- Email: [email protected]
What is the limit to case detection under the DOTS strategy for tuberculosis control?
Tuberculosis (2003) 83, 35–43
Tuberculosis www.elsevierhealth.com/journals/tube
What is the limit to case detection under the DOTS strategy for tuberculosis control? Christopher Dye*, Catherine J. Watt, Daniel M. Bleed, Brian G. Williams Communicable Diseases, World Health Organization, 1211 Geneva 27, Switzerland
Summary In year 2000, the WHO DOTS strategy for tuberculosis (TB) control had been adopted by 148 out of 212 countries, but only 27% of all estimated sputum smear-positive patients were notified under DOTS in that year. Here we investigate the way in which gains in case detection under DOTS were made up until 2000 in an attempt to anticipate future progress towards the global target of 70% case detection. The analysis draws on annual reports of DOTS geographical coverage and case notifications, and focuses on the 22 high-burden countries (HBCs) that account for about 80% of new TB cases arising globally each year. Our principal observation is that, as TB programmes in the 22 HBCs have expanded geographically, the fraction of the estimated number of sputum smear-positive cases detected within designated DOTS areas has remained constant at 40–50% although there are significant differences between countries. This fraction is about the same as the percentage of all smear-positive cases notified annually to WHO via public health systems worldwide. The implication is that, unless the DOTS strategy can reach beyond traditional public health reporting systems, or unless these systems can be improved, case detection will not rise much above 40% in the 22 HBCs, or in the world as a whole, even when the geographical coverage of DOTS is nominally 100%. We estimate that, under full DOTS coverage, three-quarters of the undetected smear-positive cases will be living in India, China, Indonesia, Nigeria, Bangladesh and Pakistan. But case detection could also remain low in countries with smaller populations: in year 2000, over half of all smear-positive TB cases were living in 49 countries that detected less than 40% of cases within DOTS areas. Substantial efforts are therefore needed (a) to develop new case finding and management methods to bridge the gap between current and target case detection, and (b) to improve the accuracy of national estimates of TB incidence, above all by reinforcing and expanding routine surveillance. r 2003 Elsevier Science Ltd. All rights reserved.
Introduction The 22 high-burden countries (HBCs) that account for 80% of new tuberculosis (TB) cases annually have made, or are making, firm plans to reach the WHO targets of 70% case detection and 85% cure by *Corresponding author. Tel.: +41 22 791 2904; fax: +41 22 792 4268 E-mail address: [email protected] (C. Dye).
2005.1 Most countries implementing the DOTS strategy have shown that they can achieve high cure rates (global average 80% for the 1999 cohort), but the prospects for detecting 70% of sputum smear-positive cases are less certain. WHO’s monitoring and surveillance project has charted the progress in case detection under DOTS for the 6 years 1995–2000, showing a linear increase in the proportion of smear-positive cases detected that reached 27% in 2000.1,2 If the same rate of progress
1472-9792/03/$ - see front matter r 2003 Elsevier Science Ltd. All rights reserved. doi:10.1016/S1472-9792(02)00056-2
36
is maintained, the target of 70% case detection will be reached in 2013. However, this conditional forecast is no more than statistical extrapolation, and is made without considering any further details of DOTS implementation. The goal of this paper is to develop a more informed prognosis for DOTS expansion by examining a wider range of statistics supplied by national TB control programmes (NTPs). The main line of investigation explores how gains in case detection have been made as DOTS programmes have expanded geographically within countries. The results suggest that the maximum case detection rate under DOTS will be significantly below the target level of 70%, unless DOTS programmes can more effectively penetrate the variety of health systems that now diagnose and treat TB, and reach populations that currently have little or no access to health services.
C. Dye et al. i.e. 6.7 million out of 8.4 million cases in 2000.4 We first establish the quantitative relationship between case detection and DOTS coverage for each of the 22 HBCs in the years 1995–2000. We then use this quantitative relation to forecast how the case detection rate will change with anticipated improvements in DOTS coverage up to 2010. We assume that DOTS coverage will improve as stated in national plans for TB control,1 and that the current relationships between coverage and case detection will be preserved, unless NTPs adopt a more aggressive approach to case finding within designated DOTS areas. We then extend the arguments to 166 countries that adopted DOTS in the period 1997–2000, and to 202 countries that have notified cases to WHO.
Results Methods We use 6 consecutive years (1995–2000) of data reported to WHO by NTPs1 to investigate the relationship between the geographical coverage of DOTS programmes within countries and the estimated case detection rate. ‘‘DOTS coverage’’ is the percentage of a national population that has access to DOTS, i.e. living in the catchment area of administrative districts that provide DOTS services. The ‘‘case detection rate’’ is the number of sputum smear-positive patients notified under DOTS in a given year divided by the estimated number of new smear-positive cases arising in that year. The case detection rate measures case finding under DOTS nationally; the ratio of detection to coverage measures the case detection rate within DOTS areas only. Detection under DOTS implies that patients receive the full package of support and treatment embodied in the DOTS strategy, including diagnosis by sputum smear microscopy and standard short-course chemotherapy preferably administered under direct observation.3 We emphasize that patients who are undetected (i.e. not notified) by public health systems, including DOTS programmes, are not necessarily untreated; the great majority of TB patients probably receive antiTB drugs in some combination. The problem is that the outcomes of treatment outside DOTS programmes are at best uncertain and often poor, with higher proportions of patients defaulting, failing treatment or dying.1 Our focus is on the 22 HBCs that account for an estimated 80% of all new TB cases arising each year,
Figures 1 and 2 show the case detection rate vs DOTS coverage for the 22 HBCs. The first nine countries have case detection rates substantially lower than 70% in DOTS areas (Fig. 1). The estimated ratios of detection to coverage are highly variable, and under 30% in Bangladesh, Indonesia, Nigeria and the Russian Federation. Where there is a discernible relationship between these two variables (i.e. for countries except Nigeria and Pakistan), it is approximately linear. Even though some countries, notably India and the Philippines (Fig. 3), have rapidly increased both coverage and detection over the 6-year period to 2000, the relationship between these two indicators remains linear within the accuracy of the data. 2001 data for India suggest 22% smear-positive case detection for 44% DOTS coverage,1,5 and the inclusion of these additional data increases the average ratio of detection/coverage from 38% to 46%. However, for none of the countries depicted in Fig. 1 do the data persuasively show a tendency for case detection to increase towards the target level of 70% (diagonal line) as DOTS coverage increases. In short, there is little evidence that DOTS programmes are improving case finding locally as they expand geographically. The 13 countries shown in Fig. 2 all appear to have higher rates of case detection within DOTS areas than those shown in Fig. 1. South Africa, Thailand, Viet Nam and Myanmar have case detection rates approaching or exceeding 70% with DOTS available in most of the country. DR Congo appears to have exceeded the 70% target, but this is likely to be a reporting anomaly: all cases notified by DR Congo are classified under DOTS, and some of these
DOTS strategy for tuberculosis control
37
China
30 25 20 15 10 5 0
40 30 20 10
10
20
30
40
0
50
10 0
20 10 20
40
60 50 40 30 20 10 20
Philippines
40
60
80
50 40 30 20 10 0
100
30 20 10 0 100
6 5 4 3 2 1 0 2 4 6 8 DOTS coverage (%)
40
60
80
100
Russian Federation
7
0
20
DOTS coverage (%)
ss+ case detection under DOTS (%)
ss+ case detection under DOTS (%)
40
20 40 60 80 DOTS coverage (%)
60
Pakistan
50
100
70
DOTS coverage (%)
60
80
0 0
DOTS coverage (%)
70
60
Ethiopia
70
50
40
DOTS coverage (%)
0
0
30
0
ss+ case detection under DOTS (%)
ss+ case detection under DOTS (%)
ss+ case detection under DOTS (%)
20
30
40
Bangladesh
30
20
50
10 20 30 40 50 60 70
Nigeria 40
10
60
DOTS coverage (%)
DOTS coverage (%)
0
70
0
0 0
ss+ case detection under DOTS (%)
Indonesia
50 ss+ case detection under DOTS (%)
ss+ case detection under DOTS (%)
ss+ case detection under DOTS (%)
India
35
10
10 8 6 4 2 0 0
5 10 DOTS coverage (%)
15
Figure 1 Low case detection in DOTS programmes: smear-positive case detection rate under DOTS vs DOTS coverage for nine HBCs, 1995–2000. The diagonal line marks 70% case detection.
