Deficiencies in the quality of pre-hospital care constitute a serious public health problem that has largely been neglected by the scientific community. Trauma and complications of acute disease produce medical emergencies outside of the hospital setting. Treating patients with these conditions involves an inherent trade-off between providing treatment on-site and reducing time to hospital care. My dissertation compares two models of providing pre-hospital care, and highlights a data-driven approach to identifying potentially fraudulent ambulance claims.
Chapters 1 and 2 compare effects of Advanced Life Support (ALS) and Basic Life Support (BLS) on outcomes after out-of-hospital medical emergencies. Most Medicare patients seeking emergency medical transport are treated by ambulance providers trained in ALS. Evidence supporting the superiority of ALS over BLS is limited. I analyzed claims from a 20% sample of Medicare beneficiaries from non-rural counties between 2006-2011 with cardiac arrest, major trauma, stroke, acute myocardial infarction (AMI), or respiratory failure. To address unmeasured confounding, I exploited variation in geographic penetration in ALS rates across counties, using instrumental variables analysis. In particular, I predicted the probability of ALS use for each patient as a function of ALS rates in each county for patients with other diagnoses, using a multilevel, multivariate model. Survival to 90 days for trauma, stroke, cardiac arrest, and AMI patients was higher with BLS than ALS; respiratory failure patients did not exhibit differences in survival. I conducted a secondary analysis based on propensity score-based balancing weights, and this produced generally similar results. I concluded ALS is associated with substantially higher mortality for several acute medical emergencies compared to BLS, and may harm patients through delayed hospital care and iatrogenic injury.
In Chapter 3, I link patient demographic information and ambulance, outpatient, and inpatient claims to look for the inconsistency of having a claim for an ambulance transport with seemingly no real patient - a 'ghost'. I find 1.9% of emergency transports have this inconsistency. I estimate the distribution of ghost ride rates by suppliers and separately, by counties, using an expectation-maximization algorithm. I find the ghost rides are not evenly distributed across counties or suppliers. Although it is not possible to conclusively distinguish billing anomalies due to fraud from data entry errors and similar explanations, this type of analysis may provide useful starting points for further investigation of Medicare fraud. / Health Policy
| Identifer | oai:union.ndltd.org:harvard.edu/oai:dash.harvard.edu:1/17467334 |
| Date | 17 July 2015 |
| Creators | Sanghavi, Prachi |
| Contributors | Zaslavsky, Alan M. |
| Publisher | Harvard University |
| Source Sets | Harvard University |
| Language | English |
| Detected Language | English |
| Type | Thesis or Dissertation, text |
| Format | application/pdf |
| Rights | open |
Page generated in 0.0018 seconds