Article Text

Download PDFPDF

Unemployment as a risk factor for AIDS and death for HIV-infected patients in the era of highly active antiretroviral therapy
  1. C Delpierre1,2,
  2. L Cuzin3,
  3. V Lauwers-Cances1,
  4. G D Datta4,
  5. L Berkman2,
  6. T Lang1
  1. 1
    Inserm U558; IFR126, Toulouse, France
  2. 2
    Harvard School of Public Health, Department of Society, Human Development and Health, USA
  3. 3
    Unit of Infectious and Tropical Diseases, COREVIH, Hôpital Purpan, France
  4. 4
    Harvard School of Public Health, Department of Epidemiology and Center for Society and Health, USA
  1. C Delpierre, Harvard School of Public Health, Department of Society, Human Development and Health, Landmark Center, 401 Park Drive, Boston MA, USA; cyrildelpierre{at}yahoo.fr; cdelpier{at}hsph.harvard.edu

Abstract

Objectives: To assess the association between social situation and disease progression among patients diagnosed with HIV infection since the advent of highly active antiretroviral therapy (HAART), taking late testing into account.

Methods: Prospective cohort study of adults diagnosed with HIV since 1996 in six large HIV reference centres in France. Associations between social situation and death, disease progression and treatment initiation were assessed using Cox regression model. Analysis was restricted to 5302 patients (77.9% of the sample) for whom the status at HIV diagnosis (late or not late) was known.

Results: 134 people (2.5%) died and 400 presented with a new AIDS defining event (7.5%). In multivariate analysis, probabilities of death (HR 3.75, 95% CI 2.11 to 6.66) and disease progression (HR 1.59, 95% CI 1.17 to 2.15) were higher for non-working patients and for late testers (HR 9.18, 95% CI 4.32 to 19.48 for death) and lower for treated patients (HR 0.18, 95% CI 0.08 to 0.41 for death and HR 0.29, 95% CI 0.20 to 0.42 for disease progression). The probability of receiving antiretroviral treatment was not associated with employment status but was higher for late testers, for those living in a stable relationship and lower for those diagnosed after 2000.

Conclusion: Among patients diagnosed for HIV infection in the HAART era, poor social situation is an independent risk factor of mortality and morbidity, and is not explained by delayed access to diagnosis or treatment.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Gains in survival by the use of highly active antiretroviral therapy (HAART) may not be realised by all people with HIV. Studies have documented worse survival amongst persons with low socioeconomic position.13 For some of these studies, this association was only observed since the HAART era,2 3 and could be the result of a differential access to treatment according to social characteristics.1 2 Other studies have found this association even after adjustment on treatment prescription.47

Most of these studies were conducted only in patients with AIDS, in patients initiating HAART or taking together persons with late and early diagnosis in spite of the strong association that exists between mortality and late diagnosis.

The main purpose of this study was to assess the association between social situation and disease progression among patients diagnosed with HIV infection since the advent of HAART between 1 January 1996 and 1 July 2006, taking late testing into account. A secondary objective was to assess the association between social situation and the use of HAART to study if morbidity and mortality differences were explained by a differential access to treatment.

METHODS

Information was collected from six large HIV reference centres in France (Toulouse, Nantes, Nice, Marseille, Tourcoing, Paris), which maintain through an electronic medical record (EMR) prospective cohorts of all HIV-1 infected patients who seek care in the centres and who provided written consent.8 9

Individuals’ characteristics at the time of HIV diagnosis included age, route of HIV infection, employment status (working, non-working (on employment benefit, unemployed) and others (disabled, retired, student)), living or not in a stable partnership, having children or not and tobacco consumption.

The clinical data consisted of the dates of HIV diagnosis, of any first AIDS defining event (ADE), of death, CD4 cell count at diagnosis and chronic hepatitis B or C. The date of the first treatment initiation (bitherapy or HAART defined as treatment with three or more antiretroviral drugs) was recorded. Patients were classified as “late testers” if they presented with either an ADE or a first CD4 cell count below 200/mm3 during the year of HIV diagnosis. If information about CD4 cell count was not documented they were excluded from analysis.

Two end points were selected: all causes of death and disease progression defined as the first new ADE or death. For patients diagnosed late for HIV infection because of an ADE at HIV diagnosis, disease progression was defined as death or as the first new ADE following the one present at HIV diagnosis.

Statistical analysis

χ2 squared tests were used to compare categorical variables, and t test or non-parametric Wilcoxon tests were used for continuous variables.

