AHRQ Product Helps Guide Policy Decisions About Emergency Care in New Hampshire

March 2006

Medicaid officials in New Hampshire are using an AHRQ product to help guide policy decisions about emergency care. The emergency department algorithm developed by former AHRQ Visiting Scholar John C. Billings is enabling them to identify potentially unnecessary or avoidable emergency department visits and to estimate the likely costs and benefits of proposed changes in Medicaid requirements.

Christine Shannon, Bureau Chief of New Hampshire's Bureau of Health Care Research, says, "We are excited about this tool. It has already helped us provide analysis of proposals made by elected and appointed officials. We intend to continue using it."

The Billings algorithm is an automated tool that allows its user to discriminate among emergency department visits—to see which ones were probably not urgent and which conditions could have been prevented or treated elsewhere than in the emergency department. Billings developed the algorithm with the help of a consensus panel of physicians, who reviewed some 6,000 emergency department records, judging the urgency and necessity of each visit. Their review yielded estimates of the probability that a visit for a given diagnosis will fall into one of several categories: non-emergent, emergent and in need of emergency care, emergent but preventable, and emergent but treatable in a primary care setting.

New Hampshire Medicaid staff economist, Thomas Grannemann, PhD, senior health policy analyst, and his colleagues applied the Billings algorithm to Medicaid claims for emergency department visits in the year 2002. They estimated that, for three communities—Manchester, Nashua, and Concord—Medicaid had paid more than $1.5 million, including federal and state outlays, for potentially avoidable or primary-care-treatable visits. A suggestion was made that these expenditures might be reduced by means of a "community access" program whereby Medicaid patients would be redirected to other sources of care.

Based on detailed information that the Billings algorithm enabled them to develop about potentially unnecessary visits, the Medicaid analysts were able to show that program savings would likely fall short of program costs. This was a key factor in the decision not to proceed with the proposed program.

The New Hampshire Medicaid analysts also evaluated a proposal to impose copays on emergency department visits by Medicaid patients. They used the Billings algorithm to identify conditions for which at least 90 percent of emergency department visits are non-urgent, and for which Federal Medicaid rules allow copays. They showed that the New Hampshire Medicaid program would actually lose money by instituting a program based on this standard. Setting the standard any lower would mean charging Medicaid patients for a significant number of visits that were actually emergencies. This information was a factor in the Medicaid program's decision not to propose emergency room copayments based on diagnosis and a later decision to shelve plans for any emergency room copayments for Medicaid recipients.

Grannemann and his team were delighted to find a software tool they could use with their existing statistical program. Grannemann notes, "The researchers on Billings' team were most helpful in providing the limited but essential technical support we needed." He urges AHRQ to go beyond funding academic research to supporting applications like the Billings algorithm, including support for concept and development, as well as ongoing support of application tools.

Impact Case Study Identifier: 
AHRQ Product(s): Research
Topics(s): Safety Net
Geographic Location: New Hampshire
Implementer: State of New Hampshire
Date: 03/01/2006

Billings J. Using administrative data to monitor access, identify disparities, and assess performance of the safety net. Chapter 5 in Monitoring the Health Care Safety Net: Book III: Tools for Monitoring the Health Care Safety Net. Agency for Healthcare Research and Quality, Rockville, Maryland. December 2003.

Page last reviewed October 2014