Skip to main content
  • Research article
  • Open access
  • Published:

Impact of variability in adherence to HIV antiretroviral therapy on the immunovirological response and mortality

Abstract

Background

Several previous studies have shown relationships between adherence to HIV antiretroviral therapy (ART) and the viral load, the CD4 cell count, or mortality. However, the impact of variability in adherence to ART on the immunovirological response does not seem to have been investigated yet.

Methods

Monthly adherence data (November 1999 to April 2009) from 317 HIV-1 infected patients enrolled in the Senegalese ART initiative were analyzed. Latent-class trajectory models were used to build typical trajectories for the average adherence and the standardized variance of adherence. The relationship between the standardized variance of adherence and each of the change in CD4 cell count, the change in viral load, and mortality were investigated using, respectively, a mixed linear regression, a mixed logistic regression, and a Cox model with time-dependent covariates. All the models were adjusted on the average adherence.

Results

Three latent trajectories for the average adherence and three for the standardized variance of adherence were identified. The increase in CD4 cell count and the increase in the percentage of undetectable viral loads were negatively associated with the standardized variance of adherence but positively associated with the average adherence. The risk of death decreased significantly with the increase in the average adherence but increased significantly with the increase of the standardized variance of adherence.

Conclusions

The impacts of the level and the variability of adherence on the immunovirological response and survival justify the inclusion of these aspects into the process of patient education: adherence should be both high and constant.

Peer Review reports

Background

Two decades ago, the advent of highly active antiretroviral therapy (HAART) has improved the health status of many people living with HIV and has significantly reduced HIV-linked death rates [1, 2]. To date, several studies have shown the relationships between adherence to HAART and each of: plasma viral load, CD4 recovery, the progression toward AIDS, or mortality [38]: however, the impact of the variability in adherence to HAART on the immunovirological response does not seem to have been studied yet.

For various reasons, unlike the case in high-income countries, the rate of access to HAART in most Sub-Saharan African countries has long been low [9, 10]: less than 4% [9]. However, great efforts have been recently made to improve that access. Senegal was one of the first Sub-Saharan African countries to initiate a policy of universal access to HAART. Indeed, the Initiative Sénégalaise d’Accès aux Antirétroviraux (ISAARV) was launched in 1998 [11]. Not long after, an operational research project was designed to follow-up the patients, evaluate the level of adherence to HAART, and find the reasons of non-adherence [12]. ISAARV and the follow-up project have been the object of several studies over various time periods. These studies have analyzed the determinants of adherence, the levels of adherence, and the link between the level of adherence and the immunovirological response or mortality [1315].

The present work examines, first, not only the level of adherence to HAART (i.e., the average adherence) but also the variability of adherence over time. It examines then the impact of this variability on the viral load, the CD4 cell count, and mortality with adjustment on the average adherence.

Methods

The data source

The original dataset is that of the ANRS 1215 cohort. This cohort included 404 patients with HIV-1 infection receiving HAART within the context of ISAARV [12, 16]. These patients were included between August 1998 and April 2002. The detailed inclusion criteria may be found in previous studies on the ISAARV cohort [12, 15, 16]. In short, 80 patients were enrolled between January 2000 and April 2001 if they were HAART-naive and had a CD4 cell count < 350 cells/mL and a plasma viral load > 3×104 copies/mL. The 324 others were enrolled between August 1998 and April 2002 if they had < 350 CD4 cells/mL (<200 CD4/mL after October 2000) and a plasma viral load > 105 copies/mL (asymptomatic patients) or > 104 copies/mL (paucisymptomatic patients); symptomatic patients free from major opportunistic infections were included whatever the CD4 cell count or the plasma viral load.

The investigation regarding the adherence to HAART started on November 1999 for the first 180 patients enrolled in ISAARV and on May 2004 for the 224 others. At first, adherence-related data were reported for 330 patients (the others died before data collection). However, later, the data of 13 patients were found irregular and unsuitable for analysis. The final analysis concerned thus 317 patients: 175 women and 142 men. The mean age was 37.5 years (interquartile range (IQR): 31–43 years). The time on HAART was censored at 108 months. The median time on HAART was 92 months (IQR: 84–105 months).

