Methods
Monitoring progress towards the Millennium Development Goals
MDGs for tuberculosis control
The MDG framework consists of a hierarchy of indicators that measure progress towards “targets”, which are the specific achievements needed to satisfy higher “goals”. Those most directly relevant to TB control are Goal 6 (to combat HIV/AIDS, malaria and other diseases) and Target 8 (to have halted by 2015 and begun to reverse the incidence of malaria and other major diseases, including TB). Among the indicators for Target 8 are two groups that can be used to evaluate the implementation and impact of TB control:
Indicator 23: between 1990 and 2015, to halve the prevalence and death rates associated with tuberculosis; and
Indicator 24: by 2005, to detect 70% of new smear-positive TB cases arising annually, and to successfully treat 85% of these cases.
The MDG indicators exclude HIV- positive TB patients, mainly to avoid double-counting in death statistics (deaths of HIV-positive people are recorded as AIDS deaths by WHO). However, we routinely calculate all TB indicators with and without HIV- positive TB patient, because TB control programmes need to know both.
This report focuses on the five principal indicators: incidence, prevalence, deaths, case detection and treatment success. The objective of reducing incidence is made explicit by Target 8; the targets for case detection and treatment success have been set by WHO’s World Health Assembly; 3 the targets for prevalence and deaths are based on a resolution of the year 2000 meeting of the Group of Eight (G8) industrialized countries, held in Okinawa, Japan.
Data collection and verification
Every year, WHO requests information from TB control programmes (or relevant public health authorities) in 211 countries or territories via a standard data collection form. The latest form was distributed in mid 2004. The section dealing with monitoring and surveillance asked for the following data: TB control strategies implemented up to the end of 2003; TB case notifications in 2003; and treatment outcomes for TB patients registered during 2002, following definitions given in Table 1. The most recent form can be downloaded from www.who.int/tb.
The data collection form is a tool for collecting aggregated national data. The process of national and international reporting is quite distinct from WHO’s recommendations about procedures for recording and reporting data within NTPs. The information gathered from the form includes a core set of data (questions remain more or less the same each year), plus new or timely information (questions may change from year to year). In the latest form, there are new questions about TB/HIV collaboration, about financing (the second year of collection but somewhat expanded), and about the outcomes of re-treatment, for patients who have received two or more courses of anti-TB drugs.
Completed forms are collected and reviewed at all levels of WHO – in WHO country offices, regional offices and at headquarters – and an acknowledgement form that tabulates all data submitted and shows WHO’s calculations of principal indicators, is sent back to the national correspondent in order to complete any missing responses and to resolve any inconsistencies.
In the WHO European Region only, data collection and verification are performed jointly by the regional office and a WHO collaborating centre, EuroTB (Paris), using a different format. EuroTB subsequently publishes an annual report with additional analyses, using more detailed data for the European Region (see: www.eurotb.org).
High-burden countries and WHO regions
Much of the data submitted to WHO is shown, country by country, in the annexes of this report. The analysis and interpretation that precedes these annexes focuses on 22 HBCs and the six WHO regions. The 22 HBCs account for approximately 80% of the estimated number of new TB cases (all forms) arising worldwide each year. These countries are the focus of intensified efforts in DOTS expansion (Annex 1). The HBCs are not necessarily those with the highest incidence rates per capita; many of the latter are medium-sized African countries with high rates of TB/HIV coinfection.
The WHO regions are the African Region, the Region of the Americas, the Eastern Mediterranean Region, the European Region, the South-East Asia Region and the Western Pacific Region. All essential statistics are summarized for each of these regions and globally. However, to make clear the differences in epidemiological trends within regions, we divide the African Region into countries that have low and high rates of HIV infection (boundary at an estimated infection rate of 4% in adults aged 15–49 years), and include those countries in the Eastern Mediterranean Region which are actually on the African continent (Djibouti, Somalia and Sudan) in the low-HIV Africa group. Furthermore, we distinguish central from eastern Europe (countries of the former Soviet Union plus Bulgaria and Romania), and combine western European countries with the other established market economies. The countries within each of the resulting nine regions are listed in the legend to Figure 6.
