20. Vaccine–Preventable Diseases

A study in Tamil Nadu evaluated immunization costs and coverage using a longitudinal panel dataset of immunization program costs (Brenzel 1995). Data were collected from a stratified, random sample of facilities between 1989 and 1991 for the North Arcot District Polio Control Program.^a The sample included 120 observations of 59 different health centers: 17 followed for three years (29 percent), 27 followed for two years (46 percent), and 15 with a single observation (25 percent). Total immunization costs included the cost of labor, vaccines, injection supplies, transportation, and overhead and the value of equipment, vehicles, and buildings.

During this period, coverage rates for FICs increased from 5 to 77 percent. The table shows that the cost per dose and the cost per FIC increased during this period.^b Changes in the cost per dose were highly statistically significant, whereas no statistical differences were apparent in the cost per FIC during the study period.

The study used data from the health facility sample to explain the determinants of immunization costs, which were modeled as a function of outputs, input prices, and other production-related variables that influence the cost function with respect to outputs. A random effects estimation was performed on the analysis sample relating the natural logarithm of health facility costs to the type of polio vaccine in use, estimated target population, and size of geographical area serviced by the health facility and natural logarithms of the number of FICs per facility, the number of hours spent by a village health nurse on immunization services per facility, and the number of small pieces of equipment used for immunization service delivery.

The analysis revealed a significant association between facility cost and the number of FICs, the hours worked by village health nurses, the area served, and the type of polio vaccine. When calculated using mean values, the marginal cost per FIC was Rs 24.43 (US$1.30) lower than the associated average cost per FIC of Rs 183 (US$9.80), implying that the average cost curve lies above the marginal cost curve for the sample of health facilities in India. A declining relationship is apparent between costs and coverage for this sample of facilities, calling into question assumptions of constant returns to scale. The results suggest that, in India, average cost-effectiveness ratios would overestimate total resource needs. Using a single-point estimate of average unit costs to determine the use of scarce public health resources will result in suboptimal resource allocations.