Background: The Epidemiological Transition in the NIS Countries

The countries that comprise the former Soviet Union vary widely in levels of socioeconomic development and societal modernization. Sizeable variations in demographic characteristics reflect this diversity. The fertility levels of the former Soviet countries span most of the range of variation observed around the world. While the European populations of the former Soviet republics have largely completed the "demographic transition" from natural to controlled fertility, the Central Asian populations are either in the early or intermediate stages of the transition. A similar differentiation in mortality could be expected in terms of the "epidemiological transition" from exogenous to endogenous causes of death.¹

The status of the epidemiological transition in the former Soviet countries is not well documented, let alone understood. Because of censorship of and restrictions on access to data on mortality by cause of death over most of the period from the 1930s to the 1980s, Bednyy's (1979, 1984) work appears to be the only widely accessible treatment in the Russian literature on variations in cause-specific mortality among the former Soviet republics in the postwar period prior to the Gorbachev era. Since the relaxation of restrictions on cause-of-death data in 1987, several papers have explored certain aspects of the topic, typically for a single point in time and with highly aggregated cause-of-death categories (Andreyev et al., 1990; Krumin , 1990; Vishnevskiy et al., 1990, 1991). Despite widely differing methodologies, certain common findings are emerging from this work.

All of the former Soviet countries have followed the universal tendency for mortality to decline as infectious diseases are brought under control while death rates from degenerative diseases rise. What is exceptional in the former Soviet countries and some of their East European neighbors is that a subsequent increase in mortality from causes other than infectious disease has brought about overall rises in mortality from all causes combined. Another distinctive characteristic of the former Soviet case is the presence of unusually high levels of mortality from accidents and other external causes, which are typically associated with alcoholism (see Shkolnikov and Nemtsov, in this volume). The variations among republics conform broadly to expectations that mortality from infectious, digestive, and respiratory system disease is highest in the less-developed Central Asian republics and lowest in the Baltic countries and Belarus.

The next two sections of this chapter review data and measurement issues, respectively, that are encountered in examining mortality patterns in the NIS countries. Next we present results for mortality levels and cause of death. We then briefly examine mortality trends during the 10-year period, 1979-1989, preceding the breakup of the Soviet Union. The final section presents a discussion of the results.

Data Issues

Measurement Issues

Because infants as a group are the focus of particular concern with regard to health priorities, infant mortality warrants inclusion in the present analysis. Given the unusual definitions and measurement errors associated with former Soviet statistics on infant mortality, as discussed above, it was necessary to adjust the reported infant mortality rates to facilitate their meaningful analysis. A variety of adjustments of the official Soviet infant morality rates have been advanced in the literature, ranging from the 15-25 percent correction proposed by some Western investigators (Anderson and Silver, 1986; Davis and Feshbach, 1980) to corrections nearly four times greater in the recent Russian literature³ (Andreyev and Ksenofontova, 1991; Baranov et al., 1990). Perhaps the most serious shortcoming of our recent practice at the U.S. Bureau of the Census has been our use of a common adjustment factor for all republics, estimated from data for the Soviet Union as a whole, which was dictated by the absence of the necessary data at the republic level. Such data are now in our possession, and we proceed to develop separate adjustments for the republics.

The data at our disposal consist of the distributions of infant deaths by sex and age in months for the 15 former Soviet republics in 1990, together with births by sex and calendar month in 1989-1990. From these data we computed male and female probabilities of dying by age in months. When the former Soviet data are examined in relation to the corresponding schedules for countries with reliable data, two features are notable (see Figure 5-1 for an illustration with respect to Germany). The most striking feature is the much lower level of mortality in the first month relative to subsequent months in the former Soviet countries. In addition, the former Soviet data contain a number of instances—in various Transcaucasian and Central Asian republics—in which the probabilities of dying increase with age after the first month. This pattern may reflect delays in reporting infant deaths to official authorities, since former Soviet regulations stipulated a generous 3-month time limit. At any rate, these results are consistent with Andreyev and Ksenofontova's (1991) assertion that the infant mortality data for months 4 through 10 are the most reliable; by the same token, they confirm our judgment that the former Soviet infant mortality rates for the first month require adjustment.

Figure 5-1

Infant mortality by age, males, NIS countries compared to Germany.

