Mortality and Risk Stratification of HIV Infected Individuals

Study 1

In a Danish population-based cohort of all HIV-infected persons aged 25-65 who started ART between January 1, 1998, and July 1, 2009, (2267 persons) with median age 40, 74% male, and a median 5 years of follow-up, Kaplan-Meier tables were computed of various HIV-infected risk groups. This group was then compared to a general population group without a diagnosis of HIV, who had no known comorbidity or drug or alcohol abuse (Group 0). Those labeled as “Group 1” included HIV-infected patients with a CD4+ count ≥200 at last measure, no detectable viral load, no AIDs-defining illness (ADI), nor with any comorbidity, drug, or alcohol abuse (including a suspected IV drug use transmission). In this analysis, 58.7% those with HIV met these criteria. The SMR for this group compared to the similarly selected general population was 2.02 for ages 25-45 and 1.14 for ages 45-65. See Figure 1. Crude mortality rate for ages 25-45 was 1.47/1000 compared to 1.27/1000 in the general population, and it was 5.82/1000 at ages 45-65 compared to 4.23/1000.¹⁰

View Figure

• Download as Powerpoint
• Download Figure

Though this was a relatively small study group and a more homogeneous population overall, the strength from an insurance perspective was the comparison to a general population group selected to be without comorbidities and likely closer to representing an insurance population than other studies. Their choice of a CD4+ count ≥200, compared to the usually accepted goal of 350 or 500, raises the question of potentially even lower SMRs at those figures, although this is speculative.

Study 2

In the Collaboration of Observational HIV Epidemiologic Research Europe (COHERE) cohort, all-cause mortality in 80,642 individuals from 23 cohorts and 31 European countries starting ART between 1998 and 2008 was compared with the general population. Median duration of follow-up was 3.5 years. The overall mortality rate was 11.5/1000 with an SMR of 4.2 compared to the general population (SMR 8.5 at ages 18-39, 4.2 at ages 40-59, and 1.7 at ages 60 and up), however among 35,316 individuals who were non-IV drug users with CD4+ counts ≥500, the SMR was 0.9 for men (CI 0.7-1.3) and, after 3 years of ART, was 1.1 in women (CI 0.7-1.7). Even among those with a prior AIDS-defining illness, but with CD4+ count ≥500, the SMR was 1.2 (0.8-2.0) for ages 40-59.¹¹

Study 3

In a well-constructed study designed in part by those in the life insurance industry, the ART Cohort Collaboration (ART-CC) data were used to assess the insurability of HIV positive people treated with ART in Europe. In this study, 34,680 patients were identified who were not infected via IV drug use, were Hepatitis C negative, and who started ART between 1996 and 2008. Data were available to estimate death rates up to 10 years after starting ART and 174,906 person-years of follow-up were registered. In this group, 70% were male, 65% aged 30-49, and 52% with heterosexual transmission, with the majority coming from France or the Netherlands. Relative mortality compared to an insured population in each country, using Poisson models, was estimated. Expected numbers of death were calculated based on “standard” insured lives insurance tables in France, Netherlands, and UK, or on adjustments to population mortality tables in Italy, Spain, and Switzerland, as commonly used in those countries by pricing actuaries.

The overall mortality was 7.1/1000/year. As seen in most other studies, CD4+ count and viral load at 6 months were the most strongly prognostic factors. Mortality rate ratios also decreased with age and with ART duration and were higher in those who started ART before 2001. The relative mortality ratio for those ages 20-39, who started ART after 2000, had 6 month CD4+ counts ≥350, 6 month HIV RNA <10,000 copies/ml, had no AIDs diagnosis, and were during the first 3 years of ART, compared to insured lives, was 459%. Sub-groups at older ages and longer ART duration had more favorable mortality ratios. For example, with the above features but at age 40-49 the MR was 324% in the first 3 years, 223% during years 4-6, and 165% after 7 or more years on ART. At ages 50-59, these values were 250%, 172%, and 128%, respectively.¹⁴

