Nombela, Nohelly MPH*; Kouadio, Bertin MPH†; Toure, Siaka MD, MPH†; Seyler, Catherine MD, MSc†; Flori, Yves-Antoine PhD*; Anglaret, Xavier MD, PhD*†
Highly active antiretroviral therapy (HAART) is the only intervention that can improve the long-term prognosis of HIV-infected patients at advanced stages of immunosuppression. Since 1998, the international HIV drug access campaigns, which aims at lowering the price of drugs for treating HIV-infected patients in developing countries, thus has legitimately focused on the price of antiretroviral drugs.1-3 Meanwhile, there has been little research carried out on the consumption of nonantiretroviral drugs by HIV-infected patients in low-resource settings.
By nonantiretroviral drugs for treating HIV morbidity, we are not only referring to drugs for treating the AIDS-defining opportunistic diseases that occur at advanced stages of immunodepression. In sub-Saharan Africa, the incidence of early severe HIV morbidity (ie, morbidity occurring in patients with a CD4 count >200 cells/mm3) is higher than in industrialized countries.4 The prognosis of patients who are close to starting HAART may thus depend on access to drugs for treating malaria, tuberculosis, or severe bacterial diseases (ie, drugs that are not considered as classic “HIV drugs”).4,5 Conversely, in low-resource settings, HAART is still initiated at advanced stages of immunosuppression in a great proportion of patients.6,7 HAART should thus reduce the need for drugs for the AIDS-defining diseases occurring in patients with a low CD4 cell count but not eliminate it.
Now that HAART is becoming increasingly available, improving knowledge of nonantiretroviral drug consumption could allow us to identify better the nonantiretroviral drugs that significantly contribute to increase the overall cost of care. In this study, we estimated the consumption of nonantiretroviral drugs in terms of frequency of drug administration and cost of drugs in a cohort of HIV-infected adults followed up at all stages of immunosuppression in Côte d'Ivoire between 1996 and 2001.
Between 1996 and 1998, 2 Agence Nationale de Recherche sur le SIDA (ANRS)-funded randomized trials were carried out in Abidjan: (1) a trial assessing the tolerance and efficacy of cotrimoxazole prophylaxis in preventing severe morbidity in HIV-infected adults at World Health Organization (WHO) clinical stages II and III (Cotrimo-CI ANRS 059 trial)8 and (2) a trial assessing the tolerance and efficacy of a short course of zidovudine (AZT) treatment in reducing the mother-to-child transmission of HIV (Ditrame-CI ANRS 049 trial).9 The Cotrimo-CI trial was stopped on March, 17 1998, showing a significant reduction of severe morbidity in the cotrimoxazole group. At this date, all patients from the Cotrimo-CI trial continued to be followed up at the same center and under the same procedures, and this standardized follow-up was opened up to all women from the Ditrame-CI trial as soon as their follow-up in the Ditrame-CI trial terminated. The long-term follow-up of patients from the Cotrimo-CI and the Ditrame-CI trials was supported by the ANRS as the “Cotrame ANRS 1203 Therapeutic Cohort Study.”6,10 The protocols of the Cotrimo-CI, Ditrame-CI, and Cotrame studies have been approved by the Ethics Committee of the Ivoirian Ministry of Health and the Institutional Review Board of the ANRS.
Patients were eligible for the present study if (1) they started cotrimoxazole prophylaxis at any time between their inclusion in the Cotrimo-CI or Cotrame study and the date of study termination and (2) they were seen again at least once at the Cotrame study center before the date of study termination. The date of study initiation was April 23, 1996, and the date of study termination was December 31, 2001 or the date of initiation of antiretroviral treatment (ART) of those patients who started ART before December 31, 2001.
The Cotrimo-CI and Cotrame studies have been previously described.6,10 In summary, the following procedures were unchanged during the overall study period: follow-up in 1 medical center, including a daycare hospital unit and an independent central pharmacy; blood cell count and CD4 cell count at baseline and every 6 months; monthly visit to the center; open access to the center at any time in case of a medical problem; management of morbidity episodes through standardized algorithms; systematic procedures of vital status investigation whenever subjects did not attend monthly visits11; and free-of-charge care, including transport, consultations, investigations, hospitalizations, and drugs.
