Our findings regarding the relationship between 911 and mortality are more subtle. First, we are unable to establish a direct reduced-form statistical relationship between the level of 911 in a given county and patient mortality. Of course, this may be due to the fact that the overall mortality rate is relatively low (approximately 7%) and only a small portion of our sample resides in counties with no 911 technology (approximately 20%), making it difficult to infer the impact of the technology level on the mortality rate. However, our analysis of the impact of 911 on response time suggests an alternative strategy: we use the adoption of 911 as an instrument for an individual’s response time in the patient mortality regressions. In particular, we show that 911 technology affects response time, and we can assume that 911 adoption is unrelated to the severity of a particular patient. Our preliminary instrumental variables analysis of the effect of response time on mortality finds that shorter response times do indeed reduce mortality. While this analysis is still exploratory, we believe that the use of county-level infrastructure as an instrument for individual-level services is a potentially fruitful approach for further exploration.
Beyond its direct effects on response time and mortality, a second role of the emergency response system is to allocate patients to hospitals. From a hospital’s perspective, the emergency response system affects both the size and characteristics of its pool of emergency patients; the sensitivity of the allocation process to the hospital characteristics will also interact with the incentives of a hospital to adopt certain technologies. We thus take several preliminary steps towards exploring these effects.
Our first result about allocation is that patient severity affects the allocation of patients to higft-technology hospitals. Our results about allocation have implications for our ability to draw inferences about the benefits of hospital technology through reduced-form analyses of the direct effect of technology on patient outcomes. This issue has been recognized by several authors, such as McClellan and Newhouse (1997), who argue that patient allocation to hospitals with different technologies is endogenous and so must be treated with an instrumental variables approach. Consistent with this view, our estimates provide direct evidence about the relationship between patient severity and allocation. payday loans direct lender