083-094 chapter 5.indd

D ISCUSSION
For over nine years, the Global Project has collected data from areas representing 39% of the world’s sputum-positive TB cases. The number of countries participating in the project has increased nearly threefold since the first report. Performance criteria for the Supranational Laboratory Network have been developed, four new laboratories are candidates to join, and nine rounds of proficiency testing have been completed. Guidelines for the surveillance of drug resistance in tuberculosis have been revised, and a fourth version of software to analyse drug resistance has been developed. Most importantly, global results of the project are fuelling discussions about policy implications. While coverage of the project is increasing on the whole, it varies widely. The areas represented in this project are those with at least the minimum requirements to conduct surveillance, and it is likely that the worst situations have not yet been uncovered. While the majority of countries surveyed report low to moderate levels of resistance, several countries have serious drug resistance problems that may jeopardize the control of TB. Paramount to the success of TB control is the expansion of good TB control policies worldwide, accompanied by expanded surveillance of drug resistance to identify areas that require directed initiatives to manage resistance.
The data reported in this third phase of the Project have reinforced many of the conclusions drawn in its first and second reports, and contribute to a more in-depth analysis of dynamics and trends. The median prevalences of any resistance and MDR among new cases of TB in this phase of the project were 10.2% and 1.1%, respectively. Despite the inclusion of different countries in each phase of the project, the medians for most resistance parameters were similar in all reports, but the outliers varied. MDR was again found to be highly prevalent in several areas of the Russian Federation and other countries of the former Soviet Union (Kazakhstan, Aral Sea regions of Uzbekistan and Turkmenistan, Estonia, Latvia, and Lithuania) as well as in some areas outside of this region, including Israel, Ecuador, and Liaoning and Henan Provinces of China. While it is important to follow trends in prevalence of MDR-TB, other resistance patterns must also be watched and closely monitored, as they are potential precursors of MDR. A prevalence of non-MDR INH resistance greater than 10% was found in fifteen countries. High prevalence of non-MDR RMP resistance were less frequent, but also of concern. Though the Global Project has been operating since 1994 very few countries have reported data for all nine years. Thus, for many countries, it is too early to interpret trends. Data from repeated surveys employing comparable methodologies over several years are essential to determine with any certainty in which direction prevalence of drug resistance is moving.
It is also important to note that data reflect TB programmes at various stages of implementation; thus trends must be interpreted in the context of additional relevant programme indicators. Programme improvement can affect the prevalence of resistance in several ways. A better programme can result in the reduction of the overall number of re-treated cases; however, difficult (resistant) cases may persist. Thus, in some instances an increase in MDR prevalence in a population may reflect a stable number of MDR cases but a decrease in the overall re-treatment population. It is also possible that, as systems improve, coverage and reporting of culture and DST may be more regular resulting in increases in reported case numbers. Improvement in laboratory proficiency, particularly the sensitivity and specificity of drug susceptibility testing, may also affect the observed prevalence of resistance. The scenarios outlined above highlight the importance of evaluating trends in prevalence of drug resistance within the context of relevant programme developments.
MAGNITUDE OF RESISTANCE BY REGION
AFRICAN REGION
All WHO regions were represented in this phase of the Global Project. In the
African region, areas representing 49.5% of new smear positive TB patients and 37% of countries have been surveyed since 1994. The African region has the fewest settings for which trends can be identified. Only Botswana, Sierra Leone, and Mpumalanga Province, South Africa, have carried out repeat surveys. In general, drug resistance in the region is low, but the trends in Botswana and Mpumalanga Province in South Africa indicate that it is increasing. Botswana in particular showed a significant increase in prevalence of any resistance. Though drug resistance was not significantly associated with HIV in the most recent survey, given the high prevalence of HIV in Botswana, even relatively small increases in resistance could have a significant impact on the population if patients with DISCUSSION
HIV infection become coinfected with drug-resistant M. tuberculosis.77 In Botswana the number of TB cases notified almost doubled between 1995 and 2001; the case detection of smear-positive TB cases reached 75% in 2001 and the treatment success rate was 77%. According to these indicators, Botswana should be able to reach WHO targets for TB control; however increasing incidence, as well as drug resistance, may jeopardize efforts to control TB in the country. In this regard, it will be extremely important to monitor the prevalence of drug resistance in Botswana over time, as an indicator of resistance behaviour in sub-Saharan African countries with high HIV prevalence. Sierra Leone, with two data points in the first and second reports, showed very little change in prevalence of resistance. However, in view of the social unrest in the country, it is possible that the NTP may have had difficulty implementing a high quality DOTS programme, and this may have affected the prevalence of resistance. Reported prevalence of resistance from recent surveys in Algeria and the Gambia was very low, and only slightly higher in Zambia, confirming the low levels of resistance in the region reported in previous phases in the project. A survey in the city of Kinshasa, Democratic Republic of Congo, reported results for combined cases only. However, the prevalence of MDR appeared to be relatively high for the region with 39% prevalence of any resistance and three-quarters of MDR cases resistant to all four drugs. It will be important to conduct a nationwide survey, as urban centres in general report higher prevalence of resistance than the national average. A survey is planned in Bas-Congo.
