Antiretroviral drug supply challenges in the era of scaling up ART in Malawi

The number of people receiving antiretroviral treatment (ART) has increased considerably in recent years and is expected to continue to grow in the coming years. A major challenge is to maintain uninterrupted supplies of antiretroviral (ARV) drugs and prevent stock outs. This article discusses issues around the management of ARVs and prevention of stock outs in Malawi, a low-income country with a high HIV/AIDS burden, and a weak procurement and supply chain management system. This system for ARVs, paid for by the Global Fund to Fight AIDS, Tuberculosis and Malaria, and bypassing the government Central Medical Stores, is in place, using the United Nations Children’s Fund’s (UNICEF’s) procurement services. The system, managed by a handful of people who spend limited time on supply management, is characterized by a centrally coordinated quantification based on verified data from all national ART clinics, parallel procurement through UNICEF, and direct distribution to ART clinics. The model worked well in the first years of the ART programme with a single first-line ARV regimen, but with more regimens becoming available (e.g., alternative first-line, second-line and paediatric regimens), it has become more difficult to administer. Managing supplies through a parallel system has the advantage that weaknesses in the national system have limited influence on the ARV procurement and supply chain management system. However, as the current system operates without a central warehouse and national buffer stock capacity, it diminishes the ability to prevent ARV stock outs. The process of ordering ARVs, from the time that estimates are made to the arrival of supplies in health facilities, takes approximately one year. Addressing the challenges involved in maintaining ARVs through an efficient procurement and supply chain management system that prevents ARV stock outs through the establishment of a dedicated procurement team, a central warehouse and/or national buffer stock is a priority.

ART facilities is expected to increase from 400 to more than 700 [4].
Th is article will discuss operational challenges and operational research directions around the management and prevention of stock outs of ARVs in Malawi, a lowincome country in southern Africa with a weak PSM and health system.

ARV drug supply management system in Malawi
In many low-and middle-income countries, the capacity of the PSM systems has always been weak [5]. Th e management of supplies for what is essentially a chronic disease needing lifelong therapy is becoming increasingly diffi cult with further scale up of HIV care and treatment.

Responsibility for procurement and distribution
During the design of the plan for the national roll out of the ART programme in Malawi, the weakness of the PSM system was well understood, and the challenge for the HIV programme managers was to establish a PSM system for ARVs that performed better than the general existing supply management systems. Th e debate about whether to improve the national PSM system as part of the HIV programme design or to develop a parallel PSM system for ARVs was overtaken by events because in 2002, a World Bank mission advised that the government Central Medical Stores should not be used for the procurement of HIV and AIDS supplies with funding from the Global Fund to Fight AIDS, Tuberculosis and Malaria (the Global Fund) [6]. Th e United Nations Children's Fund (UNICEF) was proposed instead, and UNICEF, through its supply division in Copenhagen, Denmark, has been procuring and distributing ARVs in Malawi since 2004.

Public health approach
Malawi's Ministry of Health worked together with partners, including UNICEF, to develop the PSM system for ARVs that has been in place since 2004. Th e system has functioned relatively well and there have been no major stock outs of ARVs in Malawi, unlike many other medicines and medical supplies. It was clear from the onset of the national ART programme that it could only be successful if it was based on the realities in the health sector, in other words, by following a public health approach, i.e., a simplifi ed, standardized and structured approach [7].
Th e fi rst two years of experience have been reported earlier [8] and, here, we briefl y describe the model.

National, simple and standardized
At the outset in 2004, the ART programme focused on the roll out of one fi rst-line ART regimen using a fi xeddose generic combination of stavudine, lamivudine and nevirapine. Th is regimen was the only one available in all but a few health facilities in the country providing ART services. Two alternative fi rst-line ARV regimens (in case of toxicity) and a second-line ARV regimen (in case of treatment failure) were available in only four ART clinics in the country.
ARVs were distributed using a pre-packed kit system with "starter packs" and "continuation packs", which simplifi ed the supply management [8]. Using this approach, the ART programme was scaled up quickly and without any drug stock outs in its fi rst four years [7].