The patients were first seen two weeks, one month, and two months after HAART initiation; then at least every two months. They had to obtain the drugs from a single centre (Fann Hospital, Dakar). At each drug delivery, the pharmacist estimated adherence by counting the number of remaining pills and by interviewing the patients about the reasons for non-adherence. The adherence to each drug was calculated as the number of pills taken divided by the number of pills prescribed over the last month. The overall adherence over the last 30 days was the arithmetic mean of the distinct drug adherences.

Besides, every six months, each patient had laboratory investigations that included plasma HIV viral load and CD4 cell count. The detailed investigations can be found in a previous publication [16].

Ethics

The study was conducted with the approval of the Senegalese Ministry of Health (Conseil National de la Recherche en Santé No. 0017 MSP/DS/CNRS and Direction de la Santé No. 0760 MSP/DS/CNRS). All the patients gave written informed consents for participation in the study.

Statistical analyses

All data were censored at death or last visit before the end of the 108th month after HAART initiation.

Starting from the 12th month, a monthly moving average and a monthly moving variance of adherence to HAART were calculated using, at each new monthly time point, the adherence over the previous twelve months. These two values are highly linked: a high average adherence is usually associated with a low variance of adherence; thus, they carry the same information concerning the impact of adherence on the immunovirological response. A measure of variance that is independent of the average was thus necessary. This measure was designed as follows. The averages of adherence were first sorted by increasing order then grouped into 604 small-amplitude classes of nearly 30 averages. The corresponding variances were sorted in increasing order within each class and assigned a rank. The empirical repartition function of the variances within each class of averages was obtained by dividing the ranks of the ordered variances by 1 + the class size. The quantiles of this empirical repartition function were then assimilated to the quantiles of a normal distribution. This provided standardized variances of adherence to HAART.

A latent-class trajectory model was used to distinguish typical trajectories [1719] of the moving averages of adherence to HAART. In this model, in accordance with previous analyses by Bastard et al. [15], the number of latent-classes was fixed to three, which is sufficient to show the trends without increasing the complexity of the model. This model considered time as a random effect and a random intercept for “patient”.

To model the change in the moving average over time, several functions were considered but, to estimate the model parameters, the adopted model was the one that fitted the best the moving averages without presenting algorithm convergence problems. This model considers time t (in months) as covariate and a function of time h T (t) that allows for several inflexion points. It can be written as follows.

Let ŷ it be the average adherence predicted for patient i for month t since HAART initiation. For T {12, 24, 36, 48, 60, 72, 84, 96}, let h T be a function defined by . Thus, with p ij as the posterior probability for patient i to belong to group j and as the mean of a Gaussian distribution conditionally on the fact that i belongs to group j. The value attributable to typical trajectory j at month t is thus the mean of fj(t) weighted by p ij values.

A typical trajectory of average adherence was attributed to each patient according to the maximum a posteriori probability (MAP) rule [19]. The same model was also applied to the standardized variances of adherence to obtain, first, typical trajectories of variance in adherence, then a classification of the patients according to these typical trajectories.

To investigate the link between adherence to HAART and the viral load, percentages of patients with undetectable viral loads (<1000 copies/mL of blood) were calculated per six-month periods, for each average-standardized variance pair. The link between these percentages and adherence was explored using a mixed logistic regression that considered the classes of adherence averages and the standardized variances as well as their interactions as fixed effects and the six-month period as random intercept.

To investigate the link between adherence to HAART and the CD4 cell count, a six-month rate of change (increase or decrease) in CD4 cell count was calculated for each average-standardized variance pair. These rates of change were transformed into monthly values by dividing by 6. The link between the latter values and adherence was explored using a mixed linear model that included the classes of adherence averages and the standardized variances as well as their interactions as fixed effects and “six-month” as random intercept.

