Trends in mortality statistics in England and Wales with particular reference to AIDS from 1984 to April 1987

Death certification should be able to provide accurate data on the number of deaths due to AIDS as a basis for predicting future deaths from the syndrome. Trends in deaths from other causes may identify conditions that have not been recognised to be associated with HIV infection. Mortality statistics with reference to AIDS in England and Wales were completed from death certificates. Increases in deaths from selected causes likely to be associated with AIDS or HIV infection suggested that in some patients with HIV infection, AIDS was not stated on the death certificate or subsequently notified by the doctor who signed the certificate. From calculations of excess deaths between the beginning of 1985 and the end of April 1987, compared with 1984 at least 495 deaths possibly associated with HIV infection were estimated to have occurred among men aged 15-54 during that period. In 261 AIDS or HIV infection was stated on the original or amended death entry as the cause of death, and of these 198 were included in the estimated number of excess deaths.Accurate notification of the underlying cause of death and associated diseases is required for the precise monitoring of trends in mortality from AIDS and possible identification of unrecognised conditions associated with HIV infection.     

Estimating the size of the HIV epidemic by using mortality data

Evidence that more people are dying as a result of HIV infection than is reflected by the number of deaths among reported cases meeting the WHO definition of AIDS is derived from mortality data. Ninety-five causes of death likely to be associated with HIV infection were selected. Standardized mortality ratios due to these causes increased for single men aged 15-54 years from 100 in 1984 to 118 in 1987. The age, sex, marital status, temporal and geographic distribution of these excess deaths suggest that they are HIV-associated. It is estimated that 58% of excess deaths due to HIV-related causes were among cases reported to the CDSC AIDS Surveillance Programme in 1987. Some of these deaths may have been among HIV-positive people who did not meet the WHO definition at the time of death. There is a need for surveillance to be extended to include HIV-positive people who die before meeting the WHO definition if the full extent of the HIV epidemic is to be identified.     

Impact of HIV infection on mortality in young men in a London health authority.

OBJECTIVE--To determine the number of deaths attributable to HIV infection among men aged 15-64 in a geographically defined population in the United Kingdom. DESIGN--Retrospective review of death certificates and linkage with local and national HIV and AIDS surveillance data. SETTING--Riverside District Health Authority, London. MAIN OUTCOME MEASURES--Numbers of deaths attributed to HIV infection in male residents of Riverside aged 15-64 and 15-44 over a six month period. Proportion of attributed deaths were (i) identified from death certificates by the Office of Population Censuses and Surveys as being due to HIV infection and (ii) reported as cases of AIDS or HIV related deaths to the Public Health Laboratory Service Communicable Disease Surveillance Centre. RESULTS--34 of 213 (16%) deaths in men aged 15-64 and 27 of 69 (39%) deaths in men aged 15-44 were attributed to HIV infection. Six of 33 (18%) attributed deaths were identified by the Office of Population Censuses and Surveys and 32/34 (94%) were reported to the Communicable Disease Surveillance Centre. CONCLUSIONS--HIV infection was the leading cause of death in male residents of Riverside aged 15-44 and the third commonest cause of death in those aged 15-64. Most individuals dying of known HIV infection were reported to the Communicable Disease Surveillance Centre but identification of the true cause of death from the process of death certification was poor. Measures to improve the certification of HIV and AIDS or the use of AIDS surveillance information correctly to code the cause of death needs to be considered to ensure that the true impact of HIV infection is reflected in routine mortality statistics.     

Incidence of AIDS and excess of mortality associated with HIV in haemophiliacs in the United Kingdom: report on behalf of the directors of haemophilia centres in the United Kingdom.

