The treatment of HIV-infected patients with highly active antiretroviral therapy (HAART) results in reduction in mortality and morbidity. Effective lifelong and continuous use of these drugs may be compromised by issues of (1) adherence to medications and (2) drug-related adverse events. Cost is also an important issue to consider for both developed countries, trying to limit the increase in health expenditure, and developing countries. Structured treatment interruptions (STIs) have been viewed as a possible response to these issues insofar as they do not compromise the immunological, virological, and ultimately clinical outcomes. A series of randomized trials have explored different modalities of treatment interruption that can be classified into 2 categories: CD4-guided interruptions1-3 and fixed intermittent or pulsed therapy.4-7
A Cochrane systematic review reported insufficient evidence to support the use of STI in patients with chronic suppressed HIV infection from a cost reduction perspective.8 The review identified 2 trials where CD4-guided STI achieved reduction in costs of antiretroviral (ARV) therapy. These trials priced the use of ARV drugs but did not, however, document the actual costs of the whole strategy.9,10 Whether the benefits from the reduced use of ARV drugs could be offset by the cost of treating additional adverse events should be questioned. We therefore conducted a cost study alongside a randomized open-label study comparing an 8-week off, 8-week on regimen with a continuous ARV regimen over a 96-week period (ANRS 106 Window).
Patient Population and Treatment Strategy
The study design, patient population, and intervention have been reported elsewhere.11 Briefly, a total of 403 patients, HIV-1-infected adults who were well tolerating their current combination ARV therapy, with a history of nadir CD4 count of 100 cells per microliter or more, a CD4 count above 450 cells per microliter at screening, and a plasma HIV-1 RNA level of less than 200 copies per milliliter at least in the last 6 months, were randomly assigned either to switch to a fixed intermittent treatment (IT) strategy alternating periods of 8 weeks off therapy with 8 weeks on therapy or to maintain their current treatment (CT) strategy. Patients were assessed at baseline, then every 8 weeks until they completed the 96 weeks of follow-up.
The primary end point of the trial was the immunological failure, defined as a confirmed CD4 cell count less than 300 cells per microliter; and the cumulative proportions of patients reaching the primary end point over 96 weeks of follow-up were compared across the 2 treatment arms. The trial demonstrated noninferiority of the IT strategy over that of CT, in the margin of a 7% predefined difference in the proportions of progressions between the 2 arms (3.6% vs. 1.5% in IT and CT arms, respectively, with the upper bound of the 95% confidence interval of the difference at 5.6%). We therefore conducted the planned cost-minimization analysis as both treatments performed similarly in respect of the outcome of interest. The economic evaluation was limited to the identification of the least costly alternative.12
Resources Use and Determination of Costs
Both hospital and ambulatory resources were collected prospectively for each patient using the case report form. All hospitalizations during the 96 weeks of follow-up were documented as diagnoses, main procedures, length of stay, and type of ward. The use of ARV medications was described as cumulated duration of prescription over the study period within each class of drug. Number of visits to general practitioners and specialists not scheduled in the trial follow-up was also recorded and biological tests prescribed in addition to those required by the protocol (CD4 cell counts or platelet counts), ambulatory corrections of lipodystrophy (cures of polyactic acid injection performed for lipoatrophy and liposuction performed for lipohypertrophy), and ambulatory treatments of mucocutaneous HIV events (mucosal candidiasis represented the principal HIV-related events with persistent lymphadenopathy which did not require treatment, we therefore limited the cost comparison to cutanomucosal events).
Costs were estimated from the viewpoint of the payer over the 96-week study period. The analysis was limited to direct medical costs (costs of care excluding costs of sick leaves and costs of loss of production). Unit costs were provided from the National Price Schedules 2007 for the costs of ambulatory visits, biological tests, and listed price for drugs.13 Costs of hospitalizations were determined on the basis of the corresponding diagnosis-related groups consistently with the viewpoint chosen. Unit costs are presented in Table 1. All values are reported in €2007.