Between April 23, 1996 and March 17, 1998, patients received a placebo or cotrimoxazole at a dose of 800/160 mg/d. Once the Cotrimo-CI trial was stopped, all patients from the placebo group were systematically offered cotrimoxazole. After March 17, 1998, women from the Ditrame-CI trial who joined the Cotrame cohort were prescribed cotrimoxazole whenever they met the Côte d'Ivoire national criteria for cotrimoxazole prophylaxis in adults (ie, WHO stage II, III, or IV and/or CD4 count <500 cells/mm3).12 Finally, since December 1998, patients from the Cotrame cohort who were eligible for ART were referred to one of the Côte d'Ivoire/United Nations Program on HIV/AIDS (UNAIDS) drug access initiative centers.6
All nonantiretroviral drugs prescribed by the study physicians were delivered to the study center pharmacy. Whenever possible, drugs delivered to this pharmacy were purchased at the national public drug procurement agency (Pharmacie de Santé Publique de Côte d'Ivoire [PSP], Abidjan) or through an international generic supplier (MSF-Logistique, Bordeaux, France). Drugs that were not available at the PSP or at MSF-Logistique were purchased through the private drug wholesalers existing in Abidjan. Private pharmacies were used as suppliers only in case of sudden unforeseen demand.
The Cotrame database recorded every dispensation performed at the center pharmacy. Each entry included the date of dispensation, the patient to whom the drugs were dispensed (identified by an individual anonymous code), the number of units dispensed, the name of the drug, the administration route (eg, oral, intravenous), the dosage, and the supplier of the drug.
For a given drug, the total number of dispensations during the study period was the number of time this drug was dispensed, the price of a given dispensation was the number of units dispensed multiplied by the real price actually paid for the drug that was dispensed, and the total cost during the study period was the sum of the price of all dispensations. All prices were expressed in 2001 $US.
Baseline was considered to be the date of the first prescription of cotrimoxazole. The follow-up period of the study was defined as the interval between baseline and December 31, 2001, or the date of death if before December 31, 2001; the date of ART initiation if before December 31, 2001; or the date of last contact with the study team in all patients who did not start ART before December 31, 2001, who were never seen by the study team after December 30, 2001, and who were not known to be dead.
The drugs dispensed were first described in terms of total number of dispensations and total cost during the overall follow-up interval. This description was made overall and by drug characteristics (supplier, administration routes, therapeutic groups, and drugs). The number of dispensations “per person per year” (DPPY) and cost “per person per year” (CPPY) were then calculated by dividing, for each patient, the total number of dispensations and the total cost of drugs delivered to the patient by the patient's follow-up time. The DPPY and CPPY were then described overall, by drug characteristics, and by patients' baseline characteristics. The association between the CPPY and baseline CD4 cell count was analyzed by multivariate linear regression, adjusting on the other baseline variables associated with the CPPY with P < 0.25 in univariate analysis.
Analyses were performed with SAS statistical software version 8.02 (SAS Institute, Cary, NC).
Of the 723 patients followed in the Cotrame cohort, 592 (82%) were included in the present study. The remaining 131 were not eligible for the following reasons: 108 did not fulfill the criteria for cotrimoxazole initiation; 14 started cotrimoxazole prophylaxis after December 31, 2001, or after ART initiation; and 9 were not seen by the study team after the day they initiated cotrimoxazole.
Most participants came from the Côte d'Ivoire (88%), were illiterate or at the primary school level (60%), were employed (53%), were married or living in a free union with a permanent partner (54%), and had water and electricity at home (68%). Before inclusion in the Cotrame cohort, 70% of patients participated in the Cotrimo-CI trial and 30% participated in the Ditrame-CI trial. Other baseline and follow-up characteristics are summarized in Table 1.
Available Data on Drug Consumption
During the study period, the database from the study center pharmacy recorded 59,311 drug dispensations, of which 58,776 (99%) were included in the analyses. The remaining 535 had to be excluded for the following reasons: duplicated data (n = 196), fatal error on the drug code that could not be corrected (n = 301), and date of dispensation later than patient's date of last contact with the study team (n = 38).
The 58,776 dispensations accounted for a total cost of $190,859 US. The mean DPPY was 59 (standard deviation [SD] = 68, median = 39, interquartile range [IQR]: 18-75). The mean CPPY was $198 US (median = 93, IQR: 32-223).
Origin of Delivered Drugs
Of the 58,776 dispensations, 83% were drugs bought from the public or the nongovernmental organization (NGO) sector (43% from public health national drug supplier and 40% from international supplier of generic drugs) and 17% were drugs bought from the private sector (15% from local wholesalers and 2% from private pharmacies). Conversely, the drugs bought from the private sector accounted for 42% of the overall cost (35% from local wholesalers and 7% from pharmacies), whereas 58% of the cost was attributed to the public or NGO sector (44% from public health national drug supplier and 14% from international supplier of generic drugs).
