Network analysis of drug prescriptions

Networks exist in many different aspects of the world, at social, economical, biological, and molecular levels. Network science studies their parameters, or quantitative indicators; its instruments make it possible to draw and analyze networks from a mathematical perspective. The present study is an attempt to apply network science techniques to the drug prescription process, a typical subject of Epidemiology for Public Health studies.

A drug prescription network was created using the set of drug prescriptions written during a 6-month period by a group of 99 general practitioners (GPs) operating in Italy. In this network, named co-prescription network, each drug represented a node, and different drugs prescribed to the same patient at the same moment were considered linked together. Drug prescription data for a total of 42 965 patients and 631 232 drug packages were studied. A number of co-prescription networks were obtained and analyzed on the basis of different subsets of patients by age and gender. The network parameters were measured and compared for the various subsets.

All the drug prescription networks studied showed scale invariance behavior. The age- and gender-related co-prescription networks showed different patterns, with different levels of complexity.

The present study shows that the drug prescription process has specific network aspects and dynamics and, more generally, that it is possible to apply instruments of network science to study public health phenomena from a new, different perspective. Further studies should be encouraged and performed.