Our approach to adjusting former Soviet infant mortality rates is to accept the probabilities of dying between months 4 and 10 implied by the official data and then use these rates to estimate the probability of dying in the first month. In doing so, we make the assumption that the relationship between these probabilities conforms to the pattern found in countries with reliable data at points in their histories when their infant mortality levels (for months 4-10) were comparable to those of the former Soviet countries in 1990. The relation of the logits of these probabilities in the United States and Germany is well represented by a cubic relationship (Figure 5-2), which we have taken as the basis for our adjustments. Table 5-1 contains the adjusted and original infant mortality rates. As compared with our previous, uniform adjustment, the regionally differentiated approach is more satisfying in that greater adjustments are applied to the Central Asian republics, together with Azerbaijan, where the data are faultier, while the smallest adjustments are applied in Lithuania and Estonia, where the data are more reliable.

Figure 5-2

Relationship of age-specific infant mortality rates, probabilities of dying in first month versus months 4-9, United States and Germany, males, cubic fit.

TABLE 5-1

Adjusted and Reported Infant Mortality Rates, NIS Republics, 1990.

With regard to old-age mortality, we treat it quite differently from infant mortality. Measurement errors at old age are less straightforward to correct than those in infancy and have less impact on overall life expectancy. For these reasons, we have elected not to adjust the official mortality data at later ages. Instead, we employ a methodology that removes the effect of old-age mortality and focus our attention on younger ages.

Finally, to assess the impact of cause-specific mortality in the various former Soviet republics, we employ the measure of years of life lost (Arriaga, 1994; see also Murray and Bobadilla, in this volume). This measure defines the loss of life associated with a death as the difference between the observed age at death and the age to which the deceased could or should have lived. This measure is preferable to standardized death rates, which weight all deaths evenly irrespective of age at death, and reflects the tempo of trends in cause-specific mortality more transparently than multiple-decrement life tables. Operationally, we have adopted age 75 as the age to which persons should live in the absence of premature mortality. This facilitates comparison by applying a common standard to the various former Soviet countries and also excludes mortality in extreme old age, where measurement errors are apt to be more pronounced.

Mortality Levels of the NIS Countries

An overview of the mortality levels in various republics of the former Soviet Union is provided in Table 5-2. Only 9 of the 15 former Soviet republics appear in the table—those for which we possess cause-of-death data adequate for inclusion in the analysis that follows.⁴ The selected republics include representatives of each of the principal geographic subdivisions of the former Soviet Union: the Baltic (Lithuania), other European (Russia, Ukraine, Moldova), Transcaucasian (Azerbaijan), and Central Asian (Uzbekistan, Tajikistan, Turkmenistan, Kyrgyz) regions.

TABLE 5-2

Mortality Levels of Selected NIS Countries, 1979-1993.

Table 5-2 shows that mortality levels vary considerably across these republics. In general, the European republics, especially Lithuania, enjoy lower mortality than Azerbaijan and the Central Asian republics.

In addition to our measure of years of life lost, Table 5-2 includes estimates of life expectancy at birth.⁵ Although the latter measure is surely more familiar to most readers, its susceptibility to confounding influences is also evident. For instance, the male life expectancy for the total population of Azerbaijan is lower than both the urban and rural life expectancies. This turns out to be the result not of a mechanical or typographical error, but of Azerbaijan's unusually low reported male mortality at later ages.⁶ We share the skepticism of Anderson and Silver (in this volume) regarding the low death rates at later ages reported for the Central Asian republics and Azerbaijan. In particular, the official mortality data for rural Azerbaijan in 1989 imply a higher male life expectancy at age 70 (12.31) than that of Swedish men at this age in 1990 (11.86); the 1979 data for Azerbaijan generate an even higher life expectancy (13.06). We think it highly unlikely that Azerbaijan's elderly men have really achieved better health than their Swedish counterparts, yogurt or not.⁷ The most plausible explanation is that Azerbaijan's official mortality statistics seriously understate the country's true level of mortality.

In terms of years of life lost, the results shown in Table 5-2 are generally clear-cut. First of all, the high magnitude of loss of life across this group of countries is striking. As of 1989, among males in all these countries, along with females in Azerbaijan and the Central Asian republics, years of life lost from ages 0 to 75 amount to 15-25 percent of that 75-year time span. The levels of years of life lost in 1979 are even greater.

The regional variations in years of life lost are consistent with the levels of socioeconomic and sociocultural development of the various republics. Levels of years of life lost are lower in the European republics than in the "Southern Tier" republics of Azerbaijan and Central Asia. A Baltic republic, Lithuania, emerges with the lowest levels of years of life lost among the republics included in Table 5-2. Within republics, rural populations are generally characterized by higher levels of years of life lost than their urban counterparts. The exceptions to this pattern are the male populations of the Central Asian republics, where there is reason for suspicion about the quality of mortality statistics, especially in rural areas.

