Changes in Causes of Death Among Adults Infected by HIV Between 2000 and 2005:
The Mortalite 2000 and 2005 Surveys (ANRS EN19 and Mortavic)
[Epidemiology and Social Science]
JAIDS Journal of Acquired Immune Deficiency Syndromes:Volume 48(5)15 August 2008pp 590-598
"Between 2000 and 2005, the proportion of AIDS-related deaths continued to decrease among HIV-infected adults, but it remained the most frequent underlying cause of death, mainly related to NHL. The distribution of other causes of death was heterogeneous, whereas 3 causes increased and accounted for 40% of them: non-AIDS-defining cancer, liver-related diseases, and cardiovascular deaths. Our results may be a consequence of a suboptimal detection or management of both HIV infection and viral hepatitis coinfections, and also aging of HIV-infected individuals and a high prevalence of traditional determinants predisposing to cancers or cardiovascular diseases."
"Prevention and early detection of serious morbidity may include improved strategies to diagnose HIV infection earlier, particularly in migrants and people with poor socioeconomic conditions; maintaining efforts to reach high levels of CD4 cell count to prevent both persistent AIDS and non-AIDS-defining events related to immune suppression and including a proactive identification of obstacles to optimal response to treatment; implementing active prevention of the most frequent cancers, particularly through the promotion of smoking cessation programs; evaluating the benefits of targeted screening of cancers in patients at risk; improving HCV treatment coverage and efficacy; and improving prevention and management of diabetes and dyslipidemia to limit the occurrence of cardiovascular disease. The implementation of these recommendations will be especially important in the context of an aging HIV-infected population."
Lewden, Charlotte MD, PhD*; May, Thierry MD; Rosenthal, Eric MD, PhD; Burty, Christine MD; Bonnet, Fabrice MD, PhD*_; Costagliola, Dominique PhD#; Jougla, Eric PhD**; Semaille, Caroline MD, MPH; Morlat, Philippe MD, PhD*_; Salmon, Dominique MD, PhD; Cacoub, Patrice MD, PhD***; Chene, Genevieve MD, PhD*_; on behalf of the ANRS EN19 Mortalite Study Group and Mortavic1
From the *INSERM, U897, Bordeaux, France; Universite Victor Segalen Bordeaux 2, Institut de Sante Publique, d'Epidemiologie et de Developpement, Bordeaux, France; Centre Hospitalier Universitaire Brabois, Vandoeuvre-Les-Nancy, France; Centre Hospitalier Universitaire L'Archet, Nice, France; _Centre Hospitalier Universitaire, Bordeaux, France; INSERM, U720, Paris, France; #UPMC Univ Paris 06, UMR S 720, Paris, France; **INSERM, CepiDc, Le Vesinet, France; Institut de Veille Sanitaire, Saint-Maurice, France; Centre Hospitalier Universitaire Cochin-Tarnier, Paris, France; Centre Hospitalier Universitaire La Pitie Salpetriere, Paris, France; ***Universite Pierre et Marie Curie-Paris 6, CNRS, UMR 7087, Paris, France.
Received for publication January 25, 2008; accepted May 6, 2008.
Supported by Agence Nationale de Recherches sur le Sida et les Hepatites Virales (through a grant from the CSS5).
These results were presented partly at the 14th Conference on Retroviruses and Opportunistic Infections, February 25-28, 2007, Los Angeles, CA.
Background: The multicenter national Mortalite 2005 survey aimed at describing the distribution of causes of death among HIV-infected adults in France in 2005 and its changes as compared with 2000.
Methods: Physicians involved in the management of HIV infection notified deaths and documented the causes using a standardized questionnaire similar to the previous survey performed in 2000.
Results: Overall, 1042 deaths were notified in 2005 (vs 964 in 2000): with median age, 46 years (vs 41 years); men, 76%; and median last CD4 cell count, 161/mm3 (vs 94). The proportion of underlying causes of death due to AIDS decreased (36% in 2005 vs 47% in 2000), and the proportion of cancer not related to AIDS or hepatitis (17% vs 11%), liver related disease (15% vs 13%: hepatitis C, 11%, and hepatitis B, 2%), cardiovascular disease (8% vs 7%), or suicide (5% vs 4%) increased. Among the 375 AIDS-related deaths, the most frequent event was non-Hodgkin lymphoma (28%). Among cancers not related to AIDS or hepatitis, the most frequent localizations were lung (31%) and digestive tract (14%). Among the 154 liver-related deaths, 24% were due to hepatocarcinoma.