To investigate the link between adherence to HAART and mortality, the values of the moving averages and standardized variances of adherence were considered as explanatory variables in a time-dependent Cox model [2022]. The change in patient adherence (i.e., the mean and the standardized variance) was not considered constant over time.

All statistical analyses were performed using packages from R (version 3.0.2) software [23].

Results

Adherence trajectories

The left panel of Figure 1 shows the three typical trajectories of the average adherence to HAART: i) one trajectory of “constantly high” (cH) average that groups 69.47% of the moving averages and whose average is close to 95%; ii) another trajectory of “high but slowly decreasing” (HsD) average (before a sharp increase over the last year) that groups 17.36% of the moving averages and whose average ranges from 90% (up to the fourth year) to 65% (on the seventh and eighth year); iii) a third trajectory of “decreasing then rapidly increasing” (DrI) average that groups the remaining 13.17% of the moving averages and whose average falls sharply from 80% over the first year to less than 50% on the third year before increasing up to 90% over the seventh and eighth year. Regarding the second trajectory, the sample size was not large and most of the deaths within this group (12 among 54 persons of this group) occurred before the last year. Hence, the patients still alive in this group were those who had best improved their adherence during the very last year, which explains the sharp increase of adherence during the last year.The right panel of Figure 1 shows the three typical trajectories of the standardized variance of adherence to HAART. Though irregular along time, these trajectories are somewhat ordered. They may be labelled “low” (grouping 22.59% of the moving variances), “moderate” (grouping 49.08% of moving variances) and “high” (grouping 28.32% of the moving variances).

Figure 1
figure 1

Trajectories of adherence averages (left panel) and standardized variances (right panel).

Table 1 shows the distribution of the 317 patients in the nine average-standardized variance groups according to the MAP rule. About 98% and 90% of the MAPs were higher than 0.85 regarding, respectively, the average adherence latent trajectories and the standardized variance in adherence latent trajectories.

Table 1 Distributions of the patients according to the classes of average and standardized variance of adherence to HAART

Among the 221 patients with cH average adherence, nearly six out of ten patients had a moderate standardized variance. The distributions of HsD and DrI subjects over the standardized variance of adherence groups were obviously more uniform than that of cH. The distributions of the patients in the different groups of average adherence or standardized variance of adherence had no significant links with some patient baseline characteristics (at initiation of HAART) such as sex, stage of HIV infection, or age (see Additional file 1 for details).

Impact of the variability in adherence to HAART on the viral load

Panel a of Figure 2 shows the mean percentages of undetectable viral loads according to the nine average/standardized variance groups. The calculation of these means did not include the percentages found at HAART initiation which were nearly 0 in all groups. The mean percentage increased together with the average adherence. The significance of this association was confirmed by the results of the mixed logistic regression. Table 2 shows clearly that the probability of undetectable viral load increased when the status shifted from DrI average adherence to HsD average adhrence (OR = 2.88; 95% CI: 1.52–5.45) or to cH average adherence (OR = 2.91; 95% CI: 1.74–4.86).

Figure 2
figure 2

Distributions of undetectable viral loads and CD4 cell counts according to average adherence and standardized variance of adherence groups. Panel a: Mean percentages of undetectable viral loads (with their 95% CIs) starting from the 6th month after HAART initiation. Panel b: Variations, by six-month intervals, of the CD4 cell counts starting from the 6th month after HAART initiation. cH: constantly high average adherence - HsD: high but slowly decreasing average adherence - DrI: decreasing then rapidly increasing average adherence.