The sex differential in life expectancy in the former Soviet Union has been the highest in the world for many years (Kingkade and Arriaga, 1992; see also Vassin and Costello in this volume). According to Table 5-2, men in Russia, Ukraine, and Lithuania incur twice as many years of life lost as their female compatriots. In Moldova and the Southern Tier republics, levels of years of life lost among men are greater by half than the corresponding figures for women. The rural populations of Azerbaijan and the Central Asian republics are notable for their low sex differentials in years of life lost relative to the other republics in Table 5-2.

With respect to temporal trends, the figures in Table 5-2 reflect an improvement in mortality levels from 1979 to 1989, which is in keeping with overall societal and economic trends. These two respective time points correspond to the height of stagnation in the former Soviet period (1979) and the approximate middle of perestroyka (1989). As the figures indicate, all republics participated in the overall mortality decline during this period. The greatest improvements were registered in the Southern Tier republics, excepting Turkmenistan. In the latter republic, there is little evidence of improvement among the male population, whose years of life lost in rural areas appear even to have risen slightly.

As for trends after 1989, our data are limited to the Russian Federation. While the abrupt rise in Russian mortality from 1992 to 1993 and its associated 3-year drop in male life expectancy have received much attention in the Western media, one should bear in mind that overall mortality in Russia has been rising since 1989. The 1989-1993 increase in years of life lost among Russian men was comparable in magnitude to the major declines registered in Azerbaijan and Central Asia from 1979 to 1989. In terms of years of life lost, Russian men appear to have ended up substantially worse off in 1993 than they were in 1979. Among Russian women, the deterioration in mortality since 1989 has thus far merely erased the 1979-1989 gains.

Cause of Death

Our detailed data on mortality by cause of death for the former Soviet republics in 1989 and Russia in 1993 provide an excellent basis for assessing the impact of cause-specific mortality on the overall loss of life experienced by their populations. Tables 5-3 through 5-5 present our decomposition of years of life lost by cause of death for the total, urban, and rural populations, respectively, of the nine republics. Our ten cause-of-death categories reflect our subdivision of two broad cause classes that figure prominently in the mortality patterns of the former Soviet republics: diseases of the respiratory system and external causes of death. Respiratory system diseases are often treated as a typical ''Third World" category because they include many illnesses, such as pneumonia and bronchitis, that are etiologically similar to infectious diseases; we distinguish such diseases from other respiratory conditions of more degenerative character, such as emphysema. Within the important group of external causes of death, we distinguish homicides and suicides from the remainder of accidental fatalities. In addition, we list separately two categories of alcohol-related death: alcohol poisonings and another ("Alcohol") category that includes alcoholic psychosis and alcoholic cirrhosis.

TABLE 5-3

Years of Life Lost by Cause of Death, Selected NIS Countries, 1989^a (Total Urban + Rural Population).

TABLE 5-5

Years of Life Lost by Cause-of-Death, Selected NIS Countries, 1989 (Rural Population).

According to the data shown in Tables 5-3 through 5-5, the former Soviet republics are quite diverse in their patterns of mortality by cause of death. Most notably, a sharp Northern/Southern Tier distinction emerges, reflecting the different stages of the epidemiological transition attained by the respective populations. In most of the European republics, the category accounting for the greatest loss of life is diseases of the circulatory system, while infectious respiratory disease is the leading category in Azerbaijan and the Central Asian republics. Accidents, homicides, and suicides led to substantial loss of life among men in all the republics. In the context of the pronounced infectious disease mortality in the Southern Tier republics, those categories appear less salient than in the European republics, where they are associated with more than 25 percent of overall loss of life among men. In Moldova, accidents excluding homicide/suicide represent the leading cause of mortality among men in 1989.

According to our data, deaths directly related to alcoholism generally make a negligible contribution to overall mortality levels. Only among Russian men in 1993 does the combined effect of alcohol poisonings and other directly alcohol-related diseases approach a year of life lost. Nonetheless, the increase in alcohol-related mortality among Russian men from 1989 to 1993 is certainly troubling.