In 2001 and 2002, South Africa conducted a nationwide survey by province. Each province adopted a multistage cluster sampling strategy and samples were representative of the entire province. Results of this survey indicated moderate levels of MDR-TB, ranging from 0.9% to 2.6% among new cases, with an increase in Mpumalanga Province, the only province surveyed more than once in the Global Project. Prevalence of MDR-TB in new patients in Mpumalanga province increased from 1.5% in 1997 to 2.6% in 2001, while in re-treated patients prevalence increased from 8.1% in 1997 to 13.9% in 2001. During this same period, performance indicators for TB control have progressively declined and cure rates in Mpumalanga are currently among the lowest in South Africa. Regular reports of drug shortages and high default rates from treatment over this period have given further evidence of conditions for increasing drug resistance. In contrast, data from a previous province-wide survey in Western Cape, not included in the Global Project but following the accepted methodology, indicated relatively stable levels of drug resistance. Prevalence of resistance found in the 2001–2002 survey was nearly the same as those reported in the 1993 survey.78 The TB programme in the Western Cape is currently among the best performing programmes in South Africa, which may partly explain the stable MDR-TB prevalence figures. In addition, the Western Cape was the first province in South Africa to systematically treat MDR-TB patients (since the mid-1980s). In the recent survey, it was the only province where there was not significant under detection of retreatment cases, i.e. the numbers estimated from the survey tallied with the numbers reported by the TB control programme. The above findings, when compared with those in Mpumalanga, illustrate two points: the potential impact of poor TB control, and the importance of regional surveys in large countries and countries with a high TB burden. The findings would have been obscured in a survey of South Africa as a whole. The nationwide survey in South Africa incorporated surveillance of HIV among TB patients. Prevalence of HIV among TB patients was 55.3 % nationwide, but ranged from 28.2% in the Western Cape Province to 71.9% in the Free State Province. Surveillance of HIV among TB patients in the context of a drug resistance survey is of particular importance, in order to study the relationship between the transmission of drug-resistant TB and DISCUSSION
HIV over time. Since 2001, drug susceptibility testing is conducted on all re-treatment, failure and non-converting cases and a standardized treatment regimen is provided to MDR-TB cases. Currently there are about 4000 MDR-TB patients receiving treatment. The standardized treatment regimen is priced at about US$ 3400 (non-GLC price) per patient for drugs alone. Nationwide surveys are underway in Ethiopia and Kenya. Results should be available in 2004. Mozambique, which had one of highest prevalences of MDR (3.5%) in the region in its 1998-1999 survey, is preparing a repeat survey. Senegal and Rwanda will start surveys shortly. Nationwide data from the United Republic of Tanzania, Nigeria, and Democratic Republic of Congo should be prioritized on the agenda of TB control programmes in these high-burden countries in order to detect and address emerging MDR. Surveys in Côte d’Ivoire and Uganda should be repeated in the near future. REGION OF THE AMERICAS
In the Americas, areas representing 92.6% of all smear positive cases and 41%
of countries have been covered by the project. The high population coverage is largely a result of strong laboratory networks, good surveillance coverage of the countries with higher TB burden, and commitment to surveillance in the region. To a great extent, as found in previous reports, the prevalence of MDR is low in the region as a whole; however, there are important outliers. In North America, Canada has shown a relatively steady prevalence of drug resistance among both new and previously treated cases. TB case notification has decreased since 1997 and prevalence of MDR has never risen above 1.0% among new cases, or 4% among previously treated cases. The USA has shown decreases in overall TB notifications as well as overall numbers of drug-resistant cases since 1995. Decreases in MDR and any rifampicin resistance were significant among new cases, and decreases in any resistance, MDR, any rifampicin and any isoniazid resistance were significant among previously treated cases. Cuba, a country with a history of good TB control, reported significant decreases in any resistance among both new and previously treated cases. Argentina showed a decrease in almost all resistance parameters between surveys. However, the first survey in 1994 sampled a cluster of HIV patients, among whom an outbreak of MDR had recently occurred, which probably biased the survey results. The 1999 survey reflected a change in sampling methodology adopted to minimize bias, and the lower prevalences reported in this survey are probably a more accurate representation of actual levels in the country. Uruguay showed a slight increase in all resistance parameters; however, the magnitude of overall resistance in the country is, to date, the lowest reported in the region. Preliminary results were available from Honduras and Ecuador. The sample from Honduras indicated that prevalence of drug resistance is similar to that in the majority of countries surveyed in the region. Ecuador was an extreme outlier for the region (MDR 4.92% among new cases). The high prevalence of MDR in Ecuador is most likely a result of late and partial implementation of DOTS. Interestingly, in Ecuador, MDR constitutes a higher proportion of any resistance in both new and previously treated cases than in any other country in the region. This is in contrast to Bolivia, which has a relatively high prevalence of any resistance, only a small proportion of which is MDR. Chile, which saw only slight and non-significant increases in resistance between 1997 and 2001, has employed one of the most innovative surveillance policies in the region, which may DISCUSSION