Table 2 Relationships between adherence (average and standardized variance) and the viral load according to the mixed logistic regression

Within each group of average adherence, there was an association between the mean percentage of undetectable viral load and the standardized variance in adherence. However, this association had not the same effect size within each of the three groups, which means that there was a significant interaction between the average adherence group and the standardized variance group. Indeed, in the group of cH average adherence, the odds of the viral load increased by 76% (OR = 1.76; 95% CI: 1.35–2.30) when the status shifted from “High” to “Moderate” variance and by 82% (OR = 1.82; 95% CI: 1.25–2.66) when the status shifted from “High” to “Low” variance (Table 2, bottom rows). In the group of DrI average adherence, though not significant, the associations between the standardized variance of adherence to HAART and viral load detectability showed the same trend as in the group of cH average adherence (Moderate vs. High variance: OR = 1.28; 95% CI: 0.64–2.58 and Low vs. High variance: OR = 1.73; 95% CI: 0.75–4.00). In the group of HsD average adherence, the associations between the standardized variance of adherence to HAART and viral load detectability showed an opposite trend (though not significant) to that of the other two groups of average adherence (Moderate vs. high variance: OR = 0.69; 95% CI: 0.37–1.28 and Low vs. High variance: OR = 1.71; 95% CI: 0.85–3.45).

Impact of the variability in adherence to HAART on the CD4 cell count

Panel b of Figure 2 shows the distributions of the rates of change (increase or decrease) of the CD4 cell count according to the nine average/standardized variance groups. Within the groups of cH and HsD average adherence, more than 75% of the rates were positive; i.e., were increasing. Within the group DrI average adherence, the IQRs of the rates include negative values (i.e., decreasing rates). In other words, the higher was average adherence, the higher was the rate of change of the CD4 cell count, which means that the higher was the average adherence, the faster was the immunologic recovery.Besides, panel b of Figure 2 shows that, in any given average adherence group, the rate of change of the CD4 cell count tended to decrease with the increase of the standardized variance of adherence to HAART. This trend was very clear in group DrI average, which could mean that the higher was the standardized variance of adherence, the slower was the immunologic recovery.

Table 3 shows the results of the mixed linear model regarding the trends of the monthly CD4 cell count. This trend remained positive on average whatever the average-standardized variance pair, which means an overall increase of the monthly mean of CD4 cell count. This increase was significantly more important when the status shifted from DrI average to cH average than when it shifted to HsD average (the differences being 4.70 and 4.28 CD4 cells/mm3 per month, respectively). There was still a significant relationship between the increase in the monthly mean CD4 cell count and the standardized variance of adherence after adjustment on the average adherence to HAART. Indeed, within the group DrI average, the monthly cell count increased significantly when the standardized variance decreased: the difference was 2.79 CD4 cells/mm3 per month when the status shifted from “High” to “Low” variance and 2.43 when it shifted from “High” to “Moderate” variance.

Table 3 Relationships between adherence (average and standardized variance) and the CD4 cell count according to the mixed linear model

Within the group HsD average, the monthly mean of CD4 cell count did not seem to change within High and Moderate variance; however, it was significantly higher in the High than in the Low variance group (Difference: 2.20; 95% CI: 1.20–3.21 CD4 cells/mm3 per month).

Similarly, within the group cH average, there seemed to be no significant difference between the increases in the monthly mean of the CD4 cell count in High and Moderate variance. This increase was significantly more important in the Low than in the High variance group (Difference: 0.76; 95% CI: 0.26–1.25 CD4 cells/mm3 per month).

Impact of the variability in adherence to HAART on mortality

Over the whole follow-up period, the moving average of adherence ranged from 0.91 to 99.97 (IQR: 87.30–98.41) whereas the moving standardized variance ranged from -1.85 to 1.85 (IQR: -0.55–0.51). The total number of deaths over the follow-up period was 52 (16.40% of the participants).

The results of the Cox model showed that the instantaneous risk of death was significantly associated with both the average and the standardized variance of adherence to HAART.

In the univariate analysis, i) a 10% increase in the average adherence induced a 30% decrease of the relative risk of death (HR = 0.73; 95% CI: 0.66–0.81); and, ii) a unit increase in the standardized variance induced a 50% increase in the relative risk of death (HR = 1.49; 95% CI: 1.04–2.12).