The results shown in Tables 5-3 through 5-5 offer insight into the cause-specific mortality trends underlying the decline in Russian male life expectancy from 1989 to 1993. The roles of the various causes of death in this increase are quite distinct, as the data in the tables demonstrate. Accidents, homicides, and suicides account for most of the increase in Russian male loss of life and slightly under half of the much smaller female increase. Diseases of the circulatory system also figure prominently in the increase. The causes directly related to alcoholism make minor contributions to the overall rise in loss of life, although it might be argued that relative to their (slight) 1989 levels, the role of these categories has increased substantially. In contrast, mortality from neoplasms has declined in most instances (rural women being the exception), presumably as a result of premature mortality from other causes. Curiously enough, the cause category receiving the most attention in U.S. media accounts of recent Russian demographic trends—infectious disease—turns out to have played a negligible role in the mortality increase according to actual statistics.

As to rural-urban variations, several patterns appear in our data. As expected, the predominance of infectious diseases in the Southern Tier mortality profile appears to be primarily a rural phenomenon. Loss of life due to degenerative diseases and external causes is greater among the urban populations of these republics. In contrast, in the European republics, loss of life due to these causes is not much lower, and is frequently higher, among rural than among urban populations; the rural disadvantage is often more pronounced among women than among men. Loss of life due to external causes tends to be higher in rural than in urban areas in the European republics, while the opposite is typical of the Southern Tier republics. A similar pattern obtains with respect to homicide and suicide in particular.

With regard to sex differentials, our results indicate that these are nearly always in favor of women.⁸ The cause categories associated with the greatest excess of male over female years of life lost are accidents, diseases of the circulatory system, and homicides/suicides in the Russian and other European republics.

1979-1989 Trends

We have at our disposal some data on the distribution of deaths by broad cause categories that can be analyzed relative to the 1989 data to shed some light on regional mortality trends in the former Soviet Union around the end of its existence. Tables 5-6a, b, and c present our estimates of years of life lost for these categories for the total, urban, and rural populations of our set of nine republics. Reductions in mortality from diseases of the respiratory system appear to be largely responsible for the overall declines in mortality in most of the republics during the period. While perhaps surprising at first in light of the attention given to the anti-alcoholism campaign in Soviet and Western media, this result appears entirely plausible in terms of the age pattern of respiratory system disease mortality. Because many deaths from respiratory illnesses occur in infancy and childhood, they are associated with higher levels of years of life lost than the majority of deaths due to external causes, which tend to occur at much older ages. In any case, the results in Table 5-6 indicate that reductions in mortality from injuries do in fact make a greater contribution to the mortality decline from 1979 to 1989 among men in Russia, along with Lithuania, than among those in the other republics. Changes in infectious disease mortality, both positive and negative, are also important in Moldova and the Central Asian republics.

TABLE 5-6a

Years of Life Lost by Broad Cause-of-Death Category, NIS Countries, 1979 and 1989 (Total Population).

TABLE 5-6b

Years of Life Lost by Broad Cause-of-Death Category, NIS Countries, 1979 and 1989 (Urban Population).

TABLE 5-6c

Years of Life Lost by Broad Cause-of-Death Category, NIS Countries, 1979 and 1989 (Rural Population).

Discussion

A variety of features of the mortality patterns of the former Soviet countries have been highlighted in the above analysis. These are important either as conditions that must be addressed by the public health programs of the respective countries or as elements of the context in which these programs must develop. Perhaps the most obvious aspect of the mortality profiles of the former Soviet countries is their diversity. In the Southern Tier countries examined in this analysis, Azerbaijan and the Central Asian republics, there remains substantial progress to be made in combatting diseases of infectious character. The legacy of the former Soviet period includes a health system that has had considerable success in controlling these diseases. This progress has gone furthest in the European republics.

According to our analysis, much of the progress in mortality reduction in the recent past, even in the European former Soviet republics, has also been achieved through control of respiratory and infectious disease. Unfortunately, the degenerative diseases and external causes of death that dominate the current mortality patterns of the European NIS countries have proven less responsive to traditional Soviet approaches, with the possible exception of the anti-alcohol campaign. Some attempts were made to organize, at least experimentally in certain areas, regular diagnostic checkups through which degenerative diseases might be detected in their early stages, in the hope of avoiding or forestalling their worst consequences. However, the expense of sustaining such efforts, especially if they involve upgrading of diagnostic and therapeutic technologies, would seem to rule out their implementation under the current adverse economic circumstances in most of the former Soviet Union.