A "push system" and "ceilings" for the number of new patients starting ART
To ensure some control on rational drug forecasting for new patients, ART sites were given a ceiling of the maximum number of patients they were allowed to start each month on treatment. Th ese ceilings were established on the basis of the number of inpatient beds, the population served, HIV prevalence and tuberculosis case burden (serving as a proxy for HIV/AIDS burden, given the strong association between the two diseases). Th e ceilings could be (and were frequently) adapted to cater for a higher demand and available capacity to provide ART, but only after the additional drug supplies were in place. Th ese ceilings allowed easy calculations to be made on the number of "starter packs" needed for each site.

Quarterly monitoring of patient outcomes and ARV drug stocks
At the end of every quarter, cumulative treatment outcomes of all patients started on ART at each site and the number of patients retained alive and on treatment were recorded. Th is latter count (patients alive and on treatment) captured the number needing continuation therapy, which needed to be added to the number of new patients starting treatment. ARV stocks were also counted in the pharmacies, and these data allowed a system of rational drug forecasting to be put in place to calculate the number of "continuation packs" needed for each site.

Using a parallel procurement and distribution system for ART
Th e PSM system for ARVs was set up in parallel to the government's PSM system run through Central Medical Stores. It was, and still is, characterized by a centrally coordinated quantifi cation on the basis of verifi ed data from all ART clinics in the country, procurement through UNICEF's supply division and direct distribution of drugs to ART clinics, bypassing Central Medical Stores. Distribution of a six-month supply of ARVs to ART clinics was planned to take place when remaining drug stocks were estimated to be equivalent to two months of consumption.

Operational challenges and proposed solutions or research areas
As the national ART programme has grown, it has become clear that the ARV PSM system based on the tenets we have outlined has become increasingly diffi cult to manage. Th is is due to a number of reasons, such as increasing numbers of people on ART, increasing numbers of sites providing ART, and a greater diversity of diff erent ARV regimens. We will describe the challenges faced since the early years of the model and possible solutions or research needed to strengthen the system and avoid ARV stock outs.

ARV drug regimes
By the end of December 2010, 91% of patients were still on the standard fi rst-line regimen [9]. However, the number of diff erent ARV formulations available to primary level ART sites has increased over the years to nine, and in the whole programme to 19 (11 adult and eight paediatric). Moreover, tertiary ART clinics and non-governmental organizations have been requesting an increase in the number of formulations to provide more individualized care; some specialized private and non-governmental ART clinics use a larger number of ARV formulations.
It is clear that in a growing and maturing ART programme, the very limited number of ARVs used during the fi rst years of the national programme cannot be maintained and a balance needs to be found between improving individual care and maintaining a wellfunctioning PSM system. To better inform this debate and national policy, cost-eff ectiveness studies could be carried out to compare sites using many diff erent ARV formulations with the national programme. Such research would also be useful as a guide to other countries that may be struggling with this balance.

Patient ceilings and forecasting
Th e system of patient ceilings helped to prevent stock outs in the fi rst years of the programme. However, as the numbers of people on ART increased, the bulk of the ARV consumption shifted to patients already on ART and away from patients who were newly starting treatment. Th e system of ceilings therefore became superfl uous. Programme managers also became more experienced in more accurately forecasting the number of people on ART and ARV supplies needed. However, the number of people shifting to alternative fi rst-line ART regimens (for reasons of drug toxicity) and secondline ART (for reasons of treatment failure) has been more diffi cult to predict. Th is led at one time to over-supply of second-line ARVs and contributed to shortages and even site stock outs of alternative fi rst-line ARVs, although the main factor for the stock outs were bottlenecks in the Global Fund grant disbursement.
Th ese shortages have infl uenced the erratic growth of the number of people on these regimens ( Figure 1) and made forecasting of supplies needed more complicated. A potential area for operational research could be, therefore, the development and testing of models to predict the proportion of patients on treatment who need ARV substitution because of toxicity or switching due to treatment failure to help improve forecasting of secondline and alternative ARV consumption.

