The European context

3In 1994, at the height of the epidemic, mean AIDS incidence in Western Europe, measured in terms of lifetime deaths per 100,000 population, was approximately 800 per 100,000. [2] At the time, many people feared an explosion in mortality, given the speed with which HIV/AIDS was progressing and the disease’s fatal prognosis. A very strong rise in the probability of dying from AIDS was indeed observed in all western European countries until the mid-1990s; but this was quickly followed by a spectacular fall, thanks to the advent of combination therapy from 1996 onwards (Palella et al., 1998). The rapid increase in AIDS mortality reflected – with a time lag – the spread of HIV and the onset of full-blown AIDS. In this period when no treatments were available to prevent or delay the development of AIDS in HIV-infected people, national differences in incidence of AIDS can be interpreted as reflecting national differences in HIV infection some years earlier. From 1996, the time between onset of AIDS and death lengthened considerably. Today, the incidence of AIDS is more a reflection of delays in testing and of untreated HIV seropositivity. A reduction in AIDS incidence and a drop in mortality has thus been observed in almost all western European countries (Figure 1). This fall was rapid until 1998, but then slowed down. [3] However, a distinction can be made between two groups of countries in western Europe: southern (Spain, Switzerland, France and Portugal) and northern (the Netherlands, Belgium, Germany and the United Kingdom). Countries in the first group have higher AIDS mortality than those in the second. This distinction was particularly marked in 1994, when AIDS mortality peaked in each of these southern countries (EuroHIV, 2002). Spain is marked out from most other European countries by its high levels of mortality (CNE, 2002; UNAIDS, 2010).

Figure 1

Figure 1

Estimated male and female AIDS mortality in western European countries
Note: These AIDS mortality rates are indirect standardized rates calculated from data provided by EuroHIV (the network for surveillance of HIV infection in the WHO European Region) which do not include the variable of age at death. AIDS mortality in these countries has been estimated using the ages at death recorded in Spain, a country for which the age distribution of AIDS deaths is available.

The Spanish context

4In Spain, HIV infection spread quickly and massively from the early 1980s onwards, first mainly among injecting drug users (IDUs) and then – though to a lesser extent – among male homosexuals (CNE, 2002; Ministerio de Sanidad y Consumo, 2001; Valdes, 2009). Spain’s “heroin boom” during the 1980s led to very high levels of HIV/AIDS transmission through injecting – notably through sharing of contaminated heroin needles (Gamella, 1997). This was a crucial vector in the spread of HIV/AIDS among heterosexuals, since most IDUs were sexually active young men. New prevention programmes aiming to reduce risk practices were promoted in response to this situation, like the 1997-2000 Multisectorial Mobilization Plan against HIV/AIDS (Plan de Movilización Multisectorial frente al VIH/SIDA), [4] which recommended measures whose efficacy had already been demonstrated (Drucker et al., 1998), such as methadone treatment and needle/syringe exchange programmes. Alerted by the government and by prevention campaigns, the Spanish population became aware of the problems associated with intravenous drug use (Plan Nacional sobre Drogas, 2001, p. 16). Young people then tended to focus more on other types of drugs such as ecstasy, amphetamines and drugs that are taken nasally or smoked (Gomez-Redondo and Boe, 2005; Barrio et al., 1998). Efforts at prevention alongside the wider use of combination therapy thus produced a significant reduction in both AIDS incidence and AIDS mortality (CNE, 2002; Ministerio de Sanidad y Consumo, 2001).