The total costs of ARV medications and treatments of mucocutaneous HIV events were calculated on the basis of duration of prescription and average daily dosage for each medication. As the incidence of non-HIV-related clinical events was the same in both groups, their cost was not taken into account. Protocol-driven costs (cost of physician visits and biological tests scheduled every 8 weeks in accordance with the trial protocol) were identified and reported separately (Table 3). Costs were not discounted as we used the 2007 values published by the payer and because the follow-up period did not exceed 3 years.
Discrete data were reported as frequencies and were compared by χ2 test or the Fisher exact test. Quantitative data are reported as mean ± SD and compared by the Student t test or the Wilcoxon rank sum test. All reported P values were adjusted with the stepdown Bonferroni method of Holm for multiple tests. Difference between groups was considered significant when the adjusted P value of the 2-sided test was less than 0.05. Because many of the comparisons were not prespecified, the statistical significance of the observations should be taken with caution. All analyses were performed on an intention-to-treat basis.
Of the 403 patients randomized, 12 patients did not start the assigned treatment strategy. Their characteristics did not differ from the rest of study patients, and they were not included in the main analysis. Complete cost data were available on 391 patients included in the main analysis (197 patients in the IT and 194 in the CT arms). Baseline characteristics of patients have been previously reported.5 The 2 randomized groups were well balanced for age, sex ratio, Centers for Disease Control and Prevention stage, CD4 count, concomitant infection by hepatitis B virus or hepatitis C virus, and ARV drugs used in HAART.
Use of resources during the 96-week study period is presented in Table 2. As expected, the use of ARV therapy was substantially lower in the IT group as compared with the CT group. Median exposition to ARV therapy was 49.9 weeks (51.5% of the total treatment study duration) in the IT arm and 96.1 weeks in the CT arm (100% of the total treatment study duration) as planned in the study protocol (data not shown). For each drug class, the duration (mean number of days on drug) was significantly shorter in the IT arm than in the CT arm, respectively, 596 vs. 1053 days for nucleoside reverse transcriptase inhibitor (P = 0.001), 342 vs. 555 days for nonnucleoside reverse transcriptase inhibitor (P = 0.001), and 450 vs. 740 days for protease inhibitor (P = 0.001).
Biological tests prescribed in between scheduled visits concerned a few patients only: 46 CD4 counts (37 and 9, respectively, in IT and CT arms) in 34 (8.7%) patients and 11 platelet counts (11 and 0, respectively, in IT and CT arms) in 5 (1.3%) patients, as shown in Table 2. Additional CD4 cell counts were significantly more frequent in the IT group than in the CT group (0.19% vs. 0.05%, respectively; P = 0.02). The mean number of additional platelet counts did not differ across the 2 groups.
Sixty-two (17.3%) patients were hospitalized during the follow-up period (17.3% vs. 14.4% for IT and CT arms, respectively) totalizing 86 hospitalizations (51 and 35, respectively, in IT and CT arms) as shown in Table 2. In addition, some of them were admitted as outpatients for a mucocutaneous HIV event (9.1% vs. 4.6% for IT and CT arms, respectively) or for a lipodystrophy correction (2.0% vs. 6.2% for IT and CT arms, respectively) (Table 2). All these differences were not significant.
Results of the costs are presented in Table 3. Mean costs per patient did not differ across the 2 groups except for ARV medications and additional CD4 cell count. The mean total direct medical cost in euros (€) per patient over the 96 weeks of follow-up (excluding protocol-driven costs) was 9738 in the IT arm compared with 16,162 in the CT arm (P < 0.0001), a €6424 difference almost entirely due to the difference in HAART cost. Drug costs were, respectively, €5105 and €9259 for nucleoside reverse transcriptase inhibitor (P = 0.001), €3766 and €6054 for nonnucleoside reverse transcriptase inhibitor (P = 0.001), and €4176 vs. €6439 for protease inhibitor (P = 0.001), for IT and CT arms.
The mean cost per patient of additional CD4 cell count was €1.00 in the CT arm vs. €4.06 in the IT arm (P = 0.02). The mean cost of additional platelet counts was not significantly different between the 2 groups.
Mean protocol-driven costs represented €290 in the IT arm vs. €280 in the CT arm (not significant), less than 3% of the total cost of the strategies compared.
The use of IT achieved a 40% reduction in the total direct medical cost of treatment.