prove to be a useful model for other countries. Chile performs continuous surveillance of all previously treated patients, and conducts a survey on a representative sample of new cases every three years, thus obtaining accurate information on both populations, strengthening routine patient history interviews, and identifying resistance patterns of previously treated patients early in treatment. This model can perform well in countries with long-standing TB control; however, in new programmes where patient interviews have not been well established, the risk of misclassification may be a concern. A survey in Paraguay is almost completed. Brazil, Colombia, Costa Rica, Dominican Republic, Mexico, Panama, and Peru will commence surveys shortly. A repeat survey in the Dominican Republic will be particularly useful, since an MDR prevalence of 6.6% was reported in the first survey; however, the decade lapse between surveys presents an obstacle to useful comparison of results. A second survey in Mexico will be nationwide and not partial as in the 1997 survey. Brazil will shortly undertake a nationwide survey by state, sampling separately new and previously treated populations and incorporating surveillance of HIV in TB patients. A primary objective of this survey will be the strengthening of the TB laboratory network in the country.
EASTERN MEDITERRANEAN REGION
The Eastern Mediterranean Region has both the lowest coverage of new smear-
positive cases (11.6%) and the lowest proportion of countries surveyed (21.7%). Trends are available only for the Gulf States of Oman and Qatar, both with small numbers of total cases and low to moderate levels of resistance, much of which is imported. Trends are difficult to interpret because of the small numbers of cases, though drug resistance does not appear to be a problem in either of these countries. The survey recently conducted in Egypt showed moderate levels of MDR among new cases (2.2%) but unusually high prevalences of any and RMP monoresistance. Data on resistance to rifampicin must be interpreted with some caution as quality assurance procedures have not yet been finalized by the corresponding SRL. Surveys are under way in Jordan, Lebanon, and the Syrian Arab Republic, and the Islamic Republic of Iran and Morocco are preparing for repeat surveys, with nationwide coverage in Morocco. Other Gulf States, such as Bahrain, Kuwait, and United Arab Emirates, conduct routine drug resistance surveillance of all TB patients but this information has not been yet made available and annual proficiency testing with an SRL is not currently carried out. EUROPEAN REGION
In the European Region, areas representing 47.8% of all smear-positive cases,
and 70.6% of countries have been surveyed. The high coverage of countries in the region is primarily a result of continuous surveillance of bacteriologically confirmed TB cases in much of Western Europe and parts of Central Europe. The European region displays the greatest heterogeneity of resistance parameters in the world, including both the highest and the lowest prevalences. Because of the varied history of TB control within the region, it is important to discuss the three subregions separately (Western, Central, and Eastern). In general, drug resistance is not a public health problem in Western Europe. Drug resistance prevalence is low, evidenced by an MDR median prevalence below 1%, much of which has been attributed to imported isolates and localized outbreaks. In Germany, higher proportions of any resistance and MDR were reported among new and previously treated cases in 2001 and 2002a following changes in the national surveillance system in 2001. Before 2001, drug resistance data in Germany were based on a nationally representative sample covering DISCUSSION
55% of local health departments that had elected to report drug susceptibility test results, contributing 50.2% of all reported cases. Since 2001, results of drug susceptibility testing are notifiable by law and are analysed centrally; the higher proportions observed in 2001 and 2002, therefore, do not necessarily reflect an increase over time, but may be due to the methodological change. In France, most resistance parameters among new cases are stable, and resistance in the country is relatively low. Resistance to any drug is increasing significantly in Barcelona, but individual parameters are difficult to interpret. When data were stratified by origin of birth, resistance was higher in the foreign-born population. This, coupled with an increase in immigration in Barcelona since 2000, suggests that the rising prevalence of resistance may be linked to immigration. Israel is an outlier, presenting the highest levels of resistance for most parameters. The situation of this country is unique, because of the high levels of immigration from areas of the former Soviet Union.79,80,81 Between 80% and 85% of TB patients in Israel are foreign-born, mainly from Ethiopia and countries of the former Soviet Union.82 Data from 2001 and 2002 indicate that MDR among new cases dropped to 5.8% and 2.0% respectively. Total notifications of TB cases remain low ranging from 281 to 340 cases per year, while total number of MDR cases has fluctuated between 41 in 2000 and 17 in 2002.
