Candidate for the Lee B. Lusted Student Prize Competition
Purpose:
Evidence of how efficient Mexican public institutions are to produce mammograms is limited. This paper aims to estimate a technical efficiency score of the production of mammography studies in a sample of Mexican public hospitals and assess the variation that is attributable to environmental and organizational constraints.
Method:
Parametric and semi-parametric technology analyses were performed on a sample of 38 facilities belonging to the three main public institutions in Mexico. The dependent variable in the model was the number of mammograms produced by each facility per quarter during 2011; covariates were the labor inputs, namely, the number of radiologists and diagnostic imaging technicians. Flexible production functions were tested with Corrected ordinary least-squares (COLS). The semi-parametric method identified facilities with the highest output for each level of inputs, then a stochastic frontier was built over this subset of facilities with locally weighted smoothing (LOWESS); pointwise confidence bands were estimated.
The efficient facilities identified with both algorithms were compared to their inefficient counterparts with distance functions, thus estimating the relative efficiency scores. Coherence in ranking results between methods was assessed with rank correlation tests. Lastly, the efficiency score vector was analyzed as a dependent variable in a multivariate, second-stage context that included environment variables that were likely to exert an influence on the technical efficiency level.
Result:
Under the parametric method, mean effiency score was 59%, and the semi-parametric model yielded an average of 48% (p>0.05). Both methods unveiled a non-linear relationship between labor inputs and quantity produced. The best parametric fit of the technology was delivered by the Translog model (highest R2 , second-least Akaike criterion). Hence, if all sources of inefficiency were eliminated, these sampled facilities could collectively double the quantity of breast cancer screening services provided without a major increase in the intensity of labor and infrastructure resources. The production scale was strongly and positively associated with the technical efficiency level (p<0.01), and those facilities located in low-income states exhibited lower relative efficiency scores (p<0.05).
Conclusion:
The high variability observed in productivity performance provided cogent, prima facie indication of considerable opportunity for improvement. In order to accomplish the screening coverage goals in Mexico, better understanding of the drivers that curb technical efficiency in the provision of mammography studies is strongly advocated.