Procurement processes
Th e traditional PSM system of providing health facilities with a consignment of supplies to ensure that stock positions in health facilities are maintained within the set range of "minimum-maximum stock position", based on an average monthly consumption calculated using past consumption, does not work for ARVs as the patient burden grows and the consumption continually increases. Th e Global Fund requires that all ARVs are accounted for, and thus Malawi set up a monitoring and evaluation system in which patient data are audited and physical stock checks are performed by a supervision team visiting all ART clinics on a quarterly basis [10].
UNICEF is requested by the Ministry of Health to prepare "cost estimates" for the required supplies. Once these have been received, the ministry verifi es the "cost estimates" and contacts the Principal Recipient of the Global Fund grant, the National AIDS Commission, to prepare a Disbursement Request to the Global Fund. Th e Local Funding Agency performs some checks to verify if the request is correct, following which the Disbursement Request is sent to the Global Fund for fi nal approval. Once this has been obtained, the monies are transferred to UNICEF's Supply Division and "purchase orders" are issued.
ARV shipments usually arrive some four months after the purchase orders have been placed, depending on the manufacturers' production capacity and air transport to Malawi. Supplies are then distributed by UNICEF to health facilities within a month after arrival in country. Th e process from quantifi cation to receipt of supplies at the health facilities is complex, lengthy and involves multiple players (Figure 2). Orders of ARVs that are made on the basis of stock positions and the number of people on ART at the end of June 2011 will not arrive in health facilities before May 2012, almost one year later.
Th e long procurement process and increasing quantity of ARVs pose major challenges to prevent stock outs. First, predictions have to be made for one year ahead. While a degree of confi dence is applied around some of the estimates for each site, with an obvious tendency to over-order instead of under-order, it may be necessary to increase the in-country stocks of ARVs to cover the eventuality of potential delays in the process.
Second, the current parallel supply management system operates without a central warehouse and a national buff er stock, which diminishes the in-country capacity to prevent stock-outs of ARVs. Th e buff er stock, or safety stock, is the minimum stock that is kept on hand to protect against stock out. If there is no safety stock, stock outs will occur when deliveries are delayed or when there is an unexpected increase in demand [11]. As the ART programme has grown, the quantity of ARVs stored in health facilities has increased, sometimes beyond capacity. Th ere is therefore a pressing need to have a central warehouse facility. Th is would need to be properly managed to ensure that stored drugs do not pass their expiry dates.
Th ird, if a warehouse was set up, the option of having a national buff er stock would become feasible. Experience in Malawi has shown that once the national stock position (which refl ects all stock now available and soon to be available, including stock on hand and outstanding orders [11]) drops below two months of consumption (based on drug stock data collected from all national ARV sites), the prevention of stock outs becomes very diffi cult, and a lot of time and energy is put into re distribution of supplies. We believe that the establishment of a national buff er stock that would maintain the national stock position at six months' consumption is a priority.
Such a proposal has been put forward to the Global Fund, who has agreed to fund it.
Fourth, plans have been developed to increase the frequency of distribution to health facilities from six monthly to three monthly to reduce the average quantity of ARVs held in health facilities; however, this cannot happen until a central warehouse and a national buff er stock are established. More frequent distributions will not only reduce the pressure on storage space in health facilities, but also facilitate the response to unexpected changes in demand as redistributions take place between health facilities.

Institutionalization of the PSM system for ARVs
Th e work of keeping the PSM system for ARVs on track is currently in the hands of very few people in the Department of HIV and AIDS in the Malawi Ministry of Health (monitoring and evaluation offi cer, HIV care and treatment offi cers) and the National AIDS Commission (director of fi nance, chief procurement offi cer), who spend only part of their time on supply management. Th e value of the ARVs for the national ART programme will increase from US$60 million per year in 2011 to $100 million in 2015. Th e value of these supplies is high in comparison with the government's contribution to health, which was estimated at US$65 million in 2006 [12]. Th ere is therefore an urgent need to have an appropriately qualifi ed team, including staff from the Central Medical Stores, dedicated to procurement and supply management for ARVs and other HIV commodities, possibly combined with the procurement of other supplies procured with Global Fund monies. Th is team would need to take full responsibility, and be accountable, for procurement and distribution. Once such a team was in place, it would be important to carry out operational research to document whether this led to a reduction in the quantity of ARVs expiring and the frequency of (near) stock outs of ARVs in health facilities.

