Private markets are an important part of the TB treatment landscape. This study represents, to our knowledge, the first detailed overview of the private TB drug market across countries with the majority of the world's TB burden. Whereas our previous market study focused on capturing all TB drug value, regardless of source, we now place an emphasis on defining the private sector and providing an analysis of its characteristics in terms of both products and manufacturers.
Based on the data in this study, the private TB drug market is stable, sizable in several HBCs, and demonstrates extensive variation in drug manufacturing and utilization patterns. The variation in first line dosage strengths and the low volumes or lack of availability of a number of second line agents suggest that both first line and second line treatment in the private sector diverge significantly from international and national guidelines. This, combined with the previous evidence of variable and incomplete treatment in the private sector, raises a considerable risk of treatment failure and the development of drug resistance, and strengthens the evidence supporting bold action in engaging the private sector.
In the study countries, the number of incident TB patients treated in public programs 
and the number potentially treated with private sector drugs (based on this study) are similar (67% and 66% of incident cases, respectively). In a recent survey 
, TB stakeholders provided estimates of private market size in the 22 HBCs 
. The relative ranking between countries from this exercise was similar to that from the IMS drug usage data. However, based on the IMS data, the absolute private market sizes are substantially greater than those estimated by stakeholders, likely due to repeated treatments in private and public sectors (and possibly unrecognized burdens of TB and treatment of pneumonia with TB drugs). The large size of certain private first line TB drug markets is consistent with some previous reports, e.g., that in India 86% of patients first sought help in the private sector 
, and that private practitioners provide overly lengthy regimens 
Private market size has remained stable despite the efforts by NTPs in many countries to increase their share of the TB treatment burden, and despite some indications that patient perceptions of public sector TB treatment are changing 
. This suggests that the non-NTP providers in several HBCs are fulfilling a continued need in the eyes of consumers, who choose to use the private market even though free products are available through their respective NTPs.
A variety of steps can be taken to prevent or limit irrational TB drug dispensing in the private sector 
. Regulations can be used to restrict TB medications to the public sector (e.g., in Brazil); even in this scenario, an efficient referral system is needed to reduce diagnostic delays for patients first accessing the private sector. Of note, our data show private sector TB drug volumes even in a country (Russia) thought by stakeholders to have only public sector TB drugs 
In countries with existing private TB drug markets, or for TB drugs that have other major indications (e.g., fluoroquinolones), a strictly regulatory approach may not be realistic. Other interventions are possible, however. Public sector programs can be made more extensive, appealing and flexible to reduce consumers' private sector preference 
, and innovative and strengthened PPM collaboration can be used to bring public sector treatment norms to private sector providers. The size of these responses should reflect the scale of the problem as identified here.
For PPM, subsidized drugs appear to be an important component for success in many settings 
, although some interventions concentrate primarily on advocacy around treatment norms 
; of note, the impact of this latter intervention would not be visible in our data. Increasing implementation of the International Standards for Tuberculosis Care (ISTC) 
via professional associations, with a greater emphasis (or a special campaign) on the standard regimens and dosages recommended, could also help make private TB treatment more rational.
One reason for concern about the private market is the variation in dosing practices, which has been documented previously at the provider level 
. With our data, we cannot determine exact prescribing behaviors or the rationale behind those behaviors, but we have obtained a comprehensive view of the large numbers of dosages available to providers. With the option of daily or intermittent dosing, the use of weight bands, an old and incomplete evidence base, and variable practices between countries, it becomes challenging to distinguish rational from non-rational use. Even with standardized regimens, dosage errors can be common 
, and dosage profusion only raises the likelihood that private providers may make mistakes or that their patients will be confused. Thus, the private sector is keeping alive the confusion that existed previously in the public sector. Just as WHO and the GDF helped to rationalize dosing in the public sector 
, so regulators may be able to use market authorization to rationalize private markets. Some guidance from WHO on this issue, geared specifically to an audience of national regulatory authorities, would seem appropriate. Manufacturers may also be persuaded via provider demand to make only the strengths recommended by WHO or the NTPs.
The fragmentation of the global TB drug market has been noted previously 
. We found, however, that such fragmentation is less pronounced in the FDC market, probably because FDCs are newer and more challenging to manufacture. Action at just four manufacturers (Lupin, Macleods, Wyeth and Sandoz) – e.g., to limit production to rational dosages made with high quality assurance – could reduce drug quality concerns for 69% of the private sector FDCs in these 10 countries. In addition, manufacturers could promote more uniformity of treatment via standardization of information in package inserts and roll-out of better provider educational materials.
In countries where FDCs have come to dominate the private sector, their prices have fallen to equal or even undercut the prices for loose drugs, an important learning for countries considering a public sector switch to FDCs. Indeed, the number of manufacturers is sufficient to create competition in both FDC and loose drug markets, and this competition may encourage manufacturers to differentiate their products by creating large numbers of strength variants. Confirmation of this hypothesis would require manufacturer interviews.