OBJECTIVE--To estimate the cumulative incidence of AIDS by time since seroconversion in haemophiliacs positive for HIV and to examine the evidence for excess mortality associated with HIV in those who had not yet been diagnosed as having AIDS. DESIGN--Analysis of data from ongoing national surveys. SETTING--Haemophilia centres in the United Kingdom. PATIENTS--A total of 1201 men with haemophilia who had lived in the United Kingdom during 1980-7 and were positive for HIV. INTERVENTION--None. END POINTS--Diagnosis of AIDS; death in those not diagnosed as having AIDS. MEASUREMENTS AND MAIN RESULTS--Estimation of cumulative incidence of AIDS and number of excess deaths in seropositive patients not diagnosed with AIDS. Median follow up after seroconversion was 5 years 2 months. Eight five patients developed AIDS. Cumulative incidence of AIDS five years after seroconversion was 4% among patients aged less than 25 at first test positive for HIV, 6% among those aged 25-44, and 19% among those aged greater than or equal to 45. There was little evidence that type or severity of haemophilia or type of factor VIII or IX that had caused HIV infection affected the rate of progression to AIDS. Mortality was increased among those who had not been diagnosed as having AIDS, especially among those with "AIDS related complex." Thirteen deaths were observed among 36 patients diagnosed as having AIDS related complex against 0.65 expected, and 34 deaths in 1080 other patients against 22.77 expected; both calculations were based on mortality rates observed in haemophiliacs in the United Kingdom in the late 1970s. CONCLUSIONS--Rate of progression to AIDS depended strongly on age. There is a substantial burden of fatal disease among patients positive for HIV who have not been formally diagnosed as having AIDS.     

Death certification in general practice: review of records.

The records of death that had been certified by general practitioners in one practice over 18 years were assessed in the light of the recent joint publications of the Royal College of Physicians and the Royal College of Pathologists. Over this period roughly 30% of the deaths in the practice population occurred outside hospital and a total of 262 certificates were issued. A review of 262 counterfoils of records of death certification showed that 12 counterfoils (4.6%) had no age and sex mentioned, and three counterfoils did not describe the place of death. The average age at death outside hospital was 71.6 years--the age of women being 75.1 years compared with that of men of 68.2 years. Only 2% of patients had had a necropsy. The common causes of death stated in the certificates were: cardiovascular 41%, carcinoma 35%, respiratory 15%, and stroke 8%. All contributory causes are also mentioned. Ninety seven per cent of the patients were seen after death by the doctors in the practice and 68% had been seen in the two days preceding death. We emphasise the importance of keeping accurate records of deaths in general practice for audit and research as well as for planning services for terminally ill and recently bereaved patients.     

Causes of death and case fatality rates among infants with down syndrome in metropolitan Atlanta

BACKGROUND: There is limited population-based information on the extent of underreporting of congenital heart defects (CHD) as a cause of death among infants with Down syndrome (DS) and on the variation in case fatality by presence of CHD and age at death. METHODS: Using data from the Metropolitan Atlanta Congenital Defects Program (MACDP), we identified infants with DS born 1979-2003. We used data from Georgia death certificates and the National Death Index to determine vital status and identify causes of death. Using MACDP records as a reference, we calculated the sensitivity and positive predictive value of reports of CHD as any cause of death or contributing condition in death certificates. We calculated race-specific case fatality rate by infant's age at death and presence of CHD. RESULTS: CHD was the most frequently reported cause of death from death certificates; however, a review of causes of death and birth defects data indicated a potentially greater impact of CHD among DS infant deaths than could be determined from the reported cause of death. The case fatality rate among infants with DS was significantly higher among blacks than whites, with the greatest racial disparity observed among infants without CHD who died in the post-neonatal period. CONCLUSIONS: Efforts are needed to improve reporting of causes of death related to CHD among infants with DS that would allow for a clearer assessment of determinants of case fatality among DS infants and identification of possible ways to reduce the racial disparities.     

Causes of Death among AIDS Patients after Introduction of Free Combination Antiretroviral Therapy (cART) in Three Chinese Provinces, 2010–2011

Introduction

Although AIDS-related deaths have had significant economic and social impact following an increased disease burden internationally, few studies have evaluated the cause of AIDS-related deaths among patients with AIDS on combination anti-retroviral therapy (cART) in China. This study examines the causes of death among AIDS-patients in China and uses a methodology to increase data accuracy compared to the previous studies on AIDS-related mortality in China, that have taken the reported cause of death in the National HIV Registry at face-value.

Methods

Death certificates/medical records were examined and a cross-sectional survey was conducted in three provinces to verify the causes of death among AIDS patients who died between January 1, 2010 and June 30, 2011. Chi-square analysis was conducted to examine the categorical variables by causes of death and by ART status. Univariate and multivariate logistic regression were used to evaluate factors associated with AIDS-related death versus non-AIDS related death.

Results

This study used a sample of 1,109 subjects. The average age at death was 44.5 years. AIDS-related deaths were significantly higher than non-AIDS and injury-related deaths. In the sample, 41.9% (465/1109) were deceased within a year of HIV diagnosis and 52.7% (584/1109) of the deceased AIDS patients were not on cART. For AIDS-related deaths (n = 798), statistically significant factors included CD4 count <200 cells/mm³ at the time of cART initiation (AOR 1.94, 95%CI 1.24–3.05), ART naïve (AOR 1.69, 95%CI 1.09–2.61; p = 0.019) and age <39 years (AOR 2.96, 95%CI 1.77–4.96).

