In the United Kingdom, age-specific mortality rates for most age and sex groups declined until the mid-1980s when a levelling was observed for men and women aged 15-44 . Between 1986 and 1990 there was a 6.3% increase in the mortality rate in men aged 15-44 years in England and Wales; it has since remained fairly stable . Although deaths in this age group accounts for only 3% of deaths in England and Wales, the Chief Medical Officer report On the State of the Public Health 1996 expressed the concerns that these deaths may have profound effects on individual families, that they account for many years of working life lost and that many of them may be preventable .
Since the 1980s, HIV has been an increasingly frequent cause of death in young adults worldwide [3,4]. In the United States, HIV infection was the most common cause of death among people aged 25-44 by 1993 and accounted for 19% of deaths in this age group in 1994 . However, age-specific mortality rates do not adequately describe the impact of HIV as they confine comparisons to specific age bands. Years of potential life lost (YPLL), by contrast, provides a measure reflecting the impact of all premature mortality that can be applied to assess the relative burden on society of specific causes of death [6-8].
Several international studies using YPLL have highlighted the impact of the HIV epidemic relative to other causes of death in different populations and subpopulations [9-12]. For example, HIV infection was the fourth leading cause of YPLL before 65 years of age in the United States nationally and the leading cause among men in four states in 1994 . In the United Kingdom, HIV has been shown to be an important cause of death among young adults in one area of London .
In this study, we have examined age-specific mortality rates (15-44 years) and YPLL (to age 65) associated with HIV infection compared with other causes of death in England and Wales from 1985 to 1996.
Data on causes of death
Data relating to causes of death other than HIV/AIDS used information from certificates of death between 1985 and 1996 supplied by the Office for National Statistics (ONS). Assignment of underlying cause followed the international procedures established by the World Health Organization and causes were classified using the Ninth Revision of the International Classification of Diseases (ICD9) . The causes of death used for comparison were grouped according to the pathophysiological process or organ system involved. They were chosen on the basis of associated premature mortality and public health concern from three main ICD9 chapters (neoplasms, diseases of the circulatory organs and external causes). In addition, comparison was made with meningitis, the other main infectious cause of death in young people. In this article ‚suicide‚ includes deaths undetermined whether accidentally or purposely inflicted (ICD9 codes E950-E959 and E980-E989 excluding E988.8) and ‚meningitis‚ refers to meningitis and meningococcal infection (ICD9 320-322 and 036).
Because of known reluctance to record HIV or AIDS on death certificates [13,15-17], deaths in HIV-infected individuals were assessed from routine surveillance data at the Communicable Disease Surveillance Centre (CDSC). These data were adjusted for under-reporting, which has previously been shown to be 13% [18,19].
Measures and analysis
Midyear population estimates were used for the calculation of age-specific mortality rates. Analysis of geographical distribution was based on residence stated on the PHLS surveillance form for deaths associated with HIV, and usual residence stated on the death certificate for other causes of death. The small number of deaths of people whose usual residence was reported as outside England and Wales were included in the total analysis but not in the subdivisions by geographical area. Regional comparisons were made for 1996 data using the then current district health authority boundaries. For the analysis of time trends in greater London, the local authority boundaries for the authorities that make up London were used.
The methodology proposed by ONS  for calculating working YPLL was applied for estimations of individual YPLL, defined as the difference between 65 years and the age of death minus 0.5 for deaths at ages between 15 and 65, 50 for ages below 15 and zero for deaths at ages above 65 years. For the population, YPLL from each underlying cause of death was calculated as the sum of the YPLL for all individuals who died of that cause that year.
Between 1985 and 1996, 6842225 deaths were certified in England and Wales: 3339 516 in men and 3502709 in women. Overall, only a small proportion of these deaths (3.3%) occurred in young adults aged 15-44 years: 147870 in men (4.4 %) and 78967 (2.3% %) in women. For some of the major causes of death in young adults (e.g., suicide, HIV and motor vehicle traffic accidents), most of the overall deaths occurred in this age group, whereas for other major causes of death in young adults (e.g., ischaemic heart disease, lung cancer and cerebrovascular disease), the great majority of deaths occurred in the older age groups (Table 1).
Age-specific mortality rates
The crude age-specific mortality rates for all causes of death in England and Wales in the age band 15-44 year remained fairly constant between 1985 and 1996 for both men (110/100000) and women (60/100000), whereas that in other age bands decreased. Within the age band of 15-44 year olds, an increase was seen in men 25-29 and 30-34 years whereas the mortality rates in the 15-19 and 40-44 year old men decreased.