DOTS classification
DOTS is the internationally recommended approach to TB control. It is not simply a clinical approach to patient management, but rather a strategy for TB control primarily within public health systems. Countries reporting to WHO classify themselves as DOTS or non-DOTS, referring to the elements listed in Table 2. DOTS countries must have officially accepted and adopted the strategy, and must have implemented the essential components of DOTS in at least part of the country (Annex 2). Based on NTP responses to standard questions about policy, and usually on further discussion with the NTP, WHO accepts or revises each country’s own determination of its DOTS status.
DOTS coverage
Coverage in any country is defined as the percentage of the national population living in areas where health services have adopted DOTS. “Areas” are the lowest administrative or management units in the country – townships, districts, counties, etc. If an area (with its one or more health facilities) is considered by the NTP to be a DOTS area in 2003, then all the cases registered and reported by the NTP in that area are considered DOTS cases, and the population living within the boundaries of that area counts towards the national DOTS coverage. In some cases, treatment providers who are not following DOTS guidelines (for example private practitioners, or public health services outside the NTP such as those within prisons) notify cases to the NTP. These cases are considered non-DOTS cases, even if they are notified from within DOTS areas. However, when certain groups of patients treated by DOTS services receive special regimens or management (for example nomads placed on long-course treatment), these are considered as DOTS cases. Where possible, additional information about these special groups of patients is provided in the country notes in Annex 2.
Coverage is a crude indicator, which is easy to calculate, and which is most useful during the early stages of DOTS expansion. As a measure of patient access to diagnosis and treatment under DOTS, coverage is an approximation, and usually an overestimate. Where countries are able to provide more precise information about access to DOTS services this information is reported in the country notes of Annex 2. The case detection rate (defined below) is more precise, but also more demanding of data.
Estimating TB incidence, prevalence and death rates
Estimates of incidence, prevalence and deaths are based on a consultative and analytical process; they are revised annually to reflect new information gathered through surveillance and from special studies, such as prevalence surveys. The details of estimation are described elsewhere. 5 , 6 In brief, estimates of incidence (number of new cases per year) for each country are derived by one or more of four approaches, depending on the available data:
The “St´yblo coefficient” in equation (3) is taken to be a constant, with an empirically derived value in the range 40–60, relating risk of infection (%) to the incidence of smear-positive cases (per 100 000 per year). Given two of the quantities in any of these equations, we can calculate the third, and any of these formulae can be rearranged to estimate incidence, prevalence and death rates. The available data differ from country to country but include case notifications and death records (from routine surveillance and vital registration), and measures of the prevalence of infection and disease (from population-based surveys).
For each country, estimates of incidence for each year in the period 1995–2003 are made as follows. We first select a reference year for which we have a best estimate of incidence; this may be the year in which a survey was carried out, or the year in which incidence was first estimated. We then use the series of case notifications (all forms of TB) to determine how incidence changed before and after that reference year. The time series of estimated incidence rates is constructed from the notification series in two ways: if the rate of change of incidence is roughly constant through time, we fit exponential trends to the notifications; if the rate of change varies (eastern Europe, central Europe and high-HIV Africa), we use a three-year moving average of the notification rates. If the notifications for any country are considered to be an unreliable guide to trend (e.g. because reporting effort is known to have changed), we apply the aggregated trend for all other countries with reliable data from the same epidemiological region. For China, exceptionally, we have used an assessment of the trend in incidence based on risk of infection derived from tuberculin surveys. For those countries that have no reliable data from which to assess trends in incidence (e.g. for countries such as Iraq, for which data are hard to interpret, and which are atypical within their own regions), we assume incidence is stable. Further details are available at www.who.int/tb.
For countries that have not yet measured HIV infection rates in TB patients directly, an indirect estimate can be obtained from the incidence rate ratio (IRR, the TB incidence rate in HIV-infected people divided by the incidence rate in HIV-uninfected people), as described elsewhere. 6 The prevalence of MDR-TB among previously untreated TB patients has also been estimated in a separate exercise, 7 supplemented with data from more recent surveys. 8
Estimates of incidence form the denominator of the case detection rate. Trends in incidence are determined by underlying epidemiological processes, modified by control programmes. The impact of control on prevalence is determined by the trend in incidence, and by the estimated
reduction in the duration of the condition, e.g. smear-positive disease. The impact of control on deaths is determined by the trend in incidence, and by the estimated reduction in case fatality (proportion of incident cases that ever die from TB). 5 , 6
Where population sizes are needed to calculate TB indicators, we use the latest revision of estimates provided by the United Nations Population Division, 9 even though these estimates sometimes differ from those made by the countries themselves (some of which are based on more recent census data). The estimates of some TB indicators, such as the case detection rate, are derived from data and calculations that use only rates per capita, and discrepancies in population sizes do not affect these indicators. Where rates per capita are used as a basis for calculating numbers of TB cases, these discrepancies sometimes do make a difference. Some examples of important differences are given in the country notes in Annex 2.