Conclusion

For the AIDS patients that were deceased, only those who initiated cART while at a CD4 count ≥200 cells/mm3 were less likely to die from AIDS-related causes compared to those who didn’t initiate ART at all.     

CYSTIC FIBROSIS

To determine the actual number of deaths from cystic fibrosis reported in 1957 in California, all death certificates mentioning any one of the various terms used to describe the condition were reviewed.As a cause of death, cystic fibrosis is infrequently mentioned on death certificates. In 1957, only 42 of a total of 124,082 death certificates mentioned cystic fibrosis. Half of the deaths occurred in persons less than 12 months of age. There was no significant difference by sex between deaths due to cystic fibrosis and deaths from all causes. All deaths occurred in the white population.     

Continuous Increase of Cardiovascular Diseases,Diabetes, and Non-HIV Related Cancers as Causes of Death in HIV-Infected Individuals in Brazil: An Analysis of Nationwide Data

Introduction

After antiretroviral therapy (ART) became available, there was a decline in the number of deaths in persons infected with HIV. Thereafter, there was a decrease in the proportion of deaths attributed to opportunistic infections and an increase in the proportion of deaths attributed to chronic comorbidities. Herein we extend previous observations from a nationwide survey on temporal trends in causes of death in HIV-infected patients in Brazil.

Methods

We describe temporal trends in causes of death among adults who had HIV/AIDS listed in the death certificate to those who did not. All death certificates issued in Brazil from 1999 to 2011 and listed in the national mortality database were included. Generalized linear mixed-effects logistic models were used to study temporal trends in proportions.

Results

In the HIV-infected population, there was an annual adjusted average increase of 6.0%, 12.0%, 4.0% and 4.1% for cancer, external causes, cardiovascular diseases (CVD) and diabetes mellitus (DM), respectively, compared to 3.0%, 4.0%, 1.0% and 3.9%, in the non-HIV group. For tuberculosis (TB), there was an adjusted average increase of 0.3%/year and a decrease of 3.0%/year in the HIV and the non-HIV groups, respectively. Compared to 1999, the odds ratio (OR) for cancer, external causes, CVD, DM, or TB in the HIV group were, respectively, 2.31, 4.17, 1.76, 2.27 and 1.02, while for the non-HIV group, the corresponding OR were 1.31, 1.63, 1.14, 1.62 and 0.67. Interactions between year as a continuous or categorical variable and HIV were significant (p<0.001) for all conditions, except for DM when year was considered as a continuous variable (p = 0.76).

Conclusions

Non HIV-related co-morbidities continue to increase more rapidly as causes of death among HIV-infected individuals than in those without HIV infection, highlighting the need for targeting prevention measures and surveillance for chronic diseases among those patients.     

Contribution of deaths related to alcohol use of socioeconomic variation in mortality: register based follow up study.

OBJECTIVE: To estimate the contribution of excessive alcohol use to socioeconomic variation in mortality among men and women in Finland. DESIGN: Register based follow up study. SUBJECTS: The population covered by the 1985 and 1990 censuses, aged > or = 20 in the follow up period 1987-93. MAIN OUTCOME MEASURES: Total mortality and alcohol related mortality from all causes, from diseases, and from accidents and violence according to socioeconomic position. The excess mortality among other classes compared with upper non-manual employees and differences in life expectancy between the classes were used to measure mortality differentials. RESULTS: Alcohol related mortality constituted 11% of all mortality among men aged > or = 20 and 2% among women and was higher among manual workers than among other classes. It accounted for 14% of the excess all cause mortality among manual workers over upper non-manual employees among men and 4% among women and for 24% and 9% of the differences in life expectancy, respectively. Half of the excess mortality from accidents and violence among male manual workers and 38% among female manual workers was accounted for by alcohol related deaths, whereas in diseases the role of alcohol was modest. The contribution of alcohol related deaths to relative mortality differentials weakened with age. CONCLUSIONS: Class differentials in alcohol related mortality are an important factor in the socioeconomic mortality differentials in Finland, especially among men, among younger age groups, and in mortality from accidents and violence.     

Investigation into use of drugs preceding death from asthma.

