Tuberculosis (TB)


Financing DOTS expansion

Data collection

We collected data from five main sources: NTPs, the WHO-CHOICE web site, 19 costing guidelines developed for the “Disease Control Priorities in Developing Countries” project (DCPP), 20 GFATM proposals and databases, and previous WHO reports in this series. In 2004, data were collected directly from countries by means of a two-page questionnaire included in the standard WHO data collection form. NTP managers were asked to complete three tables. The first two tables required a summary of the NTP budget for fiscal years 2004 and 2005 in US$, broken down by line item and funding source (including a column for funding gaps). The third table requested NTP expenditure data for 2003, broken down by line item and source of funding. The form also requested information about dedicated TB control infrastructure and the way in which general health infrastructure is used for TB control – for example, the number of dedicated TB beds that exist, the number of outpatient visits that patients need to make to a health facility during treatment and the average number of days for which patients are hospitalized. We also asked for an estimate of the number of patients that would be treated in 2004 and 2005. We used the WHO-CHOICE web site to identify the average costs, in international dollars (I$), of a hospital bed-day and an outpatient clinic visit in every country. The costing guidelines for the DCPP and the WHO-CHOICE web site were used to identify the purchasing power parity (PPP) exchange rates required for conversion of I$ costs to costs in US$ (for consistency with budget and expenditure data reported on the data collection form).

Data entry and analysis

High-burden countries. Data entry and analysis focused on the 22 HBCs. We created a standardized spreadsheet, with one worksheet for each country. Additional worksheets were included for summary analyses and for the data required as inputs to the analyses in each country worksheet (e.g. notification data, unit costs for bed-days and outpatient clinic visits, and the typical number of outpatient clinic visits and days in hospital for different types of patient during treatment). For each country worksheet, seven tables were created. These were:

  • NTP budget by source of funding for each year 2002–2005, with the funding sources defined according to the 2004 data collection form i.e. government (excluding loans), loans, grants (excluding GFATM), GFATM and budget gap;
  • NTP budget by line item for each year 2002–2005, with the line items defined according to the 2004 data collection form i.e. first-line drugs, second-line drugs, dedicated NTP staff, initiatives to increase case detection and cure rates, collaborative TB/HIV activities, buildings/equipment and other;
  • NTP expenditures by source of funding for 2002 and 2003, with funding sources as defined for NTP budgets;
  • NTP expenditures by line item for 2002 and 2003, with line items defined as for NTP budgets;
  • total TB control costs by funding source for each year 2002–2005, with funding sources defined as for NTP budgets;
  • total TB control costs by line item for each year 2002–2005, with the line items defined as NTP budget items, hospitalization and clinic visits;
  • per patient costs, NTP budget, available funding, expenditures and budget for first-line drugs.

Budget data for 2004 and 2005 were taken from the 2004 data collection form. Budget data for 2002 and 2003 were taken from the 2002 and 2004 annual reports, respectively. Expenditure data for 2002 and 2003 were based on the 2003 and 2004 data collection forms, respectively. Total TB control costs were estimated by adding costs for hospitalization and outpatient clinic visits to either NTP expenditures (for 2002 and 2003) or NTP budgets (for 2004 and 2005). 21 Expenditures were used in preference to budgets for 2002 and 2003 because they reflect actual costs, whereas budgets can be higher than actual expenditures (for example, when large budgetary funding gaps exist or the NTP does not spend all the available funding). When expenditures are known for 2004 and 2005, they will be used instead of budget data to calculate, retrospectively, the total cost of TB control in these years. For some HBCs, expenditures were not available for 2002 and 2003. When this was the case, we estimated expenditures based on available funding, which was calculated as the total budget minus the funding gap.

