Lusaka district, in which the capital of Zambia is situated, has a population of about 2 million people. Healthcare for the district is provided by the central Government and private for profit institutions, which are mainly situated in urban areas. The public health institutions are classified into health posts, health centres, district referral centres, general hospitals and central hospitals. Health posts are intended to cater for a population of about 500 to 100 households, while health centres serve a population of about 50,000 people in urban areas and 10,000 people in rural areas. District referral centres, general and central hospitals are meant to cater for populations of 200,000; 800,000 and above 800,000 respectively. Lusaka district has 8 health posts, 22 health centres and 2 central hospitals.

We undertook the costing study at Kanyama health centre, in Kanyama Township in the capital, Lusaka. The health centre is a public institution funded by the Ministry of Health (MOH), under the Lusaka District Health Management Team (DHMT). Its catchment area includes all the households in Kanyama Township and some other households in surrounding areas. The centre provides both outpatient and inpatient services, and has about 700 visits daily, and a total of 41 beds.

Kanyama Township had an estimated population of over 130,000 in 2006 14. Approximately 20% of the population were children below the age of 5 years. The Township has widespread poverty and unemployment and is characterised by poor sanitation and infrastructure. Urban poverty is quite high in Lusaka, with about 53% of all households living in absolute poverty 15. Unemployment is also high, and was estimated at 28% in 2005 16. Poverty mostly affects women and children, who constitute over 70% of the poor population.

The study was undertaken from the health provider perspective, and as such, only costs borne by the health centre were taken into consideration. We did not include the costs to households and social costs such as quality of life reductions caused by disease.

Data collection was guided by the Cost Analysis in Primary Health Care: A Manual for Programme Managers 17. Cost and outcome indicators were collected for the period August 2005 to August 2006. Since the health centre is a government institution, all its resources were provided by the government, or by other agencies such as non-governmental organisations and the donor community through the government. Therefore, only information about the quantities of resources used in the provision of care and on disease outcomes was obtained from the health centre records. Most of the data used in this study was collected from the MOH Health Management Information System (HMIS). This is an information system that is part of the monitoring and evaluation framework of the health sector in Zambia 18. The HMIS is a routine data collection system designed to provide information on the performance of the health system. Data such as number of visits, bed days, disease cases, and laboratory tests is collected on a monthly basis in the HMIS, a system which has been used for over a decade at Kanyama health centre.

Data on the prices and quantities of drugs and other recurrent medical supplies provided to the health centre were obtained from the DHMT pharmacy, which supplies all such items to government health centres in Lusaka District. It has a well developed system of delivery with updated documentation of all transactions to health centres. We were able to access these records to obtain updated information on supplies to Kanyama health centre, and accurate prices of most of the items provided. The prevailing market price was used as proxy for items whose prices were unavailable in the records of both the health centre and the DHMT. All costs were adjusted to 2006 prices using the 2006 Gross Domestic Products (GDP) deflator, valued at K1,175 19. Items were valued in Zambian Kwacha (K) and converted to United States Dollars (US$) at the average exchange rate in the study period, of about K3,600 to US$1 19.

Costs were classified as either economic or financial. Economic costs reflected the opportunity cost of resource use, while financial costs considered only expenditures incurred in the purchase of items. Costs were further classified as capital or recurrent. Capital costs included items such as buildings, equipment and vehicles, who's useful life was more than one year. Recurrent costs included items such as personnel and supplies, which could be replaced within a year. Items such as cutlery, buckets and garden implements, which could last for more than a year but cost less than US$100 were regarded as recurrent costs.

A physical count of all capital items such as furniture and equipment was undertaken to ascertain the exact number and condition. Only items that were functional were included for costing. The floor areas and costs of buildings were obtained from the Buildings Department at the Ministry of Works and Supply, which is the custodian of all information on government buildings. Where such costs were not available, the floor space of the buildings was measured physically and an estimated value was attached by a Quantity Surveyor at the Department.

Capital costs were annuitized to take into account the fact that such resources are bought in one year, but their useful life spans over several years 20. The annual financial cost of capital items was calculated using a straight line depreciation method, where an item's total cost was divided by the length of its useful life years. The economic cost of capital items was calculated using an interest rate of 6%, as recommended in literature 21. All five buildings of the health centre were assumed to have a useful life of 30 years, while the useful life of equipment varied from about 2 to 20 years. The useful life years for all capital items, except buildings, were obtained from the costs and prices used in the WHO Choosing Interventions that are Cost Effective (WHO-CHOICE) analysis 22.

Data were, however, sometimes not readily available and had to be estimated using proxies. For example, we did not have access to the original records of some capital and recurrent outlays, which made it difficult to obtain their original purchase prices. We therefore, made estimates which could have been over or under the true value. This was the case also for utilities, for which we did not find any records on bills paid towards water and electricity. We decided to use fixed monthly charges of US$2.78 for water 23 and US$6.50 for electricity 24 based on the prevailing rates in the study area. The utilities cost was therefore only considered as an economic and not a financial cost.

In some cases, the records were available but not in order, and the researcher could not sort them due to inadequate time and resources. It was not possible, for example, to allocate some medical supplies received by the health centre's pharmacy to the various wards or departments, as usage of these items was not specified by ward. It was thus decided to treat all such items as overhead costs, to be allocated to each of the wards according to an estimated allocation factor. The supplies that were 'ward or department specific', e.g. family planning drugs and some antibiotics were allocated directly to the necessary wards or departments.

