, when life expectancy recovered, gaining 10 years for men and 9 years for women (Figure 11).

Figure 11a

Age-group-specific contribution of main causes of death to changes in male life expectancy in Antananarivo (1) between 1976-78 and 1984-88 and (2) between 1984-88 and 1997-2000

Age-group-specific contribution of main causes of death to changes in male life expectancy in Antananarivo (1) between 1976-78 and 1984-88 and (2) between 1984-88 and 1997-2000

Figure 11b

Age-group-specific contribution of main causes of death to changes in female life expectancy in Antananarivo (1) between 1976-78 and 1984-88 and (2) between 1984-88 and 1997-2000

Age-group-specific contribution of main causes of death to changes in female life expectancy in Antananarivo (1) between 1976-78 and 1984-88 and (2) between 1984-88 and 1997-2000

Between 1976-1978 and 1984-1988, two groups of causes of death contributed massively to the decline in life expectancy: infectious diseases [7] and nutritional deficiencies. The first lowered male life expectancy by 3.5 years and female life expectancy by 2.7 years; the second reduced it by 2.5 years and 1.8 years respectively. The rise in mortality due to these two causes of death alone explains more than 70% of the fall in male life expectancy and 80% of the fall in female life expectancy. Among males, the increase in mortality due to diseases of the respiratory system also played a visible role, driving life expectancy down by 1.5 years. For both sexes, the contribution of the rise in mortality due to nutritional deficiencies and infectious diseases was greatest at early ages, particularly 1-4 years. The adverse change in mortality at those ages was alone responsible for a fall in life expectancy of 2 years among males and 2.2 years among females. At adult ages, the impact of the crisis is less clearcut among women. By contrast, it is very perceptible among men, who—like the youngest children—were the main victims of a rise in mortality due to infectious diseases and nutritional deficiencies.

36The situation was totally reversed in the second period, with a decline in mortality at all ages and for all causes, among both males and females. Children aged 1-4, who had suffered the most from the crisis, also registered the greatest gains, with a decline in mortality that extended life expectancy by 3.7 years for males and 4 years for females. This advance was mainly due to the decrease in mortality from nutritional deficiencies and infectious diseases, but it was boosted by the fall in mortality from respiratory diseases.

37Infant mortality also retreated significantly thanks to the decline in the same three causes, plus that of “other diseases”—a group in which perinatal diseases and congenital anomalies are the largest components at that age. The under-5 age group did not merely recover after the severe crisis of the mid-1980s; it also apparently benefited from the action programmes implemented in the previous two decades, most of which were aimed at combating infectious diseases. This is especially true for children below age one who had suffered relatively less than their elders from malnutrition and were the prime beneficiaries of health programmes (Appendix III).

38By contrast, over 5 years of age, the gains were very limited and barely offset the losses recorded in the previous period. Adolescents and adults in Antananarivo barely succeeded in returning to the mortality levels prevailing 20 years earlier, but there is still no sign of more fundamental health progress.

2 – Variable sensitivity of causes of death to the economic situation

39The annual change in standardized mortality rates [8] for these broad groups of causes provides a means to track more closely the relationships between health status and political-economic conditions. Figure 12 clearly illustrates the different patterns of the two main causes of life-expectancy decline: infectious diseases and nutritional deficiencies.

40Infectious diseases are by far the chief cause of male deaths. At the start of the period, they ranked on a par with cardiovascular diseases and “other diseases”. True, they were overtaken by cardiovascular diseases as the number-one cause at the end of the period. But it is their growth from the mid-1970s to the mid-1980s that led to a decline in life expectancy, and their subsequent steep fall was the main driver of its upturn. For females, the pattern is identical but at a lesser level, as their infectious mortality has been distinctly lower than their cardiovascular mortality since the mid-1990s. As regards infectious diseases, the crisis sharply aggravated this cause of death, which already played and still plays an important role in mortality. The same is true of diseases of the respiratory system, which moved on a very similar path, but at a somewhat more subdued level. This group of diseases essentially consists of infectious diseases (acute bronchitis, pneumonia, influenza); hence it is not surprising to find them following the same pattern as the main group of infectious diseases. For nutritional deficiencies, the process is different. We observe the sudden appearance of a cause of death that was practically non-existent in the early 1970s and then surged to become—at the peak of the crisis in 1986—the second cause of death among males and the third among females. This acute crisis-related mortality rapidly subsided, stagnated in the early 1990s, and has become negligible again in more recent years.