The cost of outpatient clinic visits was estimated in three steps. First, we converted I$ prices for clinic visits reported on the WHO-CHOICE web site into US$ prices using the DCPP exchange rates. Second, we multiplied the average number of visits required per patient (estimated on the WHO data collection form) by the average cost (in US$) per clinic visit, to give the cost per patient treated. Third, we multiplied the cost per patient treated by the number of patients notified (for 2002 or 2003) or the number of patients that the NTP projects will be treated (for 2004 and 2005). The cost of hospitalization was generally calculated in the same way, replacing the unit cost of a clinic visit with the unit cost of a bed-day. The procedure differed for eight countries that have dedicated TB beds, and where the total cost of these beds is higher than implied by multiplying bed-days per patient by the number of patients treated (this applied to Brazil, Cambodia, India, Nigeria, the Russian Federation, the United Republic of Tanzania, Viet Nam and Zimbabwe). We assumed that all clinic visits and hospitalization are funded by the government.

Per patient costs, budgets, available funding and expenditures were calculated by dividing the relevant total by the number of cases notified (for 2002 and 2003) and the number of patients that the NTP projects will be treated (for 2004 and 2005). Since the total costs of TB control for 2002 and 2003 were based on expenditure data, it is possible for the total TB control cost per patient treated to be less than the NTP budget per patient treated when the funding gap is large or there is an important budgetary under-spend. In addition, for 2002 and 2003, the expenditure per patient was sometimes higher than the available funding per patient. This can occur when some of the NTP budget funding gap is closed following the reporting of budget data to WHO.

All data are reported in nominal prices (i.e. they have not been adjusted for inflation) rather than constant prices (i.e. all data are adjusted to a common year of prices) for two reasons. First, this avoids adjustment of values reported in the 2002–2004 reports in this series, which makes it easier for country staff to review the data for previous years. Second, the adjustment will make only a limited difference to the numbers reported (about 5% to 2002 values and less for other years). However, as data are collected for an increasing number of years, presentation of data in constant prices will be necessary.

Following data entry, text on data sources and assumptions were added. Where there were questions about the data, these were discussed with NTP staff and the appropriate WHO regional and country office. These discussions were used to produce a final set of charts. Four of these charts appear in the profiles for each country at Annex 1: NTP budget by funding source, NTP budget by line item, total TB control costs by line item, and per patient costs, budgets, available funding, expenditures and budget for first-line drugs. These charts were selected because they illustrate the most important trends in financing, while other data are referred to in the text. A full set of charts and data is available upon request. In some instances, the review process led to revisions to data included in previous annual reports. For this reason, figures sometimes differ from those reported in the 2002, 2003 and 2004 reports.

Finally, we compared the total costs of TB control with total government health expenditures to estimate the percentage of total government health expenditures used for TB control. Total government health expenditures were estimated by multiplying the government health expenditure per capita in US$ (as estimated in the WHO national health accounts database) 22 by population size. We also explored the association between GNI per capita in 2003 and (a) government contributions to total NTP budgets and TB control costs, and (b) the cost per patient treated. Data on GNI per capita were taken from World development indicators 2004. 23

Other countries. The data provided by countries other than the HBCs were less complete, and as a consequence our analyses to date are more superficial. We used the data provided on the 2004 data collection form to assess NTP budgets by region, and compared these with the budgets reported by the HBCs. Only countries that submitted complete data of sufficient quality (e.g. subtotals and totals were consistent by both line item and funding source) were used.

GFATM contribution to TB control

We assessed GFATM funding for both HBCs and other countries, as announced after the first four rounds of funding. We assessed total approved funding at the end of 2004, how the amounts in signed grant agreements compared with those in the original proposals, disbursements to the end of 2004, the time taken between approval of a proposal and the signature of grant agreements, and the time taken between the signing of the grant agreement and the first disbursement of funds.



20 DCPP guidelines for authors, pp. 74–77, (available at, accessed 11 January 2005).

21 The exception was South Africa, because no data on hospitalization and clinic visits, or on NTP budgets, were provided in the data collection form. Costs were therefore estimated based on recent costing studies, as described in previous WHO reports in this series.