presumably come from non-DOTS areas. Tanzania, Kenya, Uganda, Mozambique and Zimbabwe have all had full DOTS coverage for more than 5 years (Table 1). In the era of HIV/AIDS, case detection rates are hard to determine precisely in these East African countries. For Brazil and Afghanistan, the variability of observations reflects difficulties in determining case detection rates with very low DOTS coverage. The observation that DOTS programmes find an unchanging fraction of smear-positive cases within designated DOTS areas holds for the 22 HBCs collectively, and for all countries. Best estimates of these fractions from the combined data are 44% and 48%, respectively (slopes of the lines in Fig. 4). The fixed ratio of detection to coverage for all notified TB cases is only 41%. Table 1 shows our assumptions about the growth in DOTS coverage for each of the 22 HBCs, based on national plans for TB control. Using these proposed changes in coverage through time, together with the fixed ratios of case detection to coverage suggested by Figs. 1 and 2, we can calculate the
expected progress in case detection to 2010 (Fig. 5). Under this set of assumptions, there will be an acceleration in the proportion of cases detected between 2001 and 2005, because of efforts to expand DOTS coverage in some large countries, including India and Pakistan. However, beyond 2005, case detection will stabilize at a maximum of 41%. This is about the same as the maximum proportion of estimated smear-positive cases reported to WHO, from both DOTS and nonDOTS areas, which appears to have converged to an asymptote of about 40% (Fig. 5). Because case detection and DOTS coverage are tightly correlated (Fig. 4), there is little statistical error in the estimated ratio of these two variables. There is substantially more uncertainty surrounding the estimated number of new TB cases arising each year, for which 95% CL have been calculated at 710%.4 Applying this error to the case detection rate gives a maximum in the range 37–45%. Figure 6 shows the distribution of cases that were not detected in 2000 among the 22 HBCs, and which
38
C. Dye et al.
40 20 0 20
40
60
80
70
ss+ case detection under DOTS (%)
60
0
60 50 40 30 20 10 0 20
60 40 20 0 20 40 60 80 DOTS coverage (%)
80
30 20 10
100
0
40 20 0
20 0 80
10 8 6 4 2 0
50 40 30 20 10 0 0
20
40
60
Mozambique
80
40 30 20 10 0
100
0
ss+ case detection under DOTS (%)
60
ss+ case detection under DOTS (%)
60 50 40 30 20 10 0
40
60
80
100
60
80
100
50 40 30 20 10 0
0
20
40
60
80
DOTS coverage (%)
DOTS coverage (%)
40
Zimbabwe
60
0
20
DOTS coverage (%)
70
10
20
50
Cambodia
20
15
60
70
30
10
Myanmar
70
20
5
70
DOTS coverage (%)
40
100
DOTS coverage (%)
60
DOTS coverage (%)
50
80
12
100
70
100
60
0
20 40 60 80 DOTS coverage (%)
ss+ case detection under DOTS (%)
40
40
Brazil
Uganda
60
0
20
DOTS coverage (%)
60
0
ss+ case detection under DOTS (%)
ss+ case detection under DOTS (%)
40
14
Thailand
ss+ case detection under DOTS (%)
60
80
100
80
60
40
ss+ case detection under DOTS (%)
80
40
50
Tanzania ss+ case detection under DOTS (%)
ss+ case detection under DOTS (%)
Viet Nam
20
60
DOTS coverage (%)
100
0
70
0
0
100
DOTS coverage (%)
0
Kenya
DR Congo ss+ case detection under DOTS (%)
ss+ case detection under DOTS (%)
South Africa 80
100
0
20 40 60 80 DOTS coverage (%)
100
Afghanistan ss+ case detection under DOTS (%)
15
10
5
0 0
5
10
15
20
DOTS coverage (%)
Figure 2 As Fig. 1, but for 13 HBCs with relatively high rates of case detection in DOTS programmes.
will remain undetected under full DOTS coverage, assuming no improvements in case finding. We estimate that over two million smear-positive cases were not detected in the 22 HBCs in 2000, three-
quarters (76%) of them in India (33%), China (18%), Indonesia (10%), Nigeria (6%), Bangladesh (5%) and Pakistan (5%). Assuming no improvements in the ratio detection/coverage, 76% of undetected cases
DOTS strategy for tuberculosis control
39
12 Smear-positive case detection under DOTS (%)
35
DOTS coverage (%)
30 25 20 15 10 5
(a)
0 1995
1996
1997
1998
1999
2000
(b)
100 Smear-positive case detection under DOTS (%)
DOTS coverage (%)
8 6 4 2 0 1995
1996
1997
1998
1999
2000
1996
1997
1998
1999
2000
50
90 80 70 60 50 40 30 20 10 0 1995
10
1996
1997
1998
1999
30
20
10
0 1995
2000
(c)
40
(d)
Figure 3 Rapid growth in DOTS coverage (a, c) and smear-positive case detection (b, d) in India (upper) and the Philippines (lower), 1995–2000.
will be living in the same six countries under full DOTS coverage. Although the majority of the world’s undetected cases live in just a few countries, low case detection is a problem for many of the countries that are now implementing DOTS. Our data suggest that 49 of 166 countries that provided DOTS in the period 1997– 2000 had case detection rates under 40% within DOTS areas. These countries are inhabited by 53% of all estimated smear-positive TB cases. Across all 166 countries, there was no systematic relationship between the case detection rate in DOTS areas and the estimated annual incidence of TB, expressed either as the absolute number of smear-positive cases or as the incidence rate per capita. Thus, low case detection is widespread; it is not a problem that is restricted to higher- or lower-burden countries.