Some East European and former Soviet health analysts have adopted the label of ''civilization diseases" to categorize the causes of death induced by factors such as stress, lifestyle, and environmental degradation, which may be viewed as unforeseen byproducts of socioeconomic development (Lisitsyn, 1982; Tolokontsev, 1987). This type of mortality is relatively unresponsive to, and its causes are seldom addressed by, traditional health care strategies. The external causes of death appear to represent this category in its purest form. Homicide and suicide, in addition, call attention to the influence of social disorganization. The prominence of these causes of death in the recent rise in Russian mortality is disturbing. We must hope that Russia and the other countries in the region will regain the stability required for further progress.

References

• Anderson, B.A., and B.D. Silver 1986. Infant mortality in the Soviet Union: Regional differences and measurement issues. Population and Development Review 12(4):705-738.
• 1989. The changing shape of Soviet mortality, 1958-1985: A review of old and new evidence. Population Studies 43(2):243-265. [PubMed: 11621871]
• Andreyev, Y.M., V.A. Biryukov, and K.Y. Shaburov 1990. Life Expectancy in the USSR and Mortality Dynamics by Cause of Death in the Regional Aspect. Paper presented at the International Conference on Health, Morbidity and Mortality by Cause of Death in Europe. December 3-7. Vilnius.
• Andreyev, Y.M., and N.Y. Ksenofontova 1991. Otsenka dostovernosti dannykh o mladencheskoy smertnosti. Vestnik Statistiki 8:21-28. [PubMed: 12178767]
• Arriaga, E.A. 1994. Measuring the Level and Change of Mortality by Cause of Death: The Use of Years of Life Lost. Paper presented at the Annual Meeting of the Population Association of America. May 5-7. Miami.
• Baranov, A.A., V.Y. Al'bitskiy, and Y.M. Komarov 1990. Tendentsii mladencheskoy smertnosti v SSSR v 70-80ye gody. Sovetskoye Zdravookhraneniye 3:3-7. [PubMed: 2142812]
• Bednyy, M.S. 1979. Mediko-demograficheskoye Izucheniye Narodonaseleniya . Moscow: Statistika.
• 1984. Demograficheskiye Faktory Zdorov'ya . Moscow: Finansy i Statistika.
• Bennett, N.G., and L.K. Garson 1983. The centenarian question and old-age mortality in the Soviet Union. Demography 20(4):587-606.
• Boyarskiy, A.Y. 1985. Kurs Demografii . Moscow: Finansy i Statistika.
• Davis, C., and M. Feshbach 1980. Rising Infant Mortality in the USSR in the 1970s . International Population Reports. Series P-95, No. 7. Washington, D.C.: U.S. Bureau of the Census.
• Dellaportas, G.J. 1965. Correlation-based estimation of early infant mortality. Public Health Report 87:275-281. [PMC free article: PMC1615942] [PubMed: 5023136]
• Goskomstat SSSR (Gosudarstvennyy komitet SSSR po statistike) n.d. Kratkaya Nomenklatura Prichin Smerti 1981 Goda, Osnovannaya na Mezhdunarodnoy Statisticheskoy Klassifikatsii Bolezney, Travm i Prichin Smerti IX Peresmotra .
• 1987. Osnovnyye Pokazateli Tablits Smertnosti i Sredney Prodolzhitel'nosti Zhizni Naseleniya 1984-1985 gg. Vestnik Statistiki 3:79-80.
• 1989. Tablitsy Smertnosti i Ozhidayemoy Prodolzhitel'nosti Zhizni Naseleniya . Moscow: Goskomstat SSSR.
• Hogan, H. 1979. Age Patterns of Infant Mortality. Unpublished manuscript. Washington, D.C.: U.S. Bureau of the Census.
• Kingkade, W.W. 1985. Evaluation of Selected Soviet Population Statistics . CIR Staff Paper No. 9. Washington, D.C.: Center for International Research, U.S. Bureau of the Census.
• 1989. Soviet Infant Mortality in International Perspective . CIR Research Note. Washington, D.C.: Center for International Research, U.S. Bureau of the Census.
• Kingkade, W.W. and E.A. Arriaga 1992. Sex differentials in mortality in the Soviet Union. Pp. 114-135 in W. Joyce, editor. , ed., Social Change and Social Issues in the Former USSR. New York: St. Martin's Press.
• Kozlov, V.I. 1989. Dolgozhitel'stvo v Azerbaidzhane . Moscow: Nauka.
• Krumin , J. 1990. The Changing Mortality Patterns in Latvia, Lithuania and Estonia: Experience of the Past Three Decades. Paper presented at the International Conference on Health, Morbidity and Mortality by Cause of Death in Europe. December 3-7. Vilnius.
• Lisitsyn, Y.P. 1982. Zdorov'ye Naseleniya i Sovremennyye Teorii Meditsiny . Moscow: Meditsina.
• Merkov, A.M. 1965. Demograficheskaya Statistika . Moscow: Meditsina. Meslé, F., V. Shkolnikov, and J. Vallin;
• 1991. Mortality by Cause in the USSR in 1970-1987: The Reconstruction of Time Series. Paper presented at the European Population Conference. October 21-25. Paris.
• Stetsenko, S.G., and I.V. Kozachenko 1984. Demograficheskaya Statistika . Kiev: Visha Shkola.
• Tolokontsev, N.A. 1987. Zdorov'ye Cheloveka i Abiologicheskiye Tendentsii v Sovremennykh Usoviakh i Obraze Zhizni. Pp. 126-139 in I.I. Smirnov, editor. , ed., Obshchestvennyye Nauki i Zdravookhraneniye . Moscow: Nauka.
• United Nations, Department of Social and Economic Affairs 1985. Handbook of Vital Statistics Systems and Methods . Volume II: Review of National Practices. New York: United Nations.
• Vishnevskiy, A.G.. V.M. Shkolnikov, and S.A. Vasin 1990. Epidemiological Transition in the USSR as Mirrored by Regional Disparities. Paper presented at the International Conference on Health, Morbidity and Mortality by Cause of Death in Europe. December 3-7. Vilnius.
• 1991. Epidemiologicheskiy Perekhod i Prichiny Smerti v SSSR. Ekonomika i Matematicheskiye Metody 27(6):1013-1021.