Copies of death certificates were provided by the Registrar General for all deaths attributed to asthma in persons aged 5 to 34 years which were registered in England and Wales in the last quarter of 1966 and the first quarter of 1967. Information was obtained from the relevant general practitioners about 177 of the 184 subjects, and necropsy data were obtained for 113 of the 124 cases in which a post-mortem examination was known to have been made. Ninety-eight per cent. of the subjects for whom evidence was obtained were known to have been suffering from asthma, and signs of severe asthma (overdistended lungs and small bronchi plugged with mucus) were found in 91% of necropsies (57% of all deaths). Evidence that death might have been due to any other pathological condition was rare. Death was sudden and unexpected in 81% of the subjects (137 out of 171), and 59% of all deaths were referred to coroners. In 39% of cases (67 out of 171) the practitioner had not regarded the patient as suffering from severe asthma in his terminal episode. Corticosteroids and sympathomimetic preparations were the only drugs to have been used by a large proportion of patients. Two-thirds of the patients had received corticosteroids before the terminal episode, but detailed information about their use provided no suggestion that excess use could have been responsible for any large proportion of the deaths. Eighty-four per cent. of the patients were known to have used pressurized aerosol bronchodilators, and several instances of their use in excess were described. Routine inquiries about their use in the hours immediately preceding death were not made, and further evidence is required before their effect can be assessed adequately.     

Widespread body pain and mortality: prospective population based study

ObjectiveTo determine whether there is excess mortality in groups of people who report widespread body pain, and if so to establish the nature and extent of any excess.DesignProspective follow up study over eight years. Mortality rate ratios were adjusted for age group, sex, and study location.SettingNorth west England.Participants6569 people who took part in two pain surveys during 1991-2.Results1005 (15%) participants had widespread pain, 3176 (48%) had regional pain, and 2388 (36%) had no pain. During follow up mortality was higher in people with regional pain (mortality rate ratio 1.21, 95% confidence interval 1.01 to 1.44) and widespread pain (1.31, 1.05 to 1.65) than in those who reported no pain. The excess mortality among people with regional and widespread pain was almost entirely related to deaths from cancer (1.55 (1.09 to 2.19) for regional pain and 2.07 (1.37 to 3.13) for widespread pain). The excess cancer mortality remained after exclusion of people in whom cancer had been diagnosed before the original survey and after adjustment for potential confounding factors. There were also more deaths from causes other than disease (for example, accidents, suicide, violence) among people with widespread pain (5.21, 0.94 to 28.78).ConclusionThere is an intriguing association between the report of widespread pain and subsequent death from cancer in the medium and long term. This may have implications for the long term follow up of patients with “unexplained” widespread pain symptoms, such as those with fibromyalgia.

What is already known on this topic

Widespread body pain, the cardinal symptom of fibromyalgia, is commonAn organic basis for symptoms is found in only a small proportion of peopleTreatment is difficult, and studies with short term follow up have shown that symptoms commonly persist

What this study adds

This was the first study with long term follow up of people with widespread pain in the communityThese people experience an increased mortality and the excess is principally related to deaths from cancer     

Reliability of recording uterine cancer in death certification in France and age-specific proportions of deaths from cervix and corpus uteri

French uterine cancer recordings in death certificates include 60% of "uterine cancer, Not Otherwise Specified (NOS)"; this hampers the estimation of mortalities from cervix and corpus uteri cancers. The aims of this work were to study the reliability of uterine cancer recordings in death certificates using a case matching with cancer registries and estimate age-specific proportions of deaths from cervix and corpus uteri cancers among all uterine cancer deaths by a statistical approach that uses incidence and survival data. Deaths from uterine cancer between 1989 and 2001 were extracted from the French National database of causes of death and case-to-case matched to women diagnosed with uterine cancer between 1989 and 1997 in 8 cancer registries. Registry data were considered as "gold-standard". Among the 1825 matched deaths, cancer registries recorded 830 cervix and 995 corpus uteri cancers. In death certificates, 5% and 40% of "true" cervix cancers were respectively coded "corpus" and "uterus, NOS" and 5% and 59% of "true" corpus cancers respectively coded "cervix" and "uterus, NOS". Miscoding cervix cancers was more frequent at advanced ages at death and in deaths at home or in small urban areas. Miscoding corpus cancers was more frequent in deaths at home or in small urban areas. From the statistical method, the estimated proportion of deaths from cervix cancer among all uterine cancer deaths was higher than 95% in women aged 30-40 years old but declined to 35% in women older than 70 years. The study clarifies the reason for poor encoding of uterus cancer mortality and refines the estimation of mortalities from cervix and corpus uteri cancers allowing future studies on the efficacy of cervical cancer screening.     