Discussion Our principal observation is that the proportion of TB cases detected in DOTS areas has changed little
as DOTS has expanded within the 22 HBCs. We estimate this fixed proportion to be 40–50%, which is approximately the same as the steady fraction of smear-positive cases notified to WHO from all sources (E40%), DOTS and non-DOTS, over the period 1995–2000. The implication is that, as DOTS reaches a nominal 100% coverage in the 22 HBCs, the case detection rate under DOTS will saturate at 40–50%, a level much lower than the 70% target. The principal difficulty is that DOTS programmes have, up to now, mostly recruited patients who would have been detected and treated anyway in the public health system.1 The quality and extent of these health systems varies greatly between countries and so, therefore, does the meaning of DOTS ‘‘coverage’’: coverage appears to be superficial in Indonesia, deeper in Thailand. DOTS has failed in some countries to reach deeply into the private sector, and in others to provide access to patients living in areas with inadequate health services. Of the undetected smear-positive cases, threequarters were living in just six countries in 2000; unless case detection under DOTS improves,
Smear-positive case detection under DOTS (%)
50 40 30 20 10 0 0
10
20 30 40 50 DOTS coverage (% )
60
70
Figure 4 Aggregate case detection rates within DOTS programmes (filled circles, smear-positive cases in 22 HBCs; open circles, smear-positive cases in 202 countries; triangles, all TB cases in 202 countries).
Smear-positive case detection (%)
100 90 80 70% target case detection
70 60 50 40 30 20
all ss+ notifications ss+ notifications under
10 0 1990
1995
2000
2005
2010
2015
Figure 5 Observed and projected smear-positive case detection rate, compared with the 70% global target. The lower set of points is derived from case notifications submitted by DOTS programmes in the 22 HBCs to WHO divided by the estimated incidence rate for these countries. Linear projection of the trend observed from 1995 to 2000 indicates thatFif the trend is maintainedF70% case detection will be reached in 2015. To reach 70% case detection by 2005, the 22 HBCs must significantly accelerate case finding, moving along the steeper line. The data in Figures 2–4 imply that the 22 HBCs will actually follow the heavier curved line, which reaches a maximum at 41% (range 36–46%) case detection. The upper set of points represent all smearpositive cases notified to WHO, by DOTS and non-DOTS programmes.
1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
0 2 2 2 9 14 30 44 76 100 100 100 100 100 100 100 100
0 49 60 64 64 64 68 72 78 83 89 90 90 90 90 90 90
0 6 14 28 80 90 98 100 100 100 100 100 100 100 100 100 100
0 47 30 40 45 45 47 58 68 79 89 100 100 100 100 100 100
0 41 65 80 90 90 92 100 100 100 100 100 100 100 100 100 100
0 39 39 48 64 63 85 90 100 100 100 100 100 100 100 100 100
0 4 2 15 17 43 90 100 100 100 100 100 100 100 100 100 100
0 2 8 8 8 8 9 27 45 64 82 100 100 100 100 100 100
0 0 6 13 22 66 77 84 100 100 100 100 100 100 100 100 100
0 0 2 2 5 5 12 30 47 65 82 100 100 100 100 100 100
0 47 51 60 60 62 70 76 82 88 94 100 100 100 100 100 100
0 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100
0 50 95 93 96 99 100 100 100 100 100 100 100 100 100 100 100
0 98 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100
0 0 0 0 3 7 7 26 44 63 81 100 100 100 100 100 100
0 0 1 4 32 59 70 85 100 100 100 100 100 100 100 100 100
0 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100
0 50 59 60 60 64 77 90 100 100 100 100 100 100 100 100 100
0 97 100 84 95 98 100 100 100 100 100 100 100 100 100 100 100
0 60 80 88 100 100 99 100 100 100 100 100 100 100 100 100 100
0 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100
0 0 6 12 11 14 15 32 49 66 83 100 100 100 100 100 100
C. Dye et al.
India China Indonesia Nigeria Bangladesh Ethiopia Philippines Pakistan South Russian DR Kenya Viet UR Brazil Thailand Uganda Myanmar Mozambique Cambodia Zimbabwe Afghanistan Africa Federation Congo Nam Tanzania
Table 1 Observed growth in DOTS coverage (%) 1995–2000, with anticipated improvements in coverage to 2010
40
three-quarters of the undetected cases will be living in the same six countries under full DOTS coverage. Although these six countries account for the majority of undetected cases worldwide, low case detection is a significant problem for many individual countries that are now implementing
DOTS strategy for tuberculosis control
41
Undetected cases ('000s) 0
100
200
300
400
500
600
700
800
India China Indonesia Nigeria Bangladesh Ethiopia Philippines Pakistan South Africa Russian Federation DR Congo Kenya Viet Nam UR Tanzania Brazil Thailand Uganda Myanmar Mozambique+ Cambodia Zimbabwe Afghanistan
Figure 6 Estimated distribution of smear-positive cases that were not detected by DOTS programmes in 2000 (filled bars), and which will not be detected under full DOTS coverage (open bars), assuming no change in the intensity of case finding within designated DOTS areas. For both series, three-quarters of undetected cases are in the top six countries.