The crude age-specific mortality rate for HIV in men aged 15-44 rose from 0.9/100000 inhabitants in 1985 to its peak, 10.3/100000, in 1994, when it accounted for 9.3% of deaths in this age group. Men aged 30-44 were most affected, with rates over 16/100000 men dying with HIV in 1994 for each 5-year age band. In women, the mortality rate for HIV in the age band 15-44 years continued to increase throughout the period from 0.02/100000 in 1985 to 1.6/100000 (2.6% of deaths) in 1996. In the age band 45-64 years, only 0.9% of deaths in men and 0.08% of deaths in women were associated with HIV in 1994.
Throughout the time period, more men aged 15-44 died of suicide and motor vehicle traffic accidents than of HIV infection. By 1994, however, HIV-associated deaths were as common in this age group as deaths from ischaemic heart disease (Fig. 1).
With the exception of suicide in men and HIV infection in both men and women, the age-specific mortality rates among 15-44 year olds between 1985 and 1996 were either stable or decreasing for major causes of death. Although suicide accounted for a greater number of deaths among men in this age group throughout the study period, the relative and absolute increase in deaths was greater for HIV infection. If deaths from suicide and HIV infection had followed the same proportional decrease as deaths from other causes among men of 15-44 years, the total crude age-specific mortality rate for men in this age group, instead of being stable, would have fallen from 107/100000 in 1985 to under 100/100000 after 1992. Of the deaths in excess of this general downward trend since 1985 in men aged 15-44, those attributable to HIV contributed more than suicide in all years from 1986 to 1996 except in 1988.
Years of potential life lost
Like trends in age-specific mortality rates, YPLL from major causes of death in young adults, with the exception of suicide and HIV infection, have either been stable or decreased during the study period. The total number of YPLL in England and Wales decreased from 1.66 million in 1985 to 1.39 million in 1996. The number of YPPL associated with HIV infection increased from 4000 in 1985 to 45000 in 1994, accounting for 3.2% of all YPLL in 1994. In men, the proportion of YPLL associated with HIV peaked with 4.6% in 1994, whereas in women the rise has continued to 1.2% of YPLL in 1996.
Compared with other important causes of death in young adults, HIV accounted for more YPLL than cerebrovascular disease and lung cancer but less than ischaemic heart disease, suicide and motor vehicle traffic accidents among males in 1994 and 1995 (Fig. 2). Among women, HIV had a bigger impact on YPLL than meningitis but smaller than breast cancer, cervical cancer, lung cancer, suicide and ischaemic heart disease in 1996 (Fig. 3). HIV has been a more important infectious cause of YPPL than tuberculosis, septicaemia, intestinal infections and meningitis in men since 1986 and in women since 1994.
In 1996 the proportion of YPPL associated with HIV varied from 0.6% in Wales to 7.2% in North Thames (Table 2). In 16 Health Authorities in Greater London, more than 10% of YPLL was accounted for by HIV; Kensington, Chelsea and Westminster were the most severely affected with 23% of YPLL in total and 30% of YPLL in men attributable to HIV. Compared with the other major causes of death investigated, HIV was the most important cause of YPLL in men in Greater London 1993-1996 (Fig. 4).
Mortality rates in young adult men in England and Wales have failed to follow the downward trend observed in other age groups. Some of the increase in crude mortality rate for men and women aged 15-44 between 1986 and 1990 can be accounted for by an increasing proportion at the upper end of that age range; however, even when adjusted for this factor, death rates have not decreased as they have for other age and sex groups . The study described here investigated the absolute and relative contribution of HIV infection to this rate.
As discussed in the Methods, deaths associated with HIV were taken from the CDSC register and deaths from other causes from ONS. Since information from different sources was compared, possible overlap between the two sources needs to be estimated and taken into account in the interpretations of the results. On the one hand, misclassification of HIV-related deaths on death certificates may result in an overestimate of the impact of other causes of death. This will have the largest effect for causes of death most commonly recorded in HIV, such as other infections (especially pneumonia) and cancer , but is not likely to have a major influence on the causes for death used for comparison in this article. On the other hand, our assumption that all deaths in HIV-infected people are caused by HIV will result in an overestimate of the impact of HIV compared with other causes of death. However, studies on mortality in people with AIDS indicate that causes of death unrelated to HIV infection are rare, accounting for 1-3% of deaths [20-22]. A direct assessment of this overlap was made for suicide, the most important non-disease-related cause of death in HIV-infected people before the onset of AIDS , on a limited dataset of all persons known to ONS to have died below the age of 60 in 1993. Of 1276 death records matching those of HIV-infected individuals recorded at the PHLS AIDS and STD Centre, only 14 records (1.1%) included ICD9 codes used for suicide and undetermined deaths.