41A fourth cause of death seems highly crisis-sensitive for males: violent deaths, which peaked sharply in 1986. The only available classification of violent deaths is by type (fracture, poisoning, etc.). Unfortunately, it does not always indicate the cause (accident, suicide, homicide), and this rules out a precise analysis of the causes of the 1986 surge. The partial statistics available show an above-average number of traffic accidents that year. In a crisis period, spending on non-essential (i.e., non-food) items is deferred. This is particularly the case with vehicle maintenance expenditures (brake linings and brake fluid for example), whether for private or public transportation.

Figure 12

Standardized mortality rates for main causes of death in Antananarivo, 1976-2000

Standardized mortality rates for main causes of death in Antananarivo, 1976-2000

42The other causes of death were minimally affected by the 1986 crisis. We note the very low level of neoplasm-related mortality in Antananarivo. The small share of cancers in mortality compared with the massive weight of infectious diseases clearly shows—even setting aside the crisis—that Madagascar is still a long way from completing the first phase of its health transition. We shall return to this point in our conclusion. However, before considering the longer-term scenario for mortality in Antananarivo, we will look more closely at the effects of the major crisis of the mid-1980s. First, we will examine the most affected group, children aged 5-9; we will then focus on the determinants of young adult male mortality.

3 – Two particularly hard hit age groups

Children aged 5-9: crisis-related excess mortality over a short period

43As a rule, mortality is lowest between ages 5 and 12, so the mortality increase that struck children aged 5-9 in particular is especially striking (Figure 10). The group most severely affected was small girls, whose causes of death deserved more detailed analysis (Figure 13). As we already saw in the case of all-age standardized mortality, the rise in mortality from 1976 to 1986 was initially due to the increase in infectious diseases (intestinal and respiratory); the number of deaths caused by these diseases rose fairly steadily until 1985 then abruptly in 1986. That same year, the sharp peak in deaths due to nutritional deficiencies explains most of the crisis-related upswing in mortality. The synergy between malnutrition and infectious diseases is now well known (Latham, 1997). Malnutrition aggravates infectious diseases by slowing the production of antibodies or impairing immune responses because of protein deficiency. Also, some situations increase the virulence of certain viruses. One example is measles, which would seem to be more dangerous when transmitted by a collateral relative than by a non-relative (Aaby, 1989). But it is hard to separate the influence of individual factors, as overcrowding is not unrelated to household income and therefore to children’s nutritional status. Not only does malnutrition increase the risk of infection but it sometimes complicates the diagnosis. This is the case, for example, with acute respiratory infections, whose clinical signs (crepitation) are harder to observe in undernourished children. Conversely, infections aggravate malnutrition. For instance, measles triggers vitamin-A deficiency; upper respiratory tract infections and gastro-intestinal infections cause iron loss; and diarrhoea is often followed by dehydration. The same is true of malarial anaemia.

Figure 13

Female mortality rates at ages 5-9 for main causes of death in Antananarivo, 1976-2000

Female mortality rates at ages 5-9 for main causes of death in Antananarivo, 1976-2000

In Antananarivo, the poor nutritional status of children initially favoured the spread of respiratory diseases, and later that of measles. These conditions then contributed to the mortality rise caused by nutritional deficiency and diarrhoea. The population’s resulting state of exhaustion offered an even more propitious breeding-ground for measles, the last epidemic of which broke out in 1988. Since then, the disease appears to have been brought under control, thanks to systematic measles vaccination in the city.