Conclusions: The heterogeneity of causes of death among HIV-infected adults was confirmed and intensified in 2005, with 3 causes following AIDS: cancers and liver-related and cardiovascular diseases.
As they live longer, individuals infected by the HIV and treated with combination antiretroviral therapy (cART) may experience a wider range of non-AIDS-related complications than in the pre-cART period. In addition to the impact of their chronic viral infection, aging and long-term exposure to treatment and to traditional risk factors such as smoking, alcohol consumption, or dyslipidemia currently contribute to a diversification of morbidity and of the causes of death.1,2 For those infected with hepatitis B or C virus (HBV or HCV), hepatic complications may occur during prolonged survival.3-6 In this evolving context, the surveillance of causes of death contributes to assess priorities in prevention, care, and future research. In 2000, the Mortalite 2000 survey showed the persistence of AIDS-related deaths and emergence of cancers and hepatitis-related deaths in France,2 as confirmed by other studies in Europe, Australia, and the United States.1,5,7-11 Between 2000 and 2005, major changes in HIV case management have occurred. Combination antiretroviral therapy has changed toward simplified and more efficacious combinations for all lines of treatment indications, with a better tolerance in the short term, although the risk of cardiovascular disease increases with longer exposure to protease inhibitors.12 In parallel, treatments were started at lower CD4 cell counts than before 2000, and strategies of treatment interruption have been introduced and evaluated as less effective than continuous treatment.13 In addition, treatment of HCV coinfection has been formally evaluated in this population of patients14 and concerned 1 in 4 HIV-infected persons in some areas in 2004.15 The Mortalite 2005 survey aimed at describing the distribution of causes of death among HIV-infected adults in France in 2005 and at comparing it with the distribution in 2000.
Between 2000 and 2005, the proportion of AIDS-related deaths continued to decrease among HIV-infected adults, but it remained the most frequent underlying cause of death, mainly related to NHL. The distribution of other causes of death was heterogeneous, whereas 3 causes increased and accounted for 40% of them: non-AIDS-defining cancer, liver-related diseases, and cardiovascular deaths. Our results may be a consequence of a suboptimal detection or management of both HIV infection and viral hepatitis coinfections, and also aging of HIV-infected individuals and a high prevalence of traditional determinants predisposing to cancers or cardiovascular diseases.
Interpretation and Relation to Other Studies
Other recent studies have reported that AIDS remains the most frequent cause of death,1,8,10,17 whereas HIV-related causes of death decrease in the cART period.7,9,18 The distribution of non-AIDS-defining causes of death largely varies according to specific characteristics of the studied population. Among individuals followed after the onset of AIDS in New York, substance abuse was the most frequent non-HIV-related cause of death between 1999 and 2004,7 whereas until 2004, cancer was the most frequent non-AIDS-defining cause of death in the Australian HIV Observational Database8 and in the US HIV Outpatient Study.9 Liver disease was the most frequent non-AIDS-defining cause of death in the cART period among HIV-infected individuals with hemophilia in Canada and in the DAD international collaboration (23% and 67% infected with HCV, respectively).5,19
In our study, 1 in 3 deaths (344, 33%) was related to a cancer, either AIDS or non-AIDS-defining cancer. When taking into account the age at death, this was not statistically different from the proportion of 27% reported in 2000. In the French general population, 39% of the deaths between the age of 35 and 54 years were related to cancer in 2004, but their localizations were different than those in HIV-infected individuals.20 Among HIV-infected individuals, NHL remains the most frequent AIDS-defining event leading to death. The use of cART is associated with an overall decrease in the risk of NHL,21,22 although to a lesser extent than other AIDS-defining events.23 The risk of NHL decreases as the CD4 cell count increases,24 although the level of CD4 allowing HIV-infected people to reach the risk of NHL reported in the general population is not known.25,26 After diagnosis, the prognosis of NHL in cART-treated patients seems to be exclusively associated with tumor-related factors.27
Among other cancers, 38% of non-AIDS non-hepatitis-related cancers were located in the respiratory tract. The risk of cancer is higher in HIV-infected adults than in the general population,25,28 and smoking plays a major role, as around half of HIV-infected adults are current smokers, a proportion that remains stable over time.29,30 Adapted smoking cessation programs in HIV-infected persons still have to be evaluated,29,31 and the benefit of specific targeted screening of cancers too.32 However, HIV itself may play a specific role in the occurrence of lung cancer, regardless of smoking status, as reported among drug users in the United States.33