Case detection rate
Smear-positive cases are the focus of DOTS programmes because they are the principal sources of infection to others, because sputum smear microscopy is a highly-specific (if somewhat insensitive) method of diagnosis, and because patients with smear-positive disease typically suffer higher rates of morbidity and mortality than smear-negative patients. As a measure of the quality of diagnosis, we calculate the proportion of new sputum smear- positive cases out of all new pulmonary cases, which has an expected value of 65–80% in areas with negligible HIV prevalence. 10 However, this report presents the numbers of all TB cases notified, smear-positive and smear-negative pulmonary cases, in addition to those in whom extrapulmonary disease is diagnosed.
The term “case detection”, as used here, means that TB is diagnosed (correctly or incorrectly) in a patient, and is reported within the national surveillance system, and then to WHO. The case detection rate is calculated as the ratio of the number of notified smear-positive cases to the number of new smear-positive cases estimated for that year. Detection is presented in two ways – as the case detection rate (countrywide) and as the DOTS case detection rate (by DOTS programmes):
The case detection rate and the DOTS case detection rate are identical when a country reports only from DOTS areas. This generally happens when DOTS coverage is 100% but, in some countries where DOTS is implemented in only part of the country, no TB notifications are received from the non-DOTS areas. Furthermore, in some countries where DOTS coverage is 100%, patients may choose to seek treatment from non-DOTS providers, who in some cases notify TB cases to the national authorities.
Both of the above definitions of the case detection rate refer to smear-positive cases, although we also present the detection rate for all forms of TB. The detection rate of all forms is similarly presented in two ways: detection by DOTS programmes, and detection countrywide.
Although these indices are termed “rates”, they are actually ratios. The number of cases notified is usually smaller than estimated incidence because of incomplete coverage by health services, under-diagnosis, or deficient recording and reporting. However, the calculated detection rate can exceed 100% if case-finding has been intense in an area that has a backlog of chronic cases, if there has been over-reporting (e.g. double-counting) or over-diagnosis, or if estimates of incidence are too low. If the expected number of cases per year is very low (especially if it is less than one), the case detection rate can vary markedly from year to year due to chance. Whenever this index comes close to or exceeds 100%, we attempt to investigate, as part of the joint planning and evaluation process with NTPs, which of these explanations is correct.
The ratio of the DOTS case detection rate to coverage estimates the case detection rate within DOTS areas (as distinct from the case detection rate nationwide), assuming that the TB incidence rate is homogeneous across counties, districts, provinces, or other administrative units. Ideally, this ratio would have a value of 70% or more as DOTS coverage increases within any country. Where the value of this indicator is much lower, it is clear that the DOTS programme has been poorly implemented, at least in some parts of the designated DOTS area. Changes in the value of this ratio through time are a measure of changes in the quality of TB control, after the DOTS programme has been established.
Treatment success
Treatment success in DOTS programmes is the percentage of new smear-positive patients that are cured (negative on sputum smear examination), plus the percentage that complete a course of treatment, without bacteriological confirmation of cure (Table 2). 11 Cure and completion are among the six mutually exclusive outcomes. 12 The sum of cases assigned to these outcomes, plus any additional cases registered but not assigned to an outcome, adds up to 100% of cases registered (i.e. the treatment cohort).
We also compare the number of new smear-positive cases registered for treatment (for this report, in 2002) with the number of cases notified as smear-positive (also in 2002). All notified cases should be registered for treatment, and the numbers notified and registered should therefore be the same (discrepancies arise e.g. when subnational reports are not received at national level). If the number registered for treatment is not provided we take, as the denominator for treatment outcomes, the number notified for that cohort year. If the sum of the six outcome categories is greater than the number registered (or the number notified), we use this sum as the denominator.
Because the number of patients presenting for a second or subsequent course of treatment, and the outcome of further treatment, are indicative of NTP performance and levels of drug resistance, we have begun to compile these data in this report. We present the numbers of patients registered for re-treatment, and the outcomes of re-treatment, for each of three registration types: re-treatment after relapse, failure or default. However, some countries do not yet compile data on cases registered for re-treatment after failure and default separately at national level. Furthermore, some countries do not have outcome data for each of these re-treament case types.