Because of the known underdiagnosis and under-reporting of HIV [15,23], a proportion of deaths associated with and attributable to HIV will not be recorded as HIV-related deaths in either of the data sources investigated. The proportion of HIV-infected individuals who die with undiagnosed HIV infection is difficult to appraise. Under-reporting of AIDS in England and Wales, however, has been estimated at 13%, so adjustment for unreported cases uses a multiplication factor of 1.15 . In applying this factor, it has been assumed the distribution of deaths in non-reported AIDS cases was the same as that for reported cases with respect to sex, age, area and year of death. It has been shown, however, that non-reported AIDS cases are, in fact, disproportionately more likely to be older and from outside London . This distortion among 13% of the AIDS deaths is unlikely to have a major influence on the geographical distribution of HIV-associated YPLL presented here, but it will tend to exaggerate the total YPLL slightly.
Under-reporting will also affect other causes of death. Difficulties in determining ‚unnatural causes‚ have been shown to result in under-reporting of suicides . These difficulties can be assumed to be particularly severe where the morbidity is high, such as in HIV-infected people. For example, intravenous drug users may have their suicides misclassified as overdose and patients with advanced AIDS may be presumed to die of infection.
Various methods have been proposed for the calculations of YPLL . We chose to apply the methodology of the ONS for calculating YPLL, which uses 65 years as the upper cut-off point . This relatively low cut-off point was used to reflect the social and economic loss associated with premature death better. A higher cut-off point would have given a greater weight to ‚natural‚ causes of death, such as cardiovascular diseases.
YPLL associated with HIV infection in men appears to have peaked in 1994 (Fig. 2). Part of the decrease since 1994 may be attributed to delay in reporting, but improved survival as a consequence of antiretroviral therapy has probably contributed. This therapeutic effect may not be sustained. The impact of HIV infection has been greatest in London, paralleling the situation established in New York  and San Francisco  by 1986. HIV was the leading cause of YPLL until 65 years among men in the seven Health Authorities covering inner London during 1993-1996, more important than, for example, all circulatory diseases or all neoplasms. Though the impact of the HIV epidemic has been much less in women, the proportion and absolute number of YPLL associated with HIV continued to increase right up to 1996 (Fig. 3).
During the study period, deaths from both HIV infection and suicide increased in men aged 15-44. Even though suicide accounted for a greater number of deaths throughout the investigation period, the largest proportional and absolute increase was seen for deaths in HIV-infected people. Although the relative risk of suicide may be increased in people with HIV, our findings from matching population death data in those under 60 years of age with HIV/AIDS registry data are consistent with previous studies [13,22] in indicating that suicide accounts for only a small proportion of all HIV-related deaths. Furthermore, HIV has a geographical distribution in England and Wales that is different from that of the increase in suicide  and it became a major problem during the second half of the 1980s, well after the rise in suicide in young men began. We feel it is unlikely that HIV has made a major contribution to suicide rates.
The contribution of HIV to premature mortality in England and Wales has been large. Since the mid-1980s, HIV has become a major cause of death in 15-44 year olds in both men and women. While part of the adverse trend in mortality in young adults since 1985 was attributable to suicide, most was due to HIV infection. It seems likely that, in spite of recent improvements in survival, premature mortality attributable to HIV infection, particularly in men, will continue to be substantial for many years to come. How HIV is transmitted is well understood and, therefore, prevention of HIV should be possible. Yet among homosexual men, indicators such as the increase in episodes of acute sexually transmitted infections and reports of HIV seroconversion suggest that substantial HIV transmission is still occurring. The consequences of not preventing such transmissions are serious both for the individuals concerned and for society.
We gratefully acknowledge the work of all those who report HIV infection, AIDS cases and related deaths. We would also like to thank the ONS for making the necessary dataset available and the staff of the HIV and AIDS Reporting Section, PHLS AIDS Centre, particularly A. Wright, J. Mithal and D. Howitt.
1. Dunnell K. Deaths among 15-44 year olds. Pop Trend 1991, 64:38-43.
2. Department of Health. On the state of the Public Health 1996. London: HSMO; 1997.
3. Selik RM, Chu SY, Beuhler JW. HIV infection as leading cause of death among young adults in US cities and states. JAMA 1993, 269:2991-2999.