As good as these data are for insurance purposes, there are limitations worth noting. As with other studies, data are for only up to 10 years of ART therapy (and the results extrapolated beyond 10 years). There were limited numbers of those age >60. In addition, these European results may not be matched in the United States, especially when considering the importance of maintaining long term treatment compliance. There were no data on smoking and other risks, which are generally found to be overrepresented in HIV positive groups, and this could have skewed the results toward higher MRs.^20,23 Finally, comparison to ultimate tables was used since the population was not tightly selected, but this may not reflect insurance practices – it assumes that the impact of additional underwriting would be the same in the HIV group as in a non-HIV infected group.

Study 4

In the UK Collaborative HIV Cohort (UK-CHIC) Study, data from 21,388 HIV-positive individuals receiving care in the UK between 2000 and 2012 were used to estimate life expectancy according to latest CD4+ count and viral suppression status. Those with suspected IVDU and mother-child transmission were excluded. In this group, 69% were men, of whom 73% were aged 30-49. Abridged life tables were constructed for men and for women from age-specific mortality rates grouped in 5-year age bands – describing the mortality experience that hypothetical cohorts of HIV-positive individuals would have if they were subjected to the mortality in the observed period. After 5 years on ART, the expected age at death of 35 year-old men varied from 54 for those with CD4+ counts <200 and no viral suppression, to 80 for those with CD4+ ≥350 and viral load suppressed. In this analysis, there was no apparent difference in mortality between those with CD4+ cell count 350-499 and those ≥500. Most of the effect of viral suppression on decreasing mortality was mediated through increasing CD4+ counts; there remained, however, an independent effect of detectable viral load on LE as well.

Crude mortality rates were 9.0/1000 person-years for men and 7.9 for women. Mortality rates for those ages 20-44, after 5 years of ART, varied from 2.7/1000 with viral suppression and CD4+ ≥350 to 38.1/1000 for those with unsuppressed HIV and low CD4+ counts.¹⁷

Study 5

Studies consistently find higher rates of many mortality risk factors in those who are HIV positive, and this may be a confounding factor in the increased mortality risk associated with HIV infection. For example, in one analysis, a simulated cohort of HIV-negative people in the United States with a similar demographic and risk profile to the HIV-infected population had an estimated 8.33 year reduction in life expectancy compared to the population as a whole.²⁰ Thus, mortality rates among the HIV population may be better than recognized after adjusting for additional factors.

To address specifically the issue of a higher prevalence of tobacco use amongst those infected with HIV, an additional analysis of the ART-CC dataset was recently published. The 17,995 individuals for whom smoking status had been collected were studied, and causes of death and overall mortality were compared with the general French population (selected as best representative of the entire group). Of those individuals, 59.8% where classified as smokers. The proportion of men was 71.3%; 70.6% had a viral load <400 copies/ml, and 56.2% had CD4+ cell counts ≥350.

The adjusted mortality ratio comparing smokers with non-smokers was 1.94, similar to that seen between smokers and non-smokers in insurance populations. Unadjusted all-cause mortality rates were 7.9 per 1000 person-years among smokers and 4.2 among non-smokers. For those with CD4+ counts ≥200 and no AIDS diagnosis, the corresponding rates were 5.4 for smokers and 2.9 for non-smokers.

Abridged life tables were used to estimate life expectancies among men – there were too few deaths among non-smoking women to obtain an adequate analysis. LE at age 35, if CD4+ count was ≥200, was 37.8 in smokers and 43.4 in non-smokers, compared to 44.4 years in the background population. At age 65, the loss of life years associated with smoking was estimated to be 6.6 years, whereas that associated with HIV was 2.9 years. The conclusion was that the risk from smoking appeared to be as great as, or greater than, that of adequately controlled HIV infection.²¹ (See Table 2)

Table 2. Limitations of Most Clinical HIV Mortality Studies

View Fullscreen