The assessment of outcomes for a given calendar year always lags notifications by one year to ensure that all patients registered during that calendar year have completed treatment. A DOTS country must report treatment outcomes, unless it is newly-classified as DOTS, in which case it would take an additional year to report outcomes from the first cohort of patients treated.
Overview of data in annexes
Annex 1 presents data on epidemiology and surveillance, and planning and financing for each of the 22 HBCs. Data on policy and strategy are collected for both DOTS and non-DOTS areas separately.
Annex 2 contains the estimates needed to evaluate MDG Target 8 and indicators 23 and 24. These data include case detection and treatment success rates to monitor DOTS implementation, and incidence, prevalence and death rates to monitor the impact of TB control.
These data are presented, for each of the six WHO regions, as follows:
- TB control policies for each country, stating which technical components of the DOTS strategy have been implemented;
- incidence, prevalence and death rates for 1990 (MDG reference year) and 2003;
- case notifications, detection rates, and DOTS coverage: nationally, and separately for DOTS and non-DOTS programmes. Notifications include new pulmonary cases (smear- positive, smear-negative and laboratory-confirmed), new extrapulmonary and re-treatment cases;
- treatment outcomes for 2002 cohorts: both the new smear-positive and the re-treatment cohorts from DOTS programmes (relapse, re-treatment after default and re-treatment after failure are presented separately where possible, as well as all re-treatment cases combined), and the new smear-positive treatment outcomes (where available) from non-DOTS programmes;
- new smear-positive notification rates by age and sex for the whole country;
- new smear-positive notifications (numbers) by age and sex, from DOTS and non-DOTS programmes;
- notification rates and numbers since 1980, for all forms of TB;
- notification rates and numbers since 1995, for new smear-positive cases;
- country notes: remarks that may help to explain data reported by selected countries (e.g. additional breakdown of cases of interest, late-reported data, reasons for incomplete data, discrepancies in estimated population sizes).
The data in Annex 2 are available as Excel spreadsheets from http://www.who.int/tb.
Footnotes
3 Resolution WHA44.8. Tuberculosis control programme. In: Handbook of resolutions and decisions of the World Health Assembly and the Executive Board. Volume III, 3rd edition (1985–1992). Geneva, World Health Organization, 1993 (WHA44/1991/REC/1).
5 Dye C et al. Global burden of tuberculosis: estimated incidence, prevalence and mortality by country. Journal of the American Medical Association, 1999, 282: 677–686.
5 Dye C et al. Global burden of tuberculosis: estimated incidence, prevalence and mortality by country. Journal of the American Medical Association, 1999, 282: 677–686.
6 Corbett EL et al. The growing burden of tuberculosis: global trends and interactions with the HIV epidemic. Archives of Internal Medicine, 2003, 163:1009–1021.
6 Corbett EL et al. The growing burden of tuberculosis: global trends and interactions with the HIV epidemic. Archives of Internal Medicine, 2003, 163:1009–1021.
6 Corbett EL et al. The growing burden of tuberculosis: global trends and interactions with the HIV epidemic. Archives of Internal Medicine, 2003, 163:1009–1021.
7 Dye C et al. Worldwide incidence of multidrug-resistant tuberculosis. Journal of Infectious Diseases, 2002, 185:1197–1202. 2003, 163:1009–1021.
8 Anti-tuberculosis drug resistance in the world. Report No.3. WHO/IUATLD Global Project on Anti-Tuberculosis Drug Resistance Surveillance. Geneva, World Health Organization, 2004 (WHO/HTM/TB/2004.343).
9 World population prospects – the 2002 revision. New York, United Nations Population Division, 2003.
10 Tuberculosis handbook. Geneva, World Health Organization, 1998 (WHO/TB/98.253).
11 TB control programmes should ensure high treatment success before expanding case detection. The reason is that a proportion of patients given less than a fully-curative course of treatment remain chronically infectious, and continue to spread TB. Thus DOTS programmes must be shown to achieve high cure rates in pilot projects before attempting countrywide coverage.
12 Treatment of tuberculosis: Guidelines for national programmes. Third edition. Geneva, World Health Organization, 2003 (WHO/CDS/TB/2003.313).