Human immunodeficiency virus infection,hepatitis B virus infection,and sexual behaviour of women attending a genitourinary medicine clinice

During the six months immediately after a public information campaign about the acquired immune deficiency syndrome 1115 women who attended a genitourinary medicine clinic in west London were tested for antibodies to the human immunodeficiency virus (HIV). Three women (0·27%) were positive, and all three were regular sexual partners of men with high risk lifestyles—two intravenous drug users and one bisexual. A consecutive series of 647 women from the cohort was tested for antibodies for hepatitis B core antigen: 27 were positive, of whom six had been born in the United Kingdom and were not known to have been at risk. The two women who were seropositive for HIV who completed a questionnaire on their sexual behaviour before they were tested reported both anal and oral receipt of semen and were in the upper fifth percentile for lifetime sexual partners. More than half (53%) of 424 women who reported that they had non-regular sexual partners never used a condom.It is concluded that heterosexual women in London are at a low risk of becoming infected with HIV.     

Causes of Death among Commercially Insured Multiple Sclerosis Patients in the United States

Background

Information on causes of death (CODs) for patients with multiple sclerosis (MS) in the United States is sparse and limited by standard categorizations of underlying and immediate CODs on death certificates. Prior research indicated that excess mortality among MS patients was largely due to greater mortality from infectious, cardiovascular, or pulmonary causes.

Objective

To analyze disease categories in order to gain insight to pathways, which lead directly to death in MS patients.

Methods

Commercially insured MS patients enrolled in the OptumInsight Research database between 1996 and 2009 were matched to non-MS comparators on age/residence at index year and sex. The cause most-directly leading to death from the death certificate, referred to as the “principal” COD, was determined using an algorithm to minimize the selection of either MS or cardiac/pulmonary arrest as the COD. Principal CODs were categorized into MS, cancer, cardiovascular, infectious, suicide, accidental, pulmonary, other, or unknown. Infectious, cardiovascular, and pulmonary CODs were further subcategorized.

Results

30,402 MS patients were matched to 89,818 controls, with mortality rates of 899 and 446 deaths/100,000 person-years, respectively. Excluding MS, differences in mortality rate between MS patients and non-MS comparators were largely attributable to infections, cardiovascular causes, and pulmonary problems. Of the 95 excessive deaths (per 100,000 person-years) related to infectious causes, 41 (43.2%) were due to pulmonary infections and 45 (47.4%) were attributed to sepsis. Of the 46 excessive deaths (per 100,000 person-years) related to pulmonary causes, 27 (58.7%) were due to aspiration. No single diagnostic entity predominated for the 60 excessive deaths (per 100,000 person-years) attributable to cardiac CODs.

Conclusions

The principal COD algorithm improved on other methods of determining COD in MS patients from death certificates. A greater awareness of the common CODs in MS patients will allow physicians to anticipate potential problems and, thereby, improve the care that they provide.     

Causes of death in men with prostate cancer: Results from the Danish Prostate Cancer Registry (DAPROCAdata)

BackgroundCurrent knowledge of the validity of registry data on prostate cancer-specific death is limited. We aimed to determine the underlying cause of death among Danish men with prostate cancer, to estimate the level of misattribution of prostate cancer death, and to examine the risk of death from prostate cancer when accounting for competing risk of death.Material and methodsWe investigated a nationwide cohort of 15,878 prostate cancer patients diagnosed in 2010–2014; with 3343 deaths occurring through 2016. Blinded medical chart review was carried out for 670 deaths and compared to the national cause of death registry. Five death categories were defined: 1) prostate cancer-specific death, 2) other unspecified urological cancer death, 3) other cancer death 4) cardiovascular disease death, and 5) other causes of death. Competing risk analyses compared Cox cause-specific and Fine-Gray regression models.ResultsChart review attributed 51.2% of deaths to prostate cancer, 17.0% to cardiovascular disease, and 16.7% to other causes. The Danish Register of Causes of Death attributed 71.7% of deaths to prostate cancer when including all registered contributing causes of death, and 57.0% of deaths when including only the primary registered cause of death. The probability of death by prostate cancer was 10% at 2-year survival.ConclusionsMore than half of the deceased men in our study cohort died of their prostate cancer disease within a mean of 2.4 years of follow up. Data from the death registry is prone to misclassification, potentially overestimating the proportion of deaths from prostate cancer.     