DOTS: in our estimation, over half of all smearpositive TB cases were living in 49 countries that detected less than 40% of cases within DOTS areas during 2000. To verify and explain these observations, we need to examine the reasons why the estimated case detection rates under DOTS might be low. There are broadly five possibilities: 1. The allegedly missing TB cases do not exist. The conclusion that case detection rates are low in some countries rests on the supposition that we have correctly estimated the annual incidence of smear-positive cases, the denominator of the case detection rate. Confidence limits (95%) on the incidence estimates for each of the 22 HBCs are typically 730%, so upper and lower estimates differ by a factor of two. But the 95% CL on global estimates of TB incidence are only 710%. This relatively small error follows from the assumption that the larger errors for
individual countries are independent of each other, and therefore tend to cancel when the estimates for different countries are combined. The upper confidence limit on global case detection in Fig. 5 is 45%, which still leaves a 25% gap to the 70% target, implying that a significant fraction of smear-positive cases will remain undetected by DOTS programmes. The difference in errors surrounding country and global estimates means that we can be more certain that a fraction of TB cases is undetected, but less certain about which countries they inhabit. The data from which incidence is estimated are of variable quality, but good for some of the HBCs where there appears to be a large deficit in reported cases. For example, China carried out a nationwide survey of disease prevalence in 2000.6 The measured smearpositive prevalence was 125/100,000 and yet only 22 new smear-positive cases were reported per 100,000 population under DOTS in that year.
42
2.
3.
4.
5.
C. Dye et al.
It is highly unlikely that prevalence is over five times incidence in China: we have estimated smear-positive incidence to be 46/100,000 in 2000, suggesting that 24/100,000 were not detected by the public TB dispensaries.4 Cases do not present to any health facility, public or private. We guess that about 20% of all TB patients receive no anti-TB drugs at all, but there are presently too few data to estimate this figure accurately.4 Certain sub-populations in some countries clearly have limited or no access to health services, including facilities that can diagnose and treat TB. The whereabouts and size of these sub-populations have not been systematically documented, but they must account for at least a small number of missing patients. Cases do not present to the public health system, and are not reported by the private sector. There is ample evidence, both circumstantial and specific, that many TB patients are never seen by the public health systems that report to WHO; they may be treated, with unknown drug regimens producing unknown outcomes, but never notified. The circumstantial evidence includes the finding that 75% of households in India prefer to use the private sector for treatment of major illnesses.7 More specifically, a selection of studies in India has shown that over half of all TB patients first visit, and are first treated in, the private sector. Similar results have been obtained in Pakistan, Philippines, Viet Nam and Uganda. The year 2000 Chinese national prevalence survey found that only 12% of TB patients were diagnosed at a TB dispensary;6 many cases diagnosed in hospitals are not referred to the public dispensaries.8 A similar prevalence survey in the Philippines found that only 13% of TB symptomatics used government health centres, the point of entry to the NTP and the route to notification.9 Cases present to the public health system, but not to DOTS programmes. Just over 1 million smear-positive cases were reported by DOTS programmes in 2000, and approximately 500,000 from non-DOTS programmes. Therefore, at least half a million smear-positive cases remain undetected by DOTS programmes each year. On past performance, these half million cases will be the ones that raise case detection under DOTS from 27% to 40–50% within the next few years. Symptomatics present to the public health system, including DOTS programmes, but are wrongly diagnosed. Incidence estimates may be too high where X-ray diagnoses are not bacteriologically confirmed; and smear-positive cases may be too few where sputum smear microscopy
is not done, or done poorly. Although both kinds of error are manifest as low case detection, they point to defects in the diagnostic process, rather than to a failure to make contact with health services. In 1999, the Russian Federation reported a total of 124,044 cases of TB of which only 44,838 were bacteriologically confirmed, and many fewer were examined by sputum microscopy.10 Several central and western European countries use sputum culture rather than smear microscopy for bacteriological confirmation of TB, and hence report relatively low smear-positive case detection rates. But the deficit of smear-positive cases in low-incidence industrialized countries is a relatively minor problem, both for these countries, and for global TB control. We draw two conclusions, with respect to measurement of the numerator and denominator of the case detection rate. On the numerator, it is clear that a significant gap does exist between cases detected under DOTS and the 70% global target. Even if there is some uncertainty about the size of the gap, it is likely to remain significant under full DOTS coverage, unless a larger fraction of cases can be embraced by DOTS, or by various extensions of the DOTS strategy. The methods of extending DOTS must be as diverse as the reasons why case detection is now low: these methods will include building links between public and private practitioners, targeting populations at high risk, ensuring best use of current diagnostic methods as well as introducing new ones, and providing health facilities where none have previously existed. On the denominator, it is true that better estimates are needed of the burden of TB in some countries. The valuable disease prevalence surveys now underway in Vietnam, Cambodia and Malaysia may or may not reveal discrepancies between current estimates and the true burden of disease. But population-based surveys are not the long-term answer to measuring TB incidence; the ultimate solution for HBCs is the same as the approach currently taken in low-incidence countries, namely comprehensive routine surveillance. Surveys of disease burden are never done in industrialized countries, partly because incidence is low, but more importantly because we have confidence that most TB cases are reported. The drive to improve surveillance in HBCs should begin with careful assessments of which sub-populations do, and do not, have access to health facilities that can correctly diagnose and treat TB. This approach will give progressively better estimates of incidence at the same time as strengthening routine
DOTS strategy for tuberculosis control
surveillance. Monitoring the number and performance of health facilitiesFwhether all facilities report, how many cases per head of population, their catchment areas, the variation from year to year and from one facility to anotherFshould also provide a pragmatic approach to setting casefinding targets for national control programmes.
References 1. World Health Organization. Global tuberculosis control: surveillance, planning, financing. WHO Report 2002. WHO/ TB/2002.295, Geneva, 2002. 2. Dye C, Watt CJ, Bleed D. Low access to a highly effective therapy: a challenge for international tuberculosis control. Bull World Health Org 2002;80:437–44. 3. An Expanded DOTS framework for effective tuberculosis control. Unpublished Document WHO/CDS/TB/2002.297. Geneva: World Health Organization, 2002.
43
4. Corbett EL, Watt C, Walker N, Maher D, Raviglione MC, Williams BG, Dye C. The growing burden of tuberculosis: global trends and interactions with the HIV epidemic. Arch Intern Med 2003; in press. 5. Tuberculosis ControlFIndia.Directorate General of Health Services, Ministry of Health and Family Welfare. http://www.tbcindia.org/rntcp.asp. Accessed 28 June 2002. 6. Report on Nationwide Random Survey for the Epidemiology of Tuberculosis in 2000. Beijing: Ministry of Public Health, 2000. 7. Uplekar M, Pathania VK, Raviglione MC. Private practitioners and public health: weak links in tuberculosis control. Lancet 2001;358:912–6. 8. Chen X, Zhao F, Duanmu H, Wan L, Wang L, Du X, Chin DP. The DOTS strategy in China: results and lessons after 10 years. Bull World Health Org 2002;80:430–6. 9. Tropical Disease Foundation Inc. Final Report. 1997 National Tuberculosis Prevalence Survey. Makati Medical Center, Philippines, 1997. 10. Shilova MV, Dye C. The resurgence of tuberculosis in Russia. Philos Trans R Soc of London B 2001;356:1069–75.