Dependency on a single donor for drug ordering and procurement
Adequate funding for ARVs is a conditio sine qua non. Paying for ARVs by patients is associated with higher loss to follow up [13] and poorer treatment outcomes, and WHO advises therefore that ART should be free of charge in low-and middle-income countries [14].
Countries with a small economy, like Malawi, are not able to contribute signifi cantly to the procurement of ARV supplies. Malawi's gross domestic product is US$318 (per capita) [15] and the government contribution to health was approximately US$65 million ($5.4 per capita) in 2005-2006. Th is is insuffi cient to procure a considerable proportion of the annual need of ARVs.
Th e vast majority of ARVs are procured with funds from the Global Fund and the lack of existing alternative funding sources makes the programme very susceptible to any bottlenecks in grant disbursement. Unforeseen delays in the PSM system over the past two years has led to periods of very low ARV stock levels, and even a few episodes of stock outs in a small number of health facilities. Most of these episodes were related to the late disbursement of funds, and virtually all steps prior to the disbursement of funds (see Figure 2) have contributed at least once to these delays. As a fi rst step, it is important for all stakeholders to agree on what are the key steps and bottlenecks required to move a drug order through to drug distribution in the fi eld, and then to fi nd potential solutions to these barriers. Time-related targets should be set for each step, and then operational research should be conducted every six months to determine if targets were reached and reasons for not reaching targets.
Another recent and interesting development is the establishment by the UN Foundation of the "Pledge Guarantee for Health" to help grant recipients to accelerate health commodity procurement. Th e pledge will support a bank to issue a letter of credit to a supplier on behalf of a grant recipient, which in principle should facilitate and speed up the procurement of supplies [16]. Th is mechanism has yet to be used.

Changes in recommended ART regimens
WHO launched revised guidelines for prevention of mother to child transmission and ART in 2010 [2,17]. Th ese guidelines advise changing the eligibility criteria to start ART and propose a move away from stavudinecontaining ART regimens to regimens with less toxicity. Th e number of people eligible for ART increases by an estimated 45%, and the recommended ARV regimens are at least twice as expensive in comparison with the stavudine-containing regimens. Any associated changes in national recommendations as a result will require careful monitoring to ensure that past regimens will still be used until the stock is fi nished and before the expiry date and that the new regimen will be available in suffi cient quantities to cover the needs (with progressive inclusion on new regimens). Malawi will then move on changing the ARV regimens.
Th is process, and the scale of the change, has no precedent. It will require a phasing out of the old fi rstline regimen so as not to waste drugs and introducing the new regimen with all that this entails for patient and community understanding and healthcare delivery. Opera tional research to carefully document such processes will be essential so that lessons can be learnt for other countries embarking on the same course and for additional regimen changes in the future.

Lessons learnt
From the start of ART scale up six years ago, important lessons have been learnt. Table 1 gives an overview of the challenges, proposed way forward and the ways in which operational research can guide future policies. First, an increase in the number of diff erent ARV formulations has led to a more complicated PSM system and has increased the risks of stock outs. Th e national ART programme therefore needs to maintain a careful balance between the optimization of individual care and a limited number of standard regimens to simplify and/or facilitate the supply management system. Second, minimum national stock positions of ARVs of two months' consumption often lead to very low stock positions in individual ART sites and additional work due to redistribution and repositioning of ARVs. National minimum stock positions should therefore increase. As the storage capacity for medicines at health facilities is limited and calculated based on stock positions of three months' consumption, the most feasible option is therefore to increase the national storage capacity through the establishment of a central stock. Increasing the national stock positions in country to six months of consumption will allow supplies to arrive four months later than planned without causing supply interruption. Increasing the national stock position further would increase the robustness of the supply system. However, increasing the minimum national stock position from two months to six months of consumption would cost US$18 million on the basis of the non-stavudine-based fi rst-line ARV regimen.
Th ird, increasing the frequency of deliveries from the central warehouse to ART sites would further decrease pressure on storage capacity in ART sites. Th ese two measures would not only improve supply management of ARVs, but would also mitigate the eff ects of delays in grant disbursement.
Fourth, the supply management of ARVs is carried out by a few individuals who are not fully dedicated to this activity. Given the increasing volumes of ARVs and other HIV commodities, a dedicated full-time team of staff experienced in supply management is needed to further develop the ARV PSM system and ensure its insti tutionalization.
Fifth, disbursement of grants needs to be streamlined.