Data from countries in Central Europe show relatively low prevalences of drug resistance, with indications of an increase in resistance in a few countries. Though based a Based on unpublished data from Robert Koch Institute in Germany. on only two data points, Poland showed a significant doubling in all resistance parameters among new cases; however, the prevalence of MDR-TB remains low at 1.2% among newly detected cases. Slovenia shows stable prevalence of any resistance with very few MDR cases detected. Slovakia has shown steady but non-significant increases in resistance parameters since reporting began in 1998. Both countries have long histories of good TB control. The Czech Republic reported a steady but non-significant increase in prevalence of any resistance and a recent, non-significant increase in MDR among new cases. In general, overall drug resistance does not appear to be a problem at this time; prevalence is low and total MDR cases are very few. It is likely that resistance will continue to be low in Central Europe.
Considerably higher prevalences of drug resistance, including MDR-TB, have been reported in Eastern Europe. The first phase of the Global Project identified drug resistance as a major public health problem in areas of the former Soviet Union. The second report reiterated these findings, and evidence from the third phase indicates that drug resistance is of serious magnitude and extremely widespread, and that there are high proportions of isolates resistant to three or four drugs.
The prevalence of MDR among new cases in Estonia increased between 1994 and 1999, as did case numbers, and significantly decreased in 2000; however, preliminary data from 2001 and 2002 indicate a slight increase in MDR among new cases. This increase, coupled with decreasing overall notifications of new cases, results in a prevalence similar to that observed in 1999, around 17%. MDR case numbers appear to be relatively steady, and translate to fewer than 70 new MDR cases per year. Recent data also indicate a further decrease in MDR among previously treated cases, both in prevalence and total case numbers, appearing to reflect a real sustained decrease over three years. This is encouraging and may be the result of well implemented DOTS in the recent past and national DOTS Plus coverage, though only a sustained decrease over the coming years DISCUSSION
will confirm this. With a high prevalence of MDR in the population, it is expected that interrupting the chain of transmission to produce a decrease in prevalence of resistance will take some time. Lithuania showed steady and important increases in most resistance parameters; however, only increases in MDR among previously treated cases and any resistance, any INH, and MDR among combined cases were significant. Additionally, overall re-treatment notifications have almost doubled since 1999. DOTS is implemented in 95% of the country, and Lithuania has plans to apply to the GLC once DOTS coverage is complete. In Latvia, new case notifications have increased steadily since 1996 as have total number of cases with any drug resistance; this is reflected in a slight but steady increase in prevalence of any resistance since 1998. Prevalence of MDR has fluctuated slightly since 1998 showing a small decrease in 2000; however, the total number of reported new cases with MDR was relatively steady from 1998 through 2000. Both the total number of re-treatment cases and the number of cases with drug resistance including MDR decreased in Latvia until 1999 and then increased in 2000. This may be a result of more complete reporting of DST results of all re-treatment cases; prior to 2000, only DST results for relapse cases were reported. Surveys have been implemented in 3 oblasts of the Russian Federation. Tomsk Oblast presented a high prevalence of MDR among new cases with significant and increasing trends in most resistance parameters. Though prevalence of MDR appears to be increasing among previously treated cases, decreases in the total number of re-treatment cases since the first survey in 1998 were reported in 2002, while numbers of MDR cases fluctuated slightly over time. Trends must be interpreted with caution, as surveillance methodologies have changed over the years (methods of case detection and involvement of rural areas), and implementation of DOTS has increased. In order to determine drug resistance trends with any certainty, surveillance of drug resistance must continue. DOTS Plus was begun in Tomsk in late 2000. Data from the second Global Report indicated that prevalence of MDR was increasing in Ivanovo Oblast, and recent unpublished data from 2002 further support this trend. Kazakhstan recently undertook a nationwide survey, which found a very high prevalence of MDR (14.2%) among new cases, with the majority of these cases resistant to four drugs. The sample size was based on new cases; however, during the survey intake period approximately equal numbers of new and previously treated cases presented at diagnostic units, and 47% of the total sample was composed of previously treated cases. This scenario suggests that there are at least 3000 MDR patients diagnosed in the country each year, 70% of whom are resistant to four first-line antituberculosis drugs. DOTS was implemented from 1998, and while notification rates continue to rise, TB mortality rates have steadily decreased. The use of second-line drugs appears to be extensive and a clear management strategy for MDR cases needs to be developed. A survey of the