In the multivariate analysis, all things being equal, i) a 10% increase in the overall adherence induced a 30% decrease of the relative risk of death (HR = 0.73; 95% CI: 0.66–0.81); and, ii) a unit increase in the standardized variance induced a 45% increase in the relative risk of death (HR = 1.45; 95% CI: 1.03–2.06).

In the multivariate analysis, a model that included a term of interaction between the average and the standardized variance of adherence did not show a better fit than a model without this interaction. However, it helped showing that the higher was the average adherence, the higher was the effect of the standardized variance on the risk of death.

Discussion

The ANRS 1215 cohort provided the advantage of a long follow-up of patients with HIV receiving HAART; this allowed the use of latent-class trajectories for the average and the variance of adherence to HAART over time. The study was able to establish associations between adherence (through its average or variance) with the viral load, the CD4 cell count, and mortality. The present study showed that cH was the dominant trajectory. This may be explained by a bias of longer patient survival but also, past the first year of treatment, by the ongoing advances in HAART (better tolerance, less pills to be taken, fixed-dose combinations, and once-a-day pill).

A few previous studies have already used trajectory groups of average adherence to HAART [15, 24]. Here, the study goes beyond the simple average and identifies independently trajectory groups of variance of adherence within each average adherence group.

The present results are in agreement with others regarding the impact of the level of adherence to HAART on the immunovirological response or on mortality. Indeed, part of the present results confirm those of Haubrich et al. [4] and of Mannheimer et al. [6] regarding the increase in the percentage of undetectable viral load (meaning viral disappearance from blood) and the increase in the CD4 cell count together with the increase in the average adherence. Another part of the present results confirm those of Palella et al. [3], Nachega et al. [7], and Abaasa et al. [8] regarding the link between the average adherence and patient survival. Moreover, in a recent study conducted in Burkina Faso [25], the authors expressed adherence with a score (ranging from 0 to 10 points) and considered it in a Cox model as a time-varying covariate to point out that the less adherent patients had a higher risk of death. By classifying adherence into two categories (optimal: 8–10 points and sub-optimal: 0–7 points), the authors showed also that patients with optimal adherence had the best CD4-cell-count recovery.

Nevertheless, the results reported here present real novelties regarding the impact of the variance in adherence to HAART on the immunovirological response and on survival. Indeed, whatever the level of the average adherence: i) the rate of CD4 recovery decreased when the variance increased; ii) the probability of achieving an undetectable viral load increased when the variance was low; and, iii) the risk of death increased along with the increase of the variance.

Various methods have been used to measure adherence [2628] and each has its own advantages and limits. The present study shows one approach of dissociating the average adherence from the variance of adherence. A Moving average and variance over 12-month blocks of time was used because the viral load and the CD4 cell count at a given time point are deemed less dependent on current than on past adherence. Besides, the classic average adherence showed high frequencies circa 50% but low frequencies circa 0% and 100%. A transformation of the variance was thus necessary to dissociate the information it provides from that provided by the mean. A standardized measurement of the variance keeps the variances within the same order of magnitude whatever the means. This method provides a more efficient measurement of the variability in adherence that is independent of the average adherence and allows focusing on the impacts of the variance that are not repeats of the impacts of the mean. However, though this method allowed showing significant impacts of adherence variability on some elements of the immunovirological response, the interpretation of the size effects remained difficult.

The latent-class trajectory approach with the average and the variance of adherence offers the advantage of showing that the effects of the variance may change according to the level of the average and avoids enforcing a specific form to the link between average and variance. However, this approach may induce some loss of power regarding the significance of the effects.

Conclusions

There are significant relationships between the variance of the adherence to HIV antiretroviral therapy and the viral load, the CD4 cell count, and mortality. The impacts of the two components (level and variability) of adherence on the immunovirological response and survival justify the inclusion of these aspects into the process of patient education: adherence should be both high and constant; a high but irregular adherence is not favourable to an efficient HAART.