Consumption by Types of Drugs
Most of the drugs distributed were for oral use (70% of dispensations and 77% of cost), whereas drugs for parenteral use accounted for 14% of dispensations and 7.5% of the overall cost, followed by the topical cutaneous drugs (9.7% of dispensations and 12.2% of cost) and intrarectal or vaginal drugs (3.6% of dispensations and 1.8% of cost) as well as ophthalmic, intranasal, intra-auricular, or intrabuccal drugs (2.7% of dispensations and 1.5% of cost).
As shown in Figure 1, the overall cost was concentrated in a small proportion of drug dispensations, because 16% of the overall dispensations accounted for 50% of the overall cost and 50% of the overall dispensations accounted for 77% of the overall cost. The most frequently delivered therapeutic groups were the anti-infectious group (32.8% of overall dispensations and 66.1% of overall cost) and the analgesic, anti-inflammatory, and antispasmodic group (22.4% of overall dispensations but only 4.9% of overall cost). Table 2 shows for each therapeutic group and class the consumption in terms of cost (percentage of the overall cost, mean and median CPPY) and dispensations (percentage of overall number of dispensations, mean and median DPPY).
Table 3 details for each class of anti-infectious drugs (antibacterial, antiparasitic, antifungal, antituberculous, and antiviral) the percentage of the overall cost and the percentage of the overall number of dispensations as well as the percentage of the cost within each class and the percentage of the number of dispensations of each subclass. Among the antifungal drugs, the most frequently dispensed were nystatin (41% of antifungal drug dispensations) and miconazole (34%), whereas the most costly was fluconazole, representing 84% of the antifungal cost but only 15% of antifungal dispensations. Among the antibacterial drugs, the most frequently dispensed were A-penicillins (23% of antibacterial drug dispensations), nitroimidazoles (21%), fluoroquinolones (16%), amoxicillin-clavulanate (5%), and M-penicillins (5%), whereas the most costly were amoxicillin-clavulanate (35% of the antibacterial drug cost), fluoroquinolones (14%), A-penicillins (14%), third-generation cephalosporins (11%), and first-generation cephalosporins (8%). Finally, almost half of the cost and dispensations of the antiparasitic drug class was for antimalarial drugs. The most frequently dispensed antimalarial drug was quinine (64% of antimalarial drug dispensations and 36% of antimalarial drug cost), whereas the most costly antimalarial drug was halofantrine (14% of dispensations and 50% of cost).
Cost by Baseline CD4 Cell Count
In univariate analyses, the CPPY was significantly higher when the patient had no remunerative activity, was living with a partner at home, had a low body mass index, and had a high WHO clinical stage or a low CD4 count. All other baseline variables, including participation in the Cotrimo-CI or Ditrame-CI trial, were not associated with the CPPY. In multivariate analyses, 4 variables remained associated with the CPPY: a low CD4 cell count (P < 0.001), a high WHO clinical stage (P < 0.001), the absence of remunerative activity (P = 0.003), and the presence of a partner at home (P < 0.001).
The mean CPPYs were $83 US, $101 US, $186 US, $233 US, and $459 US for patients with baseline CD4 counts ≥500 cells/mm3, 350 to 499 cells/mm3, 200 to 349 cells/mm3, 100 to 199 cells/mm3, and <100 cells/mm3, respectively. Figure 2 shows the percentage of CPPYs spent on the antibacterial drug class, on the antifungal drug class, on the other anti-infectious drugs, and on the other non-anti-infectious drugs overall for each of these CD4 groups.
In patients with a baseline CD4 count ≥500 cells/mm3, the most costly classes of drugs were the antibacterial (mean CPPY = $30 US), the antifungal (mean CPPY = $16 US), the analgesic (mean CPPY = $6 US), the antidiarrheic and/or antiemetic (mean CPPY = $5 US), the antihistamine and/or antipruritic (mean CPPY = $5 US), and the antiparasitic (mean CPPY = $5 US) classes.
In patients with a baseline CD4 count <100 cells/mm3, the most costly classes of drugs were the antifungal (mean CPPY = $208 US), the antibacterial (mean CPPY = $49 US), the antiparasitic (mean CPPY = $31 US), the antidiarrheic and/or antiemetic (mean CPPY = $31 US), the antihistamine and/or antipruritic (mean CPPY = $19 US), and the analgesic and/or anti-inflammatory and/or antispasmodic (mean CPPY = $14 US) classes.
These estimations of the cost of drugs in HIV-infected adults in sub-Saharan Africa before the ART period have various limits.
First, the drug prices are those of a given country during the period from 1996 to 2001 and through selected suppliers.
Second, these data are not representative of the situation of all HIV-infected patients and all HIV centers in sub-Saharan Africa. They are from 1 ambulatory clinic with a selected population, standardized algorithms of treatment reflecting the policy of a single team, and free access to drugs. These conditions differ substantially from those faced by most patients in rural and urban populations in sub-Saharan African countries. Such a favorable context could have increased the demand for health care and the number of prescriptions, leading to an overestimation of the costs and pharmaceutic utilization. Conversely, the early treatment of diseases could have avoided later complications that would have required more costly treatments.
