Diffusion of medical technology across hospitals and traits of fast adopting hospitals

The quality and cost of hospital care depends not only on what technologies have been developed, but on how quickly the most promising and cost effective technologies diffuse across hospitals.  We seek to better understand what hospital characteristics are associated with fast adoption of new technologies.  We are also interested in the degree of correlation between diffusion measures of different types of technologies.  If overall hospital administration quality determines the rate of diffusion, there should be strong correlation between diffusion measures across different types of technologies.  On the other hand, if department culture and individual doctor decisions drive adoption, there will be less correlation between hospitals’ diffusion scores across different types of technology, particularly if different departments or different practitioners use the technology. 

For the analysis, we employ 2004 to 2011 hospital discharge data from California’s Office for Statewide Health and Planning, which includes information about each patient’s diagnoses and procedures.  We develop several measures of technology diffusion for two types of technologies: minimally invasive surgeries and procedural techniques developed for reducing catheter infections.  For our study of minimally invasive procedures, we look at four laparoscopic procedures that either became available in the minimally invasive format or expanded in this format during our time period. The percent minimally invasive among these four categories was 3.4% in 2004, and had increased to 25.6% by 2011.  We develop measures of diffusion that take into account both the timing of adoption and the degree of adoption (percent of procedures minimally invasive) within a hospital.  As an alternative measure, we also look at how quickly catheter infections dropped between 2004 and 2011.  Catheter infections dropped very rapidly nationwide because of effective protocols, such as checklists, that became common practice over a fairly short period of time.  We look at the timing of the catheter infection improvements as an alternative measure of diffusion for comparison.

Preliminary results indicate that there is not a strong correlation between diffusion scores for different types of procedures.  This indicates that the diffusion of technology likely depends more on department and doctor characteristics than hospital administration.  However, there are still some hospital characteristics that consistently predict faster adoption.  The initial analysis suggests that hospitals which are members of larger systems and hospitals with more privately insured patients tend to adopt new technology earlier and at higher rates of adoption.  For-profit status is associated with lower rates and later adoption.