Hepatitis C virus was involved in 78% of liver-related deaths, and the proportion of hepatocarcinoma increased over time (16% in 2000 and 24% in 2005). Excessive alcohol consumption was reported in half of these cases, and two-thirds were infected by HIV through injecting drug use. In France, 90% of HIV-infected adults contaminated through injecting drug use are HCV infected.15 Despite improvement in the management of HIV-HCV coinfection, not all HIV/HCV-coinfected patients who would be eligible actually receive treatment for HCV.34,35 An improvement is expected in the future because new treatments are currently evaluated. Anti-HCV treatment should be largely proposed even in patients with cirrhosis, provided that they have no decompensation because early HCV viral kinetics allows to predict sustained virological response and to stop therapy in case of defavorable prognosis factors. In addition, most French HIV-infected patients are coinfected by genotype 4 that is associated with poorer outcome.36
The proportion of cardiovascular-related deaths only slightly increased. Improvement of antiretroviral strategies and management of dyslipidemia may have slowed an initially worse trend.12 Nevertheless, the relative contribution of HIV infection, antiretrovirals, and traditional risk factors in the occurrence of metabolic-related diseases is still debated.37
Although our observations have been made in the French context of free access to care, we believe that some characteristics of deceased patients reflect suboptimal management. The proportion of patients who died within 6 months of HIV diagnosis (9%) remained stable between 2000 and 2005. In London between 1998 and 2003, around one-fifth of all deaths occurred among individuals who had first been diagnosed with HIV within the 6 months before death.10 In France, more than 1 in 3 patients attending HIV care for the first time have advanced HIV infection,38 and delayed access to care was more frequent in migrants, as reported in other settings.39 In addition, 1 in 3 HIV-infected adults who died in 2005 was in poor socioeconomic conditions. The association between socioeconomic conditions and premature mortality is a constant in the general population40 and applies to HIV-infected populations.41
Over time, the latest CD4 cell count before death increased, as also reported by others.9 Nevertheless, despite 81% of our series had received cART, latest CD4 cell count was less than 200/mm3 among half of them. Median latest CD4 cell count was lower in people who died of AIDS and was around 200/mm3 in those who died of non-AIDS non-hepatitis-related cancers and of ESLD. Recent communications underlined that the risk of serious non-AIDS-defining events is higher with lower CD4 cell counts,42-44 and mortality rates in treated HIV-infected adults may reach those of the general population in the long term only when CD4 cell counts are above 500/mm3.45 Even if they are mostly treated, patients still experience immunodeficiency. In 2005, median CD4 cell count of patients followed in the French hospital database on HIV (n = 48,410) was 443/mm3 and 11% had CD4 less than 200/mm3.46
Strengths and Limitations
A larger number of wards participated in the current study compared with 2000. Nevertheless, we compared the distribution of causes of death regardless of participating wards. The overall distribution of the underlying causes of death in 2005 did not differ in the subgroup of wards that participated in the 2 surveys. Therefore, selecting only these wards would not have changed the results. Wards that participated in the Mortalite 2005 survey represented around 78,000 HIV-infected patients, whereas based on currently available data,47-49 we may consider that 100,000 individuals were living with HIV in France in 2005.
Few surveys like Mortalite 2000 and Mortalite 2005 are specifically implemented to analyze the causes of death in HIV-infected individuals. This survey is based on specific declarations of physicians and active monitoring and has advantages over existing observational cohorts. First, it may be more representative of the population of HIV-infected patients who died than studies of the causes of death in observational cohorts, which may select the patients under follow-up. Second, the determination of causes of death is based on detailed and standardized documentation that is rarely available in cohorts.
Implication for Prevention, Care, and Research
Given the confirmed emergence of heterogeneous causes of death and their link with specific characteristics of patients, surveillance of the causes of death over time among HIV-infected individuals across different geographical areas is important to adapt recommendations to specific context. In addition, it seems highly desirable to implement collection of all causes of severe morbidity in existing observational studies on HIV to allow earlier interventions in HIV case management. Indeed, there is still a potential to improve antiretroviral management in HIV-infected individuals because AIDS-related deaths still occur. Prevention and early detection of serious morbidity may include improved strategies to diagnose HIV infection earlier, particularly in migrants and people with poor socioeconomic conditions; maintaining efforts to reach high levels of CD4 cell count to prevent both persistent AIDS and non-AIDS-defining events related to immune suppression and including a proactive identification of obstacles to optimal response to treatment; implementing active prevention of the most frequent cancers, particularly through the promotion of smoking cessation programs; evaluating the benefits of targeted screening of cancers in patients at risk; improving HCV treatment coverage and efficacy; and improving prevention and management of diabetes and dyslipidemia to limit the occurrence of cardiovascular disease. The implementation of these recommendations will be especially important in the context of an aging HIV-infected population.