References

  1. Mocroft A, Ledergerber B, Katlama C, Kirk O, d’Arminio MA, Knysz B, et al.: Decline in the AIDS and death rates in the EuroSIDA study: an observational study.Lancet 2003, 362:22–9. 10.1016/S0140-6736(03)13802-0

    Article  CAS  PubMed  Google Scholar 

  2. Braitstein P, Brinkhof MWG, Dabis F, Schechter M, Boulle A, Miotti P, et al.: Mortality of HIV-1-infected patients in the first year of antiretroviral therapy: comparison between low-income and high-income countries.Lancet 2006, 367:817–24.

    Article  PubMed  Google Scholar 

  3. Palella FJ Jr, Delaney KM, Moorman AC, Loveless MO, Fuhrer J, Satten GA, et al.: Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection.N Engl J Med 1998, 338:853–60. 10.1056/NEJM199803263381301

    Article  PubMed  Google Scholar 

  4. Haubrich RH, Little SJ, Currier JS, Forthal DN, Kemper CA, Beall GN, et al.: The value of patient-reported adherence to antiretroviral therapy in predicting virologic and immunologic response.AIDS 1999, 13:1099–107. 10.1097/00002030-199906180-00014

    Article  CAS  PubMed  Google Scholar 

  5. Bangsberg DR, Perry S, Charlebois ED, Clark RA, Roberston M, Zolopa AR, et al.: Non-adherence to highly active antiretroviral therapy predicts progression to AIDS [Letter].AIDS 2001, 15:1181–3. 10.1097/00002030-200106150-00015

    Article  CAS  PubMed  Google Scholar 

  6. Mannheimer S, Friedland G, Matts J, Child C, Chesney M: The Consistency of Adherence to Antiretroviral Therapy Predicts Biologic Outcomes for Human Immunodeficiency Virus-Infected Persons in Clinical Trials.CID 2002,34(8):1115–21. 10.1086/339074

    Article  Google Scholar 

  7. Nachega JB, Hislop M, Dowdy DW, Lo M, Omer SB, Regensberg L, et al.: Adherence to highly active antiretroviral therapy assessed by pharmacy claims predicts survival in HIV-infected South African adults.J Acquir Immune Defic Syndr 2006, 43:78–84. 10.1097/01.qai.0000225015.43266.46

    Article  PubMed  Google Scholar 

  8. Abaasa AM, Todd J, Ekoru K, Kalyango JN, Levin J, Odeke E, et al.: Good adherence to HAART and improved survival in a community HIV/AIDS treatment and care programme: the experience of The AIDS Support Organization (TASO), Kampala Uganda.BMC Health Serv Res 2008, 8:241. 10.1186/1472-6963-8-241

    Article  PubMed  PubMed Central  Google Scholar 

  9. World Health Organization: Scaling up antiretroviral therapy in resource-limited settings: guidelines for a public health approach. Geneva, Switzerland: WHO; 2002:1–165.

    Google Scholar 

  10. Yazdanpanah Y: Costs associated with combination antiretroviral therapy in HIV-infected patients.J Antimicrob Chemother 2004, 53:558–61. 10.1093/jac/dkh142

    Article  CAS  PubMed  Google Scholar 

  11. Desclaux A, Ciss M, Taverne B, Sow PS, Egrot M, Faye MA, et al.: Access to antiretroviral drugs and AIDS management in Senegal.AIDS 2003,17(suppl 3):95–101.

    Article  Google Scholar 

  12. Taverne B, Desclaux A, Sow PS, Delaporte E, Ndoye I: Evaluation de l'impact bioclinique et social, individuel et collectif, du traitement ARV chez des patients VIH-1 pris en charge depuis 10 ans dans le cadre de l'ISAARV - Cohorte ANRS 1215 [Report]. Dakar: CNLS, CRCF, IRD, ANRS; 2012.

    Google Scholar 

  13. Lanièce I, Ciss M, Desclaux A, Diop K, Mbodj F, Ndiaye B, et al.: Adherence to HAART and its principal determinants in a cohort of Senegalese adults.AIDS 2003,17(suppl 3):103–8.

