Cost analysis of an integrated disease surveillance and response system: case of Burkina Faso, Eritrea, and Mali

Background Communicable diseases are the leading causes of illness, deaths, and disability in sub-Saharan Africa. To address these threats, countries within the World Health Organization (WHO) African region adopted a regional strategy called Integrated Disease Surveillance and Response (IDSR). This strategy calls for streamlining resources, tools, and approaches to better detect and respond to the region's priority communicable disease. The purpose of this study was to analyze the incremental costs of establishing and subsequently operating activities for detection and response to the priority diseases under the IDSR. Methods We collected cost data for IDSR activities at central, regional, district, and primary health care center levels from Burkina Faso, Eritrea, and Mali, countries where IDSR is being fully implemented. These cost data included personnel, transportation items, office consumable goods, media campaigns, laboratory and response materials and supplies, and annual depreciation of buildings, equipment, and vehicles. Results Over the period studied (2002–2005), the average cost to implement the IDSR program in Eritrea was $0.16 per capita, $0.04 in Burkina Faso and $0.02 in Mali. In each country, the mean annual cost of IDSR was dependent on the health structure level, ranging from $35,899 to $69,920 at the region level, $10,790 to $13,941 at the district level, and $1,181 to $1,240 at the primary health care center level. The proportions spent on each IDSR activity varied due to demand for special items (e.g., equipment, supplies, drugs and vaccines), service availability, distance, and the epidemiological profile of the country. Conclusion This study demonstrates that the IDSR strategy can be considered a low cost public health system although the benefits have yet to be quantified. These data can also be used in future studies of the cost-effectiveness of IDSR.


Background
Communicable diseases remain the most common causes of death, illness and disability in African countries. Lopez et al. (2006) reported that one-third of the deaths in lowand-middle income countries in 2001 were from communicable and parasitic diseases and maternal and nutritional conditions [1,2]. In addition, the economic cost in terms of prevention, treatment, and loss of productivity is enormous [3][4][5]. Although a number of studies on economic evaluation of interventions against communicable diseases have been reported in the literature [6,7], most of these studies in sub-Saharan Africa have focused on individual disease-specific intervention programs, such as prevention or treatment of malaria, measles, meningitis, tuberculosis and HIV/AIDS [5,[8][9][10][11][12][13][14][15]. Relatively few studies have looked at the economics of integrating resources for disease surveillance and public health response activities [16].
Surveillance is an important component of disease prevention and control programs. It is useful in early detection of unusual events for effective and timely action, monitoring and evaluation of interventions and guiding selection of appropriate corrective measures [17]. In 1998, the Regional Committee of the World Health Organization Africa region (WHO-AFRO) adopted a strategy called Integrated Disease Surveillance and Response (IDSR) [18]. Under the IDSR strategy, countries address improvements in infrastructure capacities and support activities and select a number of priority diseases and health risk conditions from a list of the 19 communicable diseases that affect African communities (Figure 1) [19][20][21]. By December 2007, considerable progress had been achieved, with 43 of the 46 countries having assessed their national surveillance system and developed plans of action; 41 countries had already adapted the technical guidelines to meet their own public health priorities and situations and then launched IDSR activities at their district levels; and 33 countries had trained staff on IDSR in at least 60% of their districts (Table 1).
In order to better understand the investment and implementation costs of this IDSR strategy, the IDSR multi-partner task force that guides the implementation of this regional strategy recommended that the partners undertake cost analyses and cost-effectiveness studies. Therefore, the purpose of this study was to analyze the incremental costs of establishing and subsequently operating activities for detection and response to the priority diseases under the IDSR.

Study countries
The study was conducted in Burkina Faso, Eritrea, and Mali, countries where infectious diseases such as cholera, malaria, meningococcal meningitis and yellow fever are either epidemic or endemic (Table 2). Burkina Faso, with a population of 13.2 million, is divided into 13 health regions, 55 districts and has more than 1,232 primary health care centers. Mali, with about 13.5 million inhabitants, has nine regions, 57 districts and over 709 fully operational primary health care centers. Eritrea, with an estimated population of 4.4 million, is divided into six regions, 57 districts and has 664 primary health care centers. These three countries were selected for this study because each had fully established IDSR leadership and structures at the national level by 2002, with implementation at regional and district levels in 2003 and 2004, respectively.