Overall, 341 wards participated in the survey, representing around 78,000 HIV-infected patients with at least 1 contact in 2004. They reported 1042 deaths in 2005 (vs 964 in 2000). Among them, detailed documentation was available for 1013 (97%): 76% were men; median age was 46 years, with interquartile range (IQR) of 40-54 (vs age of 41 years, with IQR of 36-49 in 2000); and known duration of HIV infection was 12 years, with IQR 6-17 (vs duration of 8 years, with IQR of 4-12) (Table 1). Overall, 87% had received antiretroviral treatment (vs 86%); 47% had plasma HIV RNA of less than 500 copies per milliliter (vs 33%); and median CD4 cell count was 161/mm3, with IQR of 41-340 (vs 94, with IQR of 19-260). Median time between last CD4 measurement and death was 2 months. CD4 cell count was below 200/mm3 in 55% (vs 68%) and above 500/mm3 in 12% (vs 9%).
Underlying Causes of Death
The most frequent underlying causes of death were an AIDS-defining illness (n = 375; 36% in 2005 vs 47% in 2000), cancer not related to AIDS or hepatitis (n = 173; 17% vs 11%), end-stage liver disease (ESLD; n = 154; 15% vs 13%: hepatitis C, 11%, and hepatitis B, 2%), cardiovascular disease (n = 88; 8% vs 7%), suicide (n = 50; 5% vs 4%), and other infections (n = 46; 4% vs 7%) (Fig. 1). Among the 375 AIDS-related deaths, the most frequent contributing events were non-Hodgkin lymphoma (NHL; 28%) and progressive multifocal leukoencephalopathy (PML; 14%) (Fig. 2). Cancers not related to AIDS or hepatitis were mainly the following: pulmonary (31%), gastrointestinal (14%), ear-nose-throat located (7%), and anal (6%). Among the 154 liver-related deaths, 37 were due to hepatocarcinoma (HCV: 28, HBV: 6, HCV and HBV: 3), 89 ESLD due to HCV, 11 ESLD due to HBV, and 11 alcohol related. Among the 88 cardiovascular-related deaths, 33 were related to coronary artery disease, 22 to stroke, 12 to heart failure, 6 to pulmonary hypertension, 5 to venous thrombosis, 4 to valvular disease, 2 to arterial thrombosis, 1 to aortic aneurysm, and 3 to other causes. In 4 patients in 2005 (vs 10 in 2000), antiretroviral treatment was considered the underlying cause of death, that is, lactic acidosis (n = 4) associated with pancreatitis in 1 patient and with renal deficiency in another patient. In 14 additional cases in 2005 (vs 47 in 2000), antiretroviral treatment was mentioned as a contributor, with the following underlying causes of death: AIDS (n = 5), liver-related disease (n = 3), endocarditis, cancer, digestive tract-related disease, injury, suicide, and unknown (n = 1 for each cause).
FIGURE 1. Underlying cause of death in HIV-infected adults: Overall distribution in 2000 (n = 964) and 2005 (n = 1042), and most recent CD4 cell count by cause; Mortalite 2000 and 2005 surveys, France.
FIGURE 2. AIDS-defining illnesses among people with AIDS-related underlying cause of death (n = 337) according to time since HIV diagnosis; Mortalite 2005 survey, France.
Among the 1042 deaths reported by 168 wards, 889 were identified by 117 wards that also participated in 2000 survey. The distribution of causes of death reported in 2005 by the wards that participated in the 2 surveys (2000 and 2005) was similar to the overall distribution. The 153 cases documented and reported by new participating wards were more frequently related to AIDS (39% vs 36%), to non-AIDS non-hepatitis-related cancers (24% vs 15%), and to liver disease (17% vs 14%), and less frequently related to cardiovascular disease (4% vs 9%) (P = 0.006).