    Article  Google Scholar 

  14. Etard JF, Lanièce I, Fall MBK, Cilote V, Blazejewski L, Diop K, et al.: A 84-month follow up of adherence to HAART in a cohort of adult Senegalese patients.Trop Med Int Health 2007, 12:1191–8. 10.1111/j.1365-3156.2007.01910.x

    Article  PubMed  Google Scholar 

  15. Bastard M, Fall MBK, Lanièce I, Taverne B, Desclaux A, Ecochard E, et al.: Revisiting long-term adherence to highly active antiretroviral therapy in Senegal using latent class analysis.J Acquir Immune Defic Syndr 2011,57(1):55–61. 10.1097/QAI.0b013e318211b43b

    Article  PubMed  Google Scholar 

  16. Etard JF, Ndiaye I, Thierry-Mieg M, Guèye NFN, Guèye PM, Lanièce I, et al.: Mortality and causes of death in adults receiving highly active antiretroviral therapy in Senegal: a 7-year cohort study.AIDS 2006, 20:1181–9. 10.1097/01.aids.0000226959.87471.01

    Article  PubMed  Google Scholar 

  17. McLachlan G, Peel D: Finite Mixture Models. New York: Wiley; 2000.

    Book  Google Scholar 

  18. Skrondal A, Rabe-Hesketh S: Generalized Latent Variable Modeling: Multilevel, Longitudinal, and Structural Equation Models. Boca Raton, FL: CRC Press; 2004.

    Book  Google Scholar 

  19. Nagin D: Group-based Modeling of Development. Cambridge, MA: Harvard University Press; 2005.

    Book  Google Scholar 

  20. Cox DR: Regression models and life tables (with discussion).J R Statist Soc B 1972, 63:269–76.

    Google Scholar 

  21. Crowley J, Hu M: Covariance analysis of heart transplant survival data.J Am Stat Assoc 1977, 72:27–36. 10.1080/01621459.1977.10479903

    Article  Google Scholar 

  22. Altman DG, De Stavola BL: Practical problems in fitting a proportional hazards model to data with updated measurements of the covariates.Stat Med 1994, 13:301–41. 10.1002/sim.4780130402

    Article  CAS  PubMed  Google Scholar 

  23. R Foundation for Statistical Computing. [http://www.R-project.org/]

  24. Glass TR, Battegay M, Cavassini M, De Geest S, Furrer H, Vernazza PL, et al.: Longitudinal analysis of patterns and predictors of changes in self-reported adherence to antiretroviral therapy: Swiss HIV Cohort Study.J Acquir Immune Defic Syndr 2010, 54:197–203.

    PubMed  Google Scholar 

  25. Focà E, Odolini S, Sulis G, Calza ES, Pietra V, Rodari P, et al.: Clinical and immunological outcomes according to adherence to first-line HAART in a urban and rural cohort of HIV-infected patients in Burkina Faso.West Africa BMC Infect Dis 2014, 14:153. 10.1186/1471-2334-14-153

    Article  PubMed  Google Scholar 

  26. Farmer KC: Methods for measuring and monitoring medication regimen adherence in clinical trials and clinical practice.Clin Ther 1999,21(6):1074–90. 10.1016/S0149-2918(99)80026-5

    Article  CAS  PubMed  Google Scholar 

  27. Turner B: Adherence to antiretroviral therapy by human immunodeficiency virus-infected patients.J Infect Dis 2002,185(Suppl 2):143–51.

    Article  Google Scholar 

  28. Costagliola D, Barberousse C: Comment mesurer l’observance? In L’observance aux traitements contre le VIH/sida: mesure, déterminants, évolution. Edited by: Editions EDK. Paris: Editions EDK; 2001:33–42.

    Google Scholar 

Pre-publication history

Download references

Acknowledgments

We thank the Projet IDoL: ANR-12-BSV1-0036 for its financial support to the work of author Olayidé Boussari.