AIDS surveillance in Spain

6Since the beginning of the AIDS epidemic, physicians in Spain who care for and monitor AIDS patients have been required to declare each case by completing a single form, which is then passed to the health ministries of the autonomous communities which are independently responsible for compiling regional AIDS registers. The quality of the data collected therefore depends partly on the goodwill of physicians. The regional authorities send a file to the CNE every half-year, containing both newly declared AIDS cases and any modifications to cases previously recorded; this information is added to the National AIDS Register. In fact, some cases are identified as AIDS only after one or even two years: if a physician does not recognise an opportunistic infection at the time as being linked to the advance of HIV infection, he or she can declare it retrospectively. For this reason, the data for a given year are viewed as provisional for a period of two years. All European countries use the AIDS case definition recommended by the European Union since 1993 (Ancelle-Park et al., 1993) and record statistics on AIDS deaths in accordance with the International Classification of Diseases. However, Spain has a rather flexible approach to the principle of patients’ privacy. Unlike other European countries, which use a code guaranteeing anonymity, Spanish data give the names of people with AIDS, on both the registers of the autonomous communities and in the national register. [5]

Total AIDS incidence rate

7AIDS incidence rates can be combined to produce a measure of incidence in which age-structure effects are eliminated. We calculate the incidence of new AIDS cases recorded for a given age over a year as a proportion of the population of that age in the geographical area concerned. We use rates that consider the whole population in the denominator but not the fact that some people will have contracted AIDS previously (age-specific incidence rates). These age-specific incidence rates are summed to define a total lifetime AIDS incidence rate. This gives a rate that is slightly under-estimated, but to an extent that is negligible since the risk of contracting AIDS is relatively low with respect to the size of the study population (Bergouignan, 2004; Valdes, 2009).

Annual and cohort probabilities of dying from AIDS

8In order to measure AIDS mortality, age-specific probabilities of dying are combined on a longitudinal or cross-sectional basis. The probability of dying from AIDS by age and sex is calculated as the ratio of the annual number of deaths by age and sex to total population numbers by sex and age, estimated by INE. Thus we are measuring the risk for individuals in a birth cohort G at age x of dying from AIDS before age x + 1, by determining the ratio of recorded AIDS deaths between age x and age x + 1 to the population of age x observed at the beginning of the year. By combining these probabilities into a life table, we obtain a total rate of AIDS deaths in the absence of other causes of death.

Anticipated life expectancy gains

9“Years of potential life lost” (YPLL) is an indicator that measures years of life expectancy lost through premature death. Conventionally, it often refers to deaths that occur before age 75 (Gardner and Sanborn, 1990). This tool is used to measure the weight of one particular cause of death in overall mortality (Bisig and Paccaud, 1994). Taking the YPLL method as our inspiration, we calculated “anticipated life expectancy gains in the absence of AIDS mortality” in Spain in 1994 (the year when AIDS mortality was highest). These gains were estimated by calculating the difference between life expectancy as it would have been in Spain in 1994 if there had been no AIDS mortality (obtained by combining the age-specific probabilities of dying in the absence of AIDS) and life expectancy actually observed in Spain in 1994. Finally, comparison with anticipated life expectancy gains in the absence of road traffic accident mortality – the second leading cause of death among young people aged 24-44 in Spain in 1994, just behind HIV/AIDS (Martínez de Aragón and Llácer, 1997) – enabled us to demonstrate the impact of AIDS mortality on life expectancy at birth.

Massive cohort concentration

10As also observed for new AIDS cases, the very high cohort concentration of recorded AIDS mortality in Spain essentially reflects the effect of advances in therapeutic management, but to an even greater extent (Figure 2). There is a sharp rise in probability of dying from AIDS, with a time lag between the rise in AIDS incidence rates and the increase in probability of dying from AIDS, which is followed by a rapid fall in this probability from the mid-1990s.

Figure 2

Figure 2

Total AIDS incidence rate (per 100,000) and total probability of dying from AIDS in Spain (per 100,000) for men and women, by year and by cohort

11The fact that the widespread introduction of combination therapy had a much greater effect on AIDS mortality than on AIDS incidence is reflected in an even steeper fall in the former than in the latter (Castilla and De La Fuente, 2000; Ministerio de Saludad y Consumo, 2001).

12The risk of dying from AIDS has always been greater for men, but the relative gender gap varies from one cohort to the next, reflecting changing trends in risk. AIDS mortality increases over the cohorts, peaking among the early 1960s birth cohorts before falling dramatically. This sudden drop is explained by the diffusion of combination therapy, which reduced both AIDS incidence and mortality. Combination therapy has thus had a two-fold impact on AIDS mortality.