Cumulative survival rates were estimated using Kaplan-Meier methods and differences in survival period according to factors of interest by using log-rank statistics. Each variable statistically significant at the threshold of 0.10 in bivariate analysis was included into a Cox proportional hazards regression model. All eligible subjects were included regardless of whether they later discontinued or modified their first therapeutic regimen. Kaplan-Meier curves and Cox proportional hazards regression model were also used to analyse factors associated with the use of HAART before the event. Additional Cox regression models were constructed, including individuals with missing data for employment and marital status. The assumption of proportional hazards was validated using Schoenfeld residuals10 and by examination of log-minus-log survival plot for each covariate. Data were censured on the date of last contact until 1 July 2006.

Analyses were adjusted on period of HIV diagnosis (⩽2000 and >2000) and on the medical centre. Statistical analyses were performed using SAS (version 9.1, SAS Institute, North Carolina, USA).

RESULTS

The database contained data on 6805 patients diagnosed with HIV infection between 1 January 1996 and 1 July 2006. Men made up 69.1% (n = 4700) of the whole population. With late testing, 1503 patients (22.1%) were excluded because of lack of CD4 cell count in the year following HIV diagnosis. Compared with the others, these patients were younger, more frequently diagnosed in 1996–97, infected through intravenous drug use, hepatitis B or C virus co-infected and without employment and children.

Among the others (n = 5302), median CD4 count at HIV diagnosis was 316 cells/mm3 (Inter Quartile Range (IQR): 140–517). CD4 count was below 200 cells/mm3 for 33.1% of the population. Late testers represented 2023 patients (38.2%) because of the occurrence of an ADE in 49.3% (n = 998) in a median time of 10 days (IQR: 0–30 days) or because of CD4 cells below 200/mm3 (without any ADE) in 50.7% (n = 1025).

Mortality

The median follow up was 44 months (IQR: 17–78) and the total follow up represented 21 940 person years. Before 1 July 2006, 134 patients (2.5%) died. The overall death rate was 0.61/100 persons years. Among the 111 deaths with a specified cause (82.8%), the main cause of death was HIV/AIDS related (45.1%) followed by other causes (31.5%) and by cardiovascular diseases (13.5%), with no difference according to employment status.

Cox regression model results showed that the probability of death was higher for patients without employment (hazard ratio (HR) 3.75, 95% confidence interval (CI) 2.11 to 6.66), for patients diagnosed late (HR 9.18, 95% CI 4.32 to 19.48) and those over 50 years. The probability of death was lower for treated patients (HR 0.18, 95% CI 0.08 to 0.41). Calendar period, route of HIV infection, living in stable partnership or having children were not associated with death (table 1).

Table 1 Factors associated with the probability of death and disease progression

Morbidity and mortality

The median follow up was 41 months (IQR: 16–75) and the total follow up represented 21 108 person years. Before 1 July 2006, 400 people (7.5%) presented a new ADE (n = 301, including 35 deaths) or died (n = 99). The overall event rate was 1.9/100 persons years.

The rate of new ADE or death was higher for late diagnosed patients (fig 1). As log-log survival curves for late diagnosis suggested that the proportional hazards assumption was not realistic (not shown), this variable was included as a stratification variable in the Cox model. Model results showed that the probability of new ADE or death was higher for unemployed patients (HR 1.59, 95% CI 1.17 to 2.15), and lower for those who received antiretroviral treatment (HR 0.29, 95% CI 0.20 to 0.42). No associations were observed with age, period, route of HIV infection, living in stable partnership or having children (table 1).

Figure 1 Proportion of patients without new AIDS defining event or death over time. Kaplan-Meier curves according to the late or non-late status at the moment of HIV diagnosis.

First treatment prescription

A total of 3749 patients (70.7%) started a treatment during follow up (median time between HIV diagnosis and treatment: 4.5 months (IQR: 1.1–34.0)).

Among treated patients, 3264 (87.1%) received HAART and 485 (12.9%) received a bitherapy without differences according to employment, marital status or route of transmission.

Cox regression model results showed that the probability of receiving antiretroviral treatment was higher for patients diagnosed late (HR 2.94, 95% CI 2.69 to 3.22) and for those living in a stable relationship (HR 1.09, 95% CI 1.00 to 1.20). Conversely, this probability was lower for those diagnosed after 2000 (HR 0.83, 95% CI 0.76 to 0.90).