Aral Sea region in DOTS-covered areas of Uzbekistan and Turkmenistan indicate a very high prevalence of MDR in Karakalpakstan, Uzbekistan, and a moderately high prevalence of MDR in Dashoguz Velayat, Turkmenistan. Both areas reported high proportions of previously treated cases, with most MDR cases resistant to three or four drugs. Very high prevalences of drug resistance have now been confirmed in Estonia, Latvia, Lithuania, Tomsk and Ivanovo Oblasts in the Russian Federation, Kazakhstan and the Aral Sea regions of Dahoguz Velayat, Turkmenistan, and Karakalpakstan, Uzbekistan. Preliminary evidence suggests even higher prevalences in other areas of the former Soviet Union. TB control following international guidelines must be given high priority, and immediate action must be taken to increase the coverage of drug resistance surveillance in the former Soviet Union. Currently, surveys are being planned in Kyrgyzstan, Moldova, DISCUSSION
Georgia, Donetsk (Ukraine), Armenia and Azerbaijan as well as a nationwide survey in Uzbekistan. The Russian Federation is developing a plan to systematically survey all oblasts. Currently protocols are being developed for 10 oblasts. In addition to routine national surveillance, it will be of the utmost importance to begin to gather data on the crossover of HIV and MDR, in order to implement proper infection control measures.84 One of the identified obstacles to surveillance in the region is the lack of internationally quality controlled laboratories. In order to obtain reliable data from these areas, proficiency testing of national or regional reference laboratories must be carried out immediately. Of the highest priority is the proper implementation or expansion of DOTS and DOTS-Plus both to control MDR in areas of known high prevalence and to avoid an epidemic of potentially catastrophic proportions that could destabilize TB control in the region. SOUTH-EAST ASIAN REGION
The South-East Asian Region also has a low survey coverage of new smear-
positive cases (areas representing 14.1% of all smear-positive cases), largely as a result of low coverage of Bangladesh, India, and Indonesia. Sparse data make it difficult to assess the regional scenario with any accuracy. Resistance in the countries surveyed (Nepal, Thailand, and a few settings in India) is reportedly moderate, but with large numbers of total TB cases making the actual burden of resistant cases much higher. Additionally, a large and unregulated private sector and over-the-counter access to a wide range of TB drugs in many South-East Asian countries provides a conducive environment for the development of drug resistance. Increasing surveillance coverage should be considered a regional priority. Recently, district surveys were carried out in India, in the states of Maharashtra, Tamil Nadu, and Karnataka. Data from two surveys in Tamil Nadu (one at the state and one at district level) indicate prevalences of MDR among new cases of around 3%, representing an enormous number of MDR-TB cases. Additionally, many states in the country report high proportions of re-treatment cases in the NTP, though very little is known about prevalence of drug resistance among this population. Only well designed state level surveys, sampling new and previously treated cases separately, will be able to assist in ascertaining a baseline prevalence in these populations at the state level. India is developing a plan to conduct nationwide surveillance of drug resistance by state, starting with two states this year and gradually adding and re-surveying states over time, as has been done in China and is planned in Brazil. Another objective of nationwide surveillance of drug resistance in India is the strengthening of laboratory networks, through the implementation of an external quality assurance system for smear microscopy and QA of culture and DST in state laboratories. The plan is ambitious and will require concerted political commitment to succeed. Nepal has conducted three surveys with mixed results. Prevalences of resistance among new cases from the first and third surveys were similar; however, the second survey found considerably higher prevalence of resistance among new cases. Resistance among previously treated cases (surveyed only in the last two surveys) decreased. The reasons for the fluctuations in resistance prevalence between surveys are unknown; further surveys will be helpful in confirming the prevalence of MDR-TB. Thailand showed significant decreases in prevalences of MDR and any resistance between 1997 and 2001, though notifications of all TB cases increased considerably during this time. It is possible that the decrease in drug resistance in Thailand is a result of good TB control and will be sustained in the future; however, at this time, there are not sufficient data to predict this with any certainty. It will be important to keep surveillance DISCUSSION
of drug resistance on the agenda in order to evaluate the progress of DOTS under recent health care reforms and to follow trends in high-risk populations identified in the course of the surveys.85 Myanmar is finalizing a nationwide drug resistance survey and Indonesia has plans to conduct a partial survey in 2004. Bangladesh constitutes another important gap in drug resistance information in the region and nationwide surveillance there should be a priority. The human and financial capacity of the national reference laboratory needs to be enhanced before proficiency testing can take place and a nationwide survey implemented.
