06 Jul Predicting Death is Difficult, Making it Difficult To Save Money on End of Life Care
MedicalResearch.com Interview with:
Amy Finkelstein PhD
John & Jennie S. MacDonald Professor of Economics
MIT Department of Economics
National Bureau of Economic Research
Cambridge MA 02139
MedicalResearch.com: What is the background for this study? What are the main findings?
Response: Although only 5% of Medicare beneficiaries die in a given year, they account for almost 25% of Medciare spending.
This fact about high end of life spending has been constantly used to refer to inefficiency of the US healthcare system. A natural observation is that the fact is retrospective, and it motivated us to explore a prospective analog, which would take as an input the probability of someone dying in a given year rather than her realized outcome. We therefore used machine learning techniques to predict death, and somewhat to our surprise we found that at least using standardized and detailed claims-level data, predicting death is difficult, and there are only a tiny fraction of Medicare beneficiaries for whom we can predict death (within a year) with near certainty.
Those who end up dying are obviously sicker, and our study finds that up to half of the higher spending on those who die could be attributed to the fact that those who die are sicker and sick individuals are associated with higher spending.
MedicalResearch.com: What should readers take away from your report?
Response: Given how difficult we found it to identify beneficiaries who would almost surely die, we came to the conclusion that the discussion of “end of life spending” as an indicator for waste or inefficiency is inappropriate, as one could easily come up with examples why it could be efficient to spend more on the sick, even if such spending patterns would lead to high fraction of spending on those who (unpredictably) end up dying.
MedicalResearch.com: What recommendations do you have for future research as a result of this work?
Response: Our study, we hope, should push scholars to move away from simple statistics, which are easy to generate but hard to interpret, and instead to study carefully particular healthcare intervention, and assess at a more granular level how each interventions affects death probability and quality of life.
No disclosures. We would like to thank the National Institute of Aging for funding our work.
Predictive modeling of U.S. health care spending in late life
BY LIRAN EINAV, AMY FINKELSTEIN, SENDHIL MULLAINATHAN, ZIAD OBERMEYER
SCIENCE 29 JUN 2018 : 1462-1465
The information on MedicalResearch.com is provided for educational purposes only, and is in no way intended to diagnose, cure, or treat any medical or other condition. Always seek the advice of your physician or other qualified health and ask your doctor any questions you may have regarding a medical condition. In addition to all other limitations and disclaimers in this agreement, service provider and its third party providers disclaim any liability or loss in connection with the content provided on this website.