44Another cause of death registered a significant peak: malaria. Because of its timing (1988), it lagged behind the one caused by the food shortage, but its resurgence was not unrelated to the socio-economic situation. By 1962, WHO specialists considered the infection to have been eradicated in the highlands (Hautes Terres). This led to a gradual discontinuation of home insecticide spraying, slacker epidemic surveillance, and an end to chloroquine distribution. When the disease reappeared in 1985-1986, clinics were out of anti-malaria products and the population’s real income had fallen so low that the drugs had become, in any event, unaffordable. The disease peaked in 1988 before the authorities decided on “emergency” measures. In 1988, a new anti-malaria programme was introduced: chloroquine was made available to the population through non-medical and informal channels such as schools and food stores, while DDT spraying in dwellings began on a widespread basis. These treatments sharply reduced the number of parasite carriers (Mouchet et al., 1997). However, spraying was interrupted in 1997 despite the fact that malaria was far from eradicated. Malaria-related mortality has even been rising again in recent years. This increase may be due to different factors: first, the renovation of hydraulic infrastructure in the Betsimitatatra plain surrounding the city from south to north-west has created large tracts of water, an environment conducive to proliferation of the anopheles mosquito; second, the increasing flows of city-dwellers toward malaria-infested areas and the lesser efficiency of chloroquine have probably hastened the spread of the disease in the city.

45Overall, however—apart from malaria, whose spreads represents a danger for the future—the recent pattern in mortality at ages 5-9 years is rather positive, raising hopes that the 1980s mortality peak was only an accidental deviation from an overall downtrend.

Ages 10-49: the weight of infectious diseases among young adults

46The situation is more worrying among adolescents and young adults, particularly males (Figure 14).

Figure 14

Standardized male mortality rates at ages 10-49 for main causes of death in Antananarivo, 1976-2000

Standardized male mortality rates at ages 10-49 for main causes of death in Antananarivo, 1976-2000

47At these ages, the crisis was very violent and heavily concentrated in the years 1985-1987, with a substantial rise in mortality due to infectious and respiratory diseases, nutritional deficiencies, and violent deaths. As among younger persons, intestinal infections—most notably diarrhoea due to food supply and nutrition problems—caused a steep upsurge in mortality in 1986, and the sharp increase in malaria in 1988 took a heavy death toll (Figure 15). This age group also displays significant mortality due to tuberculosis, which rose in the 1970s and 1980s to a peak in 1984-1986 before declining. However, its stagnation since 1990 suggests that it has not been effectively curbed. Furthermore, some of the deaths attributed to tuberculosis may actually be due to AIDS. But this probably concerns only a small number of cases, as HIV prevalence in Madagascar is under 2%.

Figure 15

Standardized male mortality rates at ages 10-49 for main infectious diseases in Antananarivo, 1976-2000

Standardized male mortality rates at ages 10-49 for main infectious diseases in Antananarivo, 1976-2000

48The other causes of death have been far less sensitive to political and economic disorder. However, the recent pattern in some of them could become problematic. First, cardiovascular mortality, after a significant decline immediately after the crisis, has started rising again, as in the 1970s. Second, deaths due to mental disorders—which in this instance mainly consist of alcoholism-related disorders—have been trending up sharply in recent years. Third, violent deaths also rose steeply during the crisis. While infectious diseases are still very prevalent at these ages, the data also indicate a concurrent rise in chronic diseases such as cardiovascular diseases and in social diseases such as alcoholism. In the longer run, there is a risk of a renewed upturn in mortality if preventive and curative measures are not implemented swiftly. The insidious uptrend in these conditions is less striking than the sharp mortality peaks, but it should not be overlooked.

