Medicaid Disease Management: No Impact on Emergency Room Utilization or Inpatient Costs for Enrollees with Diabetes?

Regular readers of the Disease Management Care Blog know that Medicaid is coming. While many of the nation's Governors have declined President Obama's invitation to run Medicaid the Affordable Care Act way, others have agreed to use the ACA's generous funding to enroll millions of their indigent citizens into this vastly expanded public insurance program.

"No problem!" says the disease management vendors.  For years, they've been offering their services to state Medicaid programs and would be happy to expand their contracts.

Unfortunately, an article by Matthew Conti that was just published in the journal Health Services Research suggests that that may not be a good idea.  The article's title is Effect of Medicaid Disease Management Programs on Emergency Admissions and Inpatient Costs.  The only thing that's missing are the words "The Lack of Any" at the front of that sentence.
   
The article studied the impact of diabetes "opt-out" disease management on diabetic patients' emergency room utilization and admissions in three states' Medicaid programs: Washington (started in 2002), Texas (started in 2004), and Georgia (started in 2005).

These states with were compared to states without diabetes disease management.  These control states were selected on the basis of baseline Medicaid enrollment trends that were similar to the three study states. These control states were Hawaii, Kentucky, Massachusetts, Maryland, Maine, North Carolina, Nebraska, South Carolina and Tennessee.

To perform the comparison, Dr. Conti used the Agency for Health Care Research and Quality's (AHRQ) National InPatient Sample (NIS) from the Health Care Cost and Utilization Project ("HCUP"). These databases contain patient-level and longitudinal hospital information on inpatient stays, including cost, payer, admission type (e.g., emergency, urgent and elective), age, gender, primary payer, and total charges. The span of data that was used went from 2000 through 2008.

A complicated pre-post "difference in differences" model was used to compare baseline vs. follow-up:

1) total inpatient charges/Medicaid enrollment (which averaged $430 per diabetic enrollee, with a 95% confidence interval of $265 to $700) and

2) emergency admissions/inpatient admissions (a ratio of 0.37 per admission with a standard deviation of plus or minus 0.12)  All Medicaid enrollees with diabetes were included in the analysis, whether or not they had been enrolled or opted out.  The author used this approach figuring that if a statewide disease management program enrolled up to a third of eligible persons with diabetes (that was the case in Texas), there should have been an observable impact on the entire population. That's the approach favored by the Disease Management Purchasing Consortium.
 
The results?  No state with disease management had lower emergency room utilization or inpatient costs for their Medicaid enrollees with diabetes. The DMCB couldn't find a table with numbers, but the figures (which can't be reproduced without permission) show little impact over time.

What can readers conclude?  Assuming that, during the period of study, the three states' Medicaid programs suffered from the program's endemic issues of underpayment to providers with a relative lack of access to primary care:

1. "Blanket" call-everyone telephonic disease management cannot make up for fee-for-service Medicaid's shortfalls.  It remains to be seen if the ACA's revitalization of Medicaid will make up for this and increase the parallel impact of disease management. 

2) This also means that Medicaid's experience with disease management can't be generalized to other types of insurance with better provider payment rates and patient access to care.

That being said, the DMCB has two concerns:

1. If the DMCB is reading this right, it appears all persons of any age with diabetes were included in the study, including Type 1 diabetics.  If that's correct, that could have also blunted the impact of any disease management program, since children are over-represented in Medicaid and the impact of remote telephonic coaching in Type 1 is widely viewed (even among the disease management vendors) to be ineffective.  Insulin-requiring kids need lots of face-to-face hands-on care.

2. The DMCB is unfamiliar with the three study states' disease management programs, but if they were set up the "old fashioned way" to contact all persons with diabetes without the modern regard to future risk and "impactibility," then it's little wonder that the programs failed.  State-of-the-art population health management tailors its programs by focusing on subsets of persons with chronic conditions that are most likely to benefit.  Any impact on emergency room use or inpatient charges for these patients would be lost in the data "noise" of everyone else's utilization.

Should Medicaid programs that are facing huge jumps in enrollment abandon Medicaid as a result of this study?  Based on this study, the DMCB doesn't think so.  The findings are interesting, but more research is needed.