Tuberculosis www.elsevierhealth.com/journals/tube
What is the limit to case detection under the DOTS strategy for tuberculosis control? Christopher Dye*, Catherine J. Watt, Daniel M. Bleed, Brian G. Williams Communicable Diseases, World Health Organization, 1211 Geneva 27, Switzerland
Summary In year 2000, the WHO DOTS strategy for tuberculosis (TB) control had been adopted by 148 out of 212 countries, but only 27% of all estimated sputum smear-positive patients were notified under DOTS in that year. Here we investigate the way in which gains in case detection under DOTS were made up until 2000 in an attempt to anticipate future progress towards the global target of 70% case detection. The analysis draws on annual reports of DOTS geographical coverage and case notifications, and focuses on the 22 high-burden countries (HBCs) that account for about 80% of new TB cases arising globally each year. Our principal observation is that, as TB programmes in the 22 HBCs have expanded geographically, the fraction of the estimated number of sputum smear-positive cases detected within designated DOTS areas has remained constant at 40–50% although there are significant differences between countries. This fraction is about the same as the percentage of all smear-positive cases notified annually to WHO via public health systems worldwide. The implication is that, unless the DOTS strategy can reach beyond traditional public health reporting systems, or unless these systems can be improved, case detection will not rise much above 40% in the 22 HBCs, or in the world as a whole, even when the geographical coverage of DOTS is nominally 100%. We estimate that, under full DOTS coverage, three-quarters of the undetected smear-positive cases will be living in India, China, Indonesia, Nigeria, Bangladesh and Pakistan. But case detection could also remain low in countries with smaller populations: in year 2000, over half of all smear-positive TB cases were living in 49 countries that detected less than 40% of cases within DOTS areas. Substantial efforts are therefore needed (a) to develop new case finding and management methods to bridge the gap between current and target case detection, and (b) to improve the accuracy of national estimates of TB incidence, above all by reinforcing and expanding routine surveillance. r 2003 Elsevier Science Ltd. All rights reserved.
Introduction The 22 high-burden countries (HBCs) that account for 80% of new tuberculosis (TB) cases annually have made, or are making, firm plans to reach the WHO targets of 70% case detection and 85% cure by *Corresponding author. Tel.: +41 22 791 2904; fax: +41 22 792 4268 E-mail address: [email protected] (C. Dye).
2005.1 Most countries implementing the DOTS strategy have shown that they can achieve high cure rates (global average 80% for the 1999 cohort), but the prospects for detecting 70% of sputum smear-positive cases are less certain. WHO’s monitoring and surveillance project has charted the progress in case detection under DOTS for the 6 years 1995–2000, showing a linear increase in the proportion of smear-positive cases detected that reached 27% in 2000.1,2 If the same rate of progress
1472-9792/03/$ - see front matter r 2003 Elsevier Science Ltd. All rights reserved. doi:10.1016/S1472-9792(02)00056-2
36
is maintained, the target of 70% case detection will be reached in 2013. However, this conditional forecast is no more than statistical extrapolation, and is made without considering any further details of DOTS implementation. The goal of this paper is to develop a more informed prognosis for DOTS expansion by examining a wider range of statistics supplied by national TB control programmes (NTPs). The main line of investigation explores how gains in case detection have been made as DOTS programmes have expanded geographically within countries. The results suggest that the maximum case detection rate under DOTS will be significantly below the target level of 70%, unless DOTS programmes can more effectively penetrate the variety of health systems that now diagnose and treat TB, and reach populations that currently have little or no access to health services.
C. Dye et al. i.e. 6.7 million out of 8.4 million cases in 2000.4 We first establish the quantitative relationship between case detection and DOTS coverage for each of the 22 HBCs in the years 1995–2000. We then use this quantitative relation to forecast how the case detection rate will change with anticipated improvements in DOTS coverage up to 2010. We assume that DOTS coverage will improve as stated in national plans for TB control,1 and that the current relationships between coverage and case detection will be preserved, unless NTPs adopt a more aggressive approach to case finding within designated DOTS areas. We then extend the arguments to 166 countries that adopted DOTS in the period 1997–2000, and to 202 countries that have notified cases to WHO.
Results Methods We use 6 consecutive years (1995–2000) of data reported to WHO by NTPs1 to investigate the relationship between the geographical coverage of DOTS programmes within countries and the estimated case detection rate. ‘‘DOTS coverage’’ is the percentage of a national population that has access to DOTS, i.e. living in the catchment area of administrative districts that provide DOTS services. The ‘‘case detection rate’’ is the number of sputum smear-positive patients notified under DOTS in a given year divided by the estimated number of new smear-positive cases arising in that year. The case detection rate measures case finding under DOTS nationally; the ratio of detection to coverage measures the case detection rate within DOTS areas only. Detection under DOTS implies that patients receive the full package of support and treatment embodied in the DOTS strategy, including diagnosis by sputum smear microscopy and standard short-course chemotherapy preferably administered under direct observation.3 We emphasize that patients who are undetected (i.e. not notified) by public health systems, including DOTS programmes, are not necessarily untreated; the great majority of TB patients probably receive antiTB drugs in some combination. The problem is that the outcomes of treatment outside DOTS programmes are at best uncertain and often poor, with higher proportions of patients defaulting, failing treatment or dying.1 Our focus is on the 22 HBCs that account for an estimated 80% of all new TB cases arising each year,
Figures 1 and 2 show the case detection rate vs DOTS coverage for the 22 HBCs. The first nine countries have case detection rates substantially lower than 70% in DOTS areas (Fig. 1). The estimated ratios of detection to coverage are highly variable, and under 30% in Bangladesh, Indonesia, Nigeria and the Russian Federation. Where there is a discernible relationship between these two variables (i.e. for countries except Nigeria and Pakistan), it is approximately linear. Even though some countries, notably India and the Philippines (Fig. 3), have rapidly increased both coverage and detection over the 6-year period to 2000, the relationship between these two indicators remains linear within the accuracy of the data. 2001 data for India suggest 22% smear-positive case detection for 44% DOTS coverage,1,5 and the inclusion of these additional data increases the average ratio of detection/coverage from 38% to 46%. However, for none of the countries depicted in Fig. 1 do the data persuasively show a tendency for case detection to increase towards the target level of 70% (diagonal line) as DOTS coverage increases. In short, there is little evidence that DOTS programmes are improving case finding locally as they expand geographically. The 13 countries shown in Fig. 2 all appear to have higher rates of case detection within DOTS areas than those shown in Fig. 1. South Africa, Thailand, Viet Nam and Myanmar have case detection rates approaching or exceeding 70% with DOTS available in most of the country. DR Congo appears to have exceeded the 70% target, but this is likely to be a reporting anomaly: all cases notified by DR Congo are classified under DOTS, and some of these
DOTS strategy for tuberculosis control
37
China
30 25 20 15 10 5 0
40 30 20 10
10
20
30
40
0
50
10 0
20 10 20
40
60 50 40 30 20 10 20
Philippines
40
60
80
50 40 30 20 10 0
100
30 20 10 0 100
6 5 4 3 2 1 0 2 4 6 8 DOTS coverage (%)
40
60
80
100
Russian Federation
7
0
20
DOTS coverage (%)
ss+ case detection under DOTS (%)
ss+ case detection under DOTS (%)
40
20 40 60 80 DOTS coverage (%)
60
Pakistan
50
100
70
DOTS coverage (%)
60
80
0 0
DOTS coverage (%)
70
60
Ethiopia
70
50
40
DOTS coverage (%)
0
0
30
0
ss+ case detection under DOTS (%)
ss+ case detection under DOTS (%)
ss+ case detection under DOTS (%)
20
30
40
Bangladesh
30
20
50
10 20 30 40 50 60 70
Nigeria 40
10
60
DOTS coverage (%)
DOTS coverage (%)
0
70
0
0 0
ss+ case detection under DOTS (%)
Indonesia
50 ss+ case detection under DOTS (%)
ss+ case detection under DOTS (%)
ss+ case detection under DOTS (%)
India
35
10
10 8 6 4 2 0 0
5 10 DOTS coverage (%)
15
Figure 1 Low case detection in DOTS programmes: smear-positive case detection rate under DOTS vs DOTS coverage for nine HBCs, 1995–2000. The diagonal line marks 70% case detection.