Conclusion

49The inhabitants of Madagascar—and particularly those of Antananarivo, for whom data on mortality and its causes since 1976 are available— suffered the full brunt of the economic crisis. At its peak in 1986, it was accompanied by a severe food shortage that triggered excess mortality due to malnutrition and to infectious diseases. These phenomena were most acute among children aged 1-10 and adult males. If we compare the distribution of mortality by cause observed in 1976-1978 with that of 1984-1988, we are struck by the large role played by infectious diseases, respiratory diseases, and nutritional deficiencies in mortality during the crisis years (Table 2). Because of the worsening economic situation, the island became vulnerable, once again, to famine and epidemics (Omran, 1971). Despite the fragility of economic and political improvements, the effects of the crisis on the population’s health status have been partly erased: life expectancy has recovered, recently returning to its mid-1970s levels. While the levels are comparable, the distribution of causes of death changed notably between 1976-1978 and 1997-2000 (Table 2), with a decrease in the weight of acute and infant diseases (congenital and perinatal conditions, infectious and respiratory diseases) and a rise in that of chronic diseases —particularly cardiovascular diseases—and social diseases (violent deaths). Despite the socio-economic disruptions, Madagascar continues its health transition (Meslé and Vallin, 2002). However, its future in health terms remains very uncertain. The current economic and political situation may prevent the country from consolidating the decline in infectious mortality, which is still threatened by the resurgence of infections previously thought to have been eradicated (such as malaria and the plague), and from slowing the rise in new pathologies. On the contrary, as long as poverty persists, it is probable that no lasting decline in mortality can be achieved.

Table 2

Level and structure of mortality by broad groups of causes in Antananarivo in 1976-1978, 1984-1988, and 1997-2000

Causes of death Rate (per 1,000) Proportion (%) 1976-1978 1984-1988 1997-2000 1976-1978 1984-1988 1997-2000 Males Congenital and perinatal conditions 1.14 0.96 0.58 12.3 6.9 6.8 Infectious diseases 2.13 4.03 1.68 22.9 28.9 19.6 Diseases of respiratory system 1.11 1.98 1.03 11.9 14.2 12.0 Nutritional deficiencies 0.40 1.80 0.36 4.3 12.9 4.2 Diseases of nervous system 0.36 0.44 0.54 3.9 3.2 6.3 Diseases of digestive system 0.62 0.64 0.54 6.7 4.6 6.3 Neoplasms 0.51 0.35 0.38 5.5 2.5 4.4 Cardiovascular diseasess 1.70 2.21 2.26 18.3 15.9 26.3 Other diseases 0.66 0.70 0.48 7.1 5.0 5.6 Injury and accidents 0.66 0.83 0.73 7.1 6.0 8.5 Total 9.29 13.93 8.59 100.0 100.0 100.0 Females Congenital and perinatal conditions 0.85 0.67 0.45 10.2 6.1 6.5 Infectious diseases 1.79 3.02 1.30 21.4 27.7 18.8 Diseases of respiratory system 1.06 1.50 0.77 12.7 13.7 11.2 Nutritional deficiencies 0.38 1.22 0.33 4.6 11.2 4.8 Diseases of nervous system 0.21 0.21 0.17 2.5 1.9 2.5 Diseases of digestive system 0.29 0.31 0.26 3.5 2.8 3.8 Neoplasms 0.56 0.44 0.48 6.7 4.0 7.0 Cardiovascular diseases 2.18 2.35 2.29 26.1 21.5 33.2 Other diseases 0.43 0.53 0.36 5.1 4.9 5.2 Injury and accidents 0.36 0.37 0.34 4.3 3.4 4.9 Maternal deaths 0.24 0.30 0.14 2.9 2.7 2.0 Total 8.35 10.92 6.90 100.0 100.0 100.0 Source: Authors’ calculations from database on deaths recorded at Antananarivo BMH and census-based population estimates.

Level and structure of mortality by broad groups of causes in Antananarivo in 1976-1978, 1984-1988, and 1997-2000

In this context, in the absence of regular statistics, the observation of trends in cause-specific mortality through analysis of the BMH registers remains a valuable exercise. The observation is currently being extended to the country’s other main cities. At the same time, it is gradually being taken over by Madagascar’s national institute of statistics (Institut national de la statistique, INSTAT). As part of this initiative, INSTAT’s department of demography and social statistics (DDSS) has begun to analyse deaths in the capital for the year 2004. The regular production of these data, combined hopefully with a new wave of Demographic and Health Surveys (DHSs), should provide the tools needed to assess the country’s health status in the years ahead.