Further, it was not immediately possible to allocate certain items, such as staff time to certain departments, as their usage between departments was not clearly defined. The same personnel who tended to the paediatric ward also serviced the male and female wards. As a result, it was not easy to distinguish to which ward/department some of the staff time belonged. It was thus decided to allocate time equally to those departments that were serviced by the same personnel, as there was no clearly defined pattern of work. The same was done for equipment, where percent usage was divided equally between wards.

Figure 1 shows the health centre's organizational structure and service production model, which illustrates how the health centre used a number of inputs (such as personnel and equipment), to produce services (such as lab tests and patient care) that patients received 25.

The health centre's departments were classified into 9 patient care and 6 overhead departments. The patient care departments included all units or wards where patients directly received treatment, such as the inpatient and outpatient wards and the laboratory. The overhead departments were those that serviced the patient care departments and whose activities were not immediately identified with disease treatment 1720. These included units such as the laundry, kitchen, pharmacy and general administration.

In each department, costs were categorised as follows: Personnel cost, including salaries, benefits and allowances. Capital costs, including buildings, vehicles, furniture and equipment such as x-ray machines. Supplies, including cleaning supplies, medical and surgical supplies. Maintenance, for buildings, vehicles and general repairs. For inpatient costs, we included beds, inpatient maintenance and supplies and length of stay for each disease.

The costs were further divided into patient care costs, which were costs incurred in the patient care departments, and overhead costs, incurred in the overhead departments. Overhead departments' costs were shared by all the patient care units. These costs were allocated to each patient care department according to the estimated proportion of number of visits, bed days, personnel or floor space.

The intermediate outputs were the services provided by each patient care department. The units of analysis used to measure the output were bed days for the inpatient department and visits for the outpatient department. In the laboratory, the units of measurement were the total number of lab tests conducted in the study period. In the labour/maternity ward, the units used were visits instead of bed days. This is because the majority of the women spent a day or less in this ward before being discharged. The total number of visits and admissions included both persons attending the health centre for the first time (new attendees) and re-attendees.

Total costs were divided by total outputs to obtain average costs. The average cost of inpatient care, for example, was obtained by dividing the total cost in the paediatric ward by the total number of bed days in the reference period. Average costs of service provision are presented for all the patient care departments except Anti Retroviral Treatment (ART), for which information on usage was not available.

To decide on the introduction of pneumococcal and rotavirus vaccines, the economic burden of pneumonia and diarrhoea is needed for inclusion in cost-effectiveness studies. Our result can help estimate these costs. To do this, we measured the average treatment costs of pneumonia and diarrhoea, using total under-five pneumonia and diarrhoea visits and bed days recorded in the health centre's records, and the total outpatient and in-patient costs for treating these diseases. All resources related to the treatment of childhood pneumonia, including personnel, buildings, medical supplies, laboratory equipment and utilities were costed. Unit costs for each disease were calculated by dividing the total outpatient and inpatient costs by the corresponding visits or bed days for the condition. The average healthcare costs provided for both pneumonia and diarrhoea thus include the outpatient (cost per visit) and inpatient (cost per bed day) costs.

The annual costs incurred in the patient care departments are presented in table 1. The total direct costs amounted to US$1,284,305 with US$1,195,129 attributed to recurrent costs and US$89,176 to capital costs. Capital costs contributed about 7% of total costs.

The outpatient ward had the highest total cost at US$426,847 (33%), while the Voluntary Counselling and Testing Ward (VCT) ward had the lowest at US$9,659 (1%). The outpatient ward also had the highest recurrent cost (35%), while the inpatient ward had the highest capital cost (28%).

Table 2 presents the output recorded during the study period. Kanyama health centre had an equivalent number of 150 staff in full time employment. It recorded approximately 208,964 visits, with 2,894 admissions and 10,633 bed days. A total number of 49,628 visits and 1,309 admissions were attributed to pneumonia and diarrhoea. There were a total number of 93,717 visits and 1,424 admissions made by children under the age of 5 years. About 16,511 and 8,925 visits; and 607 and 221 admissions made by under-five children were attributed to diarrhoea and pneumonia, respectively.

There were an estimated 14 outpatient visits per day per medical personnel for the entire study period. The health centre's bed occupancy rate, defined as the bed days per number of beds was 82%; while the bed turnover, defined as the number of admissions per number of beds was 71%. The average length of stay was 4 days.

Table 3 shows the outputs in all patient care departments. The out-patient department had the most visits at 128,240 and the dental had the least with 1,091. The total number of bed days recorded in the inpatient department was 10,633. There were 13,370 lab tests undertaken during the study period.

The average costs per unit of service in all the departments are given in table 4. The cost per visit in the outpatient department was US$3, while the cost per dental visit was US$20. The VCT ward recorded the lowest cost per visit at US$1. The inpatient department recorded a cost per bed day of US$18.

Table 5 shows the average cost of pneumonia and diarrhoea treatment in under-five children. The average cost of diarrhoea treatment in children under the age of 5 years was US$26 per visit and US$78 per bed day. The cost of pneumonia was US$48 per visit and US$215 per bed day.