Study design
We conducted retrospective surveys of costs for integrating surveillance and response to the priority diseases adopted by each country (see Additional file 1, Appendix 1) at national, regional, and district surveillance offices as well as public health clinics, laboratories and pharmacies. We conducted one field test in Eritrea followed by full field studies in Burkina Faso and Mali. In Eritrea, the survey sites included the central Ministry of Health, the Anseba provincial office, and offices in the Haquaz district. In Burkina Faso, most IDSR activities were focused on the epidemic-prone diseases, with particular emphasis on detection and response to meningococcal meningitis. The data were obtained from four health regions (Bobo Dioulasso, Gaoua, Kaya, and Ouahigouya), 14 districts, and 20 primary health care centers. In Mali, the survey was con-ducted in three regions (Kayes, Mopti, and Sikasso), and included one district per region and one primary health care center per district. We consulted, in each country, with public health and disease surveillance officers to select sites that they considered representative of the national IDSR system.
We took the perspective of the government-funded health care system (i.e., we only recorded costs incurred by the governments and external partners). All cost data were recorded in local currency values and then converted into US dollar using the appropriate mean annual exchange rate. We used the general consumer price index from each country and a discount rate of 3% to adjust all costs into 2002 US dollars equivalent [22]. We also examined the effect on cost per capita estimate of using purchasing Recommended IDSR priority diseases, core functions and activities in the WHO African region Figure 1 Recommended IDSR priority diseases, core functions and activities in the WHO African region.
power parities (PPP) to convert national currencies into international dollars (PPP removes currency conversion problems due to fixed conversion rates that may not reflect actual relative costs) [22].

Cost data
We collected data associated with all "health-related surveillance" (HRS) activities (i.e., all communicable and non-communicable diseases and risk factors, including the surveillance and response activities of the IDSR tar- For each country, region/province and district, we obtained annual population data from the disease surveillance units. Program cost data were obtained from disease surveillance budget and program records, and from interviews with IDSR program coordinators and key public health staff. Whenever we found a difference between budget and reported expenditure, we used the reported expenditure. Aggregated pharmacy, clinical and medical records were collected using a structured questionnaire. The survey instrument (available from http:// www.cdc.gov/idsr/survcost.htm) guided collection of data on all the resources used, including capital (one-time investment) and recurrent (on-going) items. The capital items included building infrastructure, vehicle, equipment (e.g., refrigerators, computers, etc.), and furniture (e.g., tables, chairs, etc.). The recurrent items included personnel (salaries and benefits of surveillance officers, data managers, physicians, nurses, etc.), rent (rent, utilities, operation, and maintenance), office and laboratory supplies, transportation, public awareness campaigns and short-term training. The questionnaire also collected information on other variables related to disease surveillance activities, such as length of use of buildings, vehicles or equipment per year, and resources provided through other activities and organizations.

IDSR specific cost estimation
For each health structure level, all resources were grouped into the following major categories: personnel; transportation; office consumable goods; public awareness campaigns; drugs or treatment; laboratory supplies; and capital items (Additional file 1, Appendix 2). For each category, we identified the proportion of those cost data (such as staff workload or actual use of resources, if estimates or records were available) attributable to IDSR.

Personnel costs
When time keeping records were absent, we interviewed each staff member to estimate the breakdown of their time on all HRS, IDSR priority diseases, each IDSR activity (i.e., detection, notification, analysis, investigation, response, feedback, and support), and other ministry of health activities. We recorded the number of workers, their annual income, and the number of full time equivalents needed for administration or delivering of each HRS and IDSR activity. We then apportioned total personnel costs to each IDSR activity based on the ratio of personnel time allocated to that activity relative to all IDSR activities. We included fees of individual consultants hired for specialized services such as short-term training.

Transportation costs
We considered vehicles purchased for IDSR activities as capital items (see below). IDSR-related running costs for transporting personnel and patients, drugs, specimens, vaccines and other items, as a percentage of the total fuel and maintenance costs, were estimated based on the vehicle use-time per IDSR activity. When there were no data to apportion transport costs, we proportioned costs using the ratio of personnel time for IDSR to total personnel time for all HRS activities. We included rental vehicle and public transportation fees for IDSR-specific activities.
Office consumable costs These included office supplies and materials, facilities and equipment maintenance, and utilities costs. Office consumable costs for IDSR, as a proportion of all HRS costs, were calculated using either the ratio of IDSR personnel time to all HRS personnel time, or actual amount of resources used for IDSR-activities (if the latter were available).