A very marked age effect

13Overall, AIDS mortality rates are particularly high around age 30. In Spain, in 1994, HIV/AIDS was the leading cause of death among men aged 25-44 (21.8% of all deaths) and the second leading cause of death among women at those ages (14.9%), exceeded only by cancer (Castilla et al., 1997; Martínez de Aragón and Llácer, 1997). However, changes can be observed in the age profile of AIDS mortality: over time, there is a slight upward shift in the most affected age groups, among both men and women (CNE, 2002).

Distortions in the overall probabilities of dying at young ages

14By comparing, for each age, the risk of dying from all causes (including AIDS) with the risk that would have been observed in the absence of AIDS, we are able to highlight how AIDS distorts overall probabilities of dying for each age group. Setting aside the effect of therapeutic advances – the impact of AIDS disappeared almost entirely by the late 1990s – we can see that the disease had the greatest effect on men aged 27-36 (Figure 3). Since total mortality in these young adult age groups is usually low, the AIDS epidemic led to a marked deformation in mortality trends at these ages. Indeed, during the first half of the 1990s, AIDS mortality among men aged 30-33 was even considerably higher than all-cause mortality for women in the same age group.

Figure 3

Figure 3

Probabilities of dying (per 100,000) in Spain, by sex, cause of death and age

15Trends in mortality from all other causes (excluding AIDS) are close to those observed for all-cause mortality among men aged 27-36: they rise from the mid-1980s and then fall quickly from 1996 onwards. This could mean that AIDS mortality is underestimated, with some deaths from the disease being attributed to another cause. There may be several reasons for this:

• perhaps some AIDS-related deaths were not classified as such;
• some patients with AIDS may have been lost to observation, so that physicians were not always informed of their deaths;
• some deaths of AIDS patients may have been classified differently because they were linked indirectly to the disease (suicide, for example) or were from an associated cause (such as drug overdose).

16There is no research evidence indicating that AIDS mortality is underestimated. But the suspicion that recording of AIDS deaths in Spain may not be exhaustive means that the impact of AIDS mortality analysed here is probably a conservative estimate.

17In addition, the EuroHIV 2006 survey did not provide data that can be used to evaluate the exhaustiveness of AIDS mortality surveillance systems. National surveillance institutions inform EuroHIV about levels of AIDS under-reporting (from 0% to 25% according to country), but do not provide estimates of under-reporting of deaths (EuroHIV, 2002). Moreover, no potential under-reporting of AIDS deaths is visible in the detailed analysis of AIDS mortality in France (Bergouignan, 2005) or in Switzerland (Valdes, 2009).

Importance of the period effect of AIDS mortality

18To better appreciate the impact of AIDS on life expectancy at birth, we can draw a parallel with road traffic accidents (RTA), a cause of death that affects the same young adult ages and which, in 1994, was the second leading cause of death among the 24-44 age group in Spain, just behind HIV/AIDS (Martínez de Aragón and Llácer, 1997). Anticipated life expectancy gains in the absence of AIDS mortality and in the absence of RTA mortality are very close, both in the order of six months for men and two months for women (Table 1). However, despite the fact that RTA mortality is much higher than AIDS mortality at all ages, the resulting loss of life expectancy is comparable to the loss resulting from AIDS. This is explained by the higher concentration of AIDS deaths in the younger age groups compared with RTA deaths.

Table 1

Life expectancy and anticipated gains in the absence of one cause of death in Spain in 1994, by sex

Males Females Observed e0 73.83 80.95 e0 excluding AIDS 74.32 81.09 Anticipated gains in the absence of AIDS mortality: • gains in years 0.49 0.14 • gains in months 5.83 1.68 AIDS mortality (per 100,000) 1,258.62 278.26 e0 excluding RTA 74.37 81.13 Anticipated gains in the absence of RTA mortality: • gains in years 0.54 0.18 • gains in months 6.48 2.21 RTA mortality (per 100,000) 4,987.04 1,569.28

Life expectancy and anticipated gains in the absence of one cause of death in Spain in 1994, by sex