WESTERN PACIFIC REGION
The Western Pacific Region has very high surveillance coverage of new smear-
positive cases (areas representing 50.8% of smear positive cases), in part because of the broad coverage of China and a strong commitment to surveillance in the region. China has a progressive surveillance policy and has surveyed six of 31 provinces in the country, with a repeat survey completed in Henan, and repeat surveys planned in Guangdong, Zhejiang, and Shandong provinces. New surveys are under way in Inner Mongolia and Hunan, surveys of Beijing and Shanghai cities are due to start shortly, and surveys are planned in Xinjiang, Heilongjiang, and Chongqing. Hong Kong SAR conducts continuous surveillance. Prevalence of drug resistance ranges from low in Hong Kong SAR to extremely high in Henan and Liaoning. The decrease in any resistance among new cases in Hong Kong SAR was significant and supports documented evidence of this trend.86 Liaoning province reported an extremely high prevalence of MDR (10.8%) among new cases; however, patient reinterviews indicated an estimated 25–30% misclassification of previously treated cases as new cases, indicating that prevalence of MDR is more likely to be around 8%. Henan experienced a similar misclassification in 1996 (MDR adjusted from 16% to 10.8%). Though prevalence of MDR in the subsequent 1999 survey (7.8%) was lower than the adjusted figure reported in 1996, methodological problems preclude the acceptance that there was any real reduction in prevalence. In both settings, misclassification was difficult to avoid because of previous policies, and this underlines the importance of rechecking records and reinterviewing patients during the course of a survey. Of particular note in China is the well designed QA system that has evolved in conjunction with drug resistance surveillance. Proficiency testing of provincial laboratories that have conducted or are preparing to conduct surveys takes place annually, even after the survey has been completed. Annual PT of provincial laboratories should remain a priority; in this way China is taking steps to establish a long-term QA system for DST in the country.
Data from other countries surveyed indicate that anti-TB drug resistance elsewhere in the region is low in magnitude. Japan provided data from a 1997 nationwide sentinel survey and Mongolia from a 1999 nationwide survey, both showing relatively low prevalences of drug resistance. Singapore, which conducts surveillance of all culture-positive TB cases, also reported low prevalence of resistance. The 2001 survey in Cambodia indicated a very low prevalence of resistance in the population, which is important in view of the high number of TB cases and high HIV prevalence in the country, and may in part be a result of late introduction of rifampicin. Given the high prevalence of INH resistance, and somewhat moderate prevalence of MDR reported in the last survey, and results from recent municipal surveys in Viet Nam, it will be important to follow the prevalence of resistance in this country. Resistance in Australia, New Zealand, and the South Pacific islands appears to be largely of foreign origin and low in magnitude at this time. DISCUSSION
ECOLOGICAL ANALYSIS
The ecological analysis reiterated findings of the first two phases of the project,
namely that re-treatment is strongly associated with drug resistance. This finding highlights the importance of giving greater attention to this group of patients in terms of treatment, reporting, and representative drug resistance surveillance. In general, the ecological analysis was inconclusive with the exception of the above finding. Despite the inherent weakness in ecological analysis of aggregate data, the conceptual model can constitute a step forward for more reliable and individual data collection.