Public awareness campaign costs
We measured advertising, broadcasting and media costs for public campaigns and targeted social mobilization. IDSR costs were estimated as a proportion of total media health education costs using the ratio of IDSR personnel time to all HRS personnel time.

Treatment costs
These included all drugs and vaccines as well as other programmatic measures (e.g., treated bed nets) used in the line of controlling and preventing diseases included in the IDSR program. Total annual costs were calculated based on the procurement cost and the quantity of each specific product required for the treatment of diseases. We estimated IDSR costs using either the actual amount of resource or the ratio of IDSR personnel time to all HRS personnel time at the health facility (if the former were available).

Laboratory consumable costs
We estimated the costs of laboratory consumable materials and supplies (e.g., reagents, slides, gloves, test tubes, cotton wool swabs, blood culture bottles, aluminum foil, syringes, rapid diagnostic kits, etc) required for the purpose of various diagnostic tests for diseases included in the IDSR strategy.

Capital equipment costs
The costs of buildings, laboratory and office equipment and vehicles were depreciated at 3% annually over a 50-, 10-, and 5-year useful-life time horizon, respectively. We calculated the annualized cost using the following general equation: where K is the purchase price of the item, r represents the depreciation rate, and t is the useful-life-year. We assumed the scrap value of the capital items at the end of the useful life to be zero.
For equipment and vehicles, we apportioned out capital costs using the equipment and vehicle use-time (see above). For buildings, we proportioned capital costs using the ratio of IDSR personnel time to all HRS personnel time.

Data analysis
We entered and analyzed the data in a spreadsheet (Microsoft Excel 5.0, Microsoft Corp., Seattle), calculating averages and standard deviations per resource category and per IDSR activity. We aggregated costs of all HRS and IDSR activities across all resource categories by health structure level. Using the estimated total costs for each province and district included in the study, and population estimates for each included province and district, we calculated average annual cost per capita per year for all HRS and IDSR activities. We then used these per capita costs and the annual population estimates to calculate the total annual national IDSR program cost in each country. We also compared the per capita surveillance costs to the per capita national health expenditures [23].

Missing data
We encountered two types of missing data. The first category of missing data involved cost data for some building structures and equipment. For example, cost data were missing for approximately half of buildings in each country. The second category of missing data involved cost data for the laboratory testing and treatments from Burkina Faso. To fill in for the structure and equipment cost data, we used average cost data for similar structures and equipment at other sites (in the same country) as a proxy for the missing data. For example, when the information necessary to estimate the cost of a specific building was not available, we used the data for similar ministry buildings in the same locality or nearby health structures. For the missing cost data from Burkina Faso, we conducted two analyses: one by cost category (personnel, transport, office, etc.) excluding any cost categories for which we had no data and the other by extrapolating the relevant cost data from the other countries. Table 3 summarizes the mean annual costs by resource categories at the region, district, and primary health care center level in the three countries. Detailed costs are shown in Additional file 1, Appendix 3. As expected, because of larger populations and types of IDSR activities, regional-level costs were greater in all categories than at the level of district and primary health care center. However, the cost of running IDSR at each site varied substantially by resource-type. Since disease surveillance requires

Results
trained staff, mean annual personnel costs were among the largest components of the region (10% to 47%) and district (16% to 44%) total IDSR costs in all three countries. Based on the results from Eritrea and Mali, we estimated that the laboratory and treatment costs ranged from 4% to 35% of the total IDSR cost in Burkina Faso. The proportion of the total IDSR cost due to treatment varied considerably (2% to 13%) by health structure in Eritrea and Mali. In general, the annualized capital costs constituted 2% to 13% of the total annual cost of IDSR in Eritrea, 6% to 12% in Mali, and 8% to 15% in Burkina Faso.
The mean annual costs by health structure levels from the three countries surveyed are presented in Table 4. Eritrea had the highest total IDSR-related costs, and Burkina Faso had the lowest costs. The mean cost of IDSR, expressed as percentage of all HRS cost, also varied by health structure level (  (Fig. 2). These were possibly associated with startup costs of IDSR implementation at regional and district levels. Costs disaggregated into IDSR activities are shown in Table 5. Detailed costs by year per IDSR activity are presented in Additional file 1, Appendix 4. As expected, the surveillance activities (e.g., detection, report, and analysis) that are carried out on a routine basis absorbed more resources than the support activities (e.g., evaluation and monitoring). Outbreak investigation and treatment of confirmed cases also constituted a substantial component (23% to 67%) of the total IDSR cost. When evaluating