Footnotes

1. We use these terms to distinguish between causes of death that result from specific characteristics of an individual (endogenous factors), such as behavior and genetic makeup, and causes external to an individual (exogenous factors), such as infectious diseases. The "epidemiological transition" refers to observed changes in the epidemiologic profile of diseases afflicting given populations. Generally speaking. this worldwide trend in the primary causes of death is from infectious diseases (exogenous) to chronic diseases and accidents (endogenous).

2. For example, the formula used by Goskomstat in calculating infant mortality rates corresponds better to the annual probability of dying in the first year of life than does the practice of our own National Center for Health Statistics of dividing deaths under age 1 in a given calendar year by births in the same period. Also, something close to a continuous migration register for at least the urban population was (and to our knowledge continues to be) maintained in the former Soviet countries; such a system has never been attempted on a national scale in the United States.

3. Our own experience at the Census Bureau encompasses the better part of this range, driven largely by data availability. An adjustment intermediate between those of Davis and Feshbach on the one hand and Anderson and Silver on the other, based on relational logit models fit to life table probabilities of dying for several East European countries (Kingkade, 1985), was introduced in our world population projections in 1985. Upon the release of data on the age distribution of deaths in infancy, an adjustment of 68 percent was obtained through Hogan's (1979) infant mortality model (Kingkade, 1989), which is a modern successor to the method (Dellaportas, 1965) of Baranov and Associates, and also appears akin to the methodology sketched by Andreyev and Ksenofontova (1991). Our current approach remains in this tradition, but differs in its particulars from our previous practice.

4. For the 6 excluded republics, the data files delivered to us contained severe formatting errors that render them unusable for any analysis.

5. These life expectancies were calculated with the adjusted infant mortality rates from Table 51 and are lower than Goskomstat's official life expectancies for the given republics and dates.

6. The male mortality rates in infancy and childhood reported for the rural population of Azerbaijan exceed the rates reported for the urban population. With advancing age, the urban mortality rates increase in greater measure than the rural rates, so that by the early 30s, the urban rates exceed the corresponding rural rates; this situation persists over the remainder of the age range. Thanks to the rural-urban fertility differential, the proportions urban in Azerbaijan's population increase with age, so that the national average mortality schedule weights the rural and urban subdivisions most heavily at precisely the ages at which their mortality rates exceed those of their counterparts. In turn, this leads to the life expectancies shown in Table 5-2.

7. Azerbaijan has the distinction of being the homeland of the oldest purported centenarians in the former Soviet Union, who have been a subject of perennial interest to ethnographers, gerontologists, and others. In the 1980s, a group from the Institute of Ethnography of the Soviet Academy of Sciences set out to interview the alleged centenarians in the villages where they resided. They came away concluding that the individuals in question were nowhere near the age they claimed to be, and published a monograph containing their findings (Kozlov, 1989).

8. The exceptions are of negligible magnitude, amounting at most to 6 percent of a year of life lost.