Mark Zezza, NYSHealth

Funding Area



April 24, 2019

A study by NYSHealth Policy & Research Director Mark Zezza, published in the journal PLOS ONE, examined regional variation in Medicare spending in New York State and found that, while gaps in spending have narrowed over time, substantial variation remains across counties.

The Issue

One strategy considered for controlling health care spending is to reduce spending in high-spending areas to levels achieved in areas that use health care more efficiently. This approach is based on the premise that:

  • Substantial regional variation in health care spending exists even after controlling for differences in local prices and the health status, suggesting that regional differences largely result from differences in health care provider and patient preferences;
  • More spending is not strongly correlated with better outcomes for patients; to the contrary, at least some research suggests that better outcomes are associated with areas that use health care more efficiently and spend less;
  • Research suggests that up to 20% of health care delivered may be unwarranted and have questionable value to patient outcomes.

This study used county-level Medicare spending from 2007–2016 to examine the degree to which high-spending areas remain high-spending in New York State. The strategy of narrowing the gap in spending across regions would be most effective if high-spending regions remain high-spending over time, so that efforts can be targeted and tailored to those regions with the highest levels of spending.

What the Study Found

  • Substantial variation exists in per-capita Medicare spending across counties in New York, even after controlling for differences in local prices and health status. In 2016, there was a 34% difference between the lowest- and highest-spending counties ($7,528 in Tompkins compared to $10,063 in Sullivan).
  • In general, there was evidence for convergence in spending over time across New York State: counties that were higher-spending in 2007 generally had lower growth rates in spending from 2007 through 2016, and vice versa.
  • Even though there was evidence for convergence, a substantial number of counties in were found to be persistently high-spending over time, many of which are located downstate, including Nassau, New York, Orange, Putnam, Rockland, Suffolk, and Westchester counties.
  • The services most associated with higher total spending and variation in New York were different from those driving spending trends across the country. Whereas office visits, procedures, imaging, and tests were primary drivers of spending in New York, facility-based care such as inpatient hospital and skilled nursing facility services played a substantially larger role in the rest of the country.

Policy Implications

  • These results suggest that focusing on persistently high-cost areas is a viable strategy for reducing health care spending. If per-capita Medicare spending could be reduced by 5% in the 7 persistently high-spending counties, total Medicare spending in New York State could be reduced by more than 2% annually, or nearly $350 million in 2016.
  • Efforts to reduce unwarranted variation in spending need to be tailored to the circumstances of particular regions, as there are geographic differences in the services that drive spending variation.
  • The findings about spending convergence have particularly important implications for the specifications of alternative provider payment models. These models, such as those used by accountable care organizations, develop spending targets based on broad regional trends. This practice may deter areas with historically low levels of spending from participating, as they are likely to experience higher-than-average growth rates. Policy-makers may consider adjusting the spending targets in low-spending areas to make them more attainable, in order to encourage participation.

Read the study.