Discussion
IDSR attempts to integrate multiple, competing vertical systems in order to use surveillance and response-related resources more efficiently and reduce duplication of effort, especially at district and primary health care center levels [20,21]. In this study, we measured the incremental costs of setting-up and implementing an integrated surveillance and response strategy in Burkina Faso, Eritrea and Mali. In each country, the cost of IDSR was dependent on the health structure level. The district and primary health care center levels had much lower costs, as they usually had only lower cadre health workers and disease surveillance officers to provide services. A full understanding of the between-country differences in per capita costs of IDSR will require further study. As shown when we used PPP to convert local costs into US dollars, difference exchange rates may alter the degree of differences between countries.
The study's main limitation is the potential inaccuracy when we apportioned total cumulative surveillance activ-  Table 2). Without the laboratory and treatment costs, the mean annual all disease surveillance and IDSR program costs were 54,327 and 18,476, 31,147 and 6,594, and 4,316 and 1,056 per region, district, and primary health care center level, respectively.
Total IDSR cost per region and district in Burkina Faso † , Mali and Eritrea ities cost (e.g., personnel time and building, equipment and vehicle use-time) to IDSR-specific activities. Log books of time and expenses did not provide the level of details needed to accurately divide out costs between IDSR and other surveillance and public health activities. As explained, we used the proportion of personnel time given to IDSR to proportion other costs. Furthermore, our retrospective survey may not have fully captured all costs due to the limitations of data records (e.g., no personnel time keeping records and the usual recall bias) in these countries. It is also possible that our data collection methods missed some surveillance-related expenditures. This is because, in Africa, donors often support specific public health projects (such as surveillance for a specific disease) that run parallel to the national public health system. Such projects often have a distinct identity (i.e., names and logos), and may even have staff paid directly by donor funds. Public health staff may not consider such projects part of the general public health system when enumerating costs associated with surveillance and IDSR. Another limitation of this study is the reliance on expenditure data, which may be weakened by over-and underestimation and incomplete recording and do not reflect the whole economic cost. Further, indirect costs and productivity losses were not incorporated. Moreover, our esti-mate of IDSR cost based on 4-year data may be higher than when a longer term perspective is taken due to nonrecurring start up costs. Absolute difference in cost per capita will depend upon the exchange rates used.
This study focused only the cost of resources accrued to IDSR activities and not the impact on the indicators used by the countries to monitor and evaluate their progress with their IDSR activities. In Eritrea, for example, the completeness of reporting case-based data from the health care center to the next high level increased from 50% in 2000 to 93% by the end of 2003. In Burkina Faso, the timeliness of surveillance reporting, especially data on epidemic-prone diseases, increased from 71% in 2000 to 99% by the end of 2004. Although Mali had also achieved the 80% target for these progress indicators, the transmission of complete data on time (83%) in 2005 was lower than that in Burkina Faso and Eritrea.
There are few studies on the costs of disease surveillance, and those are often not directly comparable to our study [16,24]. For example, John et al. (1998) measured the cost of emerging childhood vaccine-preventable diseases in India [16]. They found surveillance cost $0.01 per capita (1998 US $), which is approximately equal to the costs we  Table 2) measured in Burkina Faso and Mali (Table 6). However, the program in India only included childhood vaccinepreventable diseases, while the IDSR system includes not only childhood and adult vaccine-preventable diseases but also epidemic-prone diseases and endemic epidemics such as HIV/AIDS, malaria, TB, childhood diarrhea and acute respiratory infections. We can, therefore, consider IDSR a low cost public health system although the benefits, such as cases prevented, due to the IDSR program have yet to be quantified. eases   # Surveillance activities include detection, report, analysis and feedback ¶Response activities include field investigation and laboratory confirmation of suspected cases and treatment of confirmed cases. In Burkina Faso, laboratory and treatment costs were calculated using the average annual per capita costs of laboratory and treatment for Eritrea and Mali and the population size of the health structure in Burkina Faso (see Table 2). Without laboratory and treatment costs, the annual costs (std. dev) per capita of IDSR and total national disease surveillance were $0.019 (0.005) and $0.055 (0.013), respectively. §Support activities include training, supervision, evaluation, communication and coordination. † Costs shown were converted using official exchange rate. When costs were converted using the purchasing parity power (PPP), the mean cost for Eritrea was $0.87 (0.34) and for Mali $0.06 (0.03). ‡ Source: National Health Accounts http://www.who.int/nha/en/