The ANRS 1215 study group (2009–2011)

Ibra Ndoye (National Council for the Fight Against AIDS, Dakar, Senegal); Eric Delaporte, Jean-François Etard, Martine Peeters, Alice Desclaux, Pierre de Beaudrap, Christian Laurent, Julie Coutherut, Tidiane Ndoye, Nicole Vidal, Claire Moquet, Sabrina Eymard-Duvernay, Cécile Cames, Kirsten Bork, Sabah Boufkhed, Mathilde Couderc, Amandine Cournil, Bernard Taverne (IRD, UMI 233, Montpellier 1 University, France); Assane Diouf, Mame Basty Koïta Fall, Alle Baba Dieng, Christian Eric Massidi, Adama Sarr, Khoudia Sow, Mariane Ndiaye Berthe, Saïdou Ba, Absa Ba, Catherine Lissoune Fall Sané, Sokhna Boye, Caroline Desclaux-Sall, Héléne Dior Mbodj, Coumba Gueye Cissé, Frédérique Muller, Kouro Bousso Niang, Marie-Louise Sarr, Estelle Simen, Alassane Sow, El Hadj Malick Sy Camara (Regional Center for Research and Training in Clinical Treatment, National University Hospital Center in Fann, Dakar, Senegal); Maryvonne Maynart, Isabelle Lanièce, Vanina Cilote (Department of Cooperation and Cultural Affairs, French Embassy, Dakar, Senegal); Papa Salif Sow, Ibrahima Ndiaye, Cheickh Tidiane Ndour, Viviane Pierre Marie Ciss, (National University Hospital Center in Fann, Infectious and Tropical Disease Unit, Dakar, Senegal); Ndeye Fatou Ngom Guèye, Djibril Baal, Batista Gilbert, Andréa Robalo Diassy (National University Hospital Center in Fann, Ambulatory Treatment Center, Dakar, Senegal); Jeanne Diaw (General Hospital in Grand Yoff, Dakar, Senegal); René Ecochard, (Claude Bernard Lyon I University, France); Mathieu Bastard (Epicentre, Paris, France); Kadidiatou Ba Fall, Pape Madoumbé Guèye, Pape Samba Ba, Madiouba Diawara (Principal Hospital in Dakar, Senegal); Souleymane Mboup, Pape Alassane Diaw, Halimatou Diop Ndiaye, Ndeye Coumba Touré Kane, Moussa Thiam (Le Dantec National University Hospital Center, Bacteriology and Virology Laboratory, Dakar, Senegal); Karim Diop (Ministry of Health, Division of AIDS Control, Dakar, Senegal); Bara Ndiaye (National University Hospital Center in Fann, Dakar, Senegal).

Author information

Authors and Affiliations

Authors

Consortia

Corresponding author

Correspondence to Olayidé Boussari.

Additional information

Competing interest

The authors declare they have no conflict of interest.

Authors’ contributions

RE, JFE and OB conceived the research questions, developed the study methods, and interpreted the results. JFE and MB were responsible for data acquisition. RE and OB carried out the statistical analyses. RE, OB and JI drafted the manuscript. FB and CG advised on the statistical analyses and the software codes. NF advised on the study methods and the drafting of the manuscript. All the authors participated in the critical revision of the manuscript for important intellectual content, and all approved the final version submitted for publication.

Electronic supplementary material

12874_2014_1179_MOESM1_ESM.doc

Additional file 1:Relationship between the distributions of the patients in the different groups of average adherence or standardized variance of adherence and some patient baseline characteristics.(DOC 51 KB)

Rights and permissions

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Boussari, O., Subtil, F., Genolini, C. et al. Impact of variability in adherence to HIV antiretroviral therapy on the immunovirological response and mortality. BMC Med Res Methodol 15, 10 (2015). https://doi.org/10.1186/1471-2288-15-10

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/1471-2288-15-10

Keywords