Characteristics According to the Cause of Death
In 2005, individuals who died of non-AIDS non-hepatitis-related cancers and of cardiovascular disease were older than others (Table 1). Women, people nonnative from France, and those in poor socioeconomic conditions were more frequently represented in individuals who died of AIDS than in those who died of other causes. Median duration since HIV diagnosis was shorter in those who died of AIDS because 20% of them had HIV diagnosed within 6 months of death compared with 4% of those who died of other causes. Individuals who died of liver disease had the longest duration since HIV diagnosis. CD4 cell count was lower and HIV RNA higher in people who died of an AIDS-defining cause. The median CD4 cell count was 77/mm3 (10th-90th percentile: 5-443) in patients who died of NHL and 74/mm3 (10th-90th percentile: 12-384) in those who died of PML. Among those who died of cardiovascular cause, 29% had dyslipidemia that had justified medical care. More than half of individuals who died of non-AIDS-related causes were smokers compared with 44% of those who died of AIDS. Half of those who died of liver disease had excessive alcohol consumption. In 2000, these trends in the characteristics of patients according to the cause of death were similar and median CD4 cell counts tended to be lower for all main causes of death.2
Causes of Death in Particular Categories (Late Diagnosis, CD4 Above 500/mm3)
Among 94 individuals who died within 6 months of HIV diagnosis, the diagnosis of HIV was done before death, the most frequent cause of death was AIDS (76%) and the most frequent AIDS-defining illnesses were Pneumocystis jiroveci pneumonia (27%), NHL (21%), and PML (20%) (Fig. 2).
Among 122 individuals who died with latest CD4 cell count above 500/mm3 (12%), the most frequent causes of death were cardiovascular disease (19%), cancer not related to AIDS or hepatitis (15%), suicide (11%), ESLD (10%), accident (8%), AIDS-defining illness (8%), and other infections (7%).
Comparison Between 2000 and 2005
Taking into account gender and age, the frequency of the 5 main underlying causes of death differed significantly between 2000 and 2005 (P = 0.001). These changes differed according to age (Table 2). Below 30 years, 2 in 3 deaths were AIDS related in both 2000 and 2005. The proportion of AIDS-related deaths decreased mainly above 40 years, from half in 2000 to 1 in 3 cases in 2005. From 2000 to 2005, the proportion of liver-related deaths decreased in the 30-40 years' age strata and increased from 15% to 20% in the 40-50 years' age strata. The increase in the proportion of cancer-related deaths was higher when age was higher. The increase in the proportion of cardiovascular-related deaths was moderate and mainly in the 40-50 years' age strata.
All hospital wards and networks known to be involved in the management of HIV infection in France were contacted, including wards participating in the Mortalite 2000 survey and physicians in penitentiary medicine. Moreover, French societies of intensive care, pneumology, and hepatology specialists were contacted, and they invited their members to participate. Apart from this enlargement of physicians who were contacted, the design was similar to the survey performed in 2000.2
Briefly, physicians prospectively reported death cases in HIV-infected adults (18 years or older) every 3 months in 2005, together with an abstracted cause of death. Each death case was then documented using a standardized questionnaire, including all contributing causes of death, diseases present at death, and a global assessment of the underlying cause of death. One physician was specially dedicated to the survey in the coordinating team, who oversaw harmonization of data collection. The documentation of the cause of death was preferably done by the physician in charge of the patient in the HIV reference center, and this procedure may have limited duplicates. Moreover, double reports were identified by cross-matching the dates of birth and death. Hepatitis C virus infection was defined as positive for HCV antibody or HCV RNA, excessive alcohol consumption as more than 50 g or 5 glasses per day, and poor socioeconomic conditions as the presence of at least one of the following: no health insurance, no employment, no accommodation, income below 535 per month, or immigrant in illegal situation.
Determination of the Underlying Cause of Death
Information contained in the questionnaire was used to determine the underlying cause of death according to the International Classification of Diseases, Tenth Revision rules: The underlying cause of death is the disease or injury, which initiated the train of morbid events leading to death. The algorithm of determination was adapted to specific concerns in HIV infection2 and allowed categorization of deaths as follows: AIDS-related causes according to the 1993 Centers for Disease Control clinical classification, deaths related to infection with HCV or HBV including hepatocarcinoma, cancers, and other causes not related to AIDS or HCV/HBV, and adverse effects of antiretroviral treatment. The latter was considered the underlying cause of death only when this was the explicit conclusion of the physician. AIDS-defining causes were grouped in one underlying cause of death, and the frequency of AIDS-defining diseases was described secondarily. Where the standardized questionnaire was missing, the abstracted quarterly notifications were used to establish the underlying cause of death, if possible.
For statistical comparisons, we used 5 categories of causes of death: AIDS; non-AIDS non-hepatitis-related cancers; liver diseases including viral hepatitis; cardiovascular diseases; and others. We compared the characteristics of patients who died in 2005 according to different causes of death using _2 and Kruskal-Wallis tests. To compare the distribution of causes of death between 2000 and 2005, we performed a multinomial logistic model,16 adjusted for gender and age categorized as followed: younger than 30, 30-39, 40-49, 50-59, and 60 years or older. Statistical analyses were performed using Statistical Analysis System software (version 9.0).