BURDEN OF MDR
In many countries with relatively few TB cases, a high prevalence of resistance
does not reflect high absolute numbers of cases. Conversely, a low prevalence of MDR applied to high-burden settings, such as South Africa or some provinces in China, could reflect a considerable number of MDR cases. In South Africa, 2.9% of all TB sputum-positive cases are MDR, while in Kazakhstan one in every four smear-positive pulmonary TB case is MDR. Kazakhstan has a lower burden of disease and South Africa a lower prevalence of MDR, yet both countries have an enormous burden of 3000 or more MDR cases diagnosed each year. Nepal, Thailand and Estonia are other examples. Nepal and Thailand have 3.8% and 1.8% MDR prevalence among combined cases, respectively, but have about 5 times as many MDR cases as Estonia with 19.5% MDR. Ultimately the magnitude of the problem rests on the ability of a country to treat patients effectively. The international community and partners must prioritize both technical and financial support to countries with an identified MDR-TB problem. Failure to do so will result in a situation where a substandard level of care and irrational use of second-line drugs will continue to perpetuate the transmission of, and potentially amplify further, highly drug-resistant isolates of tuberculosis. A critical mass of resistant cases has great potential not only to halt the progress of TB control but to reverse it. DOTS-Plus is a comprehensive management strategy for MDR-TB that includes the five tenets of the DOTS strategy. By definition, it is impossible to conduct DOTS-Plus in an area without having an effective DOTS-based TB control programme in place. Therefore the international community must also be willing to reject GLC applications from countries/settings where basic TB programmes are suboptimal or not fully in place, in order to prevent further amplification of resistance. Fortunately, a number of funding mechanisms, such as the Global Fund to Fight AIDS, Tuberculosis and Malaria, now require all requests for the funding of second-line TB drugs to go through the GLC for approval. SRLN
The network of supranational laboratories currently comprises 20 laboratories in
five regions; three laboratories have closed since the beginning of the project and four are candidates to join. The network has completed nine rounds of proficiency testing since 1994; cumulative results over the nine rounds generally indicate overall high performance of the network. Results of proficiency testing at the NRL level are subject to scrutiny by the corresponding SRL and then shared with the Global Project. Following an evaluation by the supranational laboratory, a decision is made on whether to carry out the survey or DISCUSSION
repeat proficiency testing. The Global Project, partners and the SRLN itself have proposed performance criteria for the network at different stages of the project in order to address substandard performance and maintain a high level of quality to ensure reliable survey results. None of these proposals has ever been strictly adopted. Therefore, it was proposed that the SRLN develop criteria for network and NRL standards, with the following components: quantitative and qualitative indicators for admission into the network, assessment of performance, and contribution to the network of existing members, and type and timing of remedial action when substandard performance is detected. The network has recently agreed such criteria and details will be published in the coming year. In the future, performance criteria for NRLs will also be developed and agreed upon by the network.
Borderline isolates detected in two rounds of proficiency testing affected the overall performance of the network in those rounds and made reproduction of those isolates for PT of NRLs virtually impossible. The SRLN agreed that efforts should be made to exclude borderline isolates from quality assurance exercises, and borderline isolates that were included in previous panels should not be included in standard calculations of sensitivity and specificity for the purposes of evaluating individual laboratories. Preliminary research has shown that at least one of the apparently borderline isolates was in fact a mixed culture containing one drug-resistant and one susceptible isolate; however, further exploration is warranted.
The contribution of the Supranational Laboratory Network to the Global Project has been paramount in achieving the high quality and coverage of surveillance of drug resistance to date. While the WHO/IUATLD Global Project coordinates the SRLN and covers a small portion of operational costs, the supranational laboratories invest their own time and resources to assist in this global endeavour. There is a need for these costs to be met internationally to stabilize and enhance the network. Recently members of the SRLN have further contributed their time and experience to several regional DST training courses. Additionally, the majority of SRLN members along with partners have become involved in a Laboratory Strengthening Subgroup organized by the DOTS Expansion Working Group. The Laboratory Strengthening Subgroup seeks to assess and develop plans for improvement of entire national laboratory networks, with an emphasis on sputum smear microscopy. A recent meeting of the SRLN and the Subgroup agreed to develop a comprehensive workplan for the coming two years, and to engage the international community in putting laboratory network improvement firmly on the TB control agenda. Improved laboratory networks will translate into improved diagnostic and treatment capacity, and more accurate surveillance of drug resistance.