Third, only drugs delivered at the study center pharmacy were taken into account. We did not have data on drugs that patients could have bought in other settings. Because drugs were free of charge, we assumed that most of the drugs consumed by patients during the study period were from the pharmacy of the study center; we may thus have underestimated drug consumption.
Despite these limitations, these data are valuable because (1) they are based on the exhaustive monitoring of drugs actually delivered to the patients and on the real cost of drugs at the center's pharmacy during the study period (to our knowledge, this type of exhaustive field practice-based report has never been done until now) and (2) they are from a care center in which the medical team aims to follow international standards given a limited budget. These conditions are shared by an increasing number of HIV centers seeking to provide care to insolvent patients as a result of new resources provided by emerging international funds. In this context, estimating the cost of drugs and discussing the weight of each drug at a given center may help to launch a discussion that is of more general interest.
First, as expected, a few expensive drugs for treating opportunistic infections occurring at the more advanced stage of HIV disease accounted for a large percentage of the overall cost, although representing only a small part of overall dispensations.13 This was the case for antifungal drugs. Conversely, the most frequently prescribed group of drugs (ie, analgesic and/or anti-inflammatory group) accounted for only 4.9% of the overall cost.
Second, sub-Saharan Africa has a specific spectrum of HIV diseases compared with that of industrialized countries. Bacterial infections and tuberculosis are the 2 most frequent causes of severe morbidity in HIV-infected adults.4,14 In our study, antituberculous drugs, whose expense has been significantly reduced because of efficient national and international tuberculosis programs, represented only 1.8% of the overall drug cost.15 Conversely, antibacterial drugs, which have never been the priority of any program, represented 26% of the overall drug cost (ie, the second most costly group).16
Third, our data illustrate the increasing cost of nonantiretroviral HIV drugs in proportion to the level of immunosuppression.17,18 The mean CPPY was almost 2-fold higher in patients with a baseline CD4 count of 200 to 349 cells/mm3 compared with those with a baseline CD4 count of 350 to 499 cells/mm3 and 2-fold higher in patients with a baseline CD4 count <100 cells/mm3 compared with those with a baseline CD4 count of 100 to 199 cells/mm3.
Fourth, the percentage of antibacterial drugs in the mean CPPY decreased from 32% to 15% in patients with a CD4 count ≥500 cells/mm3 and <100 cells/mm3 at baseline, respectively, whereas the percentage of antifungal drugs increased from 11% to 29% in the same groups. The most costly group was the antifungal drugs in patients with a baseline CD4 count <100 cells/mm3 and the antibacterial group in patients with a baseline CD4 count ≥500 cells/mm3. One may hypothesize that in adults receiving HAART in Abidjan, the CPPY of nonantiretroviral drugs is close to that we observed in the group of patients with a baseline CD4 count >500 cells/mm3. In settings where ART is largely available, reducing the cost of nonantiretroviral drugs would thus require lowering the price of the most costly drugs in patients with a high CD4 cell count, starting with antibacterial drugs.
The association of drug use with unemployment and lack of a stable partner has already been described in the different context of a study conducted in London in a population of HIV-infected adults.19 For the absence of employment, possible explanations could be that persons with a source of revenue (1) had the means to pay for care elsewhere and thus did not see the advantage of seeking care at the study center, rendering them an underrepresented group, or (2) had less time to seek care for less severe medical problems as compared with those without employment. These 2 explanations are also plausible for the association with stable partnership, which could be a source of shared revenue and compete for time expenditure at medical visits. Finally, unemployed status and/or being widowed, divorced, or single could all be associated with greater severity of illness, potentially leading to greater stigmatization and a higher risk of job discrimination and exclusion from employment.
In conclusion, these data contribute to the knowledge of the pharmaceutics of HIV infection in sub-Saharan Africa. They could be used when estimating budget requirements for HIV treatment programs as well as in further cost-effectiveness analyses that seek to prioritize health interventions. They suggest that successful ART, which would stabilize the CD4 count above 500 cells/mm3, could reduce by roughly 5-fold the cost of nonantiretroviral drugs in sub-Saharan African HIV-infected adults. Finally, they provide valuable information for decision makers and organizations seeking to improve the situation of the millions of HIV-infected people in the developing world by enhancing access to the most appropriate drugs. In the HAART era, antibacterial drugs could become the second group of drugs for which it is most worth fighting for price reductions.
© 2006 Lippincott Williams & Wilkins, Inc.