DISCUSSION

To our knowledge, this study is the first analysis of longitudinal data from France assessing the link between social situation and disease progression in the HIV infected population diagnosed since 1996. Even if the frequency of new ADE or death is limited for people diagnosed since the advent of HAART, these events are more likely among unemployed people and are strongly associated with late testing. The higher risk of disease progression observed in unemployed people, particularly the higher risk of death, was not explained by delayed access to treatment.

As HIV infection is mainly managed in hospitals (93% of HIV infected patients, including in and outpatients11), hospital data are useful to monitor the HIV epidemic but should be interpreted with caution. With late testing, we were not able to classify 22% of the population because of the lack of CD4 cell count data in the year of HIV diagnosis, which is lower than in other European studies.4 12 Some of those with missing data have characteristics associated with late testing.9 As a result, the proportion of late testers could be underestimated. They were also more at risk to be unemployed and to have a new ADE or death than not-late testers and thus should not change our results. For about 50% of the total population, data on social situation were not available. Additional Cox regression models were constructed, including individuals with missing data for employment status and marital status (data not shown). No significant changes were observed in the probability of receiving treatment. With survival, unknown employment status was associated with death as was an unknown route of infection and heterosexual women had a lower probability of death. With disease progression, the results were similar except for patients older than 60 years and for injecting drug users who were at higher risk of disease progression. Thus, it seems unlikely that these missing values might induce major bias in our results.

Because of legal restrictions, it was not possible to collect the country of birth. Migrants make up an important proportion of our population and may influence the results. Indeed, migrants have a higher probability to be late testers,9 heterosexual women13 14 or with an unknown route of HIV infection,15 and may constitute a strong part of the unemployed population.13

The rate of death should be interpreted with caution. As observed in other databases,16 it could be underestimated in our population because all deaths may not be recorded in the database.

Unemployment was associated with disease progression and death as observed in other international studies.1 2 6 As expected, in France where treatment is delivered free of charge through an universal healthcare system, the probability of receiving treatment was not lower for unemployed persons and did not explain the higher risk of disease progression in these patients. Poor social situation can influence health by different pathways, including the level of adherence. In a French study by Dray-Spira et al,5 unemployment was associated with poor treatment adherence. In another study conducted by the same author,17 migrants, who were more frequently unemployed, less educated, in precarious housing, social isolation and depressed, had a lower level of adherence.18 Moreover, social disadvantage may impact stress or depression, which in turn may have a direct effect on the immune system, CD4 cell count and viral load by altering the hypothalamic-pituitary adrenocortical system, including cortisol, as observed in a study of Leserman et al.19

We did not find any relation between living in a stable partnership and disease progression, contrary to results of two other studies.4 5 Our population included all patients diagnosed for HIV infection since 1996, whereas in one study the population analysed patients diagnosed at the early stage of primary infection5 and in the second only treated patients.4 Partnership may play a different role depending on HIV history and it is likely that this role is different at the time of the HIV diagnosis, as in our study, compared to the initiation of HAART period, as in the study of Young et al.4 The presence or absence of a stable partnership is a coarse indicator of social and emotional support and an incomplete measure of the many ways by which social relationships influence health. Some other aspects of social support could have an impact on health status as well.

Late testing was the main risk factor of death for patients diagnosed since the advent of HAART. Reducing its frequency may have a major impact for reducing morbidity and mortality and should be a major objective of public health. It is noteworthy that the link between unemployment and disease progression was still significant even after adjustment on late testing, in accordance with other studies in which no association was observed between employment status and late diagnosis.9 Thus, the higher risk of mortality among unemployed people was not explained by a higher risk of late testing.

For patients diagnosed with HIV infection since 1996, disease progression and death are limited because of the HAART use. However, poor social situation is an independent risk factor of mortality and morbidity, which are not explained by delayed access to diagnosis or treatment. Level of adherence, social support, mental health or the general physical conditions according to social conditions could explain this result. Further studies combining clinical, biological, psychosocial and socioeconomic data should be implemented to explore these different hypotheses.

Acknowledgments

We are indebted to all the participants in the cohort, and all the medical teams in the different centres who collect the information, without whom this work would not have been possible. We are grateful to the National Agency of AIDS Research (ANRS) and to the Association of Aids Research (ARS) for their financial supports.

REFERENCES

Footnotes

  • Competing interests: None.

  • Ethics approval: Ethics committee approval for this study was obtained.

  • Patient consent: Informed consent was obtained for the publication of the details in this report.