presumably come from non-DOTS areas. Tanzania, Kenya, Uganda, Mozambique and Zimbabwe have all had full DOTS coverage for more than 5 years (Table 1). In the era of HIV/AIDS, case detection rates are hard to determine precisely in these East African countries. For Brazil and Afghanistan, the variability of observations reflects difficulties in determining case detection rates with very low DOTS coverage. The observation that DOTS programmes find an unchanging fraction of smear-positive cases within designated DOTS areas holds for the 22 HBCs collectively, and for all countries. Best estimates of these fractions from the combined data are 44% and 48%, respectively (slopes of the lines in Fig. 4). The fixed ratio of detection to coverage for all notified TB cases is only 41%. Table 1 shows our assumptions about the growth in DOTS coverage for each of the 22 HBCs, based on national plans for TB control. Using these proposed changes in coverage through time, together with the fixed ratios of case detection to coverage suggested by Figs. 1 and 2, we can calculate the
expected progress in case detection to 2010 (Fig. 5). Under this set of assumptions, there will be an acceleration in the proportion of cases detected between 2001 and 2005, because of efforts to expand DOTS coverage in some large countries, including India and Pakistan. However, beyond 2005, case detection will stabilize at a maximum of 41%. This is about the same as the maximum proportion of estimated smear-positive cases reported to WHO, from both DOTS and nonDOTS areas, which appears to have converged to an asymptote of about 40% (Fig. 5). Because case detection and DOTS coverage are tightly correlated (Fig. 4), there is little statistical error in the estimated ratio of these two variables. There is substantially more uncertainty surrounding the estimated number of new TB cases arising each year, for which 95% CL have been calculated at 710%.4 Applying this error to the case detection rate gives a maximum in the range 37–45%. Figure 6 shows the distribution of cases that were not detected in 2000 among the 22 HBCs, and which
38
C. Dye et al.
40 20 0 20
40
60
80
70
ss+ case detection under DOTS (%)
60
0
60 50 40 30 20 10 0 20
60 40 20 0 20 40 60 80 DOTS coverage (%)
80
30 20 10
100
0
40 20 0
20 0 80
10 8 6 4 2 0
50 40 30 20 10 0 0
20
40
60
Mozambique
80
40 30 20 10 0
100
0
ss+ case detection under DOTS (%)
60
ss+ case detection under DOTS (%)
60 50 40 30 20 10 0
40
60
80
100
60
80
100
50 40 30 20 10 0
0
20
40
60
80
DOTS coverage (%)
DOTS coverage (%)
40
Zimbabwe
60
0
20
DOTS coverage (%)
70
10
20
50
Cambodia
20
15
60
70
30
10
Myanmar
70
20
5
70
DOTS coverage (%)
40
100
DOTS coverage (%)
60
DOTS coverage (%)
50
80
12
100
70
100
60
0
20 40 60 80 DOTS coverage (%)
ss+ case detection under DOTS (%)
40
40
Brazil
Uganda
60
0
20
DOTS coverage (%)
60
0
ss+ case detection under DOTS (%)
ss+ case detection under DOTS (%)
40
14
Thailand
ss+ case detection under DOTS (%)
60
80
100
80
60
40
ss+ case detection under DOTS (%)
80
40
50
Tanzania ss+ case detection under DOTS (%)
ss+ case detection under DOTS (%)
Viet Nam
20
60
DOTS coverage (%)
100
0
70
0
0
100
DOTS coverage (%)
0
Kenya
DR Congo ss+ case detection under DOTS (%)
ss+ case detection under DOTS (%)
South Africa 80
100
0
20 40 60 80 DOTS coverage (%)
100
Afghanistan ss+ case detection under DOTS (%)
15
10
5
0 0
5
10
15
20
DOTS coverage (%)
Figure 2 As Fig. 1, but for 13 HBCs with relatively high rates of case detection in DOTS programmes.
will remain undetected under full DOTS coverage, assuming no improvements in case finding. We estimate that over two million smear-positive cases were not detected in the 22 HBCs in 2000, three-
quarters (76%) of them in India (33%), China (18%), Indonesia (10%), Nigeria (6%), Bangladesh (5%) and Pakistan (5%). Assuming no improvements in the ratio detection/coverage, 76% of undetected cases
DOTS strategy for tuberculosis control
39
12 Smear-positive case detection under DOTS (%)
35
DOTS coverage (%)
30 25 20 15 10 5
(a)
0 1995
1996
1997
1998
1999
2000
(b)
100 Smear-positive case detection under DOTS (%)
DOTS coverage (%)
8 6 4 2 0 1995
1996
1997
1998
1999
2000
1996
1997
1998
1999
2000
50
90 80 70 60 50 40 30 20 10 0 1995
10
1996
1997
1998
1999
30
20
10
0 1995
2000
(c)
40
(d)
Figure 3 Rapid growth in DOTS coverage (a, c) and smear-positive case detection (b, d) in India (upper) and the Philippines (lower), 1995–2000.
will be living in the same six countries under full DOTS coverage. Although the majority of the world’s undetected cases live in just a few countries, low case detection is a problem for many of the countries that are now implementing DOTS. Our data suggest that 49 of 166 countries that provided DOTS in the period 1997– 2000 had case detection rates under 40% within DOTS areas. These countries are inhabited by 53% of all estimated smear-positive TB cases. Across all 166 countries, there was no systematic relationship between the case detection rate in DOTS areas and the estimated annual incidence of TB, expressed either as the absolute number of smear-positive cases or as the incidence rate per capita. Thus, low case detection is widespread; it is not a problem that is restricted to higher- or lower-burden countries.