We report here the results of the first cost study comparing intermittent and continuous ARV therapy based on data collected prospectively in a randomized controlled trial including a large number (N = 391) of patients. The intermittent strategy seemed to be cost saving as it achieved a median cost reduction of €6424 per patient without a significant difference in the primary study outcome through 96 weeks. The total direct medical cost of the strategies was almost entirely due to the costs of the ARV treatments. Their mean cost per patient was more than 10 times higher than all other items of resource utilization, underlying the high sell price in developed countries of combined ARV therapies. The amount of ARV drugs used within each class was reduced by 40% in the IT group compared with the CT group.
As shown in the ANRS 106 Window trial,11 patients experienced significantly more HIV-related clinical events in the IT arm than in the CT arm (14% vs. 7%), mostly mucocutaneous candidiasis and lymphadenopathy, but there were no AIDS-defining events and no difference across the 2 arms in the incidence of non-HIV-related events (12% vs. 12%). These results differ from those of the TRIVACAN and SMART CD4 cell-guided interruption trials in which higher incidence of disease progression and death was observed in the intermittent group.2,3 In our study, costs led by all these adverse events were not significantly different across the 2 groups neither for inhospital nor for ambulatory care. Therefore, the additional costs of care for patients treated in the IT arm (costs of HAART excluded) did not offset the reduction in costs given by the treatment interruptions.
Specific Features and Limitations of This Study
This study concerned only patients with controlled HIV infection who were followed in the strict setting of a controlled trial with a controlled protocol of medical follow-up (medical history, physical examination, and biological test scheduled every 8 weeks). Except for a very small number, all patients received at least a triple combination of ARV drugs during the study period. This population of well-controlled patients has predominated with the advent of more therapeutic options; they experienced only minor adverse events (no AIDS-defining events), and few were ever hospitalized. This is a possible explanation for the absence of difference in follow-up costs.
The widespread use of the IT strategy could be questioned. First, it is uncertain whether the level of clinical support needed to ensure the proper adherence to an IT strategy could be offered in routine practice even if its cost (protocol-driven costs) is not expensive. Second, at 96 weeks, the proportion of patients who maintained CD4 cell count above 450 cells per microliter was lower in the IT group, which renders doubtful the possibility of lifelong use of the IT strategy. We did not include the costs of concomitant medications such as analgesics, psychotropic drugs, and other anti-infectious drugs, which may have slightly underestimated the total cost in both arms. We assumed a health care system perspective instead of a societal perspective. As there was no difference in the numbers of physician visits, hospital admissions, and hospital days between IT and CT groups, we assumed that patient's time cost and production costs did not differ. This however has not been assessed directly.
Also, the true equivalence in effectiveness of both strategies could be questioned. First, there was a moderate but significant decrease in median CD4 cell count in the IT group, the “cost” of which was not measured. Second, although our trial did not identify safety problems, which is consistent with the results of all other published trials using fixed interruptions not exceeding 16 weeks,4-7 it may have missed an increased risk of events with a low incidence but a high economical cost. Indeed, the SMART trial, which, however, used CD4-guided interruptions of much longer duration, evidenced an increased risk of both HIV-related morbidity and non-HIV-related morbidity that could not be detected in the Window trial, which was not powered for such end points.
The ethical implications of stepping down in therapeutic effectiveness to decrease costs in low-income countries are not within the scope of this study and would certainly need careful discussion.
Previous comparative economic evaluations or cost-effectiveness studies of HAART are available in several countries (France, Italy, United Kingdom, United States, Canada, South Africa, Russia, Thailand, India, and Taiwan), but they all compared new HAART strategies with pre-HAART therapies in time-sequential analysis using cohort data or computer simulations. Valid comparisons are not possible because of differences in methods (particularly in study populations, characteristics, or nature of the outcome of interest), and reviews cannot estimate any synthetic result.14,15 However, all studies concluded to an increase in total cost in the HAART strategy with benefits in terms of life expectancy, CD4 cell count, and viral load. Access to HAART was ever considered as a major political choice for decision makers, particularly in developing countries that try to negotiate lower price for ARV drugs or enacted laws that permit to make or import generic version of patented ARV drugs.
