Wednesday, November 28, 2012

Another Study Shows Disease Management Works

Here's more evidence that disease management really works.

The study, Online Disease Management of Diabetes: Engaging and Motivating Patients Online With Enhanced Resources-Diabetes (EMPOWER-D), was a prospective randomized clinical trial that is posted online over at JAMIA.

The Disease Management Care Blog summary:

Researchers at the Palo Alto Medical Foundation used the electronic health record (EHR) to look for active (seen once in the last 12 months) patients over 18 years of age with  Type 2 diabetes and an A1c greater than 7.5%.  If the primary physician approved, patients were then asked to complete a questionnaire and keep an appointment with a research assistant for additional review and discussion. Once patients agreed, they were entered into the study.

6,907 potential study subjects were identified, 1,594 agreed to complete the questionnaire and see the research assistant, 768 met additional research criteria. 415 agreed to be enrolled in the study, and 379 completed most of the 12 months of follow-up.

Patients were randomly allocated into one of two treatment tracks and followed for 6 and 12 months. The usual care (UC) track provided reminders about preventive care in addition to their usual visits with their physicians.  The intervention (INT) track had the usual physician visits, plus:

1. access to a "nurse care manager" (NCM) who provided advice and protocol-based medication changes,
2. Wireless uploads of glucometer readings into the EHR,
3. an EHR-based patient-specific summary "dashboard," that included risk scores, preventive care updates and a care plan,
4. web-based insulin, exercise & nutrition logs,
5. secure EHR-based messaging with the physician and NCMs and
6. patient-specific text and video offerings targeted by the NCMs.

The principal outcome measure was the A1c test, which is an indicator of overall blood glucose control. A level of 7 or lower is considered to be satisfactory control. At 6 months, INT patients statistically significantly decreased their A1c by 1.3 vs. 0.7 in the UC group.

At 12 months, there was still a difference favoring the INT group, but it did not achieve statistical significance: 1.1 vs. 0.7. 

Once the study was completed, the authors went back and looked at the proportion of patients that had decreased their A1c by at least 0.5.  At 6 months, it was 70% vs. 53% for  at 6 months, and 70% vs. 55% at 12 months for the INT and UC groups, respectively.  Both differences were statistically significant. 

Patients in the INT group were also more likely to lower their cholesterol, go through a medication adjustment, experience lower "treatment distress," have greater knowledge of their diabetes and be more satisfied with their care.  There was no difference in blood pressure control or the number of physician visits.  Overall health care costs or insurance claims expense were not measured.

Disease Management Care Blog take-aways:

You Want Evidence? Thanks to this high quality randomized controlled clinical trial conducted in a real world setting, the evidence base supporting the use of remotely placed nurse care managers continues to build.   Kudos to Palo Alto for simultaneously taking good care of their patients and conducting impressive research.

Patients At Risk: The combination of a) a steady percent of INT patients keeping their A1c 0.5 at 6 and 12 months plus b) a simultaneous overall average decline in the A1c makes the DMCB think that there was a subcohort of patients that "back-slid" and affected the group mean.  Commercial population health management service providers are working hard at prospectively identifying these higher-risk individuals for additional interventions.

Doing something is better than nothing: While the modest A1c decrease in the UC patients could have been due to regression to the mean, the DMCB wonders if they also benefited from being identified and monitored. 

Can Finally Point to Something Good About the EHR: While the EHR continues to disappoint in terms of consistently improving quality or reducing costs, this study demonstrates an important upside: it can be used to efficiently recruit potential research subjects.  That's important because thousands of candidate patients are needed to find hundreds of study participants
Physician Utilization Did Not Decline: Unfortunately, this study did not address multiple measures of utilization, so the DMCB doesn't know what to make of any decrease in potential costs versus the cost of the the program.

ACOs Take Note: While the Palo Alto System is not representative of most health care settings in the U.S., it does hold important lessons for Accountable Care Organizations. Given their contractual responsibility to improve diabetes quality as well as managed insurance risk, this study says they ignore the potential of remotely-based and technology-backed nurse care managers at their peril


Unknown said...

Good stuff. BTW, your loyal readers in geriatrics will always be interested in knowing how old the study population is (54 in this case, but you forced me to go dig it up).

Jaan Sidorov said...

Good point!

Ed Rodgers said...

Sorry for the latent comment / question but I have a question on how you define "works".

If costs were not compared this could be significantly more expensive. Knowing if this prevented expensive (financially or to the QOL of the patients) interventions I think would be of great interest to saying it "worked".

Secondly if some patients indeed did "back-slide", then again this would be partial failure.

Finally and most painfully - it would take years to answer what this could possibly mean to long term health benefits and improvements to QOL or Overall Survival.

Jaan Sidorov said...

Ed is absolutely correct: the "works" is a function of diabetes control, not cost control. My enthusiasm was more a function of the outcomes being demonstrated (despite the back sliding) in a randomized control study.

Costs are a different challenge and yes, it can take years. My thinking is that even if there isn't a reduction in claims expense, the value (defined as outcome per dollar spent) is still attractive.