DRUG RESISTANCE PATTERNS
Among the advantages of being able to analyse DST results obtained from 1994
till 2002, from over 90 000 isolates and over 100 sites on all continents, is that even the most infrequent drug resistance types (combinations) can be detected with a high degree of sensitivity. This is not always true of the data from individual sites, where the number of cultures examined is less than 1000, given that some drug resistance types show prevalences of 0.1% or even lower. The total number of isolates examined is sufficiently high to guarantee statistical significance of both new cases and previously treated cases, even though all settings within some regions such as the Eastern Mediterranean and South-East Asia are not necessarily representative of the regions as a whole. The consistency of the findings argues for the robustness of the following conclusions.
DISCUSSION
Increase in drug resistance rates
Clinical TB drug resistance develops as a result of spontaneous genetic mutations
in M. tuberculosis, which are selected if antituberculosis drugs are used inconsistently or inappropriately, e.g. if monotherapy is applied. In patients with drug-resistant tuberculosis, additional drug resistance may develop if a prescribed multidrug regimen includes the drugs these patients are already resistant to. In this situation, some of these patients may end up effectively receiving monotherapy. In this respect the findings of worldwide drug resistance surveys are revealing, in that the prevalence of drug resistance is significantly higher among previously treated patients than among new patients in all regions. The only logical inference is that present treatment practices create significant numbers of new resistant cases and amplify already present resistance.
This analysis shows a remarkable consistency, both globally and regionally, in the distribution of the major drug resistance types, as well as in the increase in drug resistance prevalence among previously treated cases relative to new cases. It should be noted that prevalence of drug resistance observed in previously treated cases is higher than in new cases in all regions. However, the size of the difference varies between regions. Since this difference is in great part directly related to the quality of drug treatment, this apparent characteristic could well lead to the development of an indicator that would measure the quality of treatment practices. Amplification of drug resistance proportions
It is well understood that the creation of drug resistance follows a sequential
process, i.e. monoresistance > double resistance > triple resistance > quadruple drug resistance. The addition of a new drug to a failing drug regimen is an effective way of amplifying the drug resistance problem. Monoresistance can only be selected in the presence of a drug concentration leading to the selection of pre-existing mutant bacilli, whereas resistance to two drugs cannot be created simultaneously in the presence of effective concentrations of two drugs. This is because the number of bacilli present in the lesions (108) is usually much lower than the theoretically required bacillary load needed to produce double resistance, i.e. in the case of SM and INH, 2.29 x 10-8 and 2.56 x10-8 respectively or 10-16. Results obtained in this study show that the proportions of monoresistance are lower in patients having re-treatment, whereas double resistance remains essentially unchanged. Triple and quadruple resistance are higher by about the same proportion as monoresistance is lower. Amplification caused by re-treatment is the easiest way to interpret these changes, i.e. the selective pressures of treatment create double resistance from mainly monoresistant H/S cases. The absence of a significant change in double resistance proportions can be explained by selective pressure, leading to an increase in triple and quadruple drug resistance modes thus balancing the inflow from the monoresistance mode. Since resistance in re-treatment cases mostly reflects the quality of recent treatment, these results could lead to the development of an indicator, based on the extent of amplification. The difference between previously treated and new case triple and quadruple resistance proportions could constitute such an indicator. More analysis is needed to validate this proposed indicator. Major drug resistance pathways
Lastly, we have seen that monoresistant H/S, double resistant HS, triple resistant
DISCUSSION
HSR and quadruple resistant HSRE cases account for about 85% of all resistant cases. This suggests that, globally as well as in most regions, the main pathway of drug resistance amplification is H, S > HS > HSR > HSRE. Other pathways can and do exist but their contribution to the drug resistance problem is relatively minor. We can therefore state that monoresistance to H or to S is the foundation for the acquisition of additional drug resistance.
Implications
The above analysis has shown that there is circumstantial but compelling evidence
that either monotherapy or “effective” monotherapy, or both, are more widespread than commonly thought. These results corroborate recently emerging evidence that standard re-treatment regimens containing first-line drugs for failures of standard treatment should be abandoned in some settings. In a recent study, Quy et al87 showed the limitations of the 2SRHZ/6HE regimen in preventing the amplification effect even in patients with primary resistance other than MDR-TB, including single drug resistance. From the analysis of the above data, there might also be wider implications for the standard 2SRHZ/6HE regimen at least in those countries with high rates of H/S monoresistance and HS double resistance. One possible way of breaking the amplification juggernaut would be to replace S in standard regimens and/or to add a third drug to the continuation phase.

Source: http://minhkhue.free.fr/doc/tuberculosis/resistance2005/9241562854_chap5.pdf