Discussion Our principal observation is that the proportion of TB cases detected in DOTS areas has changed little
as DOTS has expanded within the 22 HBCs. We estimate this fixed proportion to be 40–50%, which is approximately the same as the steady fraction of smear-positive cases notified to WHO from all sources (E40%), DOTS and non-DOTS, over the period 1995–2000. The implication is that, as DOTS reaches a nominal 100% coverage in the 22 HBCs, the case detection rate under DOTS will saturate at 40–50%, a level much lower than the 70% target. The principal difficulty is that DOTS programmes have, up to now, mostly recruited patients who would have been detected and treated anyway in the public health system.1 The quality and extent of these health systems varies greatly between countries and so, therefore, does the meaning of DOTS ‘‘coverage’’: coverage appears to be superficial in Indonesia, deeper in Thailand. DOTS has failed in some countries to reach deeply into the private sector, and in others to provide access to patients living in areas with inadequate health services. Of the undetected smear-positive cases, threequarters were living in just six countries in 2000; unless case detection under DOTS improves,
Smear-positive case detection under DOTS (%)
50 40 30 20 10 0 0
10
20 30 40 50 DOTS coverage (% )
60
70
Figure 4 Aggregate case detection rates within DOTS programmes (filled circles, smear-positive cases in 22 HBCs; open circles, smear-positive cases in 202 countries; triangles, all TB cases in 202 countries).
Smear-positive case detection (%)
100 90 80 70% target case detection
70 60 50 40 30 20
all ss+ notifications ss+ notifications under
10 0 1990
1995
2000
2005
2010
2015
Figure 5 Observed and projected smear-positive case detection rate, compared with the 70% global target. The lower set of points is derived from case notifications submitted by DOTS programmes in the 22 HBCs to WHO divided by the estimated incidence rate for these countries. Linear projection of the trend observed from 1995 to 2000 indicates thatFif the trend is maintainedF70% case detection will be reached in 2015. To reach 70% case detection by 2005, the 22 HBCs must significantly accelerate case finding, moving along the steeper line. The data in Figures 2–4 imply that the 22 HBCs will actually follow the heavier curved line, which reaches a maximum at 41% (range 36–46%) case detection. The upper set of points represent all smearpositive cases notified to WHO, by DOTS and non-DOTS programmes.
1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
0 2 2 2 9 14 30 44 76 100 100 100 100 100 100 100 100
0 49 60 64 64 64 68 72 78 83 89 90 90 90 90 90 90
0 6 14 28 80 90 98 100 100 100 100 100 100 100 100 100 100
0 47 30 40 45 45 47 58 68 79 89 100 100 100 100 100 100
0 41 65 80 90 90 92 100 100 100 100 100 100 100 100 100 100
0 39 39 48 64 63 85 90 100 100 100 100 100 100 100 100 100
0 4 2 15 17 43 90 100 100 100 100 100 100 100 100 100 100
0 2 8 8 8 8 9 27 45 64 82 100 100 100 100 100 100
0 0 6 13 22 66 77 84 100 100 100 100 100 100 100 100 100
0 0 2 2 5 5 12 30 47 65 82 100 100 100 100 100 100
0 47 51 60 60 62 70 76 82 88 94 100 100 100 100 100 100
0 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100
0 50 95 93 96 99 100 100 100 100 100 100 100 100 100 100 100
0 98 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100
0 0 0 0 3 7 7 26 44 63 81 100 100 100 100 100 100
0 0 1 4 32 59 70 85 100 100 100 100 100 100 100 100 100
0 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100
0 50 59 60 60 64 77 90 100 100 100 100 100 100 100 100 100
0 97 100 84 95 98 100 100 100 100 100 100 100 100 100 100 100
0 60 80 88 100 100 99 100 100 100 100 100 100 100 100 100 100
0 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100
0 0 6 12 11 14 15 32 49 66 83 100 100 100 100 100 100
C. Dye et al.
India China Indonesia Nigeria Bangladesh Ethiopia Philippines Pakistan South Russian DR Kenya Viet UR Brazil Thailand Uganda Myanmar Mozambique Cambodia Zimbabwe Afghanistan Africa Federation Congo Nam Tanzania
Table 1 Observed growth in DOTS coverage (%) 1995–2000, with anticipated improvements in coverage to 2010
40
three-quarters of the undetected cases will be living in the same six countries under full DOTS coverage. Although these six countries account for the majority of undetected cases worldwide, low case detection is a significant problem for many individual countries that are now implementing
DOTS strategy for tuberculosis control
41
Undetected cases ('000s) 0
100
200
300
400
500
600
700
800
India China Indonesia Nigeria Bangladesh Ethiopia Philippines Pakistan South Africa Russian Federation DR Congo Kenya Viet Nam UR Tanzania Brazil Thailand Uganda Myanmar Mozambique+ Cambodia Zimbabwe Afghanistan
Figure 6 Estimated distribution of smear-positive cases that were not detected by DOTS programmes in 2000 (filled bars), and which will not be detected under full DOTS coverage (open bars), assuming no change in the intensity of case finding within designated DOTS areas. For both series, three-quarters of undetected cases are in the top six countries.
DOTS: in our estimation, over half of all smearpositive TB cases were living in 49 countries that detected less than 40% of cases within DOTS areas during 2000. To verify and explain these observations, we need to examine the reasons why the estimated case detection rates under DOTS might be low. There are broadly five possibilities: 1. The allegedly missing TB cases do not exist. The conclusion that case detection rates are low in some countries rests on the supposition that we have correctly estimated the annual incidence of smear-positive cases, the denominator of the case detection rate. Confidence limits (95%) on the incidence estimates for each of the 22 HBCs are typically 730%, so upper and lower estimates differ by a factor of two. But the 95% CL on global estimates of TB incidence are only 710%. This relatively small error follows from the assumption that the larger errors for
individual countries are independent of each other, and therefore tend to cancel when the estimates for different countries are combined. The upper confidence limit on global case detection in Fig. 5 is 45%, which still leaves a 25% gap to the 70% target, implying that a significant fraction of smear-positive cases will remain undetected by DOTS programmes. The difference in errors surrounding country and global estimates means that we can be more certain that a fraction of TB cases is undetected, but less certain about which countries they inhabit. The data from which incidence is estimated are of variable quality, but good for some of the HBCs where there appears to be a large deficit in reported cases. For example, China carried out a nationwide survey of disease prevalence in 2000.6 The measured smearpositive prevalence was 125/100,000 and yet only 22 new smear-positive cases were reported per 100,000 population under DOTS in that year.