Others have noted the poor availability of certain second line drugs, notably capreomycin and clofazimine, despite their relative desirability within their respective classes of MDR-TB drugs 
. Our data on second-line drugs, though incomplete, are consistent with these statements. Several WHO-recommended drugs are not sold, and fluoroquinolones (and, to a lesser extent, streptomycin) dominate the landscape. Thus, it would be challenging for private providers to construct a full MDR-TB regimen in the private sector, even if they had the requisite training and laboratory facilities. However, the presence of many second line drugs suggests that drug supplies may be available for possible scale-up of MDR-TB treatment, either in the public or private sector. In the largest private TB drug market, India, all second line drugs are available, and volumes may expand greatly when rapid point-of-care tests such as GeneXpert® MTB/Rif are scaled up in the private sector.
Significant fluoroquinolone usage for TB does exist in several countries, although given the much smaller volumes of other second line agents, the fluoroquinolones are likely being used as monotherapy or as an add-on to first-line therapy rather than as part of a designated second-line therapy. (Based on the past history of TB regimens, private providers may also consider streptomycin, even more than fluoroquinolones, as a first line drug.) Thus, it seems that most MDR-TB patients are not receiving sufficient MDR-TB treatment in either the public 
or private sector; the vast majority of the estimated 440,000 new MDR-TB cases each year 
are likely not diagnosed or treated with rational MDR-TB regimens. The limited private sector usage of most second line drugs may be slowing the emergence of additional second line drug resistance, although the finding of significant fluoroquinolone usage for both TB and non-TB indications is consistent with the increasing incidence of fluoroquinolone-resistant TB in certain countries 
. As the demand for both fluoroquinolones and other second line drugs expands in the private sector, regulatory and oversight approaches will become all the more important to protect both lives and drugs.
A secondary goal of this study was to evaluate the robustness of IMS data coverage for the TB drug markets in HBCs. IMS databases are a commercial product, and sampling is conducted if and when the effort and cost required can be supported by the commercial value of the resulting data. As a result, data coverage is incomplete in lower income markets. These limitations include a complete lack of sampling (in 11 of the HBCs), less complete sampling than in higher income countries (for the remaining 11 HBCs), no medical audit (in 4 of the remaining HBCs), smaller medical audits, and certain additional data gaps.
Not surprisingly, IMS coverage is relatively low in the covered HBCs (sampling 1–20% of all outlets, compared to 72% of wholesalers in the USA or 94% of hospital beds in the UK, for example). However, comparison with manufacturer data indicates an acceptable level of accuracy (see Methods
). Certain channels are missed completely but relatively minor (see Results
and File S1
Medical audits, when available, proved reliable for first line drugs due to large volumes and few alternative indications. However, they were insufficient to provide reliable TB usage estimates for a number of the second line drugs in countries with less private sector TB treatment. For example, in some countries data on the usage of fluoroquinolones for TB was swamped by their usage for other indications, often in other formulations.
Every effort was made to exclude any private sector channels that were under the influence of the relevant NTP, and any challenges were judged to be manageable (see Results
and File S1
). We also considered that data collection may be biased away from outlets that serve lower income areas and clients, thus leading to an under-estimate of TB drug volume. (The reverse bias is also possible but arguably less likely.) This issue is a greater concern when measuring TB drug consumption because of the extreme concentration of TB in low income communities. It was not possible to evaluate this possibility, or its potential impact, although we note that IMS data in large countries are collected independently in multiple regions to minimize geographical bias. In the future, it may be worth embedding a simple, one-off income or poverty survey in the IMS data collection mechanism to determine whether any anti-poor bias exists.
On balance, these data issues suggest that under-estimation is more likely than overestimation, and that the major conclusions of this study are not likely to be affected by any of the data uncertainties. IMS is increasing efforts to strengthen data collection in India given the country's growing economic position; however, sample size in rural areas, where many TB patients are treated in the private sector, may continue to be limited. For many other HBCs, a lack of financial incentive for IMS, combined with the complexity of the various supply chains, may continue to be a barrier to data-strengthening efforts. A recent public-private partnership between the Medicines for Malaria Venture (MMV) and IMS is now, however, seeking to fill this gap in Sub-Saharan Africa (Renia Coghlan, pers. comm.).
The 10 HBCs varied significantly in the size of their private markets for TB drugs. Particularly large and stable private sectors for first line drugs were present in 4 HBCs, where carefully documented public sector treatment numbers may obscure the amount of private sector treatment occurring beforehand. In considering interventions to address private sector TB treatment, the size of the response should be commensurate with the size of the problem. Greater government and international support for implementing PPM, expanding the reach of public programs, improving regulatory oversight for both marketing approvals and quality assurance, and expanding public sector MDR-TB diagnosis and treatment could have a great impact in strengthening TB treatment outcomes in the entire health sector.