Increase in non-AIDS related conditions as causes of death among HIV-infected individuals in the HAART era in Brazil

Background

In 1996, Brazil became the first developing country to provide free and universal access to HAART. Although a decrease in overall mortality has been documented, there are no published data on the impact of HAART on causes of death among HIV-infected individuals in Brazil. We assessed temporal trends of mortality due to cardiovascular diseases (CVD), diabetes mellitus (DM) and other conditions generally not associated with HIV-infection among persons with and without HIV infection in Brazil between 1999 and 2004.

Methodology/Principal Findings

Odds ratios were used to compare causes of death in individuals who had HIV/AIDS listed on any field of the death certificate with those who did not. Logistic regression models were fitted with generalized estimating equations to account for spatial correlation; co-variables were added to the models to control for potential confounding. Of 5,856,056 deaths reported in Brazil between 1999 and 2004 67,249 (1.15%) had HIV/AIDS listed on the death certificate and non-HIV-related conditions were listed on 16.3% in 1999, increasing to 24.1% by 2004 (p<0.001). The adjusted average yearly increases were 8% and 0.8% for CVD (p<0.001), and 12% and 2.8% for DM (p<0.001), for those who had and did not have HIV/AIDS listed on the death certificate, respectively. Similar results were found for these conditions as underlying causes of death.

Conclusions/Significance

In Brazil between 1999 and 2004 conditions usually considered not to be related to HIV-infection appeared to become more likely causes of death over time than reported causes of death among individuals who had HIV/AIDS listed on the death certificate than in those who did not. This observation has important programmatic implications for developing countries that are scaling-up access to antiretroviral therapy.     

Plasma cholesterol,coronary heart disease,and cancer in the Renfrew and Paisley survey.

The relation between plasma cholesterol concentration and mortality from coronary heart disease, incidence of and mortality from cancer, and all cause mortality was studied in a general population aged 45-64 living in the west of Scotland. Seven thousand men (yielding 653 deaths from coronary heart disease, 630 new cases of cancer, and 463 deaths from cancer) and 8262 women (322 deaths from coronary heart disease, 554 new cases of cancer, and 395 deaths from cancer) were examined initially in 1972-6 and followed up for an average of 12 years. All cause mortality was not related to plasma cholesterol concentration. This was largely a consequence of a positive relation between cholesterol values and mortality from coronary heart disease being balanced by inverse relations between cholesterol and cancer and between cholesterol and other causes of death. These changes were highly significant for coronary heart disease and cancer in men and significant for coronary heart disease and other causes of death in women. The inverse association between cholesterol concentration and cancer in men was strongest for lung cancer, was not merely a function of the age at which a subject died, was present for the incidence of cancer as well as mortality from cancer, and persisted when new cases or deaths occurring within the first four years of follow up were excluded from the analysis.     

Late deaths and survival after childhood cancer: implications for cure.

OBJECTIVES--To investigate causes of death and survival in subjects who had survived at least five years after diagnosis of childhood cancer; to compare observed mortality with that expected in the general population; and to compare results with a corresponding cohort diagnosed earlier. DESIGN--Retrospective cohort study. SETTING--Population based National Register of Childhood Tumours. SUBJECTS--9080 five year survivors of childhood cancer diagnosed in Britain during 1971-85, of whom 793 had died. Comparison with corresponding cohort diagnosed during 1940-70. MAIN OUTCOME MEASURES--Cause of death established from all available sources of information (including hospital and general practitioner records and postmortem reports) and underlying cause of death coded on death certificate. RESULTS--Of the 781 deaths for which sufficient information was available, death was attributed to recurrent tumour in 578 (74%) cases, treatment related effect in 121 (15%), second primary tumour in 52 (7%), and other causes in 30 (4%). Comparison of observed mortality with that expected in the general population indicated a fourfold excess of deaths from non-neoplastic causes. The risk of dying of recurrent tumour in the next 10 years after surviving five years from diagnosis during 1940-70 and 1971-85 fell from 12% to 8%. The risk of dying from a treatment related effect increased slightly from 1% to 2%. CONCLUSION--Improvements in five year survival after childhood cancer have been accompanied by a reduction in risk of dying from recurrent tumour during the subsequent 10 years and by a slight increase in risk of dying from treatment related effects. The results provide information relevant to decisions concerning balance between effective treatments and their potentially harmful effects.