42
2.
3.
4.
5.
C. Dye et al.
It is highly unlikely that prevalence is over five times incidence in China: we have estimated smear-positive incidence to be 46/100,000 in 2000, suggesting that 24/100,000 were not detected by the public TB dispensaries.4 Cases do not present to any health facility, public or private. We guess that about 20% of all TB patients receive no anti-TB drugs at all, but there are presently too few data to estimate this figure accurately.4 Certain sub-populations in some countries clearly have limited or no access to health services, including facilities that can diagnose and treat TB. The whereabouts and size of these sub-populations have not been systematically documented, but they must account for at least a small number of missing patients. Cases do not present to the public health system, and are not reported by the private sector. There is ample evidence, both circumstantial and specific, that many TB patients are never seen by the public health systems that report to WHO; they may be treated, with unknown drug regimens producing unknown outcomes, but never notified. The circumstantial evidence includes the finding that 75% of households in India prefer to use the private sector for treatment of major illnesses.7 More specifically, a selection of studies in India has shown that over half of all TB patients first visit, and are first treated in, the private sector. Similar results have been obtained in Pakistan, Philippines, Viet Nam and Uganda. The year 2000 Chinese national prevalence survey found that only 12% of TB patients were diagnosed at a TB dispensary;6 many cases diagnosed in hospitals are not referred to the public dispensaries.8 A similar prevalence survey in the Philippines found that only 13% of TB symptomatics used government health centres, the point of entry to the NTP and the route to notification.9 Cases present to the public health system, but not to DOTS programmes. Just over 1 million smear-positive cases were reported by DOTS programmes in 2000, and approximately 500,000 from non-DOTS programmes. Therefore, at least half a million smear-positive cases remain undetected by DOTS programmes each year. On past performance, these half million cases will be the ones that raise case detection under DOTS from 27% to 40–50% within the next few years. Symptomatics present to the public health system, including DOTS programmes, but are wrongly diagnosed. Incidence estimates may be too high where X-ray diagnoses are not bacteriologically confirmed; and smear-positive cases may be too few where sputum smear microscopy
is not done, or done poorly. Although both kinds of error are manifest as low case detection, they point to defects in the diagnostic process, rather than to a failure to make contact with health services. In 1999, the Russian Federation reported a total of 124,044 cases of TB of which only 44,838 were bacteriologically confirmed, and many fewer were examined by sputum microscopy.10 Several central and western European countries use sputum culture rather than smear microscopy for bacteriological confirmation of TB, and hence report relatively low smear-positive case detection rates. But the deficit of smear-positive cases in low-incidence industrialized countries is a relatively minor problem, both for these countries, and for global TB control. We draw two conclusions, with respect to measurement of the numerator and denominator of the case detection rate. On the numerator, it is clear that a significant gap does exist between cases detected under DOTS and the 70% global target. Even if there is some uncertainty about the size of the gap, it is likely to remain significant under full DOTS coverage, unless a larger fraction of cases can be embraced by DOTS, or by various extensions of the DOTS strategy. The methods of extending DOTS must be as diverse as the reasons why case detection is now low: these methods will include building links between public and private practitioners, targeting populations at high risk, ensuring best use of current diagnostic methods as well as introducing new ones, and providing health facilities where none have previously existed. On the denominator, it is true that better estimates are needed of the burden of TB in some countries. The valuable disease prevalence surveys now underway in Vietnam, Cambodia and Malaysia may or may not reveal discrepancies between current estimates and the true burden of disease. But population-based surveys are not the long-term answer to measuring TB incidence; the ultimate solution for HBCs is the same as the approach currently taken in low-incidence countries, namely comprehensive routine surveillance. Surveys of disease burden are never done in industrialized countries, partly because incidence is low, but more importantly because we have confidence that most TB cases are reported. The drive to improve surveillance in HBCs should begin with careful assessments of which sub-populations do, and do not, have access to health facilities that can correctly diagnose and treat TB. This approach will give progressively better estimates of incidence at the same time as strengthening routine
DOTS strategy for tuberculosis control
surveillance. Monitoring the number and performance of health facilitiesFwhether all facilities report, how many cases per head of population, their catchment areas, the variation from year to year and from one facility to anotherFshould also provide a pragmatic approach to setting casefinding targets for national control programmes.
References 1. World Health Organization. Global tuberculosis control: surveillance, planning, financing. WHO Report 2002. WHO/ TB/2002.295, Geneva, 2002. 2. Dye C, Watt CJ, Bleed D. Low access to a highly effective therapy: a challenge for international tuberculosis control. Bull World Health Org 2002;80:437–44. 3. An Expanded DOTS framework for effective tuberculosis control. Unpublished Document WHO/CDS/TB/2002.297. Geneva: World Health Organization, 2002.
43
4. Corbett EL, Watt C, Walker N, Maher D, Raviglione MC, Williams BG, Dye C. The growing burden of tuberculosis: global trends and interactions with the HIV epidemic. Arch Intern Med 2003; in press. 5. Tuberculosis ControlFIndia.Directorate General of Health Services, Ministry of Health and Family Welfare. http://www.tbcindia.org/rntcp.asp. Accessed 28 June 2002. 6. Report on Nationwide Random Survey for the Epidemiology of Tuberculosis in 2000. Beijing: Ministry of Public Health, 2000. 7. Uplekar M, Pathania VK, Raviglione MC. Private practitioners and public health: weak links in tuberculosis control. Lancet 2001;358:912–6. 8. Chen X, Zhao F, Duanmu H, Wan L, Wang L, Du X, Chin DP. The DOTS strategy in China: results and lessons after 10 years. Bull World Health Org 2002;80:430–6. 9. Tropical Disease Foundation Inc. Final Report. 1997 National Tuberculosis Prevalence Survey. Makati Medical Center, Philippines, 1997. 10. Shilova MV, Dye C. The resurgence of tuberculosis in Russia. Philos Trans R Soc of London B 2001;356:1069–75.