To our knowledge, this is the first report of a cost study of STIs for HIV-infected patients performed prospectively in a large patient population. We found that reducing by 50% the exposition to ARV treatment and subsequently its total cost did not increase the expenses related to the management of adverse events or nonroutine consultations that may have been induced by this strategy.
Contributions of authors-study concept and design: J.P.A., J.M.M., B.M., and I.D.-Z.; acquisition of data: P.T., I.C., and G.J.; analysis and interpretation of the data: J.P.A., I.C., P.T., G.J., and I.D.-Z.; drafting of the manuscript: G.J., I.C., and I.D.-Z.; critical revision of the manuscript for important intellectual content: J.P.A., J.M.M., and B.M.; statistical analysis: I.C.; administrative, technical, or material support: J.P.A.; and study supervision: I.D.-Z.
1. Ananworanich J, Gayet-Ageron A, Le Braz M, et al, and the Staccato Study Group, the Swiss HIV Cohort Study. CDA-guided scheduled treatment interruptions compared with continuous therapy for patients infected with HIV-1: results of the Staccato randomised trial. Lancet
2. Danel C, Moh R, Anzian A, et al, for the Trivacan ANRS 1269 trial group. CD4-guided structured treatment antiretroviral strategy in HIV-infected adults in west Africa (Trivacan ANRS 1269 trial): a randomised trial. Lancet
3. The Strategies for Management of antiretroviral Therapy (SMART) Study Group. CD4+ count-guided interruption of antiretroviral treatment. N Engl J Med
4. Dybul M, Nies-Kraske E, Daucher M, et al. Long-cycle structured intermittent versus continuous highly active antiretroviral therapy for the treatment of chronic infection with human immunodeficiency virus: effects on drug toxicity and on immunologic and virologic parameters. J Infect Dis
5. Alexander TH, Ortiz GM, Wellons MF, et al. Changes in CD4+ T-cell differentiation phenotypes during structured treatment interruption
in patients with chronic HIV-1 infection. J Acquir Immune Defic Syndr
6. Florence E, Garcia F, Plana M, et al. Long-term clinical follow-up, without antiretroviral therapy, of patients chronic HIV-1 infection with good virological response to structured treatment interruption
. Clin Infect Dis
7. Plana M, Garcia F, Oxanius A, et al. Relevance of HIV-1 specific CD4+ Helper T-cell response during structured treatment interruption
in patients with CD4+ T-cell nadir above 400/mm3
. J Acquir Immune Defic Syndr
8. Pai NP, Tulsky JP, Lawrence J, et al. Structured treatment interruptions (STI) in chronic suppressed HIV infection
in adults. Cochrane Database Syst Rev
. 2005;(4). CD005482. DOI: 10.1002/14651858.
9. Cardiello PG, Hassink E, Ananworanich J, et al. A prospective, randomized trial of structured treatment interruption
for patients with chronic HIV type 1 infection. Clin Infect Dis
10. Nuesch R, Ananworanich J, Sirivichayakul S, et al. Development of HIV with drug resistance after CD4 cell count-guided structured treatment interruptions in patients treated with highly active antiretroviral therapy after dual-nucleoside analogue treatment. Clin Infect Dis
11. Marchou B, Tangre P, Charreau I, et al, and the ANRS 106 Study Team. Intermittent antiretroviral therapy in patients with controlled HIV infection
12. Drummond MF, O'Brien BJ, Stoddard GL, et al. Methods for the Economic Evaluation of Health Care Programmes
. 2nd ed. Oxford, England: Oxford University Press; 1998:11.
13. Prise en charge thérapeutique des personnes infectées par le VIH. Recommandations du groupe d'experts. Paris, France: Médecin-Sciences Flammarion; 2004. ISBN: 2-257-10198-7. Tableau 4-V, Antirétroviraux en pharmacie de ville: prix moyen d'un traitement mensuel:54.
14. Levy AR, James D, Johnston KM, et al. The direct costs of HIV/AIDS care. Lancet Infect Dis
15. Hellinger FJ. Economic models of antiretroviral therapy: searching for the optimal strategy. Pharmacoeconomics