4. Mann J. AIDS - the second decade: a global perspective. J Infect Dis 1992, 165:245-250.
5. Centers for Disease Control and Prevention. Update: mortality attributable to HIV infection among persons aged 25-44 years -United States, 1994. MMWR 1996, 45:121-125.
6. Perloff JD, LeBailly SA, Kletke PR, Bedetti PP, Conelli JP. Premature death in the United States: years of life lost and health priorities. J Public Health Pol 1984, 5:167-184.
7. Romeder JM, McWhinnie JR. Potential years of life lost between ages 1 and 70: an indicator of premature mortality for health planning, Int J Epid 1977, 6:143-151.
8. Centers for Disease Control and Prevention. Premature mortality in the United States: public health issues in the use of years of potential life lost. MMWR 1986, 35:1S-11S.
9. Selik RM, Chu SY. Years of potential life lost due to HIV infection in the United States. AIDS 1997, 11:1635-1639.
10. Obiri GU, Fordyce EJ, Singh TP, Forlenza S. Effect of HIV/AIDS versus other causes of death on premature mortality in New York City, 1983-1994. Am J Epidemiol 1998, 147:840-845.
11. Saunders LD, Rutherford GW, Lemp GF, Barnhart JL. Impact of AIDS on mortality in San Francisco, 1976-1986. J Acquir Immune Defic Syndr 1990, 3:921-925.
12. Bindels PJ, Reijneveld SA, Mulder-Folkerts DK, Coutinho RA, van den Hoek AA. Impact of AIDS on premature mortality in Amsterdam 1982-1992. AIDS 1994, 8:233-237.
13. Aldous J., Hickman M, Ellam A, Gazzard B, Hargreaves S. Impact of HIV infection on mortality in young men in a London Health Authority. BMJ 1992, 305:219-221.
14. World Health Organization. Manual of the International Statistical Classification of Diseases, 9th Revision, Vol. 1. Geneva: WHO; 1977.
15. Bobby JJ, Spencer PD, Wyatt JC, Farmer RDT. AIDS death in the UK - how complete are the figures? Public Health 1988, 102:519-524.
16. Hessol A, Buchbinder SP, Colbert D, et al.. Impact of HIV infection on mortality and accuracy of AIDS reporting on death certificates. Am J Public Health 1992, 82:561-565.
17. McCormick A. Unrecognised HIV related deaths. BMJ 1991, 302:1365-1367.
18. Office for National Statistics. Mortality statistics, general, 1993, 1994 and 1995, England and Wales. Series DH1 no. 28. London: The Stationary Office.
19. King MB. AIDS on the death certificate: the final stigma. For debate. BMJ 1989, 298:734-736.
20. Laurichesse HA. Mortimer J, Evans BG, Farrington CP. Pre-AIDS mortality in HIV-infected individuals in England, Wales and Northern Ireland, 1982-1996. AIDS 1998, 12:651-658.
21. Gold J, Li Y, Kaldor JM. Premature mortality in Australia 1983-1992, the first decade of the AIDS epidemic. Med J Aust 1994, 161: 652-656.
22. Chu S, Buehler JW, Lieb L, et al. Causes of death among persons reported with AIDS Am J Public Health 1993, 83:1429-1431.
23. Evans B, McCormick A. Completeness of reporting of AIDS by clinicians. J Roy Statist Soc A 1994, 157:105-114.
24. Report of an Expert Group (Chairman: Day NE) convened by the Director of the Public Health Laboratory Service on behalf of the Chief Medical Officers. The incidence and prevalence of AIDS and prevalence of other HIV disease in England and Wales for 1995 to 1999: projections using data to the end of 1994. Commun Dis Rep 1996: 6:R1-R23.
25. Evans BG. The extent of underreporting of AIDS cases in England, Wales and Northern Ireland. DMed thesis, Oxford University 1995.
26. O‚Donnell I. Farmer R. The limitations of official suicide statistics. Br J Psych 1995, 166:458-461.
27. Gardner JW, Sanborn JS. Years of potential life lost (YPLL) - what does it measure? Epidemiology 1990, 1: 322-329.
28. Charlton J, Kelly S, Dunnell K, Evans B, Jenkins R. Suicide deaths in England and Wales: trends in factors associated with suicide deaths. Pop Trend 1993, 71: 34-42.
